commonwealth of puerto rico puerto...

COMMONWEALTHOFPUERTORICO

PUERTORICOENERGYCOMMISSION

INRE:REVIEWOFRATESOFTHEPUERTORICOELECTRICPOWERAUTHORITY

MATTERNO.:CEPR‐AP‐2015‐0001

SUBJECT: EXPERT REPORT: STATE OFPREPA’SSYSTEM,LOADFORECAST,CAPITALBUDGET,FUELBUDGET,PURCHASEDPOWERBUDGET,OPERATIONSEXPENSEBUDGET.

EXPERTREPORTOF

JeremyI.Fisher,PhDArielI.Horowitz,PhD

SynapseEnergyAssociates485MassachusettsAve.

Suite#2Cambridge,MA02139

RevisedNovember23,2016

FisherandHorowitz/SynapseEnergyEconomics 1of218

TableofContentsI. INTRODUCTION.........................................................................................................................................9

A. Purposeofreport........................................................................................................................9 B. Qualificationsofauthors..........................................................................................................9 C. Organizationofreport...........................................................................................................10

II. SUMMARYOFFINDINGSANDRECOMMENDATIONS............................................................12 A. Informationreviewedforthisreport..............................................................................12 B. MajorfindingsregardingPREPA’spresentandfuturespendingneeds..........12

1. PREPA’sreliabilityhassufferedsinceFY2014.............................................12 2. PREPAoperatesinacost‐constrainedenvironment.................................14 3. PREPA’scapitalandlaborbudgetsmaynotsupportasafeandreliable

system.............................................................................................................................14 4. PREPA’snon‐laboroperationsbudgetsarepoorlyallocated................15 5. PREPA’srecordkeepingissub‐par...................................................................15 6. PREPA’sbudgetallocationprocessisopaqueanddistortionary.........16 7. Fundamental needs should drive revenue requirements; PREPA’s

budgetsconstrainedbyaperceptionofrevenues......................................16 8. Significant investment and development of core competencies are

neededtoimprovetheconditionandperformanceofPREPA’ssystem...........................................................................................................................................17

9. PREPA’scurrentstatewillrequiretime,effort,andfundstorepair..18 10. Role of energy efficiency, renewable energy, and distributed energy

resources.......................................................................................................................18 11. PREPAmustbegintothinkintermsofitsactualneedsforoperationof

asafeandreliablesystem......................................................................................19 C. RoleoftheCommission.........................................................................................................20 D. Summaryofbudgetadjustments......................................................................................21 E. Summaryofotherrecommendations.............................................................................23

III. THESTATEOFPREPA’SSYSTEM....................................................................................................26 A. PREPA’sgenerationsystemcrisis.....................................................................................26

1. PREPA’sassessmentoftheforcedoutagecrisis..........................................28 2. Forcedoutagesimposecostsonsystem.........................................................31

B. LowreliabilityforPREPAcustomers..............................................................................32 1. ReliabilityTrends......................................................................................................34

C. PREPA’sMATScomplianceconundrum.........................................................................37


IV. LOADFORECAST.....................................................................................................................................41 A. Summaryofissueandfindings..........................................................................................41 B. Theneedforandsignificanceofthesalesforecast...................................................41

1. PREPArequiresanaccurateandreasonablesalesforecast...................41 2. PREPA’ssalesforecastimpactsmanyaspectsofitsexpectedspending

andrevenue.................................................................................................................42 3. StateofPREPA’ssales,load,andcustomerbase.........................................44

C. PREPA’ssalesforecastisill‐conceivedandill‐supported......................................46 1. PREPAdidnotprovideadequatesupportforitssalesforecast...........46 2. PREPA’sforecastingis,ingeneral,notpredictive.......................................51 3. PREPA’sconsiderationofenergyefficiencyisinappropriate................58 4. TheelasticityofdemandinPREPA’ssystemisunclear...........................60 5. Useofthesalesforecastintheratecasefilingisinconsistent..............61

D. Recommendation.....................................................................................................................62 1. PREPA’sFY2017salesforecastshouldstandwithoutadjustments..62 2. PREPAmustimproveitsforecastingmethodologyanddocumentation

...........................................................................................................................................64 V. CAPITALBUDGET...................................................................................................................................65

A. Summaryofissueandfindings..........................................................................................65 B. PREPA’suseofcapital............................................................................................................65

1. Overview.......................................................................................................................65 2. PREPA’sexpectedcapitalexpendituresbyarea..........................................67 3. IssuesinreviewingPREPA’scapitalbudgetestimates.............................73 4. RelationshipbetweenthecapitalbudgetandtheIntegratedResource

Plan..................................................................................................................................78 5. Overviewoffindings................................................................................................80

C. GenerationandProductioncapitalbudget...................................................................87 1. Overview.......................................................................................................................87 2. Aguirresteamunits..................................................................................................88 3. CostaSursteamunits..............................................................................................94 4. PaloSecosteamunits..............................................................................................98 5. SanJuanSteamunits..............................................................................................101 6. AguirreandSanJuanCombinedCycleUnits...............................................106 7. CambalacheandMayagüez.................................................................................111

D. AguirreOffshoreGasPortcapitalbudget.....................................................................114 1. IntroductionandOverview.................................................................................114 2. AOGPprojectboundaries.....................................................................................115


3. Cost,contractingandfinancingatAOGP.......................................................116 4. CommissionIRPFindingswithrespecttoAOGP.......................................127 5. RecommendationwithrespecttoAOGPcapital........................................130

E. TransmissionandDistributioncapitalbudget..........................................................133 1. Overview.....................................................................................................................133 2. Transmissionsystemcapitalbudget...............................................................135 3. Distributionsystemcapitalbudget..................................................................140 4. Recommendations with respect to Transmission and Distribution

Capital...........................................................................................................................150 F. Othercapital–TransportationandComputerEquipment..................................151

1. Overview.....................................................................................................................151 2. Vehicles........................................................................................................................151 3. Computerequipment.............................................................................................152 4. Recommendations with respect to transportation and computer

equipment...................................................................................................................152 G. Recommendation...................................................................................................................152

VI. FUELANDPURCHASEDPOWERBUDGETS..............................................................................155 A. Summaryofissueandfindings........................................................................................155 B. PREPA’suseoffuel................................................................................................................155 C. PREPA’sPowerPurchaseandOperatingAgreements(PPOAs)........................157 D. PREPA’sfuelandpurchasedpowerbudgetingprocess........................................159

1. Fuelforecasts............................................................................................................159 2. PPOAcontractterms..............................................................................................161 3. UseofPROMODandconcernsassociatedtherewith...............................162

E. EvaluationofPREPA’spurchasedpowerandfuelbudgets.................................168 1. Purchasedpower.....................................................................................................168 2. Fuel................................................................................................................................172

F. Recommendation...................................................................................................................180 1. Fuelbudget.................................................................................................................180 2. Fuelpriceforecasts.................................................................................................180 3. Purchasedpowerbudget.....................................................................................180 4. Renewableenergycontractstructuresandstatusupdates..................180

VII. OPERATINGEXPENSEBUDGET.....................................................................................................181 A. Summaryofissueandfindings........................................................................................181 B. ConstituentelementsofPREPA’soperatingexpenses..........................................182

1. Laborexpenses.........................................................................................................183 2. Non‐laborexpenses................................................................................................184


3. Expensesbyfunctionalarea...............................................................................185 C. PREPA’sFY2017operatingbudgetiseffectivelyunsupported.........................186

1. PresentationofPREPA’soperatingbudgetinPREPA’sratecase.......186 2. Financialmodelbreakdownofexpensesbyareaisadhoc..................188 3. PREPAfailedtoprovidesupportingdocumentationfortheoperations

budget,evenafterseveralrequests.................................................................194 D. PREPA’soperatingbudget isnotconsistentwithhistoricalpatternsorwith

operationofasafeandreliablesystem........................................................................196 1. PREPA’s operational spending fell sharply between FY2013 and

FY2016.........................................................................................................................196 2. PREPAdescribed itsownbudget formationprocessasonebasedon

principlesofausterity............................................................................................201 3. IncreasesinAdministrativeandGeneralspending..................................201 4. PREPAhasexplicitlystatedthatstaffandbudgetrestrictionshaveled

todeclinesinsafetyandreliability..................................................................202 5. Implicationsforbenchmarkingstudy............................................................204

E. CalculationofadjustmentstoPREPA’soperatingbudget....................................205 1. Reallocation of budgets based on expected labor and non‐labor

proportionsofspendingbyarea.......................................................................206 2. Correctiveadjustments.........................................................................................210 3. OtheradjustmentstoPREPA’snon‐laborexpensesbudget.................212 4. Separationoftransmissionanddistributionbudgets.............................213 5. Summaryofadjustments.....................................................................................214 6. Achievabilityofadjustedspending..................................................................215

F. Recommendation...................................................................................................................216 1. Additionoffundstooperationsbudget.........................................................216 2. Increasedbudgetoversight.................................................................................216

VIII. CERTIFICATION.....................................................................................................................................217


TableofTables

Table1.VariablesincludedinPREPA’scandidateregressionmodelsforforecastingofresidentialsales..........50

Table2.Historicactualandmodeledresidentialsalesandgrowthratesthereof...........................................................57

Table 3. PREPA's sales forecasts by class, and representations thereof in the financial model and billingdeterminantsworkpaper...........................................................................................................................................................................62

Table4.DeviationsofPREPA'ssalesforecastsfromactualvaluesafteroneandthreeyears...................................64

Table5.TotalcapitalrequestedbyPREPAinFY2017throughFY2019,byareaandsubarea.DoesnotincludeAOGP...................................................................................................................................................................................................................68

Table6.PREPADirectorates....................................................................................................................................................................70

Table7.CapitalspendingatPREPAgeneratorsbyfiscalyear..................................................................................................88

Table8.HistoricandexpectedcapitalspendingatAguirreSteamUnitsbyFiscalYear,millionsofdollars.......92

Table9.HistoricandexpectedcapitalspendingonindividualprojectsatAguirreSteamUnitsforprojectswithspendingfromFY2017toFY2019,millionsofdollars.................................................................................................................94

Table10.HistoricandexpectedcapitalspendingonindividualprojectsatAguirreCombinedCycleforprojectswithspendingfromFY2017toFY2019,millionsofdollars......................................................................................................97

Table11.Historicandexpectedcapitalspendingon individualprojectsatSan JuansteamforprojectswithspendingfromFY2017toFY2019,millionsofdollars..............................................................................................................105

Table12.HistoricandexpectedcapitalspendingonindividualprojectsatAguirreandSanJuanCCsforprojectswithspendingfromFY2017toFY2019,millionsofdollars...................................................................................................109

Table13.AguirreOffshoreGasportcapitalbudget–2014,2016........................................................................................116

Table14.AOGPbudgetandspending...............................................................................................................................................126

Table15.Transmissionsystemcapitalbudget,millionsFY2017‐FY2019......................................................................138

Table16.Distributionsystemcapitalbudget,millionsFY2017‐FY2019.........................................................................143

Table17.FindingsfromPREPACMSSmartMeterPilot:September2016......................................................................149

Table18.HistoricalandmodeledEcoEléctricacapacityfactorsbymonth.....................................................................165

Table19.Modeledandreportedrenewableenergycontractcosts....................................................................................167

Table20.PREPA'sbudgetedversusactualspendingonpurchasedpowerinFY2017..............................................170

Table21.PREPA'sas‐filedandmostrecentPPOAbudgetexpectations...........................................................................171

Table22.PREPA'sbudgetedversusactualspendingonfuelinFY2017...........................................................................173

Table23.HistoricalpercentagesofannualspendingonfueloccurringinJulyandAugust.....................................175

Table24.PREPA'sas‐filedFY2017fuelbudgetandauthor’scalculationsofestimatesofexpectedspending.............................................................................................................................................................................................................................177

Table25.ComparisonofproposedadjustmentstoPREPA'sFY2017fuelbudget.......................................................178

Table26.PREPA'sproposedspendingonnon‐laborexpensesbycategoryofspending..........................................184

Table27.PREPA'sproposedallocationofoperationalexpensesbyfunctionalarea..................................................185

Table28.PercentagesofFY2014laborandnon‐laboroperationalexpendituresbyfunctionalarea................190

Table29.ComparisonofPREPA'sFY2014actualoperationalspendingandFY2017proposedbudgets.........190


Table30.DeviationsbetweenPREPA'sFY2014operationalspendingasreportedindifferentROIs................193

Table31.AllocationofFY2014operationalspendingbyfunctionalareaasreportedindifferentROIs...........194

Table32.ReallocationofPREPA'sFY2017operationallaborexpensebudget.............................................................208

Table33.ReallocationofPREPA'sFY2017operationalnon‐laborexpensebudget....................................................209

Table34.BenchmarkingandcorrectionofPREPA'sFY2017operationallaborexpensebudget.........................210

Table35.BenchmarkingandcorrectionofPREPA'sFY2017operationalnon‐laborexpensebudget...............212

Table36.OtheradjustmentstoPREPA'sFY2017non‐laboroperationalexpensebudget......................................212

Table37.AllocationofT&Doperationalbudgetintotransmissionanddistributionbudgets...............................214

Table38.SummaryofFY2017operationalbudgetadjustments.........................................................................................214


TableofFigures

Figure1.Totalenergy(MWh)losttoforcedoutagesbyquarter,calendaryear..............................................................28

Figure 2. Steam unit availability (in MW relative to max capacity) vs. historic generation operations andmaintenance,byfiscalyear.......................................................................................................................................................................30

Figure3.PREPAcapitalbudgetandspendingforgeneration,historicandratecase,byfiscalyear(millions).31

Figure4.PREPAmonthlySAIFI..............................................................................................................................................................35

Figure5.PREPAmonthlySAIDI..............................................................................................................................................................36

Figure6.PREPAmonthlyCAIDI.............................................................................................................................................................36

Figure7.PREPA'ssalesbycustomerclassovertime...................................................................................................................45

Figure8.PREPA'sactualtotalsystemsalesforFYs2010through2016andcontemporaneousforecastsoftotalsales.....................................................................................................................................................................................................................52

Figure9.PREPA'sactualandmodeledresidentialsalesforFYs2010through2016....................................................53

Figure10.PREPA'sactualandmodeledcommercialsalesforFYs2010through2016...............................................54

Figure11.PREPA'sactualandmodeledindustrialsalesforFYs2010through2016...................................................54

Figure12.PREPA'snormalizedactualandmodeledresidentialsalesforFYs2010through2016........................55

Figure13.PREPA'snormalizedactualandmodeledcommercialsalesforFYs2010through2016......................56

Figure14.PREPA'snormalizedactualandmodeledindustrialsalesforFYs2010through2016..........................56

Figure15.PREPAtotalcapitalbudgetandspendingforallareas,historicandratecase,byfiscalyear(millions).DoesnotincludeAOGPspending...........................................................................................................................................................69

Figure16.Historiccapitalbudgetoverspending(underspending)bydirectorate,byfiscalyear...........................71

Figure17.Historiccapitalbudgetoverspending(underspending)byDepartmentintheGeneration,byfiscalyear......................................................................................................................................................................................................................72

Figure18.Historiccapitalbudgetoverspending(underspending)bydirectorate,byfiscalyear...........................78

Figure19.NaturalgasfractionoffuelconsumedatCostaSur5&6sinceApril2016..................................................95

Figure20.CapacityfactoratPaloSecounitsFY2012‐2016and“limiteduse”designationforMATScomplianceonUnits1&2..................................................................................................................................................................................................99

Figure21.HistoricandexpectedcapitalspendingonindividualprojectsatPaloSecoforprojectswithspendingfromFY2017toFY2019,millionsofdollars..................................................................................................................................101

Figure 22. Capacity factor at San Juan 7& 8 FY2012‐2016, relative to “limited use” designation forMATScompliance....................................................................................................................................................................................................104

Figure23.Historictransmissionanddistributionbudgetandactualspendingbyfiscalyear,2010to2015;ratecasebudget....................................................................................................................................................................................................134

Figure24.Historicover(under)spendingintransmissionanddistribution.................................................................135

Figure25.PREPA’sTransmissionSystemvintage2013,fromURS2013Report.Includesplannedtransmissionthrough2018,vintage2013..................................................................................................................................................................137

Figure26.CompositionofPREPA'sfuelmixovertime,onaheatcontentbasis...........................................................157

Figure27.PREPA'sfuelpriceforecastsascomparedtoactualvalues..............................................................................161


Figure28.Historicalandmodeledfuelmix(%naturalgasbyheatcontent)burnedatCostaSurunits5and6.............................................................................................................................................................................................................................164

Figure29.BudgetedandactualspendingonPREPA'sPPOAswithAESandEcoEléctrica,FYs2010through2016.............................................................................................................................................................................................................................169

Figure30.PREPA'stotaloperationalspendingbyfunctionalarea,FYs2010‐2016...................................................197

Figure31.PREPA'soperationallaborspendingbyfunctionalarea,FYs2010‐2016..................................................198

Figure32.PREPA'soperationalnon‐laborspendingbyfunctionalarea,FYs2010‐2016........................................199

Figure33.PercentbywhichPREPAunder‐oroverspent itsoperationalbudgets inFYs2010‐2012andFYs2014‐2016,bytypeofexpenseandfunctionalarea..................................................................................................................200

Figure34.Percentageoflaborspendingbyfunctionalarea,withproposedallocationfactors.............................207

Figure35.Percentageofnon‐laborspendingbyfunctionalarea,withproposedallocationfactors...................207


I. INTRODUCTION

A. Purposeofreport

SynapseEnergyEconomics,Inc.(“Synapse”)staffwereretainedbythePuertoRicoEnergyCommission(“CEPR”orthe“Commission”)toreviewelementsofthePuertoRicoElectricPowerAuthority’s(“PREPA”orthe“Company”)firstratecase,asinitiallyfiledMay27,2016,and to make recommendations to the Commission with respect to PREPA’s revenuerequirements. We were tasked with assessing PREPA’s capital budget requirement,operationalandmaintenanceexpenses,fuelcostestimates,andpowerpurchaseagreementcosts,aswellasPREPA’sestimatedsalesforecastforthefiscalyear2017.

Thisreporthastwoprimarypurposes.First,itfulfillsourobligationtoassess,tothebestofourability,PREPA’sfundamentalrevenuerequirements,theprudenceofPREPA’sdecisions,thebasisofPREPA’scostsandcostestimates,andtheexecutionofprojectsasincurredinrates. Secondly, and to some extent more importantly, it seeks to provide a detailedbackgroundonthegenesisofthecostsincurredbyPREPA,theprocessusedbyPREPAtomakebudgetarydecisions,andthemechanismsbywhichPREPAestimatescosts.

Asafirstratecasefiling,theCommissionhasnoestablishedrecordofcapitaloroperationalrequirements, performance, or cost containment. The Commission and public have littleknowledgeofPREPA’sspendingandprioritization,andremarkablylittleinformationaboutPREPA’sgenerators,transmissionsystem,orreliability.PREPAasanentityhasoperatedfordecadeswithouteithertransparencyoroversight,andisinadeepfinancialcrisis.ThisratecaseprovidedoneofthefirstopportunitiestounderstandthestateofthesystemandhowPREPAoperatesbothday‐to‐dayandonastrategicbasis.

As a newly regulated utility, PREPAwas seemingly unaware of its obligation to providerecordstotheCommissionandstaff,aswellasintervenors.Asaresult,theorganizationwasill‐prepared to provide clean, auditable records. PREPA’s testimony in this rate caseprovided the barest of insight into a struggling utility. It is the intent of this report toconsolidate disparate records, files, discussions and testimony into a description of theutilitytheCommissionregulatestoday.

Attheendoftheday,theCommissionmustdecideifPREPA’sestimatedcostsinthisratecaseareprudentlyincurred,ifthecostsarereasonable,andwhatconditions—ifany—areplacedontheprovisionofratestoPREPA.Itistheauthor’sintentthatthisreportprovidestheCommissionconfidenceinsettingratesforFY2017,andguidancetowardsanimprovedprocessinfuturecases.

B. Qualificationsofauthors

Dr.JeremyFisherisaPrincipalAssociateatSynapse.HehasbeenemployedatSynapsesince2007 working on electricity system energy planning and currently leads its resourceplanninggroup,whichengagesintheassessmentofplanningprocessesacrossawidecohort


ofstatesandregions.Dr.Fisherhasprovidedconsultingservicesforawiderangeofpublicsector and public interest clients, including the U.S. Environmental Protection Agency(“EPA”), the National Association of Regulatory Utility Commissioners (“NARUC”), theNational Association of State Utility Consumer Advocates (“NASUCA”), National RuralElectricCooperativeAssociation(“NRECA”), thestatesofAlaska,Arkansas,Michigan,andUtah, the Commonwealth of Puerto Rico, Tennessee Valley Authority Office of InspectorGeneral(“TVAOIG”),theCaliforniaOfficeofRatepayerAdvocates(“CAORA”),theCaliforniaEnergyCommission(“CEC”), theRegulatoryAssistanceProject (“RAP”), theWesternGridGroup, the Union of Concerned Scientists (“UCS”), Sierra Club, Earthjustice, NaturalResourcesDefenseCouncil(“NRDC”),andotherorganizations.

Dr.FisherholdsamastersanddoctorateinGeologicalSciencesfromBrownUniversity,andreceivedhisbachelordegreesfromUniversityofMarylandinGeologyandGeography.HisfullcurriculumvitaeisattachedasExhibitFisherandHorowitzExhibit01.

Dr.ArielHorowitzisaSeniorAssociateatSynapse.Dr.Horowitzisanexpertindataanalysisandenergysystemsandtechnologies.AtSynapse,Dr.Horowitzconductsin‐depthresearchonspecializedelectricsystemissues,includingelectricsystemmodeling,engineering,anddatascience.DuringhertenureatSynapse,Dr.Horowitzhasledthemodelingoftheimpactsof clean air regulations for the State ofMichigan,worked onmultiple litigated resourceplanningdockets,andco‐authoredtheCommission’sexpertreportonPREPA’sintegratedresourceplan(IRP).

Dr.HorowitzholdsadoctorateinChemicalEngineeringfromTuftsUniversity,whereherresearchfocusedoncutting‐edgeelectrochemicalenergystoragetechnologies.ShereceivedherbachelordegreefromSwarthmoreCollegeinEngineering.HerfullcurriculumvitaeisattachedasExhibitFisherandHorowitzExhibit02.

C. Organizationofreport

This report covers numerous detailed topics and describes, in depth, areas of PREPA’sbudgetingandoperationalprocesses.Thereareamultitudeofareasinwhichourabilitytoassess PREPA’s budget required an assessment of process, extrapolation from relatedrecords,andcomparisonofincompletedatasets.Eachsectionofthisreportattemptstodrawindependentconclusionsbasedonthecollectedevidence.

Thereportisorganizedasfollows:

SectionII:SummaryofFindingsandRecommendationsisasummaryofourfindingsandrecommendations to the Commission. It details our findingswith respect to overarchingtopics, and provides a summary of specific recommendations with respect to the loadforecast,capitalbudget,fuelandpurchasedpowerexpenses,andoperatingexpenses.

SectionIII:StateofPREPA’sSystemprovidescriticalbackgroundonthestateofPREPA’ssystemasportrayedthroughresponsestoCommissioninquiries,data,conversationswith


PREPAstaff,andtestimony.ThissectiondetailsPREPA’scurrentgenerationreliabilitycrisis,therisinglevelofinterruptionsexperiencedbycustomers,andPREPA’sstrategytocomplywithanimportantfederalenvironmentalregulation.ThissectionprovidescontextfortheremainderofPREPA’sdecisions.

SectionIV:LoadForecastexaminesthevalidityofPREPA’sloadforecastbothintheshortterm(FY2017)andovertheextendedperiodcontemplatedinPREPA’sratecasefiling.WereviewPREPA’smethodologyandproposedloadforecastforthisratecase.

SectionV:CapitalBudgetreviewsPREPA’s$280millionFY2017capitalbudgetrequest.ThissectionexaminesPREPA’scapitalprojectsintheareasofproductionandgeneration,the Aguirre Offshore GasPort (AOGP), transmission and distribution, and other capitalprojects,includingvehicleprocurement.ThischapterreviewsPREPA’shistoricalspendingpatternsinDirectoratesandDepartments,andassesseshigh‐costindividualprojects.ThissectionalsoprovidesextensivedetailonthestateofPREPA’sgeneration,transmissionanddistributionsystemasbackgroundtoPREPA’sexpectedspending.

SectionVI:FuelandPurchasedPowerassessesPREPA’sestimated$1.47billionfuelandpurchasedpowerbudgetforFY2017.ThischapterreviewsthefuelpriceforecastusedtosupportPREPA’sestimatedfuelprocurementbudget,themodelingconductedbyPREPAtoprepare amonthly budget for fuel and purchased power, and a comparison of PREPA’sforecastsagainstcurrentpricingandmarketforwards.

SectionVII:OperatingExpense reviews the $560million operational andmaintenance(“O&M”) budget assessed by PREPA. This chapter reviews PREPA’s historical spendingpatterns and PREPA’s portrayal of its budget requirements on a going‐forward basis. ItassesseshowPREPA’sO&MbudgetsareusedbyDirectorate,andarelikelytobeusedgoingforward.


II. SUMMARYOFFINDINGSANDRECOMMENDATIONS

A. Informationreviewedforthisreport

OurreviewofPREPA’splannedspendingandrevenuerequirementisbasedoninformationprovidedbyPREPAin itsratecasepetition,aswellasasignificantandlengthyrecordofdiscovery.WereviewedmanyelementsofPREPA’sfiling, includingseveralkeyschedules(A‐1,A‐2,A‐6,E‐6,F‐3,andF‐4primaryamongthem);1thedirectandrebuttaltestimonyofPREPA’srevenuerequirementpanel2andthedirecttestimonyofthepanelofMiranda,Sales,and Sosa;3 PREPA’s business plan;4 the FortiethAnnual Report fromPREPA’s ConsultingEngineers;5andthefinancialmodelsubmittedbyPREPAinsupportofitsratecasefiling.6

WesubmittedandreviewedresponsestoseveralhundredRequirementsofInformation,aswellasreviewingmanyofPREPA’sresponsestootheroftheCommission’sadvisors.7WereviewedhistoricalrecordsrelatingtomuchofPREPA’scapitalspendingandallofPREPA’soperationalspendingoverthepastfivefiscalyears.Wealsoparticipatedintwotechnicalconference calls with PREPA. During these calls, we spoke to over a dozen of PREPA’sengineering,planning,finance,andforecastingstaff,aswellasseveralofPREPA’sadvisors.

In addition to information provided by PREPA in this case, we reviewed the testimonysubmitted by intervenors. Finally, we also relied upon PREPA’s filed IRP and theCommission’sreviewthereofinDocketCEPR‐AP‐2015‐002.

B. MajorfindingsregardingPREPA’spresentandfuturespendingneeds

Below,wesummarizethekeyfindingsofourresearchintoandanalysisofPREPA’spetitionforratereview.

1. PREPA’sreliabilityhassufferedsinceFY2014

1PREPA’sPetitionforRateReview.ASchedules,ESchedules,FSchedules,andrevisionsthereof.

2 PREPA’s Petition forRateReview, Exhibit 5, Direct Testimony of Francis X. Pampush, LucasD.Porter,andDanT.Stathos.

3PREPA’sPetitionforRateReview,Exhibit3,DirectTestimonyofSoniaMiranda,AntonioPerezSales,andVirgilioSosa.

4PREPA’sPetitionforRateReview,Exhibit3.02,PREPABusinessPlan.

5PREPA’sPetitionforRateReview,ScheduleI‐1.

6PREPA’sPetitionforRateReview,Exhibit14.02.

7WesubmittedROIsaspartoftheCommission’sFourth,Sixth,Seventh,Eighth,11th,14th,and17throundsofROIs.


AreviewofreliabilitymetricsforPREPAindicatesthattheutility’sabilitytoprovidesafeand reliable service has declined substantially since FY2014, a fact underplayed by theCompany’spresentationof this caseandgenerallynot featuredasan issue in intervenortestimony. And while reliability is not the only metric by which a utility measures itsperformance, it isa fundamentalandkeycomponentofsociety’sexpectationsforelectricservice.

Ms.Miranda’spanelmakespassingreferencetoPREPA’s“ailinggrid;”8Mr.Quintanastatesthat“insufficientrevenuesha[ve]ledtodegradationsofPREPA’sinfrastructure,”notingthat“PREPA’stransmissionsystemisdeteriorated,andthishasledtoreducedreliability.”9ButneitherwitnessdescribesthestateofPREPA’ssystemandtheextentofthecrisisfacedbytheutilitytoday.

Aspartofthepre‐filingofthiscase,PREPAsubmitteda2013reportfromtheconsultancyURS(“URS2013Report”)thatshowedPREPAconsistentlyachievingitsreliabilitygoals.10AcasualreviewofPREPA’sfilingmightgivetheimpressionthatwhiletheutilityisconcernedaboutpotentialdegradationandstaffshortages, thesystemisrelativelyrobust.AcarefulreviewofPREPA’srecordsprovidedinresponsetoprobingdiscoveryrevealadrasticallydifferentstory.PREPA’sgeneration,transmission,anddistributionsystemsarefallingapartandreliabilityissuffering.

Asof the last recordsprovidedby theCompany,PREPAhas anextraordinaryamountofbaseloadgenerationoffline.Extendedforcedoutages,startinginFY2015andFY2016,haveledPREPAtorelyonmoreexpensiveunits,andforcedtheutility touseenvironmentallynon‐compliantgeneratorsthatitpledgedwouldnotgetused.Thesevereoutages,deferredmaintenance, and a lack of experienced staff have resulted in an increasingly brittletransmission system—as witnessed by the three‐day outage just two months ago inSeptember2016.PREPA’scustomeroutagerateisfarhigherthanotherU.S.utilities,andthisratehasbeenincreasingoverthelasttwoyears.PREPA’sownrecordsshowthatthenumberofserviceinterruptionsexperiencedbyPREPAcustomersinthepastfewmonthsof2016havebeenfourtofivetimeshigherthantheaverageU.S.customer.

PREPA’smid‐levelmanagershavebeenfrankintheirdiscoveryresponsesanddiscussionson conference calls. PREPA’s fundamental infrastructure is in jeopardy due to a lack offunding and significant workforce reductions.11 The utility has shifted from performing

8PREPAExhibit3.0at211‐213

9PREPAExhibit1.0at80to84

10 URS June 2013 Annual Report (“URS 2013 Report”). PREPA Exhibit 3.02(D), Schedule I‐1.ConsultingEngineersReport.FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Pages44‐46.

11 CEPR‐RS‐03‐03, CEPR‐AH‐02‐01, discussions with generation and transmission staff duringOctober20ClarificationCall.


preventativemaintenancetotriagingoutagesastheyoccurthroughreactivemaintenance.PREPA’s internal documentation on outages is even starker: a lack of experienced staff,deferred maintenance, failures to pay vendors, and problems with procurement haveseverelyimpactedthePREPAgenerationfleet.12

We find that PREPA is in dire need of a capital infusion: monetary capital to pay formaintenance and ensure the system is operational; human capital to adequately staffgeneration and transmission equipment, and intellectual capital to strategically deployPREPA’slimitedresourcestobreakoutofacycleofdisrepairandtransitionthefleettoacleaner,morereliablesystem.

2. PREPAoperatesinacost‐constrainedenvironment

PREPA, by every measure, operates in a deeply cost‐constrained environment. Limitedaccesstocapitalmarkets,politicalpressuretosustainlowrates,andaconstrainedeconomyhave led to PREPA having relatively sparseworking funds. In light of these constraints,PREPAhasextendedmaintenanceoutagecycles,movedtoreactivemaintenance,deferredsomeinitiatives,andshedstaff.

WerecognizetheextentofPREPA’slimitedbudgetsandfinancialdistress.Thefactthattheutility has succeeded in maintaining generators and an extensive transmission anddistributionsystemunderthesecircumstancesisatestamenttotheabilityofPREPA’smid‐levelmanagerstooperateunderhighlyconstrainedcircumstances.

3. PREPA’s capital and laborbudgetsmaynot support a safe andreliablesystem

PREPA’s budgets presented in this rate case seem to envision a modest increase inoperationalandmaintenance(“O&M”)budgetssincethelastbudgetcycle,butourreviewindicatesthatPREPA’soperationalspendinghasnotbeenconsistentwithoperationofasafeandreliablesystemsinceatleastFY2014.Boththegenerationandtransmissiondirectoratehaveindicatedthatreducedstaffingandabilitytoaccessmaintenancedollarshaveseverelyhamperedtheirabilitytomaintainreliabilityandpreventthedegradationofkeyresources.

PREPA’scapitalbudgetsforthebasicupkeepofthegenerationfleetandtransmissionsystemareconsistentwithprioryearspending,despitethecurrentstateofthesystem.Whileitisdifficulttotellonagenerator‐by‐generatorsystemifPREPAhasallocatedappropriatefunds,weareconcernedthatthecausesofrecentextendedoutagesarenotnecessaryaddressedby PREPA’s near‐term capital budgets. In addition, PREPA’s reliance on a marginalgenerationfleetthatitotherwiseexpectedtoretire,oratleastsideline,isnotsupportedbythecapitalbudgetputinplaceinthiscase.Forexample,PREPAseekstorunsomeofitssteamunits in a “limited use” framework to meet environmental regulatory requirements.

12CEPR‐JF‐01‐16Attach01(PublicVersion).


However,someofthose“limiteduse”unitshaveoperatedmorethangeneratorsinPREPA’sprimary fleet, indicating that they are critical for reliability. PREPA makes no capitalinvestmentsintheseunits,despiteitsrelianceonthem.

PREPA’sbudgetsforO&MinFY2015–016declinedsubstantiallyfromFY2010–2014levels,andthatdeclinetrackedPREPA’sreducedreliabilityalmost in lockstep.Inthisratecase,PREPArequestsanO&MbudgetconsistentwithFY2015–2016.WeareconcernedthatthisbudgetisinadequatetomeetPREPA’sburgeoningmaintenancerequirements—particularlyintheareasofgeneration,transmission,anddistribution.

4. PREPA’snon‐laboroperationsbudgetsarepoorlyallocated

AmajorcomponentofPREPA’soperationalspendinglandsinAdministrativeandGeneral(A&G) functional area, and that spending in this area has increased in recent years forunexplainedcauses.AreviewofPREPA’srecordsshowsthatPREPAspenttheastonishingfigure of $165million inA&G inFY2016, ofwhich $134million fell into anundescribeddiscretionaryfund.Togivethisfigurecontext,PREPAspenttheequivalentofmorethanathirdofitsentirecapitalbudgetondiscretionaryA&Gspending.

Problematically,PREPAdescribesthatitis“aninefficientbureaucracy”thatis“overlystaffedwithnon‐value added administrativepersonnel,” and that “the executivedirectorate andexecutive team is oversized.”13 It is difficult for us to overstate how concerning this is.Moreover,wehaveabsolutelynofurtherinformationaboutwhat,exactly,PREPAspentthesefundson,althoughwecanmake inferences.Aseparatediscoveryresponse indicates thattheywerelikelyspentbythe“CorporateResponsibilities”directorate14butwedonotfindthistobeparticularlyinformativeastothenatureoftheexpenses.

5. PREPA’srecordkeepingissub‐par

Acrossseveralareas,PREPAwasunabletoprovidethebasisorjustificationforbudgetitemsorspending.Forexample,werequestedthatPREPAidentifywhichcapitalprojectshadworkordersorcontractsforengineering,procurement,andconstruction(EPC).PREPAdeniedtherequestinfull,statingthattheirrecordswerenotsufficienttoprovidesuchjustifications.PREPAexplainedthatitdoesnothaveanelectronicsystemfortrackingcapitalprojects,andthatmostinformationwasmaintainedonpaperrecords,distributedacrossregionaloffices.PREPAexplainedthatdetailsandbudgets forprojectsweremanagedata localscaleandbudgets(butnotjustifications)werefilteredupthroughacentralbudgetingprocess.

13PREPAExhibit3.0at659‐662.

14CEPR‐JF‐01‐24_Attach05.xlsx.Allspendingin“Genmiscexpcontrolledbyresp”expensesoccurredunderthedirectoratelabeled“A14‐ResponsabilidadesCorporativas”.


PREPA’s budget is allocated from a central office, but managed across a wide range ofdirectorates andmany local offices. It is not clear that PREPA has the infrastructure tomanageastrategicbudgetallocationprocess.

PREPA’s lack of reasonable documentation also renders it difficult to provide oversight.PREPA’s“detailed”explanationformulti‐milliondollarcapitalprojectswereoftenlimitedtotwotothreesentences,andinnocasedidPREPAprovideanexplanationofthebasisofitscostestimates.Forallintentsandpurposes,PREPA’scapitalbudget—forthevastmajorityof items listed—could have simply been a rough estimate based loosely on managerexperience.PREPAprovidedfewrequestsforproposals,proposalsorbids,orengineeringestimates.

6. PREPA’sbudgetallocationprocessisopaqueanddistortionary

PREPAprovidesforanapproximate$300millioncapitalbudget,butgivesnobasisforthisprecisefigure.IndiscussionswithPREPA’sfinancialdirector,welearnedthatPREPAcreatesacapitalbudget,buttheutilitycouldprovidenoclearexplanationwhythevaluewassetat$300millionversusanyothervalue.

Ahardcapbudgetcan,insomecircumstances,beareasonablemechanismofincentivizinga strategic allocationof budgets—butwe learned fromPREPA that theorganization setsbudgets in a much more rudimentary fashion, and has little infrastructure for makingstrategicallocations.ThefinancialdirectorofPREPAinformedusthatbudgetsareallocatedonprioryears’spendingorbudget,whicheverislower.Thisbudgetingprocessdecouplestheutility’sstrategicneedsfromthebudgetcreationprocessatthedirectorate level,andcreatesadistortionaryeffectwheredirectoratesseektospendbudgets—evenifthey’renotrequired—forfearoflosingfutureyearbudgets.

On a project‐by‐project basis, PREPA provided no cost‐benefit analyses, no economicassessments,andnoalternativeanalyses.These typesofanalyses,aswellasengineeringestimates,establishbaselineconfidenceinlocalbudgetingprocesses.Wehavenoneofthatconfidenceinthiscase,andinsteadmustrelyonourownexternalassessmentsofPREPA’sneedsandthevalueofPREPA’sprojects.

7. Fundamentalneedsshoulddriverevenuerequirements;PREPA’sbudgetsconstrainedbyaperceptionofrevenues

PREPA’shardcapbudgetsaredrivenbyitsperceptionthatitsrevenuestreamisnecessarilyfixed.Thisstandsinstarkcontrasttootherutilitieswhererevenuerequirementsaredrivenbyneeds,temperedbypoliticalrealities(i.e.theavoidanceofrateshock).InPREPA’scase,thepolitical realityofavoidingcost increasesappears to lead,withsystemrequirementstakingabackseat.


Tokeepitsbudgetsunderthecap,PREPAhasengagedinwhatappeartobeself‐defeatingpractices, such as deferring maintenance,15 extending outages to avoid overtime,16 andallocatingbudgetawayfromcritical,butlow‐utilizationunits.17

Inthemeantime,PREPA’sgeneration,transmission,anddistributionsystemclearlyrequiresubstantialmaintenance,overhauls,andrepairs.Asanewlyrate‐regulatedutility,arationalpathwaywouldhavebeenforPREPAtoestablishitsneedsbeforetheCommission,andthenworkwiththisCommissiontosetareasonablebudgettoensuresafeandreliableoperation,ratherthanthroughanartificialbudgetcap.

8. Significantinvestmentanddevelopmentofcorecompetenciesareneeded to improve the condition and performance of PREPA’ssystem

Quitesimply,PREPAhasdugitselfintoaholeofdisrepair,andsuffersfromshortagesofbothresourcesandskills.ThisCommission18andthelegislatureofPuertoRico19havebothsetforthambitiousvisionsforPREPA’sfuture.ThesevisionsdescribeanewPREPA,focusingonthegoalofa futurePREPAthatreliesoncleanenergygenerated fromresources thatareabundantlyavailableinPuertoRicoratherthanrequiringconstantimportsoffossilfuelsandexposingthepeopleofPuertoRicotothevagariesoftheglobaloilmarket.20Inthesevisions,empoweredcustomersinteractonequalfootingwithaflexible,transparent,andresponsivePREPA;21 demand‐side solutions contribute equitably to supply‐side resources on amodernizedgrid;22 andeverypartofPREPA’sorganizationoperatesefficiently todeliverbetterenvironmentaloutcomesandcustomerexperiencesatlowercosts.23

WesharethesevisionsforPREPA’sfuture.However,thecurrentreality isstark:manyofPREPA’sexistingunitsareinsuchapoorstateofrepairthatPREPAmustconsideritselflucky


16 URS 2013 Report. Page 57. “The Authority has adopted the policy of avoiding overtime forscheduled outages to reduce their costs. This work practice has extended the duration of theseoutagesandnegativelyimpactedavailability.”

17Forexample,PREPAdidnotidentifyanycapitaldollarsspentonmaintenanceatPaloSeco1&2overthelastsixyearsdespitethesignificantuseoftheseunits.SeeCEPR‐AH‐6‐12Attach01

18NoticeofInvestigationtoIdentifyOpportunitiestoImprovePerformanceofthePuertoRicoElectricPowerAuthority,CEPR‐IN‐2016‐0002(November15,2016).Hereafter“PerformanceCaseNOI.”

19Act57‐2014.

20Act57‐2014,StatementofMotives;PerformanceCaseNOI,SectionII.

21Id.

22Id.

23Id.


iftheyremainoperationalformorethanseveralmonthsatatime.24PREPA’stransmissionanddistributionsystemsarefallingapartquiteliterally:theyarecracking,corroding,andcollapsing.25PREPAisfailingatthebasicmandateofanelectricutility,whichistosafelyandreliably supply electricity to its customers. Neither the Commission, other governmentalauthorities, nor the public should be misled about PREPA’s current state, which isunambiguouslyoneofcrisis.

9. PREPA’scurrentstatewillrequiretime,effort,andfundstorepair

PREPAcannotescapefromthiscrisisthroughapplicationofaband‐aidoracoatofpaint.NorcanPREPAandthepeopleofPuertoRicoaffordtowaitfortheinterventionofathird‐party savior. PREPA’s situation is an emergency in the present day, and PREPA and theCommissionmustbothtakeimmediatestepstoaddressit.OneofthosestepsmustbetherecognitiononthepartofPREPA,theCommission,andthepublicofthelevelofinvestmentthatwillberequiredtoimprovePREPA’scondition.PREPAhas“deferred”maintenancesooftenandforsolongthatrequiredmaintenancehasbecomerequiredrepairs,andrequiredrepairs have become required replacements—all with a bigger price tag than themaintenancePREPApushedoffforthesakeofconstrainingitsspending.26Werecognizethatthe ratepayers of PuertoRico can ill abide increases in alreadyhigh costs for electricity.However,weemphasizethatthereissimplynowayforPREPAtoachieveasafeandreliableelectricitysystemwithoutasignificantoutlayoffunds.

TheinvestmentsthatPREPArequiresarenotsolelyfinancial,however.PREPAsuffersfromasevere lackofadequately trainedstaffandsound internalpolicies,protocols,andtools.PREPA’s recordkeeping is archaic and unreliable. Its approaches to problem‐solving areoften improvised,with results that are disastrous as often as they are admirable. It hasrepeatedly shown itself to be penny‐wise and pound‐foolish. PREPA needs guidance,training, and talent to be able to take advantage of the opportunities for improvementaffordedtoitbythisCommission.TheCommissionthereforefindsitselfinthepositionofhaving to both regulate and teach—and, equally, PREPAmust find or develop in itself acommitmenttocooperationandawillingnesstolearn.

10. Role of energy efficiency, renewable energy, and distributedenergyresources

WeshareandstronglyendorseavisionofPREPA’s future thatrestsprimarilyonenergyefficiencyanddemand‐sidemanagement,renewableenergy,distributedenergyresources.Weagreeabsolutelyandwithoutqualificationsthatincreasedutilizationoftheseresourceswillresultinlowercostsandgreaterreliabilityovertime.However,thefactisthatsafety

24PREPA’sresponsetoCEPR‐JF‐01‐16(b)oftheCommission’sSixthROI(PUBLIC;August23,2016).

25PREPA’sresponsetoCEPR‐AH‐02‐01oftheCommission’sSixthROI(PUBLIC;August29,2016).

26CAPITALBUDGET,ChapterV,Page65


and reliability must be PREPA’s first priorities, above all other considerations. Energyefficiencyandrenewableenergytaketimetoimplement;PREPA’ssystemisinadesperatestate today. Integration of very high levels of renewable energy can present technicalchallenges even for mainland utilities with abundant resources and extensive availableinterconnections. Meanwhile, PREPA is barely able to provide electric service with itspresent fleet and dispatches its unitswith software thatwas developed in 198527—wellbeforethepossibilityofhighrelianceonvariablerenewableenergywascontemplatedbyelectricutilitiesasarealisticpossibility.

WerecognizethestrongdesireonthepartofmanystakeholdersforasignfromPREPAthatit is willing and eager to move towards greater reliance on renewables. We share thisconcern.We recommend that, over themedium and long terms, the Commission shouldrequirePREPAtodevelopandimplementastrategicplanthatallowsittoimprovereliabilityand safety without closing doors to cost‐effective adoption of energy efficiency andrenewable energy. In its long‐term planning, PREPA should orient itself towards thepossibility that adoption of renewable energy and energy efficiency can allow PREPA toavoidcostlyinvestmentsintraditionalgeneratinginfrastructureandcontinuedhighlevelsofspendingonfuel.PREPA’sstrategicinvestmentsshouldbedirectedatincreasingitsabilitytointegraterenewablesasquicklyaspossible,ratherthanclosingoffavenuesfordoingsocost‐effectively.

Weemphasize,however,thatthisratecasefocusesonPREPA’scostsandrevenuesforfiscalyear 2017.Realistically, PREPA can accomplish very little in terms of implementation ofenergyefficiencyandrenewableenergyoverthenextsevenmonths.Today,PREPAandtheCommissionmustfocusonaddressingPREPA’simmediatecrisisbyestablishingsufficientrevenueforPREPAtobegintoameliorateitssystem’sstateofutterdisrepair.

11. PREPA must begin to think in terms of its actual needs foroperationofasafeandreliablesystem

OneofthegoalsofthisratecaseisforPREPAtoestablishitstrueneedsforstaffing,capital,and operational spending. PREPA needed to demonstrate that it can plot a viable pathforwardsoutofitscurrentcrisis.WehaveseensomeevidenceofthisinPREPA’sinternalcommunications,includingaclearrecognitionthatPREPArequiresimprovementsinstafftrainingandcorecompetenciesrelatedtopreventativemaintenanceandsafety.However,PREPAalsoneededtopresentandjustifyanaccurateaccountingofwhatitactuallyrequirestooperateitssystemsafetyandeffectively.

Unfortunately,PREPAdidnottakethisopportunitytodoso.Repeatedly inourreviewofPREPA’splannedspending,wefoundartificialandarbitrarybudgetcaps.PREPArestrictedits operational spending to anominal increaseoverFY2016 levels (except in the areaof

27PREPA’sresponsetoCEPR‐JF‐01‐12(a)oftheCommission’sSixthROI(PUBLIC;August23,2016).


restructuringfees).28PREPAcontinuestoshedemployeesevengivenalaborshortagethatitrepeatedlycitedasacontributoryfactorinitsdecreasedreliability.29PREPArestricteditscapitalbudgettothepointwhereanysignificantcutswouldleadtoyet‐greaterimpairmentsofsystemstability.30WeconcludethatPREPA’s“revenuerequirement”bearsverylittle,ifany,relationshiptotherevenueactuallyrequiredbyPREPAtooperateitssystemsafelyandreliably. We find this extremely problematic, for reasons expressed by PREPA itself inrebuttaltestimony:

PREPAneedsratesthatcoveritscosts.ThatisreflectedinActs57‐2014and4‐2016andinbasicprinciplesofutilityregulation.Ifautilitycannotcoveritscosts,thenthatcanjeopardizeserviceadequacy,reliability,andsafety,anddriveuptheutility’scostsof obtaining capital, lead to financial failure and bankruptcy. PREPA has beenoperatingforalongtimewithratesthatdonotcoveritscosts,andwehaveseenthatthishasledtoasystemthatnotonlyneedsmodernizationbutneedsmajorworktomaintainandimprovereliability.31

WearenotconcernedthatPREPAwillvastlyoverspenditsbudgets;rather,ourworryisthatPREPAissoentrenchedinarestrictedcashflow‐basedmindsetthatitwaseitherunwillingorunabletopresentanestimateofitsactualcostsforrepairingandmaintainingitssystem.We recognize that PREPA, today, may not have the capacity to repair itself even if theCommissionweretoallocateitunlimitedfunding.However,inordertoeffectivelyregulate,theCommissionmusthaveasoundunderstandingofthegapbetweenwherePREPAisandwhere it shouldbe,andwhatpathPREPAmust traverseon theway to improvement.Byartificiallydepressingitsrevenuerequirementforthesakeofappearances,PREPAhasnotassistedwiththisgoalinthepresentratecase.

C. RoleoftheCommission

ThisCommissionistheentitybestpositioned,byitsjurisdictionandmissionunderAct57andbyitsexpertiseinelectricitysystemsandtheregulationthereof,toguidePREPAthroughapathtoimprovement.ThisratecasehasbeenextraordinarilycomprehensiveasitistheonusoftheCommissiontogainintimatefamiliaritywitheveryaspectofPREPA’soperations.PREPAisanextraordinarilycomplexorganizationinanextraordinarilycomplexsituation.TheCommissionmustfosterafullandnuancedunderstandingofPREPA,itshistory,anditscurrentstateinordertoregulateeffectively.

28OPERATINGEXPENSEBUDGET,ChapterVII,Page182

29OPERATINGEXPENSEBUDGET,ChapterVII,Page182

30PREPA’sresponsetoCEPR‐RS‐02‐06(b)oftheCommission’sFifthROI(PUBLIC;August9,2016).

31PREPA’sPetitionforRateReview,Exhibit20,RebuttalTestimonyofGregoryRivera,lines123‐129.


In general terms, having an unbiased set of eyes on PREPA’s operations is of crucialimportance.Act57recognizesthenecessityofsuchanexternalauthority,statingthat:

PREPAhasbeenconceivedandmanagedasanindependentGovernmententitythatisnotaccountabletoitscustomers.PREPA’svisionisinaccurateandinconsistentwiththepurposesthatprompteditscreation,whichshouldbethebasisforitsoperations.Suchvisionhas,attimes,ledtotheproposalandadoptionofmeasuresthat,inpracticalterms,adverselyaffectmostofitscustomers.32

The special quality of this Commission, as opposed to other external figures, is that it istaskedwithmaintainingthebestinterestsoftheratepayersofPuertoRicoaboveallotherconcerns.Whileitseekstoconsiderinputfromallrelevantstakeholders,itisnotbeholdento any of them.However, through open and transparent regulatory processes under thepurviewofthisCommission,stakeholdershavebeengivenanopportunitytoimpactPREPA’soperationsforthefirsttimeintheutility’shistory,andhaveilluminatedseveral issuesofpublicconcern.Mostimportantly,theCommissionisnot“for”or“against”PREPA:itsonlymotivationistoseekthebestoutcomeforPuertoRico’sratepayers.Goingforward,theroleoftheCommissionmustbetomonitor(and,ifnecessary,adjust)theflowofinformationandfundsthroughPREPA,toassistPREPAinrecoveringfromthiscurrentcrisisthroughsoundregulatoryguidance,andtoenablePREPAtobestserveitsratepayersassafelyandreliablyaspossible.

D. Summaryofbudgetadjustments

Belowwesummarizeourrecommendations foradjustmentstoPREPA’sFY2017revenuerequirementorthecalculationorallocationthereof.

1. Withrespecttofuel,werecommendthattheCommissionadjustPREPA’sFY2017budgetupwardsby$461,305,000,foratotalFY2017fuelbudgetof$1,117,273,289aftersavingsfromperformanceimprovements.WecalculateandjustifythisadjustmentinFUELANDPURCHASEDPOWERBUDGETS,ChapterVI(Page155).

2. WithrespecttoPREPA’scapitalbudget,werecommendthefollowing,ascalculatedandjustifiedinCAPITALBUDGET,ChapterV(Page65)

a. TheCommissionapprovePREPA’sFY2017fullcapitalspendingatAguirre,PaloSeco,andSanJuanSteamPlants.

b. ThemaintenancecontractatSanJuanCombinedCycleberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$12millioninFY2017fromcapitaltoO&M,andthatthe

32Act57‐2014,StatementofMotives:TransformationofPREPA.


CommissionapprovetheothercapitalexpensesatSanJuanandAguirreCombinedCycleUnits.

c. ThemaintenancecontractatCambalacheberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$4millioninFY2017fromcapitaltoO&M.

d. TheCommissionrevisePREPA’srevenuerequirementforFY2017toreflecta$15millionspendingcapatAOGP,reducingFY0217revenuerequirementsby$41,339,807.ToreflecttheCommission’sorderthatPREPA“shallassume,forinformationalpurposesonly,thatAOGPwillbecomeoperationalatarealisticallyachievabledate,”werecommendincreasingtheFY2018budgetbythesame$41.3millionincrementto$454,756,927.

e. TheCommissionapprovethefullFY2017transmissionsystemcapitalbudgetasrequestedbyPREPA.

f. TheCommissionapprovethefullFY2017distributionsystemcapitalbudgetforallitemsexceptsecondarylinemeters(PID16677).

g. TheCommissionreducePREPA’smeterbudget(PID16677)from$10.6to$5.8millionforFY2017toreflecttheanticipatedcostofcontinuingPREPA’snormalAMRacquisitions.

h. WerecommendthattheCommissionapprovethefullFY2017transportationandcomputerequipmentbudget.

3. Withrespecttooperationalexpenses,werecommendthattheCommissionapplyaseriesofadjustmentstoPREPA’sFY2017operationsbudget,asjustifiedandcalculatedinOPERATINGEXPENSEBUDGET,ChapterVII(Page181).Theseadjustmentsinclude:

a. ReallocationofPREPA’sproposedbudgetbyfunctionalareatobetteralignwithhistoricalspendingpatterns,resultinginnonetchangetoPREPA’srevenuerequirement.

b. AdditionstoandsubtractionsoffundsfromPREPA’sproposedbudgetbyfunctionalareatobetteralignwithhistoricalspendingpatterns,resultinginanetadditionof$19,414,000toPREPA’srevenuerequirement.

c. Allocationofbudgetsfornewinitiativesandcoststospecificfunctionalareas,resultinginnonetchangetoPREPA’srevenuerequirement.

d. Recategorizationofmaintenancecontractsasoperationalexpenses,ratherthancapital,resultinginnonetchangetoPREPA’srevenuerequirement.


E. Summaryofotherrecommendations

BelowwesummarizeourrecommendationsthatdonotmodifyPREPA’sFY2017revenuerequirementorthecalculationorallocationthereof.

1. Withrespecttoloadforecasting:

a. TheCommissionshouldacceptPREPA’ssalesandloadforecastforFY2017only.

b. TheCommissionshouldrequirePREPAtoimproveitsloadforecastingmethodology.

c. TheCommissionshouldrequirePREPAtodevelopanewsalesforecastbasedonanimprovedmethodologypriortofilinganotherplanningorratecasewiththeCommission.

d. TheCommissionshouldrequirePREPAtoclearlydocumentanychangestoitsforecastingmodelsandsupportsuchchangeswithappropriateevidence.

e. TheCommissionshouldrequirePREPAtoadequatelydocumentitsloadforecastswithbothmethodologicaldescriptionsandcomprehensiveworkpapers.

f. TheCommissionshouldrequirePREPAtocorrectitsfinancialmodel,costofservicestudies,andallotherrelevantworkbookstorelyonitsactualby‐classforecasts.

2. Withrespecttothecapitalbudget:

a. TheCommissionorderthatPREPAwithholdfromsigningaFinalNoticetoProceedatAOGPuntilithassubmitted,andtheCommissionhasapproved,theAOGPEconomicAnalysis.PREPAshouldsubmit,ortheCommissionshouldrequirethatPREPAsubmit,theAOGPeconomicanalysisexpediently.

b. TheCommissionexaminethemaintenancecontractatSanJuanCCandtheperformanceofMHPS‐PRtodetermineifthecontractorismeetingperformanceexpectationsformaintenanceservice.

c. TheCommissionexaminetheCamabalachecontractandtheperformanceofAlstomtodetermineifthecontractorismeetingperformanceexpectationsformaintenanceserviceatCamabalache.

d. TheCommissionrequirePREPAtofilenoticeofanylong‐termcontractwithexternalserviceproviders(i.e.contractors)withapotentialnetpresentvalueof$25millionvalueorhigher.


e. TheCommissionrequirePREPAtoprovidestrategicplansforSanJuanandPaloSecosteamplants,includingthefollowingelements,ataminimum:maintenanceplan,MATScomplianceplan,aninvestmentplanformaintaining(ornot)SanJuan7‐10andPaloSeco1&2,andareliabilitystudy–i.e.whatstrainsareplacedonthesysteminthepresenceorabsenceoftheSanJuanand/orthePaloSecosteamunits.

f. TheCommissionrequirePREPAtoceasefurthersmartmeteroradvancedmeterinfrastructure(“AMI”)purchasesuntilastrategicplanjustifyingtheneedforAMIisproducedbyPREPAandapprovedbytheCommission.

g. Forunitsdesignatedas“limiteduse”forcompliancewithEPA’sMercuryandAirToxicsStandard(MATS),werecommend:

i. TheCommissionrequirePREPAtotrackandreportonprojectsatSanJuan7‐10andPaloSecoUnits1&2,regardlessofiftheseunitsareconsideredlimiteduse.

ii. TheCommissionrequirethatPREPA’snextlong‐termplanningexerciseevaluatetheextenttowhichthe“limiteduse”unitsshouldbemaintainedandcontributetopeakrequirementsonaneconomicbasis.

iii. TheCommissionrequirePREPAtrackscapitaldollarsspentateach“limiteduse”facilityseparately.

iv. TheCommissionrequirethatPREPA’slong‐termmodelingconsistentlyassessiftheselimitedusefacilitiesareavailableforpeakpurposes,andifnot,assessthevalueofmaintainingunitsthatneithercontributetopeakpurposesorprovideenergytothesystem.

3. Withrespecttofuelandpurchasedpower:a. TheCommissionshouldrequirePREPAtopreparefuelpriceforecasts,or

haveforecastspreparedonitsbehalf,atleastbiannually.b. TheCommissionshouldrequirePREPAtosubmitfortheCommission’s

reviewdetaileddocumentationandworkpapersregardingitsfuelpriceforecastingmethodologyaspartofPREPA’sinitialfilinginthenextmajorrate‐orplanning‐relatedcaseitbringsbeforetheCommission.

c. TheCommissionshouldrequirePREPAtoestablishandregularlyupdateadatabaseofitsrenewableenergycontractsandthecosts,performance,andstatusesthereof.

4. Withrespecttooperationalexpenses:a. TheCommissionshouldrequirePREPAtoadjustitsmonthlyreportformat,

suchthatfuturemonthlyreportswillincludegreaterdetailandmoreuseful


informationregardingthestatusofPREPA’soperationalbudgetsbyfunctionalarea.

b. TheCommissionshouldinvestigatePREPA’s“miscellaneous”non‐laborspendingintheAdministrativeandGeneralfunctionalarea.

5. Generally:a. TheCommissionshouldexerciseincreasedoversightofPREPA’sbudgeting

processandthestatusofitsbudgetsandspendingthroughouttheyear.


III. THESTATEOFPREPA’SSYSTEM

PREPA’s system today is in a state of crisis. Deferred and inadequate investment ininfrastructure,alossofkeystaff,andamyopicmanagementfocusonlargeriskybetshaveleft PREPA with generation and transmission infrastructure literally falling apart,unnecessarily high costs, a utility operating out of compliance with commonwealth andfederallaw,andalternativeoptionsrapidlydisappearing.

OurreviewofrevenuerequirementissuesultimatelymakesrelativelyfewadjustmentstoPREPA’scapitalandoperatingbudgetsrelativetoPREPA’srequests,butthisisnotasmuchanaffirmationofPREPA’sbudgetingprocessasitisareluctancetocutPREPA’sbudgetsinatime of crisis, a recognition that PREPA’s recent budget reductions have resulted indisastrous outcomes for the utility’s infrastructure, and a desire to see PREPA providereliableandsafeservicequickly.InassessingPREPA’sneeds,we’vecometotheconclusionthatPREPAisindireneedofacomprehensivestrategicplan,thelikesofwhichwerenotseen in either the business case supporting this rate case nor the preceding integratedresourceplan(IRP).BothPREPA’sbusinessplanandtheIRPattempttosugarcoatwhatisotherwiseadiresituation.

Overthecourseofthelasttwoyears,PREPA’sgeneratorshavefailedatanunprecedentedrate,strainingtheutility’ssystemandforcingtheutilitytorelyonhighercostgenerators.PREPA’scustomerinterruptionratesarefourtofivetimeshigherthanotherU.S.utilities,andPREPA’scostsarehigher.PREPA’sattempttomeetfederalenvironmentalregulationsthrough a massive investment in an offshore gasport and 15‐year commitment to gasdeliveries have been delayed time and again, are looking increasingly less economicallyattractive, and doubles down on the utility’s reliance on fossil fuels and inability toincorporate renewable energy.Workers suffer injuries and fatalities at an alarming rate.PREPA’smanagementisunabletothoroughlyaccountfortheuseofcapitalandoperationsbudgets,andthebudgetallocationsystemattheutilityisdistortionaryatbest.PREPA’smostexperiencedstaff,andthoseabletomakethesystemworkonhistoricallythinbudgets,areleaving.

WewerenotchargedwithtellingthewholestoryofPREPA’ssystemtoday,butcouldnotreasonably assess PREPA’s budgets and revenue requirements without an eye towardsPREPA’sperformance.ReviewingPREPA’santicipatedprojects,historicbudgetuse,andflowofinformation,wefeltstronglythatweneededtocharacterizePREPA’ssituation,andre‐emphasizetheneedforastrategicplan.

ThisratecaseisthefirstopportunityforthepublictounderstandPREPA’sperformanceandtransparency, and the results are grim. PREPA’s system today appears to be running onfumesandinouropiniondesperatelyrequiresaninfusionofcapital—monetary,human,andintellectual—torestoreafunctionalutility.

A. PREPA’sgenerationsystemcrisis


PREPA’s generation and transmission facilities are in a state of crisis. In PREPA’s filing,witnessesMiranda,SalesandSosadescribethattheutilitysuffersfrom“highforcedoutageratescausedbydeferredmaintenanceandskilledlaborleavingoperationalroles.”33

A review of the utility’s records shows a rapidly increasing generation outage rate, andcustomeroutagelevelsfourtofivetimeshigherthanotherU.S.utilities.Indeed,giventhelevelsofgenerationoutagereportedbytheutility,itisfairlyastoundingthatPREPAwasabletorestorepowerafteronlythreedaysintheSeptember21st2016Aguirreoutage.

ItisdifficulttooverstatethelevelofdisrepairoroperationalneglectatPREPA’sgenerationfacilities.Numerousreports,both internalandexternal,talkofmultiplecascadingevents,simple failures that blossom into crises, staff shortages and in some cases, staffincompetence.PREPA’ssystemtodayappearstoberunningonfumesandinouropiniondesperatelyrequiresaninfusionofcapital–monetary,human,andintellectual–torestoreafunctionalutility.

Figure1,below,showshowmuchpotentialgenerationhasbeenlosttoforcedoutagesatPREPA’ssteamplantsandtheSanJuancombinedcycle(“CC”)unitsbyquartersince2012.34The chart shows that, starting in late 2014, PREPA began losing substantial generationacrosstheirsteamfleet.Asof late2015,chronicandreoccurringoutagesatAguirre,PaloSeco and San Juan stations have led to PREPA forgoingmore than 2,000,000MWh eachquarter.Togivethisnumbersomeheft,theamountofgenerationthathasn’tbeenprovidedbythesegeneratorsbecauseofforcedoutagesexceedstheamountofgenerationthatcouldhavebeenprovidedbya2,300MWgeneratoroperatingwithamodest50%capacityfactor.

33PREPAExhibit3.0,lines367‐371

34FromCEPR‐JF‐01‐16Attach02.Forcedoutagesfromthisdatabaseincludebothcompleteoutagesaswellasde‐ratingevents,whereageneratoristurneddownwhileaproblempersistsorisfixed.MWhlostcalculatedbyPREPA.


Figure 1. Total energy (MWh) lost to forced outages by quarter, calendar year.35

AnextraordinaryamountofPREPA’sgenerationis,byallaccounts,currentlynotavailable.Theseforcedoutagesappeartobeadirectconsequenceofdeferredmaintenance,reducedcapitalspending,andashortageofexpertstaffinkeypositions.

1. PREPA’sassessmentoftheforcedoutagecrisis

DoesPREPArecognizethedepthofcrisisitisintoday?FromourdiscussionswithPREPAmanagementandreviewofdocumentation, it isclearthatPREPA’smid‐levelmanagers—both for generationand transmission/distribution—arekeenly awareof theirneeds andhavedemonstratedanabilitytoworkwithextremelylimitedresourcesspreadthinlyoverPREPA’s system. It isnot clear,however, thatPREPA’s topmanagementunderstands thelevelofcrisisorhowtostrategically invest,andthereare indications thatcompetency ismixedattheplantmanagementlevel.

An internalmanagementpresentation,apparently frommid‐2015, shows thestartof theupwardforcedoutagetrendinFigure1,above.Alreadyshowingpersistentlyhighoutagesrelative to historic trends, the slide deck underplays the emerging problem, stating that“whilePREPA’sunitshaveovertimeexperiencedaslightdecreaseinFOevents,therehasbeenarecentincreaseduring2015.”36Theslideshowsagraphicwithamonth‐to‐monthline

35CEPR‐JF‐01‐16Attach02.

36CEPR‐JF‐01‐16Attach01.ForcedOutage(“FO”)Analysis:BusinessCase.Publicversion.


witha“forcedoutagefactor”(“FOF”),whichPREPAstatesis“theprobabilitythataunitwillnotbeavailableforservice.”PREPAindicatesthatithasaveraged6.87%from2010toyear‐to‐date 2015—but then shows that the factor sits at an historic high of 27%—anextraordinary level. For any other top managers, this trend, and an examination of theoutagesalreadyincurredbymid‐2015,shouldhavetriggeredarapidmitigationprocess.ThegenerationfleetatPREPAisthebackboneofoperations.

In testimony, PREPA’s Executive Director Javier Quintana states that PREPA has made“improvements[to]generationperformance,”and“finalizinganintegratedprocessthatwillreducethenumberandseverityofforcedoutageevents,furtherstabilizingthesystem.”37Mr. Quintana’s testimony implies that, as of May 2016, PREPA’s system was improvingrelativetohistoricperformance.Thisisnotcorrect.Inmid‐2016,PREPAwasexperiencingitshistoricallyworstperformancesince2012withmassiveandchronicfailuresatAguirreandPaloSecoplants.InlateSeptember2016,afireattheAguirresubstationplungedtheislandintoathree‐dayblackout,demonstratingthatPREPA’ssystemwassofragilethatitwas unable to copewith this first contingency. From the perspective of a utility generalmanager, mitigating severe generation and transmission outages should be a top‐levelpriority—notacasualsidenote.

Whatledtothiscrisisstate?PREPAidentifiesatleastthreeprimarydriversoftheincreasingforcedoutagesatgeneratorsandinthetransmissionsystem.

Deferredmaintenance:Ms.Miranda’spaneldiscussesthatdeferredmaintenanceisinpartresponsible for forced outages.38 This is highlighted again in PREPA’s internaldocumentation, where it states a root cause of forced outages are “exceeding OEMrecommendedequipmentoverhauls schedule,” “budget constraints toperformnecessaryinspections,maintenanceandrepairs.”Problematically,PREPAstatesinthisreportthattheyare“waitingforgasconversion[atAguirre]beforeinvestinginmajorrepairs.(e.g.AG1andAG2).”39Inevitabledelaysinthatprojectmayhavecausedfarmoredeferredmaintenancethan originally anticipated by PREPA. Finally, the 2013Annual Report by URS (PREPA’sconsultingengineers)indicatesapolicychangeatPREPAthatmayhaveresultedinworseoutcomes.“TheAuthorityhasadoptedthepolicyofavoidingovertimeforscheduledoutagestoreduce theircosts.Thisworkpracticehasextendedthedurationof theseoutagesandnegativelyimpactedavailability.40

37PREPA’sPetitionforRateReview,Exhibit1.0,DirectTestimonyofJavierQuintana‐Mendez,lines313‐319.

38PREPAExhibit3.0,lines367‐368.

39CEPR‐JF‐01‐16Attach01(Publicversion).

40PREPA’sPetitionforRateReview,Exhibit3.02(D)andScheduleI‐1.ConsultingEngineersReport.FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Hereafter“URS2013Report”or“ConsultingEngineer’sreport.”


ReviewingPREPA’soperationsandmaintenance (“O&M”)expenditures in thegenerationdivision,we see that the availability of PREPA’s generation units, heremeasured by thepower(MW)availableinthesteamunitsafterforcedoutageshasdeclinedinlockstepwithspendinginO&M.Unitsbecomeunavailableasdifferentpartsfailovertime.Weconcludethat a substantial componentof the forcedoutages ingenerationand transmissionareafailuretoinvestormaintainatanappropriatetime.

Figure 2. Steam unit availability (in MW relative to max capacity) vs. historic generation operations and maintenance, by fiscal

year.41

Theproblemisnotisolatedtothegenerationsystem.Inresponsetoaqueryregardingthedegradationofthetransmissionsystem,Ms.Mirandastatedthat:

…duetolimitedcapital,transmissioninvestmentswereselectedbasedprimarilyonthe immediate impact upon system reliability of a failure to invest. Thus, PREPAinvested in transmission projects that helped to alleviate or mitigate operationalcongestionproblems,systemoverloads,andvoltageregulationproblemsatthebulktransmissionandsubtransmissionlevel.Whenundersignificantcapitalconstraints,such investments are the most urgent and failure to make them has the mostimmediateandunavoidablepotentialconsequences.42

Staffavailabilityandcompetence:Ms.Miranda’spaneldiscussesthatforcedoutagesaredue, inpart to“skilled labor leavingoperationalrolesandnotbeingreplaced.”43PREPA’sowninternaldocumentationbacksupandexpandsthiscontention,blamingtheoutageson

41DerivedfromCEPR‐JF‐01‐16Attach02.Author’scalculation.

42CEPR‐JF‐01‐01(a).



a “lossofsignificantnumberofexperiencedpersonnel,”andthat “newemployeesdonothavetherequiredexpertiseandknowledge.”Thatstaffalsofailstodocumentandassesstherootcauseofproblems,resultinginalackoffollowupandstrategicthinking.44However,theproblems extend further, into PREPA’s contractors, where “technical advisors [are] notalwaysfamiliarwithtechnology…[suchas]statorwindingsatAguirre,turbinecontrolsissueatCostaSurandvibrationsproblemsatSanJuan.”45

Decliningcapitalinvestments:BothMr.QuintanaandMs.Miranda’spaneltestimonybothindicatethat“PREPA'shistoricallylowinvestmentoncapitalexpenditureshadledtomorethan twice the number of forced outages than the U.S. industry standard.”46 PREPA’sinvestmentsatitsgenerationfleethavedeclinedsubstantiallyinrecentyears–andwhilePREPA’sgenerationspendingwassubstantiallyoverbudgetfrom2010to2013,thesystemwasmaintainedmorereliably.

Figure 3. PREPA capital budget and spending for generation, historic and rate case, by fiscal year (millions).

2. Forcedoutagesimposecostsonsystem

ThestateofPREPA’ssystemexactsasteepcostonthesystemintheshortterm,andoverthelongrun.

Intheshortterm,PREPAisforcedtouse“lessefficientdieselbackupunits,”and“require[es]increasedspinningreservelevels[atviablegenerators]tominimizepowerdisruptionstocustomers.”47 A review of PREPA’s generation patterns reveals that PREPA has, in fact,

44CEPR‐JF‐01‐16Attach01.Page6.

45Id.Page7.




increaseditsrelianceonthedieselgenerationfleetinFY2015andFY2016,doublingitsuseofdistributedgenerationturbinesandtriplingitsuseofthedieselAguirreCCplant.48

Overthelongrun,PREPA’sforcedoutagesmeanthatitiscompelledtoexpendlimitedcapitalintherestorationofitssystem,ratherthaninfundamentalimprovements.Aswewillshowinthefollowingsections,PREPA’scapitalbudgetsareorientedtowardstherestorationofexistinghardware—generation,transmissionanddistribution—leavinglittlespaceforthefundamentalsystemimprovementsthatPREPAneedstomovebeyonditscurrenthighcostfossilfleet.

WeconcludethatPREPA’sfocusintheneartermneedstobeontherestorationofthesystemthroughstrategicinvestmentsingeneration,transmission,distributionandhumancapital.

B. LowreliabilityforPREPAcustomers

PREPA’scustomersexperiencealevelofinterruptionabovethatofmostotherU.S.utilitiesbyafactoroffourormore.PREPA’ssystemischaracterizedbyfrequentinterruptions,manyofwhicharesustained.Someoftheseinterruptions—liketheSeptember21,2016outage—havetheiroriginsinlarge‐scaletransmissionandgenerationfaultsthatcascadethroughthePREPA system. Other outages occur due to the deterioration of the distribution system.PREPAstatesthat“thephysicalstructural/mechanicalandelectricaldeteriorationofcriticalcomponentsinthetransmissionandsub‐transmissionlineshavedirectlycausedsignificantgridoutages and service interruptions…Physical access to repair or replace structures isverydifficult,whichconsequentlymeansthatinadditiontothelongserviceinterruptions,the system is exposed to the potential risk of multiple contingencies and cascadingoutages.”49

IndiscussionswithPREPA,thestateofdisrepairbecomemoreapparent.Accordingtothedirectoroftransmissionanddistribution,reductionsinspendinghaveimpactedthesystemdramatically:

Thefactthatwehavehadcapitalconstraintsandhaven’tbeenabletoreplaceandconstructmanyofourtransmissionlinesisaproblemhere.Thefactisthatwe’refacingahighlevelofdeteriorationofthesystem.Justtogiveyouanidea,inthefirst110daysofthisfiscalyear,wehavefacedthirty‐eightmajortransmissionlineoutages.AndwhenIsaythirty‐eight,wemeanthatinallofthoseoutageswehadahighriskofconductorsonthefloor.That’sthetypeofsituationthatwe’refacingrightnow.50

48CEPR‐AH‐03‐07Attach01.

49CEPR‐AH‐02‐01(d).

50October20ConferenceCallat2:50.


PREPAcharacterizesthecausesofinterruptionsasfollows:51

a) TreeTrimmingConditions–about35‐45%

b) WeatherDeterioration–about15‐25%(thisincludesheavyrain,normalrain,strongwinds,etc.)

c) Structural/Mechanicalconditionsofpolesandcables–10–25%

d) NoCauseReported–15–20%(thisismainlyduetoundetectedconditionsatthemomentofthefailure)

e) Othercauses–10‐20%

PREPA’sdistribution system is spread over 16,000 miles of primary lines spread along the island.52 Understandably, a significant level of staffing is required to maintain thisinfrastructure.PREPAhasincurredasubstantial22percentworkforcereductionsince2014,with a notable impact on reliability. PREPA states that in reducing the workforce,constructioncrewshavereorientedfrompreventativetoreactivemaintenance.53Differentlystated, the utility describes itself as always playing a catch‐up game on maintenance—followingoutages,insteadofimprovingthefundamentalsystem.PREPAfurtherstatesthatthe shortage of fundsmakes it unable to execute a well-planned preventative maintenance program.54

Utilitiestypicallymeasurethesystemreliabilitybasedonstandardreliabilitymetrics.ThemostcommonlyreportedreliabilitymetricsareSAIFI,SAIDI,andCAIDI.

SAIFI, the System Average Interruption Frequency Index can be thought of as theprobability that an average customer will have an outage (of any duration) during areportingperiod.Itiscalculatedasthefractionoftotalcustomersinterruptedinaperiodover the number of customers served in that period.55 Ameasured annual SAIFI of 0.35meansthatonaverage,aboutonethirdofcustomersexperiencedaninterruptionoverthecourseofayear.

SAIDI,theSystemAverageInterruptionDurationIndexcanbeconsideredtheaveragenumber of outage minutes any given customer can be expected to experience over a

51CEPR‐AH‐02‐01(g).

52CEPR‐AH‐02‐01(e).

53CEPR‐RS‐03‐03.

54Id.

55SAIFI =Σsustainedcustomers interrupted/Σsystemcustomersserved.Note: countsinterruptionsof5minutesorlonger.


reportingperiod.Itiscalculatedasthenumberoftotalcustomerminutesinterruptedinaperioddividedbythenumberofcustomersserved.56AmeasuredannualSAIDIof50wouldmeanthattheaveragecustomerexperienced50minutesofinterruptioneveryyear.

CAIDI, the Customer Average Interruption Duration Index is the actual averageinterruptiontimeexperiencedbycustomerswhoexperienceanoutage.It iscalculatedasSAIDIdividedbySAIFI.IfthemeasuredSAIFIis50minutesperyear,andthemeasuredSAIDIis0.35peryear, thatmeans that customerswhodidexperienceoutageswereoffline fornearly2.5hours(50/0.35=142minutes).

In theURS2013Report,PREPAstated that its internalSAIDI targetwas1.8hoursor84minutes.57From2010to2013,PREPAwasabletokeepSAIDIbelowthattarget.However,thatvalueexcludedoutagesthatwerelessthanfifteenminutesindurationaswellasmajorevents such as the effects of tropical storms/hurricanes and disruptions from multiplecontingencies.58In2014,PREPAmodifieditscalculationstomeasurereliabilitytobemoreconsistentwithindustrytrends.59Overthattimeperiod,reliabilityhasalsodegraded.

1. ReliabilityTrends

PREPAprovidedreliabilitymetrics(SAIDI,SAIFI,andCAIDI)forthelastfiveyearsreportedbymonth.60PREPAstated its current reliabilitygoalsasSAIFI at0.328 interruptionsperconnectedcustomer,SAIDIat48minutesperconnectedcustomer,andCAIDIat46minutesperaffectedcustomer.61EvaluatingPREPA’ssummarydataaswellasdetailedrecordsofinterruptionsprovidedbyPREPA,weobservethatPREPAhasoftenexceededeventheseliberal goals.We compare PREPA’s reliability goals and its actual metrics in the figuresbelow.Asanote,thechartsbelowdonotshowtheimpactoftheSeptember21stAguirre2.5‐dayoutage.ItisnotclearwhyPREPAexcludedthiseventorhowitwastakenintoaccount,ifatall.

56 SAIDI = Σ sustained customerminutes interrupted/ Σ system customers served.Note:someutilitiesprovidemoregranularSAIDIvaluesatdistrictandevenfeederlevel.

57URS2013Report,pages58‐59.

58 URS June 2013 Annual Report. PREPA Exhibit 3.02(D) Consulting Engineers Report. FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Page58.

59TheInstituteofElectricalandElectronicsEngineers(IEEE)publishedastandardmethodologytocalculatereliabilitymetricstoignoremajorevents(i.e.storms).

60CEPR‐MC‐01‐011Attach01.

61CEPR‐MC‐01‐011(b)(i)‐(iii).


PREPA’smonthlySAIFItargetis0.33,.62oranexpectationthatonaverage,acustomerwillhaveanoutageeverythreemonths—orfourtimesperyear.AsshowninFigure4,PREPA’srecordedsystemSAIFIpeaks inSeptember—hurricaneseason—and is lowestduring thewinter.However,indataforthiscalendaryear(MaythroughSeptember),weobservethatSAIFIhasstayedathistorichighlevelsof0.45,orfiveandahalfoutageseveryyear.

Figure 4. PREPA monthly SAIFI.

PREPA’stargetforSAIFIis48minutespermonth,ortenhoursofoutageperyear.Inrecentmonths,PREPA’sactualhasbeencloserto16hoursperyear.Inotherwords,theaveragePREPAcustomercanplantospendnearlyafulldaywithoutpowereachyear.

62 PREPA did not state that the data provided here was by monthly reporting period, but anindependentassessmentofinterruptiondataprovidedinCEPR‐AH‐02‐07Attach02confirmsthatthesenumbersaremonthly.Infact,areviewofactualinterruptiondatasuggeststhatinterruptionsarefarworsethanportrayedbythestatisticshere.Onaverage,PREPAcustomersexperienceatleasta fiveminute outage everymonth, or a SAIFI of 11.61 per year – twelve times the average U.S.customer.


Figure 5. PREPA monthly SAIDI.

Outagedurationhasbeenincreasing,asshowninFigure6,below.Whenoutagesdooccur,theaverageoutageisovertwohourslong.ThemonthlytrendforCAIDIappearstobemovingupwardssinceJanuary2013.Thisindicatesthatwhencustomersexperienceoutages,thoseoutageslastlongerthaninthepast.

Figure 6. PREPA monthly CAIDI.

The data provided by PREPA indicates that PREPA’s targets for distribution systemreliabilityareextraordinarilyhigh(i.e.,permissive),andsinceJanuary2013,theCompanyhasexceededitsownmonthlyreliabilitytargetsin64percentofthemonthsforSAIDI,51percentforSAIFI,and68percentofthetimeforCAIDI.


An independent assessmentof outagedataprovidedbyPREPA, entailingnearly400,000outage events,63 suggests that PREPA’s distribution reliability is actually far worse thanshownhere.AreviewofthatdataindicatesthataveragePREPAcustomersexperience11.5outageseveryyear–around20hours’worth.

The outage levels experienced by PREPA customers according to these statistics areextraordinaryandarefarinexcessofsystem‐widevaluesforotherutilities.AreportfromLawrenceBerkeleyNational Laboratory (LBNL) findsnationalannual SAIDImeasures ofabout200minutesperyearin2012—includingstormevents.64PREPA’starget‐‐excludingstormevents—is600minutesperyear,abovethe75thpercentileofutilitiesmeasuredbyLBNL.Removingstormevents,utilitiesintheU.S.averagebetween75to200SAIDIminutesperyear.

Similarly,thetargetfrequencyofserviceinterruptionsexperiencedbyPREPAcustomersisabout three times that experienced by other U.S. customers. The number of serviceinterruptionsexperiencedbyPREPAcustomersinthelastfewmonthsof2016havebeenfourtofivetimeshigherthantheaverageU.S.customer.

The extremely high outage rate for PREPA is indicative of a system with substantialproblems,andconfirmsPREPA’sstaff’scontentionthatthesystemisseverelydegraded.

Changestoimprovethereliabilityofthedistributionsystemwillnotoccurovernightandwillnotbesolvedbysimplyincreasingbudgets.AnintegralprocesstoimprovethevisibilityofthedistributionsystemwillhelpPREPA’splannerstoidentifyandprioritizeareasonthePREPAsystem.

C. PREPA’sMATScomplianceconundrum

OnFebruary16,2012,theU.S.EnvironmentalProtectionAgency(EPA)issuedtheNationalEmission Standards for Hazardous Air Pollutants (called the Mercury and Air ToxicsStandard,orMATS).Thisruleistargetedatrestrictingmercuryemissionsfromsolidfuel‐andfueloil‐firedpowerplants.Therulerequireduncontrolledorinadequatelycontrolledcoal‐andoil‐firedpowerplantstoinstallandoperate(orupgradeexisting)environmentalcontrols byApril 2015, or as late asApril 2017with an extension and special reliability

63CEPR‐AH‐02‐07Attach02through07.

64Larsen,P.H.,K.H.LaCommare,J.H.Eto,andJ.L.Sweeney.AssessingChangesintheReliabilityoftheU.S. Electric Power System. Lawrence Berkeley National Laboratory. August 2015. Page 9.https://emp.lbl.gov/sites/all/files/lbnl‐188741_0.pdf. Provided as Exhibit Fisher and Horowitz,Exhibit03.


exemption from EPA.65 For most U.S. coal or oil generators, MATS compliance could beachieved through three primary avenues: install controls (primarily sulfur dioxidescrubbers,sorbentinjectionandparticulatecontrols),firewithnaturalgas(whichisexemptfromtherule),orretire.

PREPA’sfleetisinauniqueposition.FourteenofPREPA’sunits,comprisingabout2,900MWofoil‐firedcapacity,oroverhalfofPREPA’snameplatecapacity,aresubjecttoMATS.Formanyofitsunits,PREPAisrelyingontwolessoftenusedprovisionsoftherulethatprovideexemptions.

Limitedusedesignation:Unitsthatoperatebelowaneightpercentcapacityfactormaybedesignatedas“limiteduse.”EPAexemptsunitsthataredesignatedaslimiteduse,primarilyonthejustificationthattheseunitsareeffectivelybackuporpeakinggeneratorsanddonotcontributesubstantialpollution.IntheIRP,PREPAhasindicatedthatitplanstodesignateatotalofeightsteamunits“limiteduse”asitscompliancemechanism,startinginApril2015.Ms.Miranda’spaneldiscussesthatoverthelastyear,PREPAhasbeenunabletomeetthelimitedusedefinitionforthoseunitsalreadydesignatedassuch,66primarilyasafunctionofotherlargeunitoutages.

Exception forcombustion turbines:EPAdoesnot consider simple‐cycleandcombinedcycle(“CC”)stationarycombustionturbines(CTs)applicableunderthedefinitionof“electricgeneratingunit” for thepurposesof theMATSrule.67Thismeans thatPREPA’scombinedcycle units—both existing and potential new units—are not subject to MATS. PREPAmaintainssomeofthefewcombinedcycleunitsintheU.S.activelyfiredonliquidfuel,hencetherarityofuseofthisparticularprovision.

Tohandle the remainder of its fleet’s compliance, PREPAhad effectively three options –install stack controls to limit emissions, retire the units outright, or find away to bringnaturalgastothoseunits.

Earlyin2011,PREPAdeterminedthatitwouldmeetMATScomplianceprimarilybybuildinganoffshoregasportnearSalinastofeedtheAguirresteamandcombinedcycleunits.Indoingso,itwouldconvertitslargestplanttogas,andthenexpanditsfacilityatAguirre,therebyallowingmuchoftherestofitsfleettoretire.Theplanreliedonanassumptionthat(a)thegasportwouldbebuiltinatimelyfashion,(b)PREPAcouldkeepits“limiteduse”unitsto

65However,wehavefoundnoevidencethatPREPArequestedorwasgrantedsuchanextraordinarylevelofextension.WethereforeassumePREPA’sMATScompliancedeadlinetohavebeenApril2016,allowingforacommonly‐grantedone‐yearextension.


6777FR9309.


belowaneightpercentcapacityfactor,and(c)EPAwouldallowPREPAtomaintainsomenon‐MATScompliantunitsuntiltheAguirresitecouldbeexpanded.

Overthecourseofthenextthreeyears,PREPAdevelopedplansfortheoffshoregasportwithExcelerate Energy, a company that operates offshore re‐gasification platforms (i.e., fromliquefiednaturalgastovaporgas),targetingagasportinserviceby2014—atimelinethathasnowslippedto2018.Thegasport,nowcalledtheAguirreOffshoreGasport(“AOGP”)becameamajorinitiativeoftheutility,andaccountsforasubstatntialportionofPREPA’scapitalrequirementsoverthenextfewyears.

InitsavidpursuitoftheAOGPproject,PREPAlostsightofitsothercomplianceoptions.Asthe AOGP project became increasingly delayed, PREPA’s options for timely complianceevaporateduntilinMarchof2016PREPAofficiallypassedanyformofextensiongrantedtoit,implicitlyorexplicitly.Today,PREPAisoutofcompliancewithMATSatmanyofthesameunits forwhich itwas out of compliance in 2011.With current timelines, compliance atAguirrecan’tbeachieveduntillate2018—ifnotlater.PREPA’shopestobringnewgasfiredgenerationonlineafter thegasport isbuiltarealsonowdelayedandthusoverallsystemcomplianceincreasinglylookslikeitmaytakeuntil2022orlaterwithPREPA’scurrentplanofaction.

Aswewilldiscussinthisdocument,PREPA’sMATScompliancefeaturesprominentlyintheutility’s strategy of investments, from generation to transmission, and is belied by theutility’s inability to keep the Aguirre units—key to the AOGP project—online andoperational.

ArePREPA’sinvestmentsinthisratecaseconsistentwithitsMATSstrategy?Yes—butsincePREPAreliessoheavilyonlargecentralstationgenerators,andthosegeneratorsarefailing,PREPAhasneeded,andmaycontinuetoneed,topropupthoseagingsteamgeneratorsinordertopursuethisstrategy,ratherthanmovingtowardsacleaner,moreefficientfleet.

WasPREPA’smovetopursueAOGPatthecostofallotheroptionsprudent?Probablynot,althoughaneconomicanalysisofAOGPisstillpendingbeforethisCommission.

InpursuingAOGP,PREPAfailedtocompleteapermitted, licensed,andhalf‐builtpipelinefromtheEcoEléctricafacilitytoAguirre.Ms.Mirandastatesthatonedecisionwithwhichshedisagreeswithis“thecancellationofthesouthgaspipelinein2009.”Shestatesthat“basedonthepreviousstudiesthatwe(PREPA)didtojustifythisproject,thecancellationofthesouthgaspipelinein2009wasnotoptimal.”68

In pursuing AOGP, PREPA also narrowed its options for improving and expanding thecurrentMATScompliantfleet,andseekingsubstantialrenewableenergy.Instead,byseeking

68CEPR‐SGH‐001‐016(a)‐Supplemental.


topropupAguirre,PREPAcreatedafleetthat,byitsownmeasure,cannoteffectivelytakeonrenewableenergysimplybecauseitsexistinggeneratorsramptooslowly.

OverthelastfouryearsPREPAhasincreasinglycornereditselfintoaneedforAOGP.Havingshedmanyofitsotheroptions,PREPAnowfindsitselfhavingplacedabetonoffshoregas—whichmaybea“toobigtofail”proposition.

MuchofouranalysiswillgobacktoMATScomplianceandhowPREPA’sinvestmentseithersupport,ordonotsupport,PREPA’sneedtomeetthisimportantregulation,aswellastheconsequences of PREPA’s single‐minded pursuit of its strategy for doing so, even as itssystemdeterioratedaroundit.


IV. LOADFORECAST

A. Summaryofissueandfindings

Inthischapter,weaddressthesalesforecastusedbyPREPAinitsratecasefiling.PREPA’ssalesforecastisoneofthemostimportantelementsinthecalculationofPREPA’sproposedrates.TheCommissionmustdecideiftheforecastisreasonable.IftheCommissiondecidesthatPREPA’s sales forecast is reasonable, no changewill be required. If theCommissiondecidesthatPREPA’ssalesforecastisnotreasonable,itmustthendecidewhetherornottorequirethatPREPAreformulateitsrateproposalusingadifferentsalesforecast.

Below, we discuss the significance of sales forecasts to both short‐term and long‐termplanning,describehowPREPAformulatedthesalesforecastusedinitsratecasefiling,andcritiqueboththemethodologyandtheresults.OurconclusionisthatPREPA’ssalesforecastmethodology is poorly conceived and supported, and that the sales forecast wasaccompaniedbyinsufficientdocumentation.Webelievethatthecurrentforecastislikelynotreliable for FYs 2018 and 2019 and recommend that the Commission require PREPA toimprove its forecastingmethodology and prepare a new sales forecast prior to the nextmajorcaseitbringsbeforethisCommission(forexample,anIRPUpdateorthenextratecase).WeobservethatPREPA’sforecastsforoneyearout,whilenotgenerallyreliable,arenot systematicallybiased.Anassessmentof actualFY2017 salesbymonth indicates thatPREPA’sforecastforFY2017islikelyreasonablyclosetoreality.Thereforewerecommendnochangetothesalesforecastatthistime.

B. Theneedforandsignificanceofthesalesforecast

1. PREPArequiresanaccurateandreasonablesalesforecast

Thepurposeofelectricutilitiesistosupplyelectricityaccordingtothedemandofcustomers,atjustandreasonablerates.Inordertooperateasafe,reliable,andfinanciallysustainablesystem,utilitiesmusthaveareasonableexpectationofhowmuchelectricitytheywillsellinthecomingyearoryears.Utilitiesengageinloadforecastingtopredicthowmuchelectricitytheyexpecttoselloverboththeshort‐andlong‐terms.

Loadforecastsgenerallyhaveseveralconstituentelements.Themainelementisaforecastof total demand for or sales of electricity (in kWh or some multiple thereof69) and aprediction of “peak load” or “peak demand” over time, often inMW.70 Utilitiesmay alsoforecasttotalgenerationrequirements(alsoinkWh)asafunctionoftheanticipatedlevelofdemand and transmission‐ and distribution‐related losses, as well as other system

69kWh:kilowatt‐hour,aunitofenergy.Dependingontheutility,salesforecastscanalsobeprovidedinmegawatt‐hours(MWh;1MWh=1000kWh)orgigawatt‐hours(GWh;1GWh=1000MWh).

70MW:megawatt,aunitofpower.Conceptually,peakdemandrefers to themaximumamountofpowerthatautilitymustsupplyatanyonemomentofayear.


inefficiencies.Whilethesalesforecastexpresseshowmuchtotalenergytheutilityexpectstoselloverthecourseofayear,thepeakloadforecastindicatesthemaximumpowertheutilityexpectstoneedtogenerateandtransmitatanyonetimeduringthatyear.

Ultimately,allelectricutilitieshaveneedofaccurateloadforecasts.ThePuertoRicoElectricPower Authority (PREPA) is no exception. In systems with traditional volumetric rates,utilities’revenuesarelargelydependentonsales,asisthecaseforPREPA.Salesforecasts,andaccompanyingpeakloadforecasts,alsodictatemuchofutilities’budgetsasdiscussedfurther below. As such, preparation of an accurate and reasonable sales forecast is thecornerstoneoffinancialplanningforelectricutilities.

2. PREPA’s sales forecast impacts many aspects of its expectedspendingandrevenue

Asmentionedabove,thesalesforecasthasimportantimplicationsforbothautility’sincomeanditsexpenditures.Intermsofincome,revenueistied–atleastpartially‐tosalesformostutilities.ForPREPA,thesalesforecastisusedasthedenominatorinthecalculationofvariouscomponents of PREPA’s rates. When a utility collects revenues primarily through avolumetric (per‐kWh) charge, having an accurate sales forecast is necessary to enablecollectionofsufficientrevenuetocoverbothfixedandvariableexpenses.Suchexpensesmayincludefuel,whichinPREPA’scaseiscompensatedthroughafuelcostadjuster,andload‐linkedservicebenefitchargessuchasforadministrationofenergyefficiencyprograms.Ratedesign and the implications thereof are discussed inmore depth in the expert report ofCommissionadvisorPaulChernick.

Utility budgets also depend on accurate sales forecasts. Utilities’ expectations of totalspendingonfuel,purchasedpower,operationsandmaintenance,andload‐linkedprogramsmentionedabovearealldependingontheirexpectationsofsales.Increasedsalesgenerallyleadtoincreasedspendinginsomeorallofthesecategories.

ForPREPA,salesareusedasaninputintoPROMOD,71aproductioncostmodel.72Briefly,PREPAcalculates thetotalamountofenergy itneedstogenerateeveryyear,consideringbothitssalesforecastandfactorssuchaslineandnon‐technicallosses.73PREPAtheninputsits total generation requirement into PROMOD along with other data such as cost and

71CEPR‐AP‐2015‐0001,October20,2016ConferenceCall.

72Moredetailregardingthestructure,capabilities,anduseofproductioncostmodelscanbefoundin section3of the following report: Fisher, J., Santen,N., Luckow,P.,DeSisternes, F., Levin,T.&Botterud,A.,AGuidetoCleanPowerPlanModelingTools:AnalyticalApproachesforStatePlanCO2Performance Projections; Synapse Energy Economics and Argonne National Laboratory (2016),http://www.synapse‐energy.com/sites/default/files/Guide‐to‐Clean‐Power‐Plan‐Modeling‐Tools.pdf.

73BaseIRP,VolumeIII,Section1.2.1.3.


performance parameters for its generation fleet. PROMOD then determines an optimaldispatchpatternforPREPA’sgenerationfleetoverthecourseofayear,andoutputsthecosts(aswell asexpected fuel consumption,emissions,andother systembehavior)associatedwiththatdispatchplan.Adifferentsalesforecastwillthereforeleadtoadifferentdispatchpatternwithdifferentassociatedcosts.

AsdiscussedfurtherinFUELANDPURCHASEDPOWERBUDGETS(ChapterVI,Page155)below,PREPA’sfuelandpurchasedpowerbudgetsaredirectoutputsofthePROMODrunusedtoinformtheratecasefiling.74Assuch,evensmallchangesinsalesforecastscanhavesignificantinfluenceonPREPA’sfuelandPowerPurchaseandOperatingAgreement(PPOA)budgets.WhenweprobedastowhythefuelandpurchasedpowerbudgetsinthisratecaseweredifferentthanthoseintherecentlyclosedIRP,PREPArespondedthatthetwocasesuseddifferentsalesforecasts.Forexample, forFY2017thesalesforecastusedintheratecaseishigherthantheIRPbyapproximatelythreepercent.75Thepurchasedpowerbudgetin the rate case is higher by almost five percent and the fuel budget is higher by nearlyfourteenpercent.76Thesedeviationsindicatetheimportanceofanaccuratesalesforecast–evensmallchanges insalescan lead toalterations inbudgetsamounting to tensorevenhundredsofmillionsofdollars.

Peak(MW)andtotalsales(kWh)forecastscanimpactutilities’planningovertheshort‐andlong‐terms.Peakloadforecastsarethekeyinputtodeterminationsofresourceadequacyandareanimportantpartofcostallocation(shortterm)andresourceplanning(longterm).Localor system‐wide increases inpeakdemandcanpromptsystemexpansion, includingboth new generation or new transmission. In jurisdictions with renewable portfoliostandards that are tied to sales, including Puerto Rico,77 sales indicate the amount ofrenewableenergythatmustbeaddedtothesystemtoachievecompliance.

74Bycontrast,PREPA’ssalesforecastdoesnotappeartoimpactitsgeneration‐relatedoperationsandmaintenancebudget,asdiscussedinOPERATINGEXPENSEBUDGET,ChapterVII(Page182).

75IRPandBusinessPlansalesforecast:16,693GWhtotalsalesinFY2017.Ratecaseforecast:17,268GWhtotalsalesinFY2017.PREPA’sPetitionforRateReview,Exhibit3.02,slide2.PREPAconfirmedthatitusedthesamesalesforecastintheIRPandintheBusinessPlaninitsresponsetoCEPR‐AH‐01‐05(a).

76Beforeperformanceimprovement‐relatedsavings.IRPfuel($671,726,000)andpurchasedpower($662,018,000)costsfromAppendixC‐10(P3F1withupdatedfuels)oftheUpdatedFuelIRP.Ratecasepre‐performancesavingsfuel($763,695,078)andpurchasedpower($692,572,739)costsfromScheduleA‐6REV.Author’ssumoftotalfuelandpurchasedpowerexpendituresinPREPA’sresponseto CEPR‐JF‐01‐22Attach 02.xlsx (of the Commission’s Sixth ROI, August 23, 2016) confirms thatperformanceimprovement‐relatedsavingswerenottakenintoaccountinPROMOD,asthevaluesmatchthepre‐savingsvaluesinScheduleA‐6REV.

77Act82‐2010,§2.3(b).


With respect to generation and the transmission networkwewould expect a significantrelationship between capital expenditures and sales. The Commission examined therelationship between sales expectations and PREPA’s need for additional generation atlengthintheIRPproceeding.PREPAfoundthatwhendemandwasreduced,itrequiredlessreplacementgenerationintheNorththanpreviouslythought.78PREPAhasalso indicatedthatsomeelementsofitstransmission‐relatedcapitalspendingplanmaybeavoidablegivenlowersalesorload.

WeaskedPREPAwhatcapitalprojectscouldbeavoidedifloadwasreducedinthesystem.PREPAprovidedgeneralcategoriesofprojects,butnotspecificdollarvalues.79AssociatingPREPA’sdescriptionswithprojects,weestimatethatPREPAcouldavoidonlyaverysmallamountofcapitalintheplannedtransmissionanddistributionsystemcapitalspendinginFY2017.80Aswediscussinthecapitalsection,thevastmajorityofPREPA’scapitalspendingfortransmissionanddistribution(thetwoimpactedcategories)inFY2017arerehabilitationanddeferredmaintenanceprogramsandthusunlikelytobeavoidablebyloadreductionsinthenearormid‐term.

TheimportanceofPREPA’ssalesforecastfornearlyeveryelementofitsbudgetingisclear.This demonstrates both the importance of an accurate sales forecast and the difficultyinherentinchangingforecastswithoutsufficientreason.AdjustingPREPA’ssalesforecastwould require PREPA to run new model runs and potentially to re‐examine its capitalspendingplanratherthanprovidingasimpleproformaadjustment.

3. StateofPREPA’ssales,load,andcustomerbase

Inthissection,weprovideabriefdescriptionofsalesandloadonPREPA’ssystemtoprovidecontexttothevaluesshownbelow.

PREPA oversees a system of nearly 1.5 million customers, the majority of which areresidentialcustomers(1.3million).81PREPAhasjustover120,000commercialcustomers,several hundred industrial customers, and approximately 2,000 agricultural customers.TotalsalesinFY2016wereapproximately17,439GWh.82Commercialcustomersmakeup

78Notethattheseprojectsareoutsideofthetimehorizonconsideredbythisratecase.UpdatedFuelIRP,p2‐1.

79PREPA’sresponsestoCEPR‐PC‐09‐34(c),(e),and(f)oftheCommission’sTwelfthROI(5October,2016).

80 PREPA’s Petition for Rate Review, Schedule F‐3 REV. Sum of FY2017 values for projectswithPlanAreaheading“Transmission”or“Distribution.”

81PREPA’sresponsetoCEPR‐PC‐09‐42oftheCommission’sTwelfthROI(October20,2016).RefertoCEPR‐PC‐09‐42Attach01.xlsx.

82 PREPA’s response toCEPR‐AH‐06‐14of theCommission’s FourteenthROI (October21, 2016).RefertoCEPR‐AH‐06‐14Attach01.xlsx.


nearlyhalfofPREPA’ssales,at8,187GWhinFY2016.83Residentialcustomers,at6,439GWh,accountforslightlymorethanathird.84WeshowPREPA’ssalesovertimebyclassinFigure785below.

Figure 7. PREPA's sales by customer class over time.86

PREPA’ssaleshavedeclinedoverthepastseveralyears,fromnearly20,000GWh/yearinFY2010totheapproximately17,000GWh/yearmentionedabove.Decreasesinsaleshavecomenearlyevenlyfromtheresidential,commercial,andindustrialclasses.87Peakloadhasalso declined by almost 400MW, from 3,404MW in FY2010 to 3,030MW in FY2015.88Considering this trend in light of PREPA’s financial situation only underscores theimportanceofsoundloadforecasts.PREPAmustknowwhetherornotitssaleswilldeclinefurtherand,ifso,byhowmuch.Ifitforecastsdemandthatisunrealisticallyhigh,PREPAisatriskofoverbuildingitssystemandunder‐collectingtherevenuesneededtofundcapitalprojectsandday‐to‐dayoperations.Meanwhile,ifitforecastsdemandthatisunrealistically

83Author’scalculation,basedondataprovidedinCEPR‐AH‐06‐14Attach01.xlsx.

84Id.

85Inthisfigure,the“Misc.”categoryincludesthesalestopubliclighting,agriculture,and“Others”asreportedbyPREPAinCEPR‐AH‐06‐14Attach01.xslx.

86CEPR‐AH‐06‐14Attach01.xslx.

87Id.

88Id.


low,PREPA(likeallutilities)isatriskofhavinginadequateresourceswithwhichtoservetheneedsofitscustomers.

C. PREPA’ssalesforecastisill‐conceivedandill‐supported

1. PREPAdidnotprovideadequatesupportforitssalesforecast

PREPAprovidedeffectivelynosupportforitssalesforecastinitsinitialratecasefiling.TheforecastisdiscussedonlybrieflybythepanelofMiranda,Sales,andSosa:

TheloadforecastwasupdatedonApril2016.TherevisedforecastconsideredhistoricaldatauntilMarch2016andeconomicindicatorsupdatedinOctober2015.ItalsoincludedanestimateofthecostofkWhconsideringthesecuritizationcostandthemostrecentlyupdatedfuelpricefromDOE.Thisupdatedforecastwasusedintherevenuerequirementandthereforeforthisratecase.89

PREPAalsodiscusseditsforecastingmethodologyinabbreviatedandunclearterms:

PREPAestimatedtheenergyconsumptionandgenerationforthecurrentFY2015withextrapolationswhichconsiderthemonthlybehaviorofthesevariablesfromhistoricaldatafromFY1993throughFYDecember2014.WiththeestimatesforFY2015,weusedeconometricmodelstoforecasttheconsumptionandclientsbymainclass(residential,commercialandindustrial)forthenextfiveFYyears.ThemodelsincludethemainvariablesfromthePuertoRicoeconomy,thecostofkWhbycustomerclassandthebehaviorofthelastyearconsumptionforeachclass.Usually,theenergyconsumptionbykilowatthour(kWh)asadependentvariablehasacorrelationwiththemainvariablesofthelocaleconomy.ThevariablesofPuertoRicoeconomyaretheGrossDomesticProduct,theGrossNationalProductandtheDisposablePersonalIncome.90

Asdiscussedbelow,weconcludethatthisdescriptionismisleadinginanumberofways.NofurthermethodologicaldocumentationwasprovidedinPREPA’soriginalfiling.

Intermsoftheforecastitself,PREPA’sbusinessplanandfiledschedulesprovideonlytotalsystemsalesforFYs2017through2030.91GiventhatMiranda,Sales,andSosacitePREPA’s

89PREPA’sPetition forRateReview,Exhibit3,DirectTestimonyofSoniaMiranda,AntonioPerezSales,andVirgilioSosa,lines811‐815.

90PREPAEx.3.0,lines785‐795.

91PREPAEx.3.02,slide2;ScheduleF‐1REV.


econometricmodelsasprovidingforecastsfor“thenextfiveFYyears,”92thesalesforecastsforFYs2022through2030areentirelyunsupportedbyPREPA.However,whilethisissuemaybeof import forplanning andotherpurposes, it has little tono impact onPREPA’splannedspendingforFY2017,whichisthefocusofourreport.

NofurthersupportfortheloadforecastwasprovidedinPREPA’sfinancialmodelorotherfiledworkpapers,toourknowledge.PREPAelaboratedonitsloadforecastingmethodologyandprovided incompleteworkpapers in response to severalROIs from theCommission,includingCEPR‐AH‐01‐05through‐07andCEPR‐AH‐06‐14,aswellasduringtheOctober20ConferenceCall.

PREPA’sresponsestotheseROIswereunsatisfactoryinseveralways.PREPAprovidedanexpanded methodological description in response to CEPR‐AH‐01‐0593 This documentdescribesPREPA’soverallmethodologyinsimilartermstothoseincludedintestimonyandthenpresentsthe“models”usedtoforecastdemandbyclass.By“model,”PREPArefersonlytoinputsandoutputs;nofunctionalformisprovided.

For example, PREPA identifies the following as the “model” used to forecast residentialdemand:

LRkWh=ƒ(LPRM,LDPI,LLAGR)

ItdefinesLRkWhasthenaturallogarithmofresidentialconsumption,LPRMasthenaturallogarithmoftheper‐kWhcostofelectricityforresidentialconsumers,LDPIasthenaturallogarithmofdisposablepersonalincomeatrealprices,andLLAGRasthenaturallogarithmofthepreviousyear’stotalconsumptionbytheresidentialclass.94

This description is exceptionally uninformative, because it provides only inputswithoutdescribinganactualfunctionalformorprovidingcoefficients.Allthattheaboveequationindicates is that residential consumption is correlated in some way with the price ofelectricity,disposable income levels,andrecenthistory.The“model”providedbyPREPAdoesnotprovidethenatureofthecorrelation.Wewouldbeequallyjustified,basedonlyonthispoorlydefinedmodel, thatPREPAbelieves residential consumption increases as thepriceofelectricityincreases,oralternativelyconcludingthatconsumptiondecreaseswithrisingprices.Nordowehaveanyevidencetoevaluatetherelativestrengthofthemodelinputsontheforecastedresult.Wehavenogroundstoconcludewhetherpersonalincome,electricity costs, or simply the previous year’s consumption is the strongest predictor ofresidentialelectricityuse.Ultimately,weareunabletothereasonablenessofPREPA’smodel

92PREPAEx.3.0atlines790.

93PREPA’sresponsetoCEPR‐AH‐01‐05oftheCommission’sFourthROI(August9,2016).RefertoCEPR‐AH‐01‐05_Attach08.pdf

94Id.,p3.


asPREPAhasnot,infact,providedthemodelitself.ThesameistrueforPREPA’s“models”forthecommercialandindustrialclasses.

Theremainderofthemethodologicaldocumentisbothbriefandinternallyinconsistent.Forexample, the description repeats the claim that PREPA’s econometricmodels include asinputs“theelectricitypricebyrevenueclass.”95However,ourexaminationofthe“models”in question indicates that PREPA considers the cost of electricity only in its forecasts ofresidential96andindustrial97load.Grossdomesticproductandthepreviousyear’ssalesforthecommercialclassaretheonlyinputsusedtoforecastcommercialconsumption.98PREPAalsomentionsthatitforecastsloadforadditionalcustomerclasses,includingpubliclighting,agriculture, and unspecified “others.”99 The methodology PREPA uses to forecastconsumptionbytheseclassesisdescribedsimplyasan“extrapolationmethod”althoughthemethoditselfisentirelyunexplained(forexample,wedonotknowwhatdataPREPAisusingto extrapolate consumption from these classesorwhatperiodof time is coveredby thisdata). PREPA also states it assumed estimated consumption by these classes would beconstantfortheentireforecastperiod,withoutjustifyingthisassumption.

PREPA also providedworkpapers100 “related to the calculations performed for this salesforecast.”101WefindtheseworkpapersilluminatesomeaspectsofPREPA’smethodologybutultimately raise even more questions and concerns regarding PREPA forecastingmethodology.Theworkpapers reveal thatPREPAdoesnotappear toknowwhatmodelsrepresentitssystemwell.Insteadofusingasingleeconometricregressionmodelwithafixedsetofinputsforeachrateclass,weobservethatPREPAtriesmultipledifferentregressionsand thenchoosesona forecast‐by‐forecastbasis. Indeed,PREPAattempted fivedifferentregressions to model industrial consumption,102 ten different regressions to modelcommercial consumption,103 and no fewer than nineteen different regressions to model

95Id.,p1.

96Id.,p3.

97Id.,p4.

98Id.,p4.

99Id.,p5.

100RefertoCEPR‐AH‐01‐05_Attach10.xlsxthrough…_Attach14.xlsx.

101CEPR‐AH‐01‐05(d).Wenotethatproviding“related”workpapersisnotfullyresponsivetotherequest,whichaskedfor“functionalcopiesofanyworkpapers”usedinthecalculation,aswellasthesourceofallassumptions.PREPAfailedtoprovidejustificationorsourcesforanyofthevaluesthatitassumedandhardcodedintoitsworkpapers.

102CEPR‐AH‐01‐05_Attach11.xlsx;“IAU”tab.



consumptionfromtheresidentialclass.104Confusingly,someoftheregressionsappeartousethe same model or use different models but produce identical results.105 We have noexplanationforthisaberrance.

Thisobservation isechoedbyanotherwise‐crypticstatement inPREPA’smethodologicaldocumentationfortheforecastcreatedfortheIRP:“Someextrapolationmethodswereusedtodeterminethegenerationandconsumptionbyserviceclassforthefiscalyear2015.Weselectedtheextrapolationthatbestfittedtothefiscalyearbehaviorofthegeneration.”106PREPA’s methodology is, to our knowledge, unique. We know of no other utility thatattemptsthisformof“guessandcheck”loadforecastingasopposedtorelyingonasingle,well‐vetted model for each applicable cohort of consumers.107 There are three mainproblemswiththisstrategy,asenumeratedbelow.

First,theactualinputsandmodelsusedtoprepareforecastsamongwhichPREPAchoseareunclear.PREPAdidnotprovideaglossaryofthetermsusedinworkpapers,leavinguswithvery little ability to audit the basic theoretical grounding of the models themselves.Moreover, the termsused in theworkpapersdonotcorrespondto the termsused in themethodologicaldocumentation.Forexample,PREPA’smethodologicaldocumentdescribesresidentialloadasbeingafunctionof“PRM,”“DPI,”and“LAGR.”108TheindicatedmodelinPREPA’s workpapers, however, is a function of “PRES,” “YPD,” and “LAGRKWH.”109 Weconcludefromcontextthat“LAGR”isequivalentto“LAGRKWH,”“DPI”isequivalentto“YPD,”and “PRM” is equivalent to “PRES.” Several other variable names appear in PREPA’sworkpapers,andweareunabletodeducewhatPREPAmeansforthesenamestorepresent.


105Forexample,onlytenofthenineteenregressionspresentedfortheresidentialclassareunique(forecasts1through8,forecasts19,andtwoofthethreeforecaststhatarelabeled“model14”intheIAUtabofCEPR‐AH‐01‐05_Attach12.xlsx;theremaindersareduplicatesofmodels5or6).Forecasts7and8ofthetencommercialforecastsprovidedinCEPR‐AH‐01‐05_Attach10.xlsxareduplicatesofforecasts3and4.AlloftheindustrialclassforecastsprovidedinCEPR‐AH‐01‐05_Attach11.xlsxareunique.

106CEPR‐AH‐01‐05_Attach01.pdf,p2.

107See,forexample,themethodologicaldescriptionspertainingtoloadforecastingintheplanningdocuments of the Independent SystemOperator–NewEngland (https://www.iso‐ne.com/static‐assets/documents/2016/06/forecast_model_structures_2016.pdf),PacificGasandElectric(Section4.1 of the 2016 IRP, available online at https://www.portlandgeneral.com/‐/media/public/our‐company/energy‐strategy/documents/2016‐irp.pdf?la=en),andPacifiCorp(AppendixAofthe2015IRP, available online at:http://www.pacificorp.com/content/dam/pacificorp/doc/Energy_Sources/Integrated_Resource_Plan/2015IRP/PacifiCorp_2015IRP‐Vol1‐3_AllBooks_Redacted.pdf).

108CEPR‐AH‐01‐05_Attach08.pdf,p3.

109CEPR‐AH‐01‐05_Attach12.xlsx,highlightedforecastin“IAU”tab.


Second,PREPA’s strategy indicates that it, in reality,hasvery little senseofwhat factorsimpactitscustomers’consumptionpatterns.AverywiderangeofmodelsusedbyPREPAtoforecast residentialdemandare summarized inTable1.110This table tellsus thatPREPAconsideredat leasteighteendifferentmodelsusingvariouscombinationsof fivedifferentvariables (three of which, we believe, are the inputs identified in the methodologicaldocumentationfortheresidentialforecast).PREPAdoesnotprovideevidencethatanyoneofthesemodelsisactuallyaccurate,orreliable.

Table 1. Variables included in PREPA’s candidate regression models for forecasting of residential sales.111

VariablesConsidered*

Model PRES YPD PBI LAGRKWH CRES1 ln(PRES) ln(YPD)

ln(LAGRKWH)

2 PRES YPD

LAGRKWH3 ln(PRES) ln(PBI) ln(LAGRKWH)4 PRES PBI5 ln(YPD)6 YPD7 ln(PBI)8 PBI9 YPD LAGRKWH10 ln(YPD) ln(LAGRKWH)11 PBI LAGRKWH12 ln(PBI) ln(LAGRKWH)13 PRES YPD14** ln(PRES) ln(YPD)14** ln(PRES) ln(PBI) ln(LAGRKWH) ln(CRES)15 PRES PBI16 ln(PRES) ln(PBI)19 PRES PBI ln(CRES)

*Author’sinferencessuggestthesevariablenameshavethefollowingmeanings:”PRES”=costofelectricityperkWhforresidential customers; “YPD” = disposable personal income; “PBI” = GDP; “LAGRKWH” = previous years’ residentialconsumption.Wearenotawareofthemeaningofthevariable“CRES.”**Twoseparatemodels,andthreeforecasts,arealllabeledas“model14”

PREPAdoesnotappeartoknowwithcertaintywhetherornotresidentialconsumptioniscorrelatedwithelectricityprices,orwithpreviousyears’consumption,orwithanyoftheothervariablesconsidered.PREPAdoesnotknowwhetheritcanpredictconsumptionwithonly one variable, orwhether four different inputs are required. PREPA does not knowwhetherconsumptionvariesexponentiallyorlinearlywithanyoftheconsideredvariables.Instead, PREPA tries all of these possibilities, and then chooses one—without, it seems,developinganyconceptualframeworkforthefactorsthatimpactloadinPuertoRico.

110Modelsincludedinthistablearebasedonbothlinearandexponentialregressions;exponentialregressionspredictthenaturallogarithmofconsumptionusingthenaturallogarithm,denotedbyln(),ofinputvariables.

111CEPR‐AH‐01‐05_Attach12.xlsx


Finally, and most concerningly, the way in which PREPA actually selected a model torepresentfutureconsumptionforeachclassisinfactentirelyopaque.PREPA’sstatementquoted above implies to us that it chose the model for which FY 2016 results werenumericallyclosesttoactualFY2016consumption.However,ifthisiswhatPREPAintended,it is not what PREPA actually did. For the residential class, there are five models thatpredictedavalueforFY2016consumptionthatisclosertoactualFY2016consumptionthanthevaluepredictedbyPREPA’schosenmodel.112Fortheindustrialclass,onemodelprovidesacloserpredictionforFY2016thanthemodelchosenbyPREPA.113Forthecommercialclass,fiveoutoftenmodelsyieldamoreaccuratevalueforFY2016thanthemodelchosenbyPREPA.114

For each model, PREPA also provided an “R squared” value. The R squared of a modelnormallyexpressesthemodel’s“goodnessoffit,”orhowclosethemodeledvaluesaretotheactualdatainputintothemodel.AhigherRsquaredgenerallyindicatesabetterfit,withanRsquaredof1indicatingaperfectoverlapbetweenrealandmodeleddata.IdentifyingthemodelwiththehighestRsquaredvaluewouldbeoneconventionalwaytoselectaregressionmodel.However,themodelsthatPREPAchosetorelyoninconstructingitsforecastsdonotalwayshavethehighestRsquaredvaluesofallmodelsconsideredbyPREPA.115Ultimately,weareunabletorecreatePREPA’slogicforselectingoneforecastoveranotherandhavenogroundstoconcludethatsuchselectionswerenotsimplyarbitrary.

WeconcludethatPREPA’ssupport for its load forecasts isseverely lacking, includingthemethodological description, theoretical underpinnings, and provision of source data andassumptions pertaining to its forecasts. We recommend that the Commission requiresignificantlyimproveddocumentationinfutureproceedings.

2. PREPA’sforecastingis,ingeneral,notpredictive

112 CEPR‐AH‐01‐05_Attach 12.xlsx; author’s comparison of values for FY2016 in columns labeled“Real” and “Estimado según modelo.” PREPA chose Model 1, which over‐predicted FY2016consumptionby193GWh.Models7and8,bothofthetwoprovidediterationsofmodel14,andmodel19allyieldedmoreaccuratepredictionsofFY2016consumptionthanmodel1;model19predictedFY2016consumptiontowithin48GWhoftheactualvalue.

113 CEPR‐AH‐01‐05_Attach 11.xlsx; author’s comparison of values for FY2016 in columns labeled“Real” and “Estimado según modelo.” PREPA chose Model 1, which over‐predicted FY2016consumptionby45GWh.Model2under‐predictedFY2016consumptionbyapproximately4GWh.

114 CEPR‐AH‐01‐05_Attach 10.xlsx; author’s comparison of values for FY2016 in columns labeled“Real” and “Estimado según modelo.” PREPA chose Model 9, which over‐predicted FY2016consumptionby162GWh.Models1,2,3,7,and10yieldedmoreaccuratepredictionsofFY2016consumptionthanmodel9;model3over‐predictedFY2016consumptionbyapproximately19GWh.

115Forexample,theRsquaredvalueofresidentialmodel7is0.989ascomparedto0.985formodel1(themodelselectedbyPREPA).


Themultitudeofpoorly‐documentedforecastspresentedbyPREPAisevenmoreconcerningwhenconsideringthat,ingeneral,PREPAhasfailedtoaccuratelyforecastitssalesoverthelongterm.PREPA’ssalesforecastshaveconsistentlybeenaggressivecomparedtoitsactualsales,116 as shown in the figurebelow.However,wedonote thatPREPAappears tohavemoderated its expectations in recent years. PREPA’s vintage FY2010 sales forecast wasnotably aggressive compared to actual sales.117 In the period FY2011‐FY2013, PREPAforecasted that thedeclineof its saleswouldslow,and thenreverse itself, leading tonetgrowth in sales by the present day.118 Instead, PREPA’s sales cratered from FY2013 toFY2015.119PREPA’smore‐recentforecastshavereflectedthistrend,reflectingmoremodestexpectations.BoththeFY2015‐andFY2016‐vintageforecastsareessentiallyflatoverthenext several years.120 However, as discussed below, we have some concerns as to theaccuracyofeventhisrelativelyconservativeprospect.

Figure 8. PREPA's actual total system sales for FYs 2010 through 2016 and contemporaneous forecasts of total sales.121

116CEPR‐AH‐06‐14Attach01.xlsx

117Id.

118Id.

119Id.

120Id.

121Id.


In terms of the by‐classmodels considered for forecast used in this rate case,we againobservethatnoneoftheforecastsdoagoodjobofpredictingPREPA’sactualsales.PREPAappears to havemodeled FYs 2010 through 2016with each of the regressionmodels itconsidered,allowingustoeasilycomparethemodeloutputstotheactualsalesvaluesfromthoseyears.122Forbrevity,wewillnotdiscussthedetailsoftheforecastsforeachclassatlength.Wefindthat,ingeneral,noneoftheforecastsmatchactualvaluesconsistentlyovermultiple years. For each class, the forecasts tend to either under‐predict or over‐predictactual values, and uniformly fail to capture important features (such as the increase incommercialsalesfromFY2012toFY2013).123

The following figures124 showrealvaluesas compared to theoutputsofPREPA’svariousmodels.

Figure 9. PREPA's actual and modeled residential sales for FYs 2010 through 2016.125

122 CEPR‐AH‐01‐05_Attach 10.xlsx (commercial sales forecasts), CEPR‐AH‐01‐05_Attach 11.xlsx(industrialsalesforecasts),andCEPR‐AH‐01‐05_Attach12.xlsx(residentialsalesforecasts).

123CEPR‐AH‐01‐05_Attach10.xlsx.

124Blacksolidlinesdenoteactualsales;orangedashedlinesidentifytheunadjustedresultsoftheregressionmodelchosenbyPREPAtoforecastsalesoftheclassinquestion;andbluedottedlinesshowtheresultsoftheothercandidateregressionmodelsconsideredbyPREPA.ActualvalueswerenormalizedtoactualFY2010sales.

125CEPR‐AH‐01‐05_Attach12.xlsx


Figure 10. PREPA's actual and modeled commercial sales for FYs 2010 through 2016.126

Figure 11. PREPA's actual and modeled industrial sales for FYs 2010 through 2016.127

126CEPR‐AH‐01‐05_Attach10.xslx.

127CEPR‐AH‐01‐05_Attach11.xslx.


We observe easily that PREPA’s models do the best job of capturing industrial load.128Commercial load is often underpredicted while residential load is often overpredicted.However,forbothoftheseclasses,thereissignificantvariationintherelationshipbetweentherealandmodeleddata:someforecastsareentirelyabovetherealdata,someareentirelybelow,andsomemodeleddataisaboverealvaluesinsomeyearsandbelowinothers.Weobserve significant uncertainty in these results. In both the commercial and residentialforecasts,noneofthemodelsfullycapturenotablefeaturesintheactualdata.Theinaccuracyof the forecastmodels can be seenmore starkly in the following set of figures,129whichnormalizes all forecasts and actual data to their respective (real ormodeled) values forFY2010.ThisnormalizationallowsustoobservewhetherornotPREPA’smodelscapturethepatternofchangingsales,eveniftheywouldrequireadjustmentupwardsordownwardstobettermatchactualvalues.

Figure 12. PREPA's normalized actual and modeled residential sales for FYs 2010 through 2016.130

128 With the exception of a single model that consistently under‐predicts industrial load byapproximately800GWh.

129Asabove,blacksolidlinesdenoteactualsales;orangedashedlinesidentifytheunadjustedresultsoftheregressionmodelchosenbyPREPAtoforecastsalesoftheclassinquestion;andbluedottedlinesshowtheresultsoftheothercandidateregressionmodelsconsideredbyPREPA.ActualvalueswerenormalizedtoactualFY2010sales.ModeledvalueswerenormalizedtomodeledFY2010salesaspredictedbyeachregressionmodel.

130Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐01‐05_Attach12.xslx.


Figure 13. PREPA's normalized actual and modeled commercial sales for FYs 2010 through 2016.131

Figure 14. PREPA's normalized actual and modeled industrial sales for FYs 2010 through 2016.132

Asabove,themodelscapturechangesinindustrialloadacceptablywellbutdonotaccuratelypredictcommercialload.Theresidentialresultsareparticularlystark:notasingleforecastcameclosetocapturingtheactualextentofthedeclineinresidentialloadbetweenFYs2010




and2016.WeconcludefromthesefiguresthatnotonlydoPREPA’smodelsdoapoorjobofrecreatingactualhistoricaldata,theyalsofailtoaccuratelyrepresentthemagnitude(andsometimesthedirection)ofyear‐to‐yearchangesinsales.

ThisobservationraisesthequestionofhowPREPAactuallyarrivesatasalesforecast,giventhatallofitsvariousmodeledforecastsreturnresultsthatdeviatesignificantlyfromactualhistoricalvalues.FromexaminationofPREPA’sworkpapers,weobservethatitfollowsthisunusualprocedure:first,PREPAusesagivenmodeltoforecastsaleswithinagivenclassforseveral historical years and five future years. Second, PREPA calculates the year‐to‐yeargrowth rate implied by that forecast. Finally, PREPA applies that growth rate to actualhistoricaldata,regardlessofanydeviationsbetweenthemodeledandrealhistoricalvalues.Thefollowingtablesummarizesthistechnique.

Table 2. Historic actual and modeled residential sales and growth rates thereof.

FY

HistoricalData*(GWh)

ActualAnnualGrowthRate**

ModelOutputs†(GWh)

ModeledGrowthRate**

ModeledGrowthRateappliedtoHistoricalData†

(GWh)

PercentageDifference

betweenModeledandRealData**

2010 7,057 6,639 ‐5.91%2011 6,707 ‐4.95% 6,582 ‐0.86% ‐1.86%2012 6,560 ‐2.20% 6,452 ‐1.99% ‐1.64%2013 6,656 1.46% 6,482 0.47% ‐2.60%2014 6,271 ‐5.78% 6,407 ‐1.17% 2.17%2015 6,250 ‐0.34% 6,396 ‐0.16% 2.35%2016 6,355 1.68% 6,548 2.37% 3.04%2017 6,583 0.54% 6,3892018 6,626 0.65% 6,4302019 6,632 0.09% 6,4362020 6,681 0.74% 6,484

*CEPR‐AH‐01‐05_Attach12.xlsx,tab“IAU”,column“Real.”**Author’scalculation.†CEPR‐AH‐01‐05_Attach12.xlsx,tab“IAU”,column“Estimadosegúnmodelo”formodel1.

Thismethodologyisnotsound.Whileitisnotunusualforutilitiestoadjustloadforecaststomatch actual data from the recent past, PREPA’s method cannot be considered anormalizationorbenchmarking.Rather,PREPAeffectivelyconsidersthederivativeofsalesovertimetobetheactualoutputofitsforecastmodels.Astheabovefiguresdemonstrate,the year‐to‐year growth rates of the modeled data are not any more reliable than themodeleddataitself.Whenthemodeleddataitself is inaccurate, itcannotsupplyaccurategrowthrates.Simplyportingoverthesegrowthratestorealdatahasnotheoreticalbasisordefense.PREPAalsoappearstouseelementsofthistechniqueinitspreparationofapeakload forecast.133We recommend that the Commission require PREPA to cease using this

133 CEPR‐AH‐01‐05_Attach 14.xlsx. Rather than forecasting peak load directly, PREPA calculatesexpected non‐coincident peak load by customer class based on its by‐class sales forecasts and acustomerclass‐specificloadfactorvalue.Then,PREPAsumsallnon‐coincidentpeakloads,findstheyear‐to‐yeargrowthrate,andappliesthatgrowthratetothemostrecentyear’speakloaddata.This


unsupportablemethodology,infavorofdevelopingmodelsthathaverealpredictivevalueasappliedtoPREPA’ssystem.

3. PREPA’sconsiderationofenergyefficiencyisinappropriate

TheCommissionhasbeen clear in itsdesire thatPREPAbegin toexplicitly consider andpursuecost‐effectiveenergyefficiency. In itsDecember8thOrderon IRPComplianceandIntervenors’Comments,theCommissionorderedPREPAtoexplicitlymodelatrajectoryforenergy efficiency savings.134 In the IRP Final Order, the Commission ordered PREPA topursueorcooperatewithathird‐partyadministratorinthepursuitofenergyefficiency.135Despitethis,PREPAappearstohavenotconsideredenergyefficiencyinitssalesforecastinanyexplicitmanner.TheConsultingEngineer’sAnnualReport, providedas Schedule I‐1,discusses a range of energy efficiency and demand‐side management programsadministeredbyvariousgovernmentagenciesontheIsland.136Wecouldfindnoevidencethat these programs were either active or considered to be active by PREPA. The soleexceptionistheGovernmentEnergyEfficiency(EE)programcreatedbyAct57.137

In its direct testimony, PREPA claimed that it considers the Government EE programexplicitly:

IntheloadforecastofthebaseIRP,138GovernmentEnergyEfficiency(EE)asmandatedbyAct57‐2014,toreducetheconsumptionatgovernmentinstitutionswithrespectofabenchmark,wasforecastedtoachieve80percentofthemandate.139

Indescribingthesalesforecastusedfortheratecase,PREPAassertedthattheGovernmentEEprogramhasnotmetitstargets:

method is equally invalid. Moreover, the sum of non‐coincident peak values is not necessarilypredictiveofPREPA’scoincidentpeakastheseby‐classpeaks,bydefinition,donotalloccuratthesametime.

134OrderonIRPComplianceandIntervenors’Comments,CEPR‐AP‐2015‐0002(December8,2016).SectionI.1.b.

135IRPFinalOrder,SectionVII.B.1.k.

136PREPA’sPetitionforRateReview,ScheduleI‐1,p56.

137Act57‐2014,§IV.

138TheCommissionreferredtothisdocumentasthe“RevisedIRP”intheFinalOrderontheIRP.Werefertoithereasthe“BaseIRP”toavoidconfusionarisingfromPREPA’svariousreferencestothesame.

139PREPAEx.3.0,lines798‐800.


…theGovernmentEE,evenafterAct57beinginforceforafullyear,hasnotachievedtherequiredreductioninconsumption.Generallyspeaking,therewaslittleincentiveforgovernmentagenciestomanageandcontroltheirenergyconsumption,inparticularthosebelongingtothemunicipalities…The80percentestimateforcomplianceisanextremelyoptimisticviewongovernmentcompliance.140

PREPA’sclaimthattheGovernmentEEprogramhasnotachievedtherequiredreductioninconsumptionisfalse.Whilewewerenotprovidedactionplansofanyindividualgovernmententities,thestatutorygoaloftheGovernmentEEprogramistoachieveafortypercentdropin consumption from2013 levelsby2022across all governmentoffices.141This target iscommensuratewiththeten‐percent‐by‐2016goalsuggestedinPREPA’sresponsetoCEPR‐AH‐01‐07.142EfficiencysavingsdataprovidedbyPREPAdemonstratesthatthegovernmentasawholehasmorethanachievedtheinterim2016target.143Whiletheinterimgoalwastosave226GWhacrosstheentiregovernment,actualsavingshavebeenapproximately327GWh144(ornearlyfifteenpercentof2013consumption).

Onatechnicalclarificationcall,PREPAclarifiedthatwhiletheactualsavingsgoalhasbeenachieved, fewerthanhalfof the individualgovernmentofficesandentitieshaveachievedtheirindividualgoals.145Inourestimation,thisisirrelevanttothequestionofwhetherornotPREPAshouldconsidertheGovernmentEEprograminitsforecastsgoingforwards.Thereis no particular reason to assume that all government offices will achieve the statutoryrequirement at the same rate. Therefore, we find that PREPA has inappropriatelydisregardedthisprogram.ConsideringthatthetotalsavingsassociatedwiththeprogramtodaterepresentapproximatelytwopercentofPREPA’sFY2017salesforecast,146thisfailuremayhavehadasignificantimpactontheloadforecastusedintheratecase.

PREPAargued,onthesamecall,thatthesavingsfromtheprogramwereinfacttakenintoaccount because historical consumption is used in the forecasting of future sales.147 Asdiscussedabove,PREPA’smethodoftakinghistoricalconsumptionintoaccount–applying

140PREPAEx.3.0,lines802‐805,807‐808.

141Act57‐2014§IV.4.1(b).

142PREPA’sresponsetoCEPR‐AH‐01‐07(b)oftheCommission’sFourthROI(August1,2016).RefertoCEPR‐AH‐01‐07_Attach02.xlsx.

143Id.

144Id.

145October20ConferenceCall.

146Author’scalculation:quotientoftotalachievedsavingsfromPREPA’sresponsetoCEPR‐AH‐01‐07_Attach02.xlsx(327GWh)andFY2017expectedtotalsalesfromEx.3.02(17,268GWh).



thegrowthratesof inaccuratemodeloutputs torealdata– isnotrigorousordefensible.Moreover,atbestthisallowsPREPAtotakeintoaccountsavingsachievedtodate.PREPA’smethodologydemonstratesnomeanstomodifyasales forecastbasedonexpectationsoffuturesavings.WearepuzzledbythisomissionasPREPA’speakloadforecastcontainsanexplicit assumption of one incremental megawatt of savings due to demand‐sidemanagementeveryyear.148PREPA’sassumptionofDSMsavingsonpeakload(butnotontotalsales)datesfromatleast2013,asitisdescribedintheConsultingEngineer’sAnnualReport.149Despite this,wehaveseennoevidence thatPREPAorotherauthoritieson theIsland are administering, or thatPREPA is considering savings due to, activeDSMorEEprogramsapartfromtheGovernmentEEprogram.NosuchprogramsweredescribedinthisratecaseorintheIRPproceeding.150

In sum, we find PREPA’s consideration of energy efficiency in its sales forecast to beinconsistentwithitsownclaims,inconsistentwiththefacts,andmethodologicallyunsound.ConsideringtheCommission’sstrongemphasisonenergyefficiencyasanimportantpartofPREPA’sfuture,werecommendthattheCommissionrequirePREPAtoimprovethemannerinwhichitconsidersEEinfuturesalesforecasts.TheCommissionshouldrequirePREPAtoexplicitlynotepastEEsavingsbyclass,toforecastfutureEEandDSMsavingsbyclass(withappropriatedocumentationandsupport),andtoexplicitlydemonstratetheimpactoffutureEEsavingsonsalesforecastsbyclassandonatotal‐systembasis.

4. TheelasticityofdemandinPREPA’ssystemisunclear

PREPAdescribessalesinitssystemashavingasignificantlevelofpriceelasticity;thatis,PREPAassertsthatdemanddecreasesnoticeablywhenelectricitypricesincrease.AsstatedbyPREPA:

Before2006,electricitypricesdonotshows(sic)anyeffectintheenergyconsumptionpatterns,elasticityprice.Butafterthat,theconstantincreasesstarts(sic)affecttheenergydemand.ThePREPAannualenergydemandshow(sic)aprice‐elasticitycorrelatedtothefuelcoststhatdefinestheenergyspendingpatternsofourcustomers.151

There isakey issueraisedby thisassertion:PREPA is, in thisproceeding,applying forasignificant increase in rates. According to PREPA’s own statement, PREPA and theCommission should expect such an increase in rates to result in a decline in sales. This

148See,forexample,thecolumns“DemandaMax.,”“DSM,”and“DemandaMax.Adjusta”inPREPA’sresponsetoCEPR‐AH‐06‐14(a),foundinCEPR‐AH‐06‐14Attach01.xlsx.

149ScheduleI‐1,p56.

150SeeVolumeIIIoftheBaseIRP;§IV(D)oftheIRPFinalOrder;andAttachmentIofPREPA’sMotionforReconsiderationontheIRPOrder.

151CEPR‐AH‐01‐05Attach08.pdf,p1.


concern is echoed by several of the intervenors.152 However, apart from adding in therecently‐approvedtransitioncharge,153PREPAdoesnotappeartohaveconsideredincreasedratesinitsloadforecasting.Indeed,PREPA’sratecasefinancialmodel154providesforsuchfunctionality,buttheparametersdescribingelasticityaresettozero.155

IfPREPAbelievesthatitscustomersareparticularlyprice‐sensitive,itcouldhaveconductedaniterativecalculation.Forexample,PREPAcouldhaveforecastedconsumptionassumingpresent‐dayrates,andthenrecalculateditsproposedratessuchthatthoserateswouldyieldsufficient revenue given the assumed level of consumption, and then recalculatedconsumption again. If such an iterative process were to converge on a stable value forconsumptionand rates,PREPAcouldbe confident that itwouldnotunder‐collectdue toinsufficient sales.PREPAdidnotdoso; indeed, itdidnotadjust its forecastswhatsoeverbasedontheamountoftherateincreaseforwhichitactuallyapplied.

WearethereforeleftwiththeconcernthatPREPAwillunder‐collectduetoinsufficientsalesatitsincreasedrates.WebelievethattheimpactofthiseffectonFY2017willbesmall,asthefiscalyearisalreadywellunderwayandnonewrates(apartfromthetransitioncharge)havetakeneffect.However,weconsiderthequestionofelasticitytobeanotherimportantreasonthattheCommissionshoulddisregardPREPA’scurrentsalesforecastinanyconsiderationofratesforFYs2018and2019.

5. Useofthesalesforecastintheratecasefilingisinconsistent

152 CEPR‐AP‐2015‐0001, testimony of Ramón Cao Garcia on behalf of ICSE‐PR, lines 201‐203;testimony of Enrique Alberto Garcia Morelos and Reinaldo Valente Manuel Aguero on behalf ofCEMEX,lines225‐227.

153CEPR‐AH‐01‐05Attach08.pdf,p1.

154ThefinancialmodelisworkbookcontainingmanyofPREPA’sbaselineassumptionsofspendingandrevenueforFYs2017through2035underbothrestructuringandnon‐restructuringscenarios.ThecostsexaminedinourreportdonotvarysignificantlybasedonPREPA’sassumptionsregardingrestructuring. Many different version of PREPA’s financial model have been submitted to theCommissionbyPREPAthroughoutthecourseofthisproceeding.Forthepurposesofouranalysesinthisreport,wereliedontheversionpresentedasExhibit14.02ofPREPA’sPetitionforRateReview(PREPAEx. 14.02.xslm), the version of the financialmodelwith a restructuring rate change thatPREPAsubmittedaspartofitssupplementaltestimonyincompliancewiththeCommission’sOrderRequiringRevisedTestimony(September27,2016)aftertheissuanceofIRPFinalOrder.Wechosetorelyonthisversionofthefinancialmodelasitwasthemostup‐to‐datemodelavailableforthebulkofthetimeinwhichwewerecalculatingadjustments.Wedonotbelievethatfurtherupdatestothefinancialmodelsubstantiallyimpactouranalysisordiscussioninthisreport.

155PREPAEx.14.02.xslm,tab“Scenario,”lines56‐65.


Finally,wearemystifiedtoobservethat,althoughPREPAprepareditssalesforecastonaby‐classbasis,156 theby‐classsalesvaluespresented intheratecase financialmodelandthebillingdeterminantsworkpaper(whichinformstheembeddedcostofservicestudy)arenotconsistentwiththeseforecasts.Theby‐classvaluesintheseworkbooksappeartohavebeencalculatedonanadhocbasis.Intheratecasefinancialmodel,by‐classvaluesforFYs2017andthereafterarecalculatedbasedonthepercentageofFY2015salesrepresentedbyeachcustomer class.157 In the billing determinants worksheet, by‐class sales for FY2017 arecalculatedbasedonaby‐classinflatorfromFY2014values.158Thetablebelowsummarizesthesedisparatevalues.

Table 3. PREPA's sales forecasts by class, and representations thereof in the financial model and billing determinants workpaper.

ClassForecastsbyClass*

(GWh)

BillingDeterminantsWorkpaper**

(GWh)FinancialModel†

(GWh)Residential 6,388.82 6,177.45 6,242.81Commercial 8,169.85 8,347.80 8,322.16Industrial 2,330.78 2,399.15 2,333.12OtherPublicAuthorities ‐ 33.66 ‐Agriculture 26.11 26.43 26.14PublicLighting 316.06 283.83 308.70PREPAUse ‐ ‐ ‐Others 36.71 ‐ 35.39Total 17,268.33 17,268.33 17,268.33

*CEPR‐AH‐01‐06_Attach05.xslx**WP1(BillingDeterminants)REV2016‐10‐11.xlsx,“I‐3”tab,line59.†PREPAEx.14.02.xslm,“Inputs”tab,lines4‐10.

Weseenojustificationfortheseworkpaperstocontaininaccurateby‐classbreakdownsofaforecastthatPREPAoriginallycalculatedasasumofby‐classvalues.Werecommendthatthe Commission require PREPA to revise these workpapers to rely on actual by‐classforecastsratherthanimprovisedestimationsthereof.

D. Recommendation

1. PREPA’sFY2017salesforecastshouldstandwithoutadjustments

WerecommendthattheCommissionacceptPREPA’ssalesforecastwithoutmodificationforFY2017only.Ourrecommendationisbasedonthreeobservations:

156PREPA’sresponsetoCEPR‐AH‐01‐06oftheCommission’sFourthROI(August1,2016).RefertoCEPR‐AH‐01‐06_Attach05.xslx.

157PREPAEx.14.02.xslm,“Inputs”tab,lines4‐10.

158PREPA’sPetitionforRateReview,WP1(BillingDeterminants)REV2016‐10‐11.xlsx(originallyprovidedasExhibit4.0),“I‐3”tab,line59.


First,PREPA’sforecastsaregenerallyreliableforthefirstforecastedyearconsidered,albeitthey are uniformly overstated formore‐distant future years. As summarized in Table 4,PREPA’slastsixforecastshavebeen,onaverage,twopercentofffromactualvaluesofthefirst future year.159 For example, the FY2011‐vintage forecast predicted FY2012 sales towithin four‐tenths of a percent.When considering only one year in the future, PREPA’sforecasts are under‐stated nearly as often as they are over‐stated. By contrast, PREPA’sforecasts are on average five percent too high when considering the third year of theforecastedvalues.ThisobservationsuggeststhattheCommissioncansafelyrelyonPREPA’sforecastedvaluesforFY2017butshouldnotdosoforFYs2018and2019.

Second, recent monthly actuals also suggest that PREPA’s FY2017 forecast has beengenerallywithinthreepercentofrealmonthlysales.160

Third,variationinthesingle‐digitpercentagesfromforecasttoforecastisnotunusualforelectricutilities.161

Forth,andinlightoftheabove,thereisessentiallynobenefitthatcanbegainedbyadjustingtheloadforecast.Asdescribedabove,PREPA’sloadforecastimpactsbothitscalculationofratesanditsformationoffuelandPPOAbudgets.Theimpactofasmallchangeinforecastedsales is therefore not easy to predict. As PREPAwill likely have a budget reconciliationprocessofsomeform,weconcludethattheamountbywhichPREPAwouldover‐orunder‐collectduetoanyinaccuracyinitssalesforecastislikelytobesmallenoughthatitcanbereconciledthroughsuchaprocess.

159Author’scalculation,basedonsourcedatafromCEPR‐AH‐06‐14Attach01.xlsx.

160 Author’s calculation, based on values from CEPR‐AH‐01‐06_Attach 01.xlsx, CEPR‐AH‐01‐06_Attach05.xlsx,andPREPA’sJuly2016andAugust2016monthlyreports.RelevantpagesattachedasFisherandHorowitzExhibit04.PREPA’sratecaseforecastwas2.2%belowJuly’sactualmonthlysales(1549GWhforecastedversus1584GWhactual)and2.6%aboveAugust’sactualmonthlysales(1536GWhforecastedversus1496GWhactual).

161See,forexample,thevariationinsalesforecastsbetweenPacifiCorp’s2015IRPandIRPUpdateasseen in Figure 3.1 of the 2015 IRP Update (available online at:http://www.pacificorp.com/content/dam/pacificorp/doc/Energy_Sources/Integrated_Resource_Plan/2015%20IRP%20Update/2015%20IRP%20Update_20160426.pdf).Seealsovariationbetweenthe Independent System Operator – New England’s 2014 and 2015 Capacity, Energy, Load, andTransmissionforecastsoftotalsystemsales(availableonlineat:https://www.iso‐ne.com/system‐planning/system‐plans‐studies/celt/?).


Table 4. Deviations of PREPA's sales forecasts from actual values after one and three years.162

ForecastVintage%DeviationfromActualSales

inFirstFutureYear%DeviationfromActualSales

inThirdFutureYearFY2010 3.8% 5.7%FY2011 0.4% 2.1%FY2012 ‐1.6% 6.0%FY2013 3.6% 6.5%FY2014 0.7%FY2015 ‐1.6%Average(AbsoluteValue) 2.0% 5.1%

2. PREPA must improve its forecasting methodology anddocumentation

EventhoughwerecommendnochangetoPREPA’sFY2017salesforecast,werecommendthattheCommissionrequiresignificantadjustmentstoPREPA’sforecastingmethodologyanddocumentation,andthattheCommissionrequirecertaincorrectionstotheuseofthesalesforecastintheratecase.Theserecommendationsaresummarizedasfollows:

a. PREPAmustdevelopasingle,reliable,theoretically‐soundforecastingmodelforeachrateclass.Thesemodelsshouldbeabletoadequatelypredicthistoricalsalesgivenhistoricalvaluesoftherelevantparameters.

b. PREPAmustdevelopanewsalesforecastbasedonthesemodelspriortofilinganotherplanningorratecasewiththeCommission.

c. AnychangestoPREPA’sforecastingmodelsinthefuturemustbeclearlydocumentedandsupportedbyevidencethatthepriormodeldidnotsufficientlyaccountforrelevantcircumstancesordata.

d. PREPAmustadequatelydocumentitsloadforecasts,includingthroughtheprovisionofclear,comprehensive,andaccuratemethodologicaldocumentation.PREPAmustprovideworkpapersshowingallrelevantinputsandcalculations,anditmustclearlynotesourcesforallassumptionsandhardcodedvalues.

e. PREPAmustcorrectitsfinancialmodel,costofservicestudies,andallotherrelevantworkbookstouseactualby‐classforecastsinsteadofadhocbreakdowns.

162Author’scalculation,basedonsourcedatafromCEPR‐AH‐06‐14Attach01.xlsx.Percentdeviationsarebasedonactualvaluesoneandthreeyearsaftertheforecastwasprepared.Forexample, thedeviationsshownforthevintageFY2010forecastarecalculatedbasedonthedifferencebetweenmodeledandrealsalesinFY2011andFY2013.


V. CAPITALBUDGET


In this chapter, we address PREPA’s budget for spending on capital expenses. PREPA’scapitalbudgetincludestheitsplannedspendingmajorimprovementsandexpansionsofitstransmission,distribution,andgenerationinfrastructure—aswellasitsplannedspendingontheAguirreOffshoreGasport(AOGP).

TheCommissionmustdecide if thePREPA’s capital budget, bothoverall and for specificprojectswhenpossible,isreasonable.Inparticular,theCommissionmustascertainwhetherPREPA’scapitalbudgetissufficientbutnotexcessivetoallowPREPAtooperateasafeandreliable electric system. If the Commission decides that PREPA’s capital budget and theindividualprojectsthereinarereasonable,nochangeswillberequired.IftheCommissiondecides that PREPA’s capital budget or aspects thereof are not reasonable, itmust thendecidewhetherornottomakeadjustmentstothatbudget.

Below, we discuss the significance of the capital budget, describe the evidence (or lackthereof)supportingthatbudgetandcommentonseveralconcerningpatternsobservedinPREPA’shistoricalspendingandFY2017budget.WecometotheconclusionsthatPREPA’scapital budget is poorly‐supported and ‐documented and that PREPA’s prioritization ofcapital projects has historically been misguided. However, we also find that PREPA’sgeneration and transmission infrastructure are in sufficiently deteriorated state (due toperennially deferred maintenance) that significant capital spending is necessary. Werecommendseveralspecificbudgetadjustments,includingalimitonspendingatAOGPinaccordancewiththeCommission’sFinalOrderontheIRP,acancellationofPREPA’splannedspending on advanced metering infrastructure (AMI), and a recategorization of severalmaintenance contracts as operational rather than capital expenses. In addition, werecommendthattheCommissioninterrogatePREPA’scapitalbudgetingprocessfurtherinits pending investigation on PREPA’s performance. We summarize our overall findingsfurtherinSectionBLANKbelow.

B. PREPA’suseofcapital

1. Overview

a. Definitionofcapitalexpenses

Forthepurposesofthiscase,weneedtodistinguishbetweenwhatisdefinedasacapitalexpenditureandamaintenanceexpenditure.

Operatingandmaintenance(O&M)expensesarerequiredtoruntheutilityday‐to‐day.Forautility’sphysical(generation,transmission,anddistribution)infrastructure,thiscategoryincludesordinaryrepairs,whicharestandardrepairsandmaintenancerequiredtokeepanassetinoperationalcondition.Thesetypesofrepairsshouldbeaccountedforintheperiod


incurred—i.e.,expensedinamaintenancecategory.Maintenancecostsalsoincluderepairsthatmaylastmorethanayear,butshorterthanafullmaintenancecycle.

Capitalcosts,ontheotherhand,areexpensesthatincreasetheusefullifeofaunit,andaregenerallylong‐lasting.Usedtoacquireproperty,assets,orequipment,orextendthelifeofanexistingasset,capitalistypicallydepreciated(paidoff)overthelifeofanasset.Capitalinvestmentsoccurmorerarely,requiresubstantialbudgetallocations,andincreasethelifeorcapabilitiesofanasset.

Inmostutilitycircumstances,capitalbudgetsareamortizedoverthelifeoftheasset(ortheimprovement) and depreciated. The purpose of this accounting practice is to provideintergenerationalratepayerequity,andensurethatthecostsofalong‐livedassetarespreadacrossthebeneficiaries.

Whenanassetisseverelydegraded,thelinebetweenacapitalcostandanoperatingexpensemaybethin.Simplyrepairingtheassetmaybeinsufficienttokeepitoperational.Instead,itrequiresalargerlevelofinvestment.Forexample,aprojectdesignedtotemporarilypatchaboilerisalmostcertainlyanoperationalexpense(thepatchisnotdesignedtolastmultipleyears),butthereplacementoftheboilerinthenextmajoroutagecycleincursanexpenseanticipatedtolastmanyyears.

This rate case stretches the line between capital and operational expenses. As PREPArecoversfromyearsofdeferredmaintenanceanddeficientbudgets,itisrequiredtoreplacea large amount of deteriorated infrastructure—including equipment that may not haverequiredreplacementifday‐to‐daymaintenancehadbeenexecutedappropriatelyandattheappropriate time. PREPAmay requiremultiple years of capital expenses tomake up fordeferredmaintenance.Manyoftheseprojectsappeardesignedtoreplaceorrefurbishbrittleinfrastructure.

b. Capitalprojectsasincurred,ratherthaninservice

In standardutility rate cases, capitalprojectsonlybecomepartof revenue requirementsonce theproject is inservice.Utilitycommissions followa“usedanduseful”principle, inwhich capital projects must be in service, operational and providing useful service toratepayersbeforetheybecomearatepayerburden.Forsuchamechanismtowork,utilitiesmusthaveaccesstosomeformofactivecapitalmarketto fundtheproject’s fullexpensethroughconstructionandholdthedebtontheproject,graduallypaidoffbyratepayers.

InPREPA’scase,theutilityhasnoreadyaccesstolow‐costcapitalmarkets,andthusPREPAisleftintheunenviablepositionoffundingcapitalprojectswithnolenderbutratepayers.Asdiscussed by the Commission’s other experts in this case, PREPA is effectively askingratepayerstotaketheplaceofalenderandpayformanyofthecapitalprojectsinfull,asthecostsforthoseprojectsareincurred.Thisisasharpcontrasttonormalutilityoperations,andrequiresanextraordinaryamountoftransparencybetweenPREPAandtheCommission.As a regulator of a public power utility (as opposed to an investor‐owned utility), the


Commission’savailableavenuestohaveanypartyotherthanratepayersabsorbtheimpactofimprudentdecisionsarealreadylimited.TohavecapitalexpensesincurredbyratepayersasPREPAengagesinthemexposesPREPA’scustomerstotheutility’sdecisionswithlittleornobuffer.

In this rate case, PREPA has requested that ratepayer exposure to capital projectrequirementsbeimmediateandabsolute:PREPAexpectsratepayerstofundcapitalprojectsas the dollars for those projects are spent. Large capital projects are oftenmulti‐year innature, requiring a capital outlay to procure equipment, additional expenses to installequipment,andpaymentstocontractorsatsuccessfuloperation.Newgeneratorsmayhavecapitalspendingschedulesthatarespreadacrossyears,andevensmallerprojectsmaybeincurredoverextendedperiodsbeforereachingcompletion.

Because of the utility’s lack of access to capitalmarkets, PREPA’s capital budget reflectsPREPA’s spending throughmulti‐year projects, rather than at the endof theproject—asmight be expected in other utilities. This unusual processmeans that in FY2017, PREPAratepayerswillbepayingforsomecapitalprojectsinwhole,payingthelastfractionofcapitalprojectsstartedinFY2016orevenFY2015,orpayingforthefirstpartofprojectsthatmightlastthroughFY2020.

Again,thisunusualrequestbyPREPArequiresanextraordinaryamountoftransparency:isalineiteminthecapitalbudgetjustafractionofthewholeexpense,andifso,whendidtheproject start and when will it end? The Commission is charged with evaluating thereasonablenessofPREPA’sprojects–inplanning,implementation,andintotalcost.Byonlyrevealingafractionofthetotalbudgetforanygivenyear,theCommission’sjobishampered.Questionsbecomemoredifficulttoanswer:WhendidPREPAplantheproject?Whatwastheduediligenceprocessbehindtheproject?HasPREPAimplementedtheprojecteffectivelyandappropriatelycontaineditscosts?Isthebudgetfortheprojectappropriateforthescopeof work? These questions are all squarely in the Commission’s purview, but PREPA’sdocumentation regarding the costs of these projects leave the Commission with littleresolutiononthesequestionsfromthestart.

2. PREPA’sexpectedcapitalexpendituresbyarea

In Schedule F‐3 REV (“Revised”), PREPA provided an extensive table of projects thatcontributetothecapitalbudget.PREPAspecificallypulledoutthecapitalbudgetforAOGPand provided this separately from all other capital projects. PREPA’s list of non‐AOGPprojectsincludedover400individuallineitemsandincludedbudgetsrangingfrom$1,000to$20,000,000inFY2017.Nearlythirtypercent(119)oftheprojectshadzeroorablankbudget in FY2017, and over one‐third (45) of those had no budget listed in any yearconsideredinthisratecase,FY2017‐FY2019.

Ofthe357non‐AOGPprojectswithabudgetlistedinatleastoneyearbetweenFY2017andFY2019, budgets were divided relatively evenly between production, transmission, anddistribution,with a smaller budget amount allocated to the area of “office furniture and


equipment,”acategoryprimarilyorientedtowardstransportationandcomputerequipment.Table5belowshowsthebreakdownofthecapitalbudgetbyarea,sub‐area,andyear.

Table 5. Total capital requested by PREPA in FY2017 through FY2019, by area and subarea. Does not include AOGP.

FY2017 FY2018 FY2019ProductionThermalSteamGeneration* 55.6 52.5 60.9CombinedCycleandCTs** 25.4 23.2 29.7HydraulicProductionPlant 2.2 2.0 4.1ProductionTotal 83.2 77.6 94.7

Transmission115kV 37.4 59.7 62.7230kV 18.4 23.2 20.338kV 11.9 9.9 17.438KV 0.6 0.6 0.0OtherTransmission 13.0 11.4 7.5TransmissionTotal 81.3 104.7 107.9

Distribution13.2kV 10.4 13.5 14.14.16kV 0.6 0.5 0.98.32kV 31.5 24.9 38.2Distribution 30.8 25.1 39.1OtherDistribution*** 0.7 0.6 0.8DistributionTotal 73.9 64.5 93.1

OfficeFurnitureandEquipmentTransportationEquipment 19.4 12.5 16.5ComputerEquipment 13.1 6.2 7.5OtherEquipment 4.3 4.3 6.0OfficeEquipment 0.0 0.0 0.0OfficeFurnitureandEquipmentTotal 36.8 23.0 30.0

GeneralLandandBuildingBuilding 2.4 1.4 1.8Land 1.0 0.8 0.8GeneralLandandBuildingTotal 3.4 2.1 2.6

PreliminarySurveysandInvestigationPreliminarySurveys 1.6 1.3 2.0

Total 280.2 273.3 330.2*ThermalProductionPlant,byPREPA**OtherProduction,byPREPA***Blank,byPREPA

This section will discuss PREPA’s capital budgets and allocations, reviewing individualprojectsandhistoricspendingtogaugethereasonablenessofPREPA’sasks.

Wenotethatwewerenotabletoassesseachandeveryoneofthe357individualcapitalprojects,163andthusonlyfocusonsomeofthemostkeyspendingareaswithopenquestions.Forthepurposesofthisratecase,werecommendthattheCommissionacceptPREPA’sbudgetestimatesforareasthatwedonotexplicitlydiscuss.Inthefuture,weexpectthat

163402totalprojects,357withspendinginFY2017‐FY2019.


PREPAwillprovidesubstantiallymoretransparencyonitsbudgetestimationprocess, itsprioritization,andthebasisforexpectedbudgets.

a. Historicspendingpatternsincapitalbudget

PREPA’s total capital budget has been internally capped at about $300million (with theexceptionof2010),atrendwhichisstillpresentintheutility’sbudgetforthisratecase(seeFigure15,below).

Figure 15. PREPA total capital budget and spending for all areas, historic and rate case, by fiscal year (millions).164 Does not include AOGP spending.

PREPAexceededitscapitalbudgetfrom2010through2013.In2014and2015,theutilityunderspentitsanticipatedcapitalbudgetbyabout$50millioneachyear.ThistellsusthatPREPAiscapableofspendingwithinabudgetcap,althoughaswe’veindicatedelsewhereinthisreport,itisnotclearthatPREPA’sspendingin2014and2015wassufficienttomaintainsystemreliability.

PREPA’sstructureisdividedintosixteendirectoratesorresponsibilityareas,nineofwhichusecapitalbudgets.Thefollowingtabledividesoutthedirectoratesandthosethatutilizecapitalbudgetorreportcapitalspending.165

164ScheduleF‐3REVandPREPA’sresponsetoCEPR‐AH‐05‐10of theCommission’sEleventhROI(October7,2016).RefertoCEPR‐AH‐05‐10Attach03.

165Id.


Table 6. PREPA Directorates

Code NombredelaDirección DirectorateName CapitalBudgetA01 JuntadeGobierno BoardofGovernorsA02 Ejecutivo Executive XA03 AsuntosLaborales LaborMattersA04 ConsultorJurídico LegalAdviser XA05 PlanificaciónyProtecciónAmbient EnvironmentalPlanningandProtection XA06 Ingeniería EngineeringA07 Finanzas Finances XA09 RecursosHumanosyAsuntosLaboral HumanResourcesandLaborAffairs XA10 Generación Generation XA11 ServicioalCliente CustomerService XA12 TransmisiónyDistribución TransmissionandDistribution XA14 ResponsabilidadesCorporativas CorporateResponsibilitiesA16 SeguridadCorporativa CorporateSecurity

Overall, the vast majority of historic capital expenditures were used in the generationdirectorate ($100million), and transmission and distribution directorate ($150million),followed by the customer service directorate ($15‐$30 million). In PREPA’s directoratestructure, the generation directorate appears to hold responsibility for all of PREPA’sgenerationfacilities.Transmissionanddistributionattendstopoles,wires,andsubstations,whilethecustomerservicedirectoratehandlesmetersandservicedrops(i.e.,connectiontometers).

Historically,moderatebudgetswereallocatedtotheenvironmentalplanningandprotectiondivision($7‐$14million)whichconductsoverallplanningandappearstoengageinone‐offprojects,includingenvironmentalcontrols.Theforward‐goingbudgetsforthisdirectoratearesubstantiallylowerthanthelastthreeyears.

Incontrast,theformerlymodestbudgetsoftheexecutivedirectorate($5‐$17million)aresubstantially higher in this case ($20‐$30 million). From a capital perspective, thisdirectorate is responsible for transportation, computers, infrastructuremanagement,andsafety.

Figure 16, below, shows PREPA’s historic over‐ and underspending by directorate. Thegeneration directorate has themost substantial annual swings in capital budget, havingspent120‐160%ofitsallocatedbudget,butunderspendingsubstantiallyin2014and2015.Withtheexceptionofoneyear(2013),thetransmissiondirectoratehitsitsbudgettargetsfairlyaccurately,despitethesizeofitsoverallbudget.


Figure 16. Historic capital budget overspending (underspending) by directorate, by fiscal year.166

Muchofthegenerationdirectorate’sunder‐andoverspendingisattributabletosinglelargeprojects that are either canceled (leading to underspending) or go substantially overbudget—this includes boiler and turbine replacements, and some unexplained spendingpatterns. Figure 17, below, shows that much of the overspending in the generationdirectorate in 2011 and 2012 was associated with specific projects at the large centralstationgenerators(SanJuanandCostaSurin2011,andalsoAguirrein2012),andalargeallocationintoamiscellaneousfundof“technicalservices.”

166CEPR‐AH‐05‐10Attach03.Forvisualclarity,graphicexcludesA07(Finance)andA09(HumanResources)withbudgetsoflessthan$1,000,000.


Figure 17. Historic capital budget overspending (underspending) by Department in the Generation, by fiscal year.167

Lookingmoredeeplyintothetechnicalservicesdirectorate,weseethatabouthalfofoverallgeneration directorate overspending in FY2011 is attributable to a “new generation”resource.168Thatlineitemamountedto$48millionofunbudgetedspendinginFY2011andFY2012.WhenqueriedduringtheOctober20ConferenceCall,PREPAwasunabletodescribethebasisofthisspending.ItislikelythatthiscostwasassociatedwithtransitioningCostaSurtoagas‐firedplant,althoughitiscuriousthattheprojectwasunbudgeted.

Whiletherearemacropatternsabovethedirectoratelevel,itisdifficulttodissectdownintotheutility’sspendingandthedriversofthecapitalbudgetdeviationswithoutsubstantiallymoredocumentationfromPREPA—andoftenthatdocumentationwasdifficulttocomeby.Fromtheperspectiveofarateregulator—i.e.,thisCommission—itisdifficulttochoosetoallocatebudgetwithoutaclearunderstandingofbothitspurposeandtheutility’spropensityandabilitytosticktobudget.


168CEPR‐AH‐05‐10Attach03.DepartmentA10–Generation,DivisionD44‐IngenieríayServiciosTécnicos,Responsibility101‐NuevaGeneración.


3. IssuesinreviewingPREPA’scapitalbudgetestimates

PREPA’scapitalbudgetswerenottransparentlyconveyedandmanywerepoorlysupportedbytheevidentiaryrecordrequestedofPREPA.DuetothepoorrecordsprovidedbyPREPA,thelackofdocumentationonthepurposeorbasisofmanyoftheprojects,andthecomplexnatureofthetasksfacingPREPAintheirgenerationandtransmissionfacilities,weareonlyabletoprovideanaccountingofalimitednumberofprojectsanticipatedbyPREPA.Inmanyutilitycases,theindividualprojectsatexistingplants(generationortransmission)areoneitheranestablishedschedule,correspondtoknowninitiatives,oraregenerallyinlinewithprior year expenditures. In PREPA’s case, many of the projects are unique to thecircumstances of PREPA today, represent deferred maintenance projects, or are simplyundocumented.

HerewediscusssomeofthekeyissuesfacingusaswereviewPREPA’scapitalbudget.

a. Poorrecordkeepinganddescriptionshamperreview

PREPAwasunable to provide justification for the projects listed in the capital budget, afundamentalcomponenttoprovidingregulatoryoversight.Indiscovery,werequestedthatPREPAidentifywhichprojectshadworkordersorcontractsforengineering,procurementand construction (EPC) for projects. PREPA denied the request in full, stating that theirrecordswerenotsufficienttoprovidesuchjustifications.PREPA’sastonishingresponsewasasfollows:

PREPA does not have electronic systems configured to be able to assessautomatically which work orders are associated with a particular CAPEX[capital expenditure]project;most of this information is onpaper records.And thosepaper recordsarenotorganizedbyproject ina central location.Further answering, the process of preparing the CAPEX budgets start inPREPA’stransmissionanddistributiondistricts,generatingplants,customerserviceoffices,amongothers.169

PREPAfurtherexplainedthat:

The particular details of each project aremanaged locally, not at a centrallocation. Once the final complied spending limits aremet, the plan can gothrough the final approvals. Regarding projects with nomoney spent, anydocumentsreflectingthecorrespondingjustificationsandrelatedinformationis located in the multiple generation plants, distribution offices, customer

169CEPR‐AH‐02‐02(c).


serviceareas,andotherofficesordepartmentsthatareresponsiblefororusethatCAPEXproject.170

Whilewecan,andwill,debatePREPA’sapproachtobudgetprioritizationamongstprojectsanddistricts,itisunacceptablethatthefundamentalbasesofprojectscannotbeprovidedondemand.Onaverage,PREPA’scapitalprojectscomeatacostof$1,000,000peryear.171Forthevastmajorityoftheseprojects,PREPAcouldnotprovideanyformofdocumentationexplainingtheproject,justifyingtheexpense,howtheestimatewasgenerated,andthevalueof the project. This indicates a breathtakingly cavalier approach to the dispensation ofratepayerdollars.

FollowingPREPA’s failure toprovideworkordersorcontracts forprojects in thecapitalbudgetplan,weaskedPREPAtoprovideadetaileddescriptionofeachcapitalprojectwithexpectedFY2017spendingabove$1,000,000andhowcapitalcostswereestimated.PREPAprovidedaspreadsheetwithanywherefromasentencetoashortparagraphon72oftheprojects.172 Nowhere did PREPA provide an explanation of how it derived budget ofanyproject.

TheexplanationsprovidedbyPREPAwere,forthemostpart,deeplyinsufficient.A“BoilerImprovement” at Aguirre 1 with $6.5 million in spending was described—in full—as“replacementofma[j]or[sic] lifeextensioncomponentsaspartofGasnaturalconversionoutage.”173Onlythroughsiftingthroughthe2013URSReportandextrapolatingfromapriormaintenanceschedule174diditbecomeapparentthatthiswaspartofa50,000‐hourregularmajormaintenanceoutagecycle.The2013URSreportdescribesthat:

The scope of work will include an environmental outage plus boilermodificationstotheconvectionsection(superheatandreheat)headersandtubes to be compatible with gas firing. Other work will include repairs tothermal insulation, installation of new motor control center (MCC) andswitchgearforthecirculatingwaterpumps.175

170CEPR‐AH‐02‐02(c).

171Author’scalculation.AverageFY2017costofcapitalbudgetitemswithnon‐zerospending.

172CEPR‐JF‐02‐02Attach01.

173Id.PID15243.

174AH‐06‐12Attach01.

175 URS June 2013 Annual Report. PREPA Exhibit 3.02(D) Consulting Engineers Report. FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Page11.


WhilestillinsufficientforauditingPREPA’scostestimate,andthreeyearsout‐of‐date,the2013URS Report explanation is farmore informative than the information provided byPREPAinthiscase.

b. Engineeringauditrequiredtoassessprojectnecessityandbudgetallocation

WenotehereexplicitlythatPREPA’scapitalencompassesaverywiderangeofprojects,fromthe refurbishment of individual turbines or generators, to balance of plant operations,buildingnewpolesandlines,acquiringadvancedtransmissionanddistributionequipment,orderingnewutilityvehicles,computersystems,andnetworkequipment.Intheparticularabsence of workpapers justifying PREPA’s expenses, a thorough evaluation of PREPA’sspending,budgetgaps,andbudgetallocationwouldhaverequiredateamofengineersandauditors.Wehaveneitherthetimenorexpertisetoengageineachoftheseareasofinquiry.Our analysis below therefore focuses on several specific projects and an evaluation ofPREPA’scapitalbudgetsbyareaintheaggregate.

c. CapitalbudgetiscappedbasedonPREPA’shistoriclimits

PREPA’scapitalbudgetappearstohaveaninternalcapbasedonfactorsthatarenotclear.Inmaintainingthishardcap,PREPAmayhaveimposeddistortionaryeffectsonitscapitalbudget.AreviewofPREPA’scapitalbudgetsfromFY2010toFY2015indicatesthatPREPAsetsatop‐downbudgetcapapplicabletotheentiresystem.InFY2010,PREPAsetitscapitalbudgetatexactly$350million.InFY2011,2013and2014PREPA’sbudgetwassetatexactly$300million.InFY2015,PREPA’stotalcapitalbudgetfellto$244million,butwesuspectthatthisdropreflectsaninternalbudgetingmechanismandnotnecessarilyamoreefficientuseoffunds.

Figure15,above,showsPREPA’shistoriccapitalbudgetcapsandactualspending,aswellasthebudgetsproposedinthisratecase,exclusiveoftheAOGPproject.NeitherspendingnorexplicitbudgetdataforFY2016weremadeavailable;PREPAhasindicatedthatitsFY2016budgetwassetatFY2015levelsbecausetheboardfailedtoapproveaFY2016budget.176Theaggregatebudgetandspendingshowseveralremarkablefeatures.

i. PREPA’scapitalbudgetiscappedfromthetopdown

176CEPR‐RS‐16‐11(a).“TheapprovedFY2015budgetwasautomaticallyadopted(perArticle504ofthe1974TrustAgreement)astheFY2016budgetwhentheBoarddidnotapproveanewbudgetprior to thestartofFY2016.Section504of the1974TrustAgreementestablishes that if foranyreasontheBoarddoesnotadopttheAnnualBudgetbeforethefirstdayofanyfiscalyear,thebudgetfortheprecedingfiscalyearshallbedeemedtobeinforceandshallbetreatedastheAnnualBudget.”


PREPAsoughttomaintainanexact,top‐downbudgeteveryyearfromFY2010toFY2015meaningthat,regardlessoftheconditionofthesystem.PREPAexplainsthat:

Historically,therehasbeenpoliticalpressuretonotincreasePREPA’sratesinresponse to cost and investment needs and therefore PREPA has had tosacrificeneededcapitalexpendituresinordertoremainsolventandtonotrunoutofcash.177

While PREPA’s policy of maintaining a budget cap may generally be a mechanism ofnarrowingandprioritizingkeyprojects, thisapproachonlyworks ifPREPAdeploys thatlimitedcapitalstrategically,assesseswhenprojectsgooverbudget,andseekstoreallocateefficientlyacrossdepartments.

Basedonourassessmentofspecificbudgetitems,itisnotclearthatPREPAhasaneffectivemechanismforthestrategicdeploymentofcapitalunderthisbudgetcap. Instead,PREPAappearstouseageneralruleof thumbforthedeploymentofcapital. IndiscussionswithPREPA,178 the finance director explained that PREPA’s budgets are designed such thatdepartmentsareallocatedtheloweroftheirpasthistoricbudgetortheirlastyear’sactualspending,exceptinextraordinarycircumstances.

PREPA’smechanismofsimplylimitingdepartmentstohistoricbudgetsorspendingcreatesdistortionaryeffects,requiringsomedepartmentstofindcreativewaystospenddollars(orrisklosingitthenextyear)andlimitingbasicsystemneedsatotherdepartments.

OneareainwhichthisdistortionaryimpactisreadilyapparentisintheallocationofcapitaldollarstocriticaldeferredmaintenanceatPREPA’sgeneratingandtransmissionfacilities,versusPREPA’smovetostartrollingoutsmartmetertechnology.Inastrategicdocumentfor Advanced Meter Infrastructure (AMI), PREPA explains that it “allocates $10 Millionannually to the meter replacement program. This plan contemplates using this fund toimplementthereplacementofthenetwork[withAMI].”179GiventhedirestateofPREPA’ssystem,areasonablebudgetprioritizationprocesswouldhavedeterminedthatthefarmoreexpensiveAMIsystemisaluxurythatPREPAmayhavetodeferuntilthesystemisreliable,

177CEPR‐SGH‐01‐08.Page11.

178October20,2016ConferenceCall,ErnestoRamos,1:11:00‐1:15:00.ErnestoRamos.“Historically,wehavespentabout$300millionformaybethelasttwofiscalyears.Thisisrelatedtotheavailablefundsthatwehaveforconstruction.Thisconstraintisrelatedtothefundswehaveinconstructionfund.Thatlimitswhatweaskforfromplanning[department]….Theconstructionfundisdedicatedtocovercapitalexpenditures.Thedollarvalueiscreatedbyourtrustagreement.Itisnotmorethan$300[because]ofourfinancialcapacity.”

179 CEPR 161020 Request No. 8. Attach 01. Page 4. “La AEE, asigna anualmente $10Millones alprograma de remplazo de metros. Este plan contempla utilizar este fondo para implementar elremplazodelared.”


safeandaffordable,andthatthat$10millioncouldhavebeenbetterusedelsewhereevenifthatmeantspendingitonprojectsotherthanmeterreplacements.

Becausethebudgetisallocatedfromthetop,itisdifficulttodeterminetherealstateofneedforanygivenareaofinfrastructurewithoutafullengineeringaudit.

ii. PREPAhasexceededhistoricalcapitalbudgets

A review of PREPA’s historic spending indicates that, until FY2014, PREPA regularlyoverspentitscapitalbudget.AsFigure18shows,themajorityoftheoverspendfrom2010to 2012 was in the generation directorate (“A10”), followed by an overspend in thetransmission and distribution directorate (“A12”) in FY2013.180 PREPA underspent itscapitalbudgetsinFY2014andFY2015.Fromourperspective,withoutadeepdiveintoallofthesubcomponentsofthePREPAsystem,itisalmostimpossibletodiscernthebasisofanyyear’s capital budget or the reasons for an overspend or underspend. These patterns,however,areindicativeofhowPREPAmightusethebudgetsdecidedinthisratecase,andprovideinsightintoPREPA’sdecision‐makingprocesses.

180CEPR‐AH‐05‐10Attach03.Author’scalculationsandaggregationsbydirectorate,departmentandresponsibility.


Figure 18. Historic capital budget overspending (underspending) by directorate, by fiscal year.181

Thereareusefulstoriesinthecapitalbudgetsandspendingschedulesthatwewillillustratethroughoutthisreport.Forexample,thischartshowsthatthegenerationdirectoratespent120‐160% of its allocated budget in FYs 2010‐2012, while in FYs 2014 and 2015 thedirectorate underspent its budget substantially. We observe that the generationdirectorate’s overspending in FY2011 is primarily a function of $33 million in “newgeneration” for which PREPA was unable to provide an explanation,182 while itsunderspending in 2014 and 2015 relates to substantialmaintenance projects at Aguirrewhichdidnotoccur.PREPAlaterexplainedthattheseprojectsweredeferreduntiltheAOGPprojectcameonline—nowscheduledforlate2018.Intheabsenceofthesedeferrals,PREPAwouldhavenearlyhitcapitalbudgetsin2014and2015.WhilePREPA’sunderspendingongeneration may seem to be a substantial gain in the name of budget control, PREPA’sgenerationfleethassuffereddramaticlossesinthelasttwoyears—possiblyasafunctionofthisreductioninspending.

4. Relationship between the capital budget and the IntegratedResourcePlan


182October20ConferenceCall,discussionwithGenerationDirectoratemanager,


a. CorrespondencebetweenScheduleF‐3REVandIRP

AccordingtoPuertoRicoRegulation8720(March28,2016),theratefilingrequirementsforScheduleF‐3shall“listprojectedconstructionandcapitalexpenditurerequirementsforeachof the three (3) years following the test year,” and that “this schedule shall include anexplanationofhowtheprojectedcapitalexpenditurescorrespondwithPREPA’sIntegratedResourcePlan.”183

In theory, the schedule provided by PREPA should have explained which projects wereassociatedwith the IntegratedResourcePlan (IRP) and their relationship to theplan. Inpractice, providing a detailed explanation between 350 capital expenses and the IRP isunderstandably difficult. PREPA’s version of this compliancewas to state that “PREPA'scapitalimprovementprogramisconsistentwiththeIRPinthatitcontainsprojectssuchasAOGP to transitionPuertoRico's generationmix toNaturalGas, replace inefficient agingunits, bring PREPA into compliancewithMATS [Mercury and Air Toxics Standard], andimproveoverallsystemreliability.”

PREPAthenmarkedprojectsaseitherbeing“CAPEX”or“IRP,”where“CAPEX”projectsareassociatedwithPREPA’svintage‐2014MajorCapitalProjectsplanforFYs2015and2016.184Forprojectslabeledas“IRP”itwasnotalwayscleariftheyemergedfromtheIRPprocess,orweresimplyconsistentwiththeutility’spreferredresourceplan.

While the requirement to explain how projects in the capital expenditure schedulecorrespondtotheIRPwasambiguouslyworded,areasonableinterpretationisthatPREPAshouldhaveexplainedhowitscapitalspendingstrategywasinfluencedbytheIRPprocessanddemonstratedthatitspreferredportfolioandresultantactionplanfromtheIRPwerereflectedinthespendingschedule.PREPA’sdemonstrationonthisfrontfallsshort.

ScheduleF‐3REVidentifiestwenty(20)transmissionprojectswithanFY2017costof$48.2million as associated with the IRP. There are no non‐transmission projects listed asassociatedwith the IRP, and the remaining transmission projects only amount to $10.5million.

OneoftheprimaryfeaturesofPREPA’sIRPwasthedecisiontoproceedwithbuildingtheAguirreOffshoreGasport(AOGP).AreviewofPREPA’shistoricspending,however,indicatesthatthedecisiontoproceedandbeginsubstantialinvestmentsinAOGPprecededtheIRPprocessbyasubstantialmargin,andthatsinceat least2011PREPAhasmadeaseriesofdecisionsembeddingitsassumptionofAOGPasafutureresourceintoitspresentspending,includingdeferringmaintenanceatAguirrestation,optingtoreduce investmentsatunits

183Section2.08(C)ofRegulation8720,NewRegulationonRateFilingRequirementsforthePuertoRicoElectricPowerAuthority’sFirstRateCase.



PREPAassumedwouldbereplacedbyimprovedgas‐firedgenerationfromAOGP,activelyshunningalternativeoptionstothegasport,optingtopursueairpermitsontheassumptionthat the gasport would be finalized, and generally betting the entire utility’s MATScomplianceonthegasport’stimelycompletion.

b. Spending at AOGP does not reflect Commissionrequirementforeconomicassessmentbeforeproceeding

In the Final IRP Order, the Commission disapproved AOGP, and called for a $15millionbudgetcapontheproject,subjecttoCommissionreview.TheOrderrequiredthatPREPAfileaneconomic analysisofAOGP (“AOGPEconomicAnalysis”),185 and thatPREPAcouldnotexceed the cap unless expressly approved by the Commission subject to review of theanalysis.PREPAdidnotfileaneconomicanalysis,requestpermissiontoproceedwithAOGP,seekclarificationontherequirementsoftheFinalOrder,orindicateitsintenttopursuesuchananalysis in the immediate future. Instead,PREPAmotioned forreconsiderationontheFinalOrder.

OnSeptember27,2016,theCommissionrequiredPREPAtore‐fileexhibitsandtestimonywith respect to revenue requirements (“27 September Order”). The Order specificallyrequired that PREPA’s rate case reflect the Commission’s findings that “spending onpermitting,planningandengineeringrelatedtotheAguirreOffshoreGasPort(“AOGP”)shallnotexceed$15millionfromthedateoftheissuanceoftheIRPOrder,”andfurtherclarifiedthat“thespendinglimitappliestothetotalcombinedspendingassociatedwithAOGPandthegasconversions.”186

Despitehavingamendedtheratecasefilingtohypotheticallycomplywiththe27SeptemberOrder, PREPA failed to adjust their revenue requirements to meet the Commission’srequirementthattheratecasebeconsistentwiththeIRPFinalOrder.Asoftoday,PREPA’spetitionreflectsastatedintenttospendapproximately$56million187onAOGPinFY2017.188PREPA’splannedspendingonAOGPisinconsistentwiththeCommission’srequirementthatPREPAcapitsspendingat$15millionuntiltheAOGPEconomicAnalysisisapproved,orceasetheproject.

5. Overviewoffindings

BelowwepresentourgeneralfindingsfromourreviewofPREPA’scapitalbudget.

185IRPFinalOrderB.1.a.(2)

186DocketCEPR‐AP‐2015‐001.Orderrequiringrevisedtestimony.Page2.

187ScheduleF‐3REV.

188WediscussAOGP,andPREPA’s treatment thereof in thisratecase, inmoredetail inSECTIONbelow.


a. PREPAbudgetprocessispoorly‐documented

i. PREPA’sprovisionofdocumentation tosupport itscapital budget demonstrated gross managerialincompetency

Asdiscussedabovegenerally,andaswewilldiscussindepthonindividualprojects,PREPAwasunabletoprovidebasicexplanations,workplans,orotherduediligencedocumentationon many of the capital projects contemplated in Schedule F‐3 REV. This failinginappropriatelylefttheburdenofbudgetreviewandjustificationtous,ratherthanPREPA.We were compelled to assess separate records of generator operations, forced outages,historicspending,andexplanationsscatteredacrossdozensofdisparatedataresponsestoevenbegintogainacoherentviewofthestateofPREPA’ssystem,theirneeds,andthevalueof the projects requested by PREPA. PREPA’s explanation that the workplans andjustificationfortheprojectsrequestedinF‐3REVwereinpaperrecords,“notorganizedbyproject”, and scatteredacrossPREPA’s systemgivesusextraordinarily lowconfidence inPREPA’sabilitytocoherentlyassesstheirownneeds,coordinateprojects,evaluatebudgets,andstrategicallydeploydollars.

Overall, our findings are tempered by our inability to review PREPA’s due diligencemechanism,assessifanysuchmechanismswereemployedbyPREPA,orevenascertainthenatureofsomeprojects.

ii. PREPA’sprovisionofdocumentationasrequiredbyPR87202.08wasdeficient.

PR 8720 2.08(d) requires that Schedule F‐4 “shall include a detailed description andreferencestosupportingdocumentationusedindevelopingtheprojections.”Withrespectto capital expenditures, PREPA’s F‐4 filing had four short paragraphs stating thatinvestments were necessary to maintain PREPA’s system and bring the utility intocompliancewithenvironmentalregulations.PREPAprovidednosupportordocumentationforitscapitalprojections.Overthecourseofthediscoveryprocess,wewereabletoelicitcontractswithbudget figures, internalpresentationswithbudgetestimates,andPREPA’sFY2015CapitalImprovementProgram.189Thesedocumentswouldhaveprovidedsomelevelof insight into PREPA’s process. Andwhile PREPA doesmaintainmany contracts on itswebsite, it is inappropriate to believe that the Commission should be responsible forconnectingthe242contractslistedonthatpagewithPREPA’smajorcapitalprojects.

WhilewearedisappointedbyPREPA’s inability toprovidesufficientexplanationof theirbudgetcreationprocess,weareunabletojustify—atthistime—recommendingasubstantial

189CEPR‐AH‐05‐08Attach02.FY2015ProgramadeMejorasCapitales.


reductionofPREPA’scapitalrequirementsbasedonafailuretoprovidedocumentation,asitisalsoapparentthatthesystemisindireneedofmaintenanceandbasicimprovements.

b. FY2017capitalbudgetdoesnotrepresentthefullscopeofprojectscommittedtoorcompletedinFY2017

PREPA’sbudgetsasprovidedinScheduleF‐3REVprovideanestimateofcapitalonanas‐spentbasis,ratherthananin‐servicebasis.Thismakesitverydifficulttoassesstotalprojectcostsforthepurposesofbenchmarkingbudgetsandexpenditurestoindustrystandards.

PREPAhasmultipleprojectsthatlandintheFY2017budgetthatstartedinprioryearsandarereachingcompletioninFY2017,aswellasprojectsthatarejuststartinginFY2017andareexpectedtobecompletedinafutureyear.ThereisnoplaceinwhichPREPAdisclosestheall‐incostofanygivenproject,anditwasonlybyexplicitlyrequestingprojectspendingfrom FY2015 to FY2020 that wewere able to discern the total project costs of specificinitiatives.

Overall,wefindthattheFY2017capitalbudgetdoesnotappropriatelyindicatethelevelofinvestmentrequiredbyPREPA,orcommittedtointhiscase.

c. Deferred spending has led to rapid deterioration ofPREPA’sgenerationfleetsinceFY2014

PREPA’s generating fleet has deteriorated rapidly over the last two fiscal years with adramaticallyincreasedpercentageoftimeinforcedoutages.190Whiletheunderlyingcausesof these outages appear to be widespread, many of the outages can be linked to thedeteriorationofphysicalplant,includingleakingpumpsandvalves,crackedducts,brokenandboilers,pipes,fans,andgates.

The maintenance problems at PREPA’s fleet have proceeded in lock step with PREPA’sdeclininggenerationandoperationalbudgets.Inaninternalbriefingfrommid‐2015,PREPAacknowledgesthattherehasbeenarecentincreaseofforcedoutagesduring2015,191atrendthatcontinuedwellinto2016.Accordingtotheutility,oneoftheprimaryreasonsPREPA’ssystemhasdeterioratedisthatithasnotbeenabletoacquirereplacementmaterialsduetobudgetconstraints,andhasdeferredsubstantialrepairspendingtheAOGPproject.192

190CEPR‐JF‐01‐16Attach02.Author’sassessment.

191CEPR‐JF‐01‐16Attach01(Public).Page3

192Id.Page6


A surprisingly large fraction of PREPA’s fleet has had substantialmaintenance problemsthrough2015and2016,drivingupsystemcostsinPREPAbecauseofanincreasedrelianceonhigher‐costunitspreviouslyreservedforpeakingandemergencypurposes.193

TheseobservationsleadustotheconclusionthatPREPA’srecenthistoriccapitalspendinghaslikelybeeninsufficienttomaintainasafeandreliablesystem.

d. Budgetprioritiesmaybedistortedbyartificialcap

PREPA’scapitalbudgetisdesignedtohitanapproximate$300milliontarget.ThistargetisPREPA’simpressionofbudgetavailabilitythatcanbeallocatedtocapitalprojects,basedonhistoriccollections.194Significantly,PREPAalsoallocatesbudgettodepartmentsonthebasisofprioryearspendingorbudget,whicheverislower.195Thisbudgetingprocessismoreinlinewithagovernmentagencywithlittleyear‐to‐yearvariationintotalbudgetallowanceorspending needs, and less in line with a business with substantial tangible assets underchanging fuel, environmental and demand conditions. Without a carefully considereddeploymentofthatlimitedbudget,thispolicyresultsindepartmentsthatareincentivizedtouseorlosebudgetsthatarenotfullyrequired,andlimitsdepartmentsthathavesubstantialneeds.

ThisratecaseprovidesanopportunityforPREPAtomakeacaseforspendingcapitaldollarstoprovidethepeopleofPuertoRicoafunctionalandefficientelectricitysystem.BothPREPAandtheCommissionhavearoleinthendeterminingifPREPA’sneedsarelegitimateandifPREPA’s dollars are appropriately deployed. Under the current mechanism, PREPA hasprioritizedprojects,butthereareindicatorsthattheprioritizationprocessisfaulty.

Forexample,PREPA’smovetocontinueandexpandanAMIprogram,ratherthanreallocatelimited dollars to failing generating stations is a significantlymisguided use of funds. Inaddition,aswillbediscussedinthenextsection,thebudgetcapandprioritizationprocessfavorstriagespending,ratherthanreasonablepreventativemaintenance.

Overall,inPREPA’scircumstance,theutility’sbudgetshouldreflectthespendingthatPREPAneedstomaintainandsupportthesystem,withoutallowinggenerationandtransmissiontofallintodisarray.

We findthatPREPA’scapitalbudget limitasapplied inthisratecase is inappropriate. Infuturecases,theCommissionshouldrequirePREPAtoclearlylayoutbudgetrequirementson the basis of requirement to meet minimum utility standards, clearly distinguishing


194October20ConferenceCall,discussionwithfinancialdirectoronbudgetcapsandpriorities.

195Id.


between spending for emergency purposes, preventative maintenance, systemimprovement,andsystemexpansion.

e. Budget prioritization process appears to favor brittleinfrastructure to the potential detriment of operationalinfrastructure

A review of PREPA’s budget priorities in generation and transmission and distributionindicatesthattheutility’sbudgetprioritiesarefocusedoninfrastructurethatiscurrentlyfailing—revealingaspendingpatternfocusedalmostexclusivelyontriage.While it isnotinappropriate to prioritize spending with the aim of ensuring an operational system,PREPA’sbudgetprioritiesmayinfactbeneglectingkeyoperationalinfrastructure.

Forexample,atPaloSecoPlant,PREPAisinvestinginthemaintenanceofPaloSeco3&4,hedgingthatEPAwillagreetoallowtheutilitytomaintaintheseMATSnon‐compliantunitsthrough2020.Aswewilldiscusslater,PaloSeco3&4havehadnumerousextendedforcedoutagesoverthelasttwoyears.PREPAisconsequentlyspending$8.5millionatthosetwounitsinFY2017.

Incontrast,PREPAhasearmarkedPaloSeco1&2as“limiteduse”unitsunderMATSwiththe expectation that these units will not contribute meaningfully to the PREPA system.Nonetheless,PaloSeco1&2haveproventobemorereliable,orat least failing in fewerhours,thanPaloSeco3&4.ThefactthatPaloSeco3&4areactuallyavailableforuseandcontributemeaningfullytoPREPA’ssystemsuggeststhatPREPAmaywanttoensurethattheseunitsarekept inworkingorder,even if theyareonlymeant toprovidepeakingorbackup power to the system. PREPA did not identify any capital dollars spent onmaintenanceatPaloSeco1&2overthelastsixyears196nordiditdisclosethelasttimethattheseunitswereprovidedasignificantmaintenancecycle,197suggestingthattheutility iswillingtoallowtheseunitstosimplygooutofserviceduetoneglect.

WefindthatPREPA’sbudgetprioritizationprocessmayfailtoprovidesufficientcapitalforinfrastructurethatPREPAstillreliesonandprovidesservicetotheutility.

f. MostofFY2017AOGPbudgetshouldbedeferredtoFY2018incompliancewithCommission’sIRPFinalOrder

In this report, we extensively discuss PREPA’s decisions to move forward on the AOGPprojectinspiteoftheCommission’srequirementtosubmitanAOGPEconomicAssessmentorbecappedata$15millionbudget.Inlightofthefactthatwe(a)havenoAOGPEconomic

196CEPR‐AH‐06‐12Attach01



Assessmentwithwhichtojudgetheprudenceoftheproject,and(b)PREPAhasshownnosignsofimpendingsubmissionofsuchanassessment,webelievethatthiscapstillapplies.

Further,PREPAhastodatespentonlyaminisculefractionofitsanticipatedFY2017budgetforAOGP,andarecentdiscoveryresponseindicatesthatitdoesnotanticipateanyfurthermajorexpendituresuntilitsignsthenoticetoproceed,currentlytargetedforearly2017,198orthefirstquarterof2018,199dependingonwhichPREPAsourcewerelyupon.WebelievethatPREPAhasadequatetimetofileanAOGPEconomicAnalysisanddemonstrate if theAOGPprojectiseconomicallyviablewithoutsubstantiallyinterruptingthecurrentschedule.Assuming that PREPA is cooperative inmoving such an analysis forward efficiently, theprojectcouldencounternodelaysfromtheCommission’sreview.However,ifPREPAfailstosubmitsuchadocument ina timely fashionand isheldtothe$15million limit, thismaycausetheutilitytopushbackcurrentdeadlines,regardlessofthestatusofpermitsorotherlogisticalhurdles.

Werecommendthat theCommissionrevisePREPA’srevenuerequirement forFY2017toreflecta$15millionspendingcapatAOGP,andfindthatdoingsoisunlikelytosubstantiallyimpact the current project schedule. Maintaining this $15 million cap reduces FY0217revenuerequirementsby$41,339,807.

g. Maintenance contracts at San Juan Combined Cycle andCambalache should be allocated to operations andmaintenance

OurassessmentofPREPA’scapitalbudgetrevealedtwolong‐termmaintenancecontractsfor San Juan Combined Cycle (“CC”) and the Cambalache Gas Turbines (“GT”). Thesecontractsprovideforoversightstaff,regularinspections,maintenanceequipment,andtheprocurementofgeneratorreplacementparts.ThesecontractsprovidemaintenanceservicestoPREPA.

For the purposes of this rate case, we characterize these maintenance contracts as anoperationalexpenseratherthanacosttobecapitalized.Theremay,infact,becostsunderthese contracts that represent legitimate capital expenses—i.e., long‐lasting replacementparts and new equipment—but the contracts themselves generally do not specify theprovision of those parts at specific costs, and thus we believe that it is appropriate tocharacterizethecostsofthesecontractsasanoperationalmaintenancecostthathasbeenoutsourced.

WefindthatPREPA’smaintenancecontractsatSanJuanCCandCambalacheGTaremoreappropriatelycharacterizedasoperationsandmaintenanceexpenses,ratherthancapital.

198CEPR‐JF‐04‐02Attach01.pdf.ExcelerateAguirreDetailedDesignSchedule,September2016.

199CEPR‐JF‐04‐02(d)


We recommend that the Commission recharacterize these capital expenses as O&M, anadjustmentwhich shifts a total of $16million in FY2017 from the capital budget to thegeneration‐relatedoperationalexpensebudget.

h. Transmissionbudgetisreasonable

PREPA’sprimarycapitalbudgetfortransmissioninFY2017isfortherehabilitationoftwo230kVlinesandone115kVlinethattogetherformtheprimarynorth‐southcorridorfromAguirreandAEStoSanJuan,andtherehabilitationoftwo115kVlinesfromtheCambalacheplantinArecibobacktoSanJuan.Ourassessmentofthecostandscopeofthesebudgetsarein line with other utility estimates for line construction and rehabilitation. Much of theremainderofPREPA’stransmissionbudgetisallocatedsmall,singleitems,manyofwhicharedifficulttoindividuallyassess.However,thefactthatthetransmissiondirectoratehashistorically stayedwithin budgets gives us confidence that the overall budget request islikelyreasonable.

We find that PREPA’s capital budget request for transmission is likely reasonable andrecommendthattheCommissionapproveitwithoutmodification.However,PREPA’sbudgetdoesnotaccountfortheSeptember21st2016fireattheAguirresubstation,andsubsequenttransmissionfailureatbothofthe230kVlinesemergingfromthatsubstation.Whileitisnotnecessarilyappropriatetocontinuouslyupdateabudgetrequestwithinafiledratecase,thisoutagerepresentsasignificantenoughoccurrencethatitbehoovestheCommissiontogainathoroughunderstandingofitscausesandimplications.

WerecommendtheCommissionrequirePREPAtopresenttheresultsof its investigationintothisoutagetothisCommissionanddiscloseitsbudgetrequirementsfromthatevent.

i. Distributionnon‐meterbudgetisreasonable

PREPAdiscussesextensivelythatitsdistributionsystemisinasubstantialstateofdisrepairandrequiresrehabilitation.Inparticular,PREPAdisclosesthatpolesarerotting,insulationhasdegraded,andPREPA isspendingsubstantialbudgetondeferredmaintenance inthesystem.

Ourreviewofthedistributionbudgetshowsthatthelargestitemsaregenerally“blanket”estimatesforthepurchaseofequipmentonanas‐neededbasis.Withoutadditionaldetail,itisdifficulttoassessthereasonablenessoftheseblanketestimates.

PREPAplacesmeteracquisitionintoitscustomerservicedirectorate,whichwewilladdressbelow.

WefindthatPREPA’scapitalbudgetrequestfornon‐meterdistributionspendingislikelyreasonableandrecommendthattheCommissionapproveitwithoutmodification.


j. Meter acquisition budget is not reasonable, andmisallocatedtoadvancedmeters

InreviewingPREPA’scapitalbudgetplanwithrespecttometers,welearnedthatPREPAwasin the middle of an extensive, and expensive, advanced meter infrastructure (“AMI”)program.AMImeters,otherwiseknownas“smartmeters,”allowtheutilitytocommunicatebacktothecustomerorcustomer’smeteronanear‐instantbasis.Manyutilitiesthinkthattheeventualintegrationofsmartmeterswillallowtheutilitytodirectpowerflowsmoreefficiently,handledistributedgenerationmoreefficiently,fixproblemsonthedistributionsystemfaster,andimplementcomplexandinteractiveratestructures.PREPAwasunabletoprovidereasonablesubstantiationforthesmartmeterprogram,andatacostnearlydoublethecostofthecurrentadvancedmeterreading(“AMR”)infrastructurethatPREPAbeganimplementing less than a decade ago, we cannot find a reasonable justification for thisprogram.

PREPAengaged inanacquisitionof15,000smartmeters througha localwholesalerandreportedthatnearlyhalfofthefirstbatchinstalledwerenotwithinspecifications.Betweenthe extraordinary cost of even a well‐structured smart meter rollout, and the planningrequiredtoensurethatsucharolloutissuccessfulwecannotcondonethefurtherexpansionofthisprogramatthistime,oreventheacquisitionofadditionalsmartmeters.

WeproposeanadjustmenttoPREPA’srevenuerequirementtoaccountfortheutility’sneedtoreplacemeterswithoutinvestinginthissubstantialinfrastructure.

WefindthatPREPA’sallocationofbudgettonewsmartmetersorAMIismisplacedandlikelya poor allocation of funds given PREPA’s otherwise degraded state in generation,transmissionanddistribution.Werecommendthat theCommissionstrikethis itemfromPREPA’scapitalbudget.

C. GenerationandProductioncapitalbudget

1. Overview

The following section will describe PREPA’s capital budget request for generation andproductionfacilities.TounderstandPREPA’srequirements,weassessthestateofPREPA’ssystem,andhowPREPA’scapitalbudgetrequirementsaddresstheutility’smostpressingneeds.

Aswewillshowbelow,PREPA’sabilitytomoveforwardwithstrategicinvestmentsgearedtowardsacleaner,moreefficientfleetareseverelyhamperedbyastateofcrisis.Havingbeendrained of both monetary and experienced human capital, PREPA’s fleet appears to beoperatingonfumestoday.Drivenbyamonetarycrisis,poormanagement,andalackofaclear strategicvision,PREPA’sgeneration fleet requires substantial investment simply tostayoperational.


We will discuss PREPA’s crisis and budget requirements. It is not clear, however, thatPREPA’s anticipated budget is sufficiently geared to help the utility emerge from thisreliabilityshortfall.PREPA’scapitalrequestsshowatriagementality,butrisk leavingtheutility inacontinuously‐reactivemode,rather thanbeingable togetaheadof itscurrentpredicament.

Table7,below,showsPREPA’scapitalspendingbyplantforeachfiscalyear2015‐2020.Notethat2015and2016spendingonlyshowcostsassociatedwithprojectsthatstillexistin2017orbeyond.Thus,thistablemaynotrepresentallcapitalspendingatPREPA’sgeneratorsinFY2015‐2016,andnoinformationwasprovidedonaprojectspecificbasisforthoseyears.

Table 7. Capital spending at PREPA generators by fiscal year200

2015 2016 2017 2018 2019 2020

SteamGenerators

AguirreSteam $2.9 $7.4 $27.0 $23.4 $14.7 $3.2CostaSurSteam $6.4 $2.0 $12.1 $18.0 $22.2 $30.0PaloSecoSteam $0.5 $1.4 $8.5 $3.0 $2.4 $6.0SanJuanSteam $0.1 $0.0 $0.2 $0.7 $15.5 $0.0PaloSeco/SanJuan $0.2 $0.3 $0.5 $0.1 $0.0 $0.0

CombinedCycle

AguirreCombinedCycle $1.9 $3.8 $4.9 $4.7 $7.8 $6.5SanJuanCombinedCycle $12.1 $14.0 $13.0 $13.0 $14.8 $17.5

SimpleCycleGenerators

Cambalache $2.2 $2.7 $4.0 $4.0 $4.0 $10.0Mayaguez $0.3 $0.7 $0.6 $0.6 $1.0 $0.6Culebra $0.6 $0.2 $2.0 $0.0 $0.0 $0.0HydroGasPlant $1.7 $0.3 $3.4 $3.3 $5.4 $5.4

AllUnits $2.1 $3.9 $6.3 $6.5 $6.5 $7.4Substations $0.1 $0.3 $0.4 $0.3 $0.5 $0.4

Total $31.2 $37.1 $82.7 $77.6 $94.7 $87.0

With theexceptionof thehydrogasplant,Culebra,and thesubstationcategory that fallsundertheproductiondirectorate,allofwhichhave limitedinformationavailable,wewilldiscussthecapitalbudgetrequestandpertinentissuesateachofPREPA’sgenerators.

2. Aguirresteamunits

Aguirre Plant is a 1,492 MW plant near Salinas. Aguirre is comprised of threesubcomponents,201asteampowerplant(900MW),acombinedcyclepowerplant(520MW),andasimplecyclepowerblock(72MW).202TheAguirresteampowerplant,builtin1975,203


201FederalEnergyRegulatoryCommission.February2015.AguirreOffshoreGasPortProject:FinalEnvironmentalImpactStatement(FEIS).DocketNos.CP13‐193‐000andPF12‐4‐000.Section1.4.1.

202PREPABaseIRP,August17,2015.Table3‐1,andauthor’scalculationforsimplecyclepowerblockbasedonFEISandIRP.

203AguirrePowerPlant,PREPAwebsite.http://www.prepa.com/aguirre.asp


isdividedintotwoelectricalgeneratingunits(EGUs)of450MWeach,andfiredbyheavyfueloil(No.6).204

TheAguirrePlantcomplexisthelargestsingleplantinthePREPAsystem,makingupoveronequarter(26percent)ofPREPA’scapacity.205Over60percentofthecapacityatAguirrePlantisprovidedbythetwosteamunits(“AguirreSteamUnits”)whichare,individually,thelargestcentralstationgeneratorsonPREPA’ssystem.Combined,theseunitsrepresent16%ofPREPA’ssystemcapacity.

TheAguirre SteamUnits arenot compliantwithEPA’sMercury andAirToxics Standard(MATS),whichaffectsfourteenofPREPA’sgeneratingunits.PREPA’sstrategyforreachingMATS compliance at Aguirre is to switch the units from oil to natural gas through theconstructionoftheAOGPproject.

Since2012 theAguirre SteamUnitshaveoperatedat a roughly50%capacity factor (i.e.providing about half of their absolute generation available),206 although this value fellsubstantiallyinFY2015andFY2016duetoextendedforcedoutages.

Inthisratecase,PREPAestimates$27millionincapitalexpendituresattheAguirreSteamUnitsinFY2017.207TheindividualprojectsrepresentedbythesecostsinmanycasesextendthroughFY2018andFY2019. In totalover the threeyearspresented in thiscase,PREPAanticipates$65millionincapitalatAguirreSteamUnits.ThemostsubstantialprojectsattheAguirreSteamUnitsaregearedtowardstowardtherehabilitationoftheturbinesandboilersattheplants.

a. IssuesatAguirreSteam

i. ForcedOutagesatAguirreSteam

The forced outage rate at theAguirre SteamUnits are substantial. Forced outages occurwhenageneratoreitherautomaticallyturnsofforisbroughtofflineforamechanical,safety,orenvironmentalproblemorviolation.Theseoutagesrepresenttimeswhentheunitcannotcontributetocapacityorreliability,andcannotprovideenergytothesystem.Ahighforcedoutagerateisindicativeofaunitthatiseitherinneedofrepair,suffersachronicoperationalproblemorerror,orispronetooperatorerror.

From2012to2014(calendaryear),theAguirreSteamUnitsaveragedasevenpercentforcedoutagerate;onaverage,theunitsweresufferinganunexpectedoutageduring612hours,or

204PREPABaseIRP,August17,2015.Table3‐1

205Totalsystemcapacity5,659MW.PREPABaseIRP,August17,2015.Table3‐1


207ScheduleF3‐REV,sortedbyprojectperindexinCEPR‐AH‐02‐03.


twenty‐fiveandahalfdayseveryyear.208In2015,AguirreSteamUnit1sufferedanextendedoutage frommid‐July to the endof the year,209 startingwithpump failure and cascadingthroughaseriesof faults inthegeneratorandboilereconomizer.210AguirreSteamUnit2then suffered an evenmore substantial outage starting December 1st of 2015when theturbinesufferedafault.TheturbineremainedofflineunderastateofrepairthroughthelastprovidedrecordinAugust2016,oranoutageof246days.

Theseextraordinaryoutagelevelsresultedincombinedforcedoutageratesinof27percentin 2015 and 46 percent in 2016 – effectively out as often as not. A unit with thisextraordinarily low ability to meet demand requirements due to outages has very littlecapacity value to the system. So, while the Aguirre Steam Units are the largest units inPREPA’ssystem,theyarealsoextremelynon‐reliable.

Inanundatedinternalbriefingdocument,PREPAexplainedthatoutagesin2013and2014atAguirre1wereprimarilycausedbecause“delays inpayingvendor lead to [a]delay inreceivingmaterials.”211PREPAdescribestheproblemsencounteredatAguirre2asa“failurein[the]transformer,”and“naturaldamageofmaterialsasaresultofnormalwearandaging.”Whilethefailuretoreceiveparts212andnormalwearandtearmayhavecausedoutagesin2013and2014,theextendedoutagesin2015and2016appeartobeaseriesofcascadingoutagesinwhichonefailureledsequentiallytoanotherfailure.

PREPA’s logofoutages213 tellsastoryofmaintenance failuresandextremedowntimeforsingleissueevents.Forexample,inearlyJuly2015,afeedwaterpump214atAguirre1brokedown,requiringPREPAtobringtheboilerdowntohalfoperationforsixdays.Fourteendayslater,PREPAtookthewholeunitofflineforadayandahalftofixaleakinthesameboiler.Forthenextfullmonth,PREPAkepttheunitathalfcapacitywhilethesamefeedwaterpumpremainedoutofservice.Justfourdayslater,nowinearlySeptember2015,theturbinebeganreportinganelectricalfailure,takingtheunitofflinefortwodays.Theunitwasbroughtbackonlineforninehours,onlytotripoffagainwhenthegeneratorfailed,takingtheunitofflineforexactlythreemonths,untiltheendofNovember2015.PREPAsuccessfullybroughttheunitbackonlinefor15hours,onlytohavetheboilerfailagain–thistimeintheeconomizer.Havingresolvedtheboilerissues,twodayslaterthefeedwaterpumpfailedagain,takingthe

208CEPR‐JF‐01‐16Attach02

209July7th,2015toDecember31st,2015

210CEPR‐JF‐01‐16Attach02.Logofforcedoutages.

211CEPR‐JF‐01‐16Attach01.ForcedOutage(“FO”)Analysis:BusinessCase.Undated.(Public).Page14.

212Id.

213CEPR‐JF‐01‐16Attach02.Logofforcedoutages.

214Feedwaterpump:apumpdesignedtomovefreshwaterintotheboilersystem.


unitofflineuntilDecember31st,2015.Overthesix‐monthperiod,Aguirre1wasavailabletogenerateforlessthanonemonth’sworthoftime,operatingathalforlessofitscapacityfortwomonths,andcompletelyunavailableforthreemonths.

Such sequential failures are not a function of normal wear and tear or aging, but areindicativeofsystematicmaintenancefailures,afailuretoperformpredictivemaintenance,operational errors, and faulty repairs. These failures indicate either a unit under severedisrepairoraninabilityofstafftomeetminimummaintenancerequirements.

ii. RelationshiptoAOGP

TheAguirreSteamUnitsareacornerstoneoftheAguirreOffshoreGasport(AOGP)project.AsPREPAexplainsintheMotionforReconsiderationoftheCommission’sFinalOrderintheIRPdocket,215theAOGPprojectisprimarilydesignedtohelpbringPREPAintocompliancewithMATS. Specifically, theAOGPproject ismeant toprovidenatural gas to theAguirreSteamUnits,whicharetheonlyunitsattheAguirrePlantcomplexwhichrequiremitigationtocomeintocompliancewiththerule.

Inthisratecase,PREPAhascoupledtheconversionoftheAguirreSteamUnitsfromoiltogaswiththeconstructionoftheoffshoregasportandassociatedfacilitiesasasinglecapitalexpenditure.ButthereareothercapitalexpendituresassociatedwithkeepingtheAguirreSteam Units operational through and beyond the conversion, including substantialexpendituresontheboilersandturbinesoftheseunits.

ThechoiceofwhethertospendsubstantialdollarsrevitalizingtheAguirreSteamUnitsiscomplicated.PREPAindicatesthatmuchofthespendingatAguirrehasbeendeferredwhilePREPA“waitedfor[the]gasconversion[fromAOGP]beforeinvestinginmajorrepairs.”216ThenetresultmayhavebeenthattheAguirreSteamUnitsbegantofailwhiletheprogramwas delayed, resulting in substantial deferredmaintenance.On one hand, PREPA shouldinvestincapitaldollarstoensurethattheAguirreunitsremainonlineduringcriticalperiods,andmakeupforcriticaldeferredmaintenanceprojects.TheoutagerateoftheAguirreSteamUnitsisunacceptable,strainsPREPA’sgrid,andrequiresPREPAtooperatemoreexpensiveunits on an ongoing basis. On the other hand, it may be unwise for PREPA to spendsubstantial capital dollars on life extension projects at the Aguirre Steam Units if theCommission concludes that theAOGPproject shouldnotproceed and theAguirre SteamUnitsrequirerelativelyrapidreplacement.

215PREPA’sMotionforReconsiderationofProvisionsoftheFinalResolutionandOrder,October13,2016,CEPR‐AP‐2015‐0002,paragraphs23,24,and28.

216 CEPR‐JF‐01‐16 Attach 01. Forced Outage (“FO”) Analysis: Business Case. Undated. (PublicVersion).page7.


UnderPREPA’sIRPplan,theutilityexpectstoretiretheAguirreSteamUnitsby2026and2027,respectively.217Withthisplan,itwouldmakesensetocontinuecapitalinvestmentstoensure reliable operation for another decade. However, if the AOGP Economic Analysisindicates that AOGP should not be pursued, PREPA should seek to make minimalinvestmentsattheAguirreSteamUnitstoensurethattheunitsareoperableandreliable,butnotinvestinacompleterenovationoftheunits.

ThepurposeofthecapitalexpendituresattheAguirreSteamUnitswithrespecttoAOGPwillremainunresolveduntilPREPAsubmitstheAOGPEconomicAnalysis.

b. CostofCapitalProjectsatAguirreSteam

Overall,PREPAanticipatesspending$27millioninnon‐AOGPspecificretrofitsattheAguirreSteamUnits inFY2017(seeTable8).AswithPREPA’sothercapitalprojects, this capitalrepresentsboththetailofspendingbeguninprioryears,aswellthestartofspendingforprojectsexpectedtoextendthroughFY2018and2019.Thesecontinuingprojects,aswellasafutureproject,areexpectedtorequireatotalof$65millionbetweenFY2017andFY2019.Combinedwiththenaturalgasconversionprojects,PREPAisplanningonspending$114millionincapitalontheAguirreSteamUnitsbetweenFY2017‐2019.Thespendingdiscussedhereonlyaccountsforprojectsthatarecurrentlyplannedinthenexttwoyears.Itislikelythat between now and 2019, PREPA will plan, scope and budget for additional capitalprojectsattheAguirreSteamUnits.

Table 8. Historic and expected capital spending at Aguirre Steam Units by Fiscal Year, millions of dollars.218

2014219 2015 2016 2017 2018 2019 2020AOGPUnit1&2Conversions 15.6 20.6 1.8 6.4 43.2 0.0 0.0BoilerImprovements 0.0 2.2 1.7 8.7 11.1 0.0 0.0TurbineImprovements 0.0 0.7 5.7 16.2 11.4 13.0 3.0BalanceofPlant 0.0 0.0 0.0 2.0 0.8 1.7 0.2TotalAguirreSteamUnitCapital 15.6 23.5 9.3 33.4 66.6 14.7 3.2

i. BasisforCapitalProjectsatAguirreSteam

217PREPABaseIRP(August17,2015).Table7‐4.

218Sources.DollarfiguresonAOGPfromCEPR‐AH‐01‐01Attach01(2017‐2020);CEPR‐AH‐05‐11(2014‐2016).Dollarfiguresforboiler, turbineandbalanceofplantfromScheduleF3‐REV(2017‐2019) & CEPR‐AH‐02‐02 (2015‐2020). Note that PREPA may have substantial historic capitalspending for boilers, turbines and balance of plant at Aguirre Steam Units, but did not provideinformationforspendingnotassociatedwithspecificcapitalprojectsidentifiedinF3‐REV.

219 Indicates 2014 or before. PREPA provided data only indicates spending in FY2015‐2017 andcumulativetodate.Author’scalculationtoderivepre‐2015spending.


PREPA’sspendingandexplanationsindicatethattheutilitystartedaseriesofsubstantialretrofitsof theboiler, turbineandgeneratoratAguirreUnit2 in2015(orbefore).Thesecostsappeartorepresentanoverhaulofkeyplantcomponents,includingarebuildoftheprimaryboilerheatexchangers,220afullrehabilitationoftheturbines,221andastatorrewindatthegenerator.222SeeTable9,below.

ItisnotcleariftheAguirre2projectsarerequiredtosimplyallowAguirretooperatereliablyover a short, near‐term basis or represent a large life‐extension effort. While PREPA’sexplanationfortheturbineandgeneratorprojectsbothindicatethattheyareforpurposesofextendingtheusefullifeofAguirre2,theturbinerehabilitationeffortcoincideswiththecompletefailureoftheturbineatAguirre2inNovember2015.223ThereforeitisclearthatPREPArequiresatleastminimumcapitaltokeeptheseunitsavailablewhiletheCommissiondeterminesiftheyshouldbecontinuedthrough2027,orreplacedmorerapidly.

According toPREPA’s records,Aguirre1will also receive substantial new retrofits at itsboiler,turbineandgenerator,amountingto$27.6ofspendingbetween2017and2019.Thedescription for these projects are effectively identical to the spending at Aguirre 2, andthereforethesamequestionsareunanswered:aretheseprojectsnecessaryifAguirreweretobemaintainedthroughaneartermreplacement–i.e.2021?

220CEPR‐AH‐06‐09.(b).1

221CEPR‐AH‐06‐09.(b).9

222CEPR‐AH‐06‐09.(b).8



Table 9. Historic and expected capital spending on individual projects at Aguirre Steam Units for projects with spending from

FY2017 to FY2019, millions of dollars.224

PID ProjectDescription FY2015 FY2016 FY2017 FY2018 FY2019 FY2020

Unit

1

15243 BoilerImprovement 0.4 ‐ 6.0 6.5 ‐ ‐14165 TurbineGeneratorImprovement ‐ ‐ 2.0 0.5 7.0 ‐14727 GeneratorImprovement ‐ 1.1 1.4 4.1 ‐ 3.0

Unit

2

14169 BoilerImprovement 0.8 1.1 2.1 4.6 ‐ ‐18651 TurbineGeneratorImprovement 0.1 3.6 6.0 ‐ ‐ ‐16749 GeneratorImprovement 0.0 ‐ 1.8 4.1 ‐ ‐

CommonExpenses

14151 SteamValveReconditioning 0.1 0.9 ‐ ‐ 2.0 ‐

14725BoilerSupportStructuresImprovement

1.0 0.6 0.6 ‐ ‐ ‐

13442 HPandIPRotorImprovement 0.4 ‐ 5.0 2.7 4.0 ‐14143 MainSteamCondenser 0.0 ‐ 0.9 ‐ ‐ ‐

15921CathodicProtectionandImprovement

‐ ‐ 0.1 0.1 ‐ ‐

14135WaterDemineralizerPlant&Polishers

‐ ‐ 0.7 0.4 ‐ ‐

14142 DischargeCanalMaintenance ‐ ‐ 0.4 0.3 0.5 ‐14125 WaterTreatmentPlantClarifiers ‐ ‐ ‐ ‐ 1.2 0.2

ii. ReasonablenessofCapitalProjectsatAguirreSteam

BasedontheextraordinaryoutagesatAguirre,itwouldappearthatsubstantialcapitalmay,in fact,be required tokeep theseunitsoperationalevenover thenear term.However, ifPREPAisunabletooperatetheseAOGPcornerstoneunitseffectively,thisunderminesthepotentialsavingsthatPREPAassertswillbemadeavailablethroughtheAOGPproject.

WehaveinsufficientrecordsandexpertisetoassessifthedollarvaluesbudgetedbyPREPAare commensuratewith the projects described here.However, PREPA’s total anticipatedspendattheAguirreSteamUnitsonadollarperkilowatt(kW)basis,averaging$24/kWfrom2017‐2019fornon‐AOGPprojects,arein linewith“run‐rate”capitaldollarsbudgetedforsteamcoalunitsatotherutilities,andarethereforenotapressingcauseforconcern.

c. Recommendation Regarding Capital Projects at AguirreSteam

We recommend the Commission approve PREPA’s FY2017 capital spending at AguirreSteam.

3. CostaSursteamunits

224ScheduleF3‐REV


CostaSurPlantisa990MWplantnearGuayanilla.TheCostaSurPlantiscomprisedoffoursteamboilerelectricalgeneratingunits, twosizedat85MWandtwosizedat410MW.225WhilenotdiscussedinPREPA’sotherliterature,PREPAmaintainstwosimplecycleturbinesattheCostaSursite,thecapacityofwhicharehavenotbeendisclosed.226TheCostaSurunitsaredesignedtobefiredbyheavyfueloil(No.6).227Since2012,CostaSurunits5&6,thelargerunitsattheplant,havebeenfiredwithapproximately60‐80%228naturalgasacquiredfromGasNaturalFenosa,themajorityowneroftheEcoEléctricapowerplant.Thefractionofgasversusoilburnedatunits5&6hasvariedovertime,butisrarelylessthan60%.

Figure 19. Natural gas fraction of fuel consumed at Costa Sur 5 & 6 since April 2016.229

PREPA’s2015IRPindicatedthattheseunitsburnaspecifiednaturalgasfraction,butclearlytheamountburnedattheplantvariessubstantially.TheIRPstatedthat“CostaSur5burned80percentofnaturalgasand20percentofNo.6fueloilandCostaSur6burns75percentofnaturalgasand25percentofNo.6fueloil.”230Withtheexceptionofonemonthinlate2013,thereisnotimeinwhichCostaSurhaveeitherhitthesetargetsorbeenrestrictedtotheselimits.

TheCostaSurPlantisthesecondlargestsingleplantinthePREPAsystem,makingupabout17percentofPREPA’scapacity.231ThesmallertwounitsatCostaSur,3&4,arenotMATScompliant.PREPAhasdesignatedtheseunits“limiteduse”asamechanismofmeetingMATS

225PREPABaseIRP,August17,2015.Table3‐1.



228CEPR‐AH‐06‐01.Percentagesbyequivalentheatcontent.

229CEPR‐AH‐06‐01.

230PREPABaseIRP,August17,2015.Table7‐5,footnote1.

231Totalsystemcapacity5,659MW.PREPABaseIRP,August17,2015.Table3‐1.


requirements.232 Under this designation, the units should remain below an eight percentcapacityfactor,startinginFY2016.InFY2016,PREPArequiredtheuseoftheseunits,withCostaSur3exceedingitsallowedutilizationwithan11.3percentcapacityfactor.233

ThelargertwounitsatCostaSurhaveoperatedwithanaveragecapacityof65%overthelastfiveyears,andarethereforethehighestutilizationunitsonthePREPAsystem.234Theseunitshavemaintainedanaverageavailabilityof97percentfromcalendaryear2012to2015,meaning that the duration and severity of their forced outages were relatively low.235PREPA’scapitalinvestmentsintheseunitshavehistoricallybeenataboutthesamelevelasatAguirre,buttheunitsareappeartobeinbettershape.Forexample,Aguirre’sSteamUnit1’slastmajormaintenancecyclewasin2012,236andCostaSur5&6receivedtheirlastmajormaintenanceoutagecyclein2013and2010,respectively.237ThiscyclestructuresuggeststhatCostaSurandAguirreshouldbeataboutthesamelevelofmaintenance,butCostaSurisinfarbettershapethantheAguirreSteamunits.

ThisratecaseenvisionsasubstantiallylowerlevelofinvestmentatCostaSurthanAguirre,even excluding theAOGP gas conversions at theAguirre SteamUnits. Overall, Costa Surappears to be preparing for a major overhaul in FY2018‐FY2020, with substantialinvestmentsinboilersandturbines.Inaddition,CostaSurisintheprocessofmodifyingitscoolingintakeanddischargesystemstomeetenvironmentalregulatoryrequirements.

a. CapitalProjectsatCostaSur

PREPAanticipatesspending$7.4milliononcapitalimprovementsatCostaSurUnits5&6inFY2017,alevelwhichwillincreasesubstantiallythroughFY2020238astheplantmovesintoaseven‐yearoverhaulcycle.

PREPAhasnotindicatedanycapitalexpendituresatCostaSur3&4,eventhoughtheseunitsareexpectedtoremainonlineandprovidingbackupcapacityinPREPA’slong‐termplans,andeventhoughtheywereutilized,albeitinalimitedfashion,inFY2016.ItisuncleartheextenttowhichtheCompanyexpectstomaintainorrelyontheseunitswithoutcontinuedmaintenancespending.

232PREPABaseIRP,August17,2015.Table7‐5,note3.


234Id.

235CEPR‐JF‐01‐16Attach02.Author’scalculations.


237Id.

238ScheduleF3‐REV.


Table 10. Historic and expected capital spending on individual projects at Aguirre Combined Cycle for projects with spending

from FY2017 to FY2019, millions of dollars.239

PID ProjectDescription FY2015 FY2016 FY2017

FY2018

FY2019 FY202016972 BoilerImprovementUnit5 3.1 ‐ ‐ ‐ 4.5 13.516973 BoilerImprovementUnit6 ‐ 0.2 2.7 7.0 6.5 ‐New SteamCoilsUnit5and6 ‐ ‐ 0.2 0.2 0.3 ‐New AirPreheaterBasketsUnit5and6 ‐ ‐ 0.6 1.0 ‐ ‐15923 WaterHeatersUnit5and6 0.2 1.1 1.5 1.7 ‐ ‐16974 TurbineGeneratorImprovementUnit5 1.1 ‐ ‐ ‐ 3.0 9.016975 TurbineGeneratorImprovementUnit6 ‐ ‐ 1.8 3.0 4.4 ‐16941 TurbineGeneratorBackUpImprovement 1.5 0.7 0.6 ‐ ‐ 4.0New FoxboroAutomationSystem ‐ ‐ 0.1 ‐ ‐ ‐16766 Mod.totheCoolingWaterIntakeStructure 0.5 0.1 1.5 3.0 3.0 3.016786 ModificationtoCoolingDischargeSystem ‐ ‐ 3.0 2.0 0.5 0.515029 BreakerReplacement‐480Vand4160V ‐ ‐ 0.2 0.2 ‐ ‐ Total 5.9 1.9 7.4 12.9 18.7 26.5

i. Basis forandReasonablenessofCapitalProjectsatAguirreSteam

PREPA’sspendingpatternssuggestthattheutilityistargetinganapproximateseven‐yearoverhaulatCostaSur5&6.ThecapitaldollarsfortheboilerandturbinerefurbishmentareinlinewithsimilarprojectsenvisionedbyPREPAandseenelsewhere.

PREPA’stotalanticipatedspendattheAguirreSteamUnitsonadollarperkilowatt(kW)basis, averaging $16/kW from 2017‐2019, are in line with “run‐rate” capital dollarsbudgetedforsteamcoalunitsatotherutilities,andarethereforenotapressingcauseforconcern.

b. RecommendationRegardingCapitalProjectsatCostaSur

WerecommendtheCommissionapprovePREPA’sFY2017capitalspendingatCostaSur5&6.

In addition,we recommend that PREPA track capital dollars spent at each “limited use”facilityseparatelythentheunitsthatareexpectedtobemaintained.ForCostaSur,PREPAshould trackandreportonprojectsatUnits3&4.Further,PREPA’s long‐termmodelingshouldconsistentlyassessiftheselimitedusefacilitiesareavailableforpeakpurposes,andif not, assess the value ofmaintaining units that neither contribute to peak purposes orprovide energy to the system. PREPA’s lack of maintenance spending at these units isinconsistentwithitscontinuedpeakuseofthefacilities.

239ScheduleF3‐REV


4. PaloSecosteamunits

PaloSecoPlant iscomprisedof four largeelectricalgeneratingunits inacomplexon thewesternsideoftheBahíadeSanJuanonToaBaja,abouttwolinearmilesfromdowntownOldSanJuan.PaloSecohasasimilar,butsmaller,setupthanCostaSur.PaloSeco1&2are85MW,andPaloSeco3&4are216MW.240LikePREPA’sotherlargesteamplants,PREPAmaintains simplecycle turbinesat thePaloSecosite, in this casesixunitsnototherwisediscussedintheliterature.241LikePREPA’sotherunits,PaloSecowasdesignedtobefiredbyheavyfueloil(No.6).242ThePaloSecoPlantisthefourthlargestplantinthePREPAsystemat602MW,makingupabout10percentofPREPA’scapacity.243

a. IssuesatPaloSeco

i. MATScomplianceatPaloSeco

NoneofthesteamunitsPaloSecoareMATScompliant,andPREPAhasnotdesignedaMATScompliancestrategytobringthePaloSecoplantintocompliance.WhilePREPAindicatesthatithasdesignatedthesmallertwounitsatPaloSecotobe“limiteduse”asamechanismofmeetingMATSrequirements,244thelargerunitsdonothaveaspecificcompliancestrategy.IntheIRP,theconsultantwritesthat“SiemensassumesthatPREPAentersintoasettlementagreementwithEPAregardingPaloSeco3&4steamunits(withatotalcapacityof432MW)allowingtheseunitstocontinueoperationburningNo.6fueloilthroughDecember31,2020,afterthatwillbeeitherreplacedordesignatedasalimiteduseunit.”245ThisstrategyreliesonPREPA’sabilitytoreplacethePaloSecounitsonanexpeditedbasis.

PREPA’sdesignationofthePaloSeco1&2unitsaslimitedusewouldrequiretheunitstoremainbelowaneightpercentcapacityfactor,startinginApril2015(FY2016).InFY2016,PREPArequiredtheuseoftheseunits,andbothPaloSeco1&2exceedingtheirallowedutilizationbyasubstantialmargin,withcapacityfactorsof39and44percent,respectively.246Considering that in prior years Palo Seco 1 & 2 have been operated as baseload orintermediateunits,withcapacityfactorsequivalentto,orexceeding,PREPA’sothersteam‐firedunits.ItisnotclearunderwhatcircumstancesPREPAwouldbeabletooperatetheseunitsundera“limiteduse”designation,consistentwiththeutility’scommitmenttoEPA.




243Totalsystemcapacity5,659MW.PREPABaseIRP,August17,2015.Table3‐1

244PREPABaseIRP,August17,2015.Table7‐5,note3.

245PREPABaseIRP,August17,2015.Section7‐5.



AsshowninFigure20,thecapacityfactorsofPaloSeco1&2haveexceededthatofPaloSeco3&4,whicharesupposedtobetheunitsPREPAisinvestedinasPaloSeco3&4comeoffline.Thisdoesnotappeartobethecase.

Figure 20. Capacity factor at Palo Seco units FY2012‐2016 and “limited use” designation for MATS compliance on Units 1 & 2.247

PREPA explains in discovery that, for the “limited use” units to be in service, the othergeneratingstationsoperatedbyPREPA,inparticulartheotherunitsatSanJuanandPaloSeco,needtobeingoodoperationalorder–whichiscurrentlynotthecase.PaloSeco3&4have had substantial outages, and PREPA does not expect Palo Seco 4 to be back in fullserviceuntilthethirdquarterofFY2017,orJanuary2017.248

ii. ForcedoutagesatPaloSeco3&4

Palo Seco 3 & 4 are a fundamental part of PREPA’s northern fleet, but have sufferedsubstantialoutages, inexcessof those seenat theAguirreSteamUnits. LikeAguirre, theforcedoutagerateatPaloSecohasincreaseddramaticallyin2014and2015.Betweenthefoursteamunits,PaloSecowasonlyavailableabouttwo‐thirds(65percent)ofthetimeincalendaryear2015,asteepdeclinefromitsavailabilityin2012and2013(97and94percent,respectively).

ItisconfoundingwhyaunitlikePaloSecocannotbekeptoperational,anditsoutageslikelypoint to largerproblemsat theplantwithmaintenanceand labor skills. FromDecember2014tothepresentday,PaloSeco4hassufferedsubstantialoutages.249StartinginDecember2014,steamturbinewasde‐rated(reducedinmaximumoutput)byabouthalfforabouttwo

247CEPR‐AH‐03‐07Attach01.Author’scalculations.

248CEPR‐JF‐01‐10(c).



monthswhile“highvibrations”causedoperationalconcerns. InFebruary2015,whiletheunitwasstillde‐rated,thesteamturbinerotorfailedandwas“severelydamaged.”250Afterassessingtheextentofdamage,PREPAtooktheunitdownforafive‐monthinspectiontoidentify possible additional damage, and began a full overhaul.251 After two months ofoperation,theturbineappearstohavefailedyetagain,andtheunitcameofflineagain.TheunithasnotoperatedthroughtheendoftherecordprovidedbyPREPA(toAugust2016),andhasremainedoutofservice.252Overall,PaloSeco4stayedonforcedoutageforoverayearandahalf,withonlytwomonthsofactualoperationinthattime.

PaloSeco3hashadasimilarlyspottyhistorywithaseriesofforcedoutagesbeginninginOctober2015.253Notably,PREPAreportsonly41daysofforcedoutagesin2012and2013,andnone in2014.Butstarting in2015,PaloSeco3began failingoften.First, thecoolingwater intake structure failed, followed by leaks in the boiler, both of which causedsubstantialde‐rates.ByNovember2015,theplantwasexperiencingdebrisandseaweedinthecoolingwatersystemandcracksintheboiler’sairducts,leavingtheunitatabouthalfcapacity.Finally,byMarch2016,thecirculatingwaterpumpshadfailedandtheplantwastakenofflineforhalfamonthforrepairs.Lessthantwoweeksaftercomingonline,anewseriesoffailuresbeganinthecondensatesystemandturbinecontrols.ThroughtheendofJuly2016,theunitcontinuedtooperateatalimitedoutput.

b. CapitalprojectsatPaloSeco

PREPAanticipatesspending$8.5milliononcapitalimprovementsatPaloSecoUnits3&4inFY2017.Mostoftheanticipatedexpendituresaretargetedforthisfiscalyear(2017),andthemostsubstantialprojectsaremeanttomitigatethecurrentdamageandforcedoutagesincurred at the plant. The largest single project, PID 13448 (“Turbine GeneratorImprovement”) isafulloverhauloftheturbineatPaloSeco4torestorethisunitbacktoservice.Thesecondlargestsetofexpendituresontravelingscreens, thecondenserwaterpump and improvements (PID 19639 and “New”) are designed to mitigate significantproblems in the intake water structures that get filled with debris and have causedsubstantialforcedoutagesatPaloSeco3.254

PREPAhasnotindicatedanycapitalexpendituresatPaloSeco1&2,eventhoughtheseunitsareexpectedtoremainonlineandprovidingbackupcapacityinPREPA’slong‐termplansas“limiteduse”units.InFY2016,despitetheunits’“limiteduse”designation,thesetwounitsproduced substantial energyatmoderate capacity factors. It isunclear if PREPAcan run

250CEPR‐AH‐06‐06(b)

251Id

252CEPR‐AH‐06‐06(a)


254CEPR‐AH‐06‐07(a‐c)


these units as “limited use,” given the requirement to have generation in the north andcontinuingoutagesatPREPA’sothernorthernpowerstations.

Figure 21. Historic and expected capital spending on individual projects at Palo Seco for projects with spending from FY2017 to

FY2019, millions of dollars.255


FY2018 FY2019 FY202013448 TurbineGeneratorImprovementUnit4 ‐ 0.3 3.0 ‐ ‐ 2.016969 BoilerImprovementUnit4 ‐ ‐ 0.3 ‐ ‐ 2.019660 PipingHangersReplacementUnit3and4 ‐ ‐ 0.3 0.3 0.3 0.516950 SteamCoilsReplacementUnits3and4 ‐ ‐ ‐ ‐ 1.2 1.219639 TravelingScreens‐PaloSeco 0.4 0.8 0.6 ‐ ‐ ‐New CondenserCirculatingWaterPump ‐ ‐ 0.7 ‐ ‐ ‐New CondenserImprovementUnit4 ‐ ‐ 1.0 ‐ ‐ ‐New LoadingPlatformforDEMIPlant ‐ ‐ 0.1 ‐ ‐ ‐12246 FeedwaterPumpTurbineImprovement ‐ ‐ ‐ 0.3 0.3 0.3New BunkerCReserveTankR3ConversiontoDiesel ‐ ‐ 2.0 2.0 ‐ ‐19621 WaterDemineralizationPlantImprovement 0.2 0.3 0.2 ‐ ‐ ‐New TreatmentPlantFilters ‐ ‐ 0.3 ‐ ‐ ‐New WasteTreatmentPlantNautilus1Replacement ‐ ‐ ‐ 0.4 ‐ ‐New WasteTreatmentPlantNautilus2Replacement ‐ ‐ ‐ ‐ 0.6 ‐ Total 0.5 1.4 8.5 3.0 2.4 6.0

c. Recommendationwith respect to capitalprojectsatPaloSeco

We recommend theCommission approvePREPA’s FY2017 capital spending at Palo Secosteamplant.

Inaddition,theCommissionshouldrequirePREPAtoprovideastrategicplanforPaloSeco,includingthefollowingelements,ataminimum:maintenanceplan,MATScomplianceplan,aninvestmentplanformaintaining(ornot)PaloSeco1&2,andareliabilitystudy–i.e.whatstrainsareplacedonthesystemin thepresenceorabsenceof thePaloSecosteamunits.

WerecommendthattheCommissionrequirePREPAtotrackandreportonprojectsatPaloSeco1&2,regardlessofiftheseunitsareconsideredlimiteduse.

5. SanJuanSteamunits

TheSanJuanSteamPlantsitsinSanJuanatthesouthernendoftheBahíadeSanJuan.Itiscomprisedof four100MWsteamunits256builtbetween1965‐1969,making it theoldest

255ScheduleF3‐REV



steamplantstillinactiveuseatPREPA.TheSanJuanSteamUnitsweredesignedtobefiredbyheavyfueloil(No.6).257

a. IssuesatSanJuansteamunits

i. ForcedoutagesattheSanJuansteamunits

LikePalo Seco andAguirre Steam, San JuanSteamhas seen a substantial and increasingforcedoutagesoverthelasttwoyears,althoughthereasonsfortheoutagesappeartobeavarietyoffailuresacrosstheindividualunits.SanJuan10wastakenoutofserviceinOctober2014anddueto“highvibrations”hasseenabouttwomonthsoffullservicesincethattimeinmid‐2015,withaneffectiveavailabilityofabout18%incalendaryear2015.Theunitisstillofflineandnotexpectedtoreturntoserviceuntil“thethirdquarterof2017,”orspring2017.258SanJuan9hasalsohadaseriesofoutagesinmid‐2015,buthasgenerallyremainedserviceableoverthelastsixmonthswithrelativelyminoroutagescomparedtoSanJuan10.

San Juan7&8havehad abetter track record thanSan Juan9&10,maintainingbetteravailabilityoverthelasttwoyears.ItisnotclearwhySanJuan7&8,whichhavegenerallybeen considered less viable units, have remained available while San Juan 9 & 10 havesuccumbedtosomanymaintenanceproblems.PREPA’sMATScompliancestrategystandsinstarkcontrasttotheconditionoftheseunits.

ii. MATSComplianceattheSanJuansteamunits

PREPAhasnoimmediatemechanismofmeetingMATScomplianceobligationsatSanJuanplantasidefromoperatingtheunitsatorbelowEPA’s“limiteduse”eightpercentcapacityfactor limit. In April 2013, PREPA submitted a letter to the Puerto Rico Planning BoardstatinganintenttodesignateSanJuan7&8as“limiteduse,”andpetitioningforleniencywith respect to San Juan 9 and 10. The letter explainedSan Juan9&10 couldnotbedesignatedaslimiteduseunitsbecausesuchareductionwouldhaveanadverseimpactontheelectricalsystemoperation.Theletterstates:

Theseunits areamong the largestoneson thenorth sideof the system. IncombinationwithSanJuanPowerPlantunits,theyprovidetherequiredloadgenerationbalanceforthereliableandstableoperationofthesystem,andalsothey have the largest frequency regulation range on the north coast of theisland. The power generation from these units is required tomaintain thereliability of the electrical system, especially under contingencies of thenorthern generating units. Based on the above, these units are consideredcriticalreliabilityunits.Theirretirementorgenerationreduction(limiteduse)


258CEPR‐JF‐01‐10(c)


willhaveanadverseimpactontheelectricalsystemoperationandreliability,as well as the people of Puerto Rico's wellbeing and national securitystability.259

Instead,PREPAofferedthatit“plan[ned]toconvertthesenitstoburnnaturalgasonaduelfuel scenariowith Bunker C fuel oil,” and that PREPAwas “conducting an evaluation toidentifythefeasibleinfrastructuretosupplyandmanagenaturalgastotheunitsinthenorthcoastoftheisland”withan“estimateddateofcompletionbythesecondquarterof2017.”260

DespiteaCommissionrequirement,PREPAdidnotprovidearesponsefromthePuertoRicoEnvironmentalQualityBoard(PREQB)orU.S.EPAoneitherPREPA’sdesignationorpetition.PREPA’splanstoextendnaturalgastothenorthwerenotseriouslyconsideredinthe2015IRP,261andwheretheywereconsidered,PREPAdidnotenvisionthisscenariountilJuly2022.

PREPA’spreferredscenariointhe2015IRPenvisionsthatSanJuan9&10wouldberetiredinDecember2020.262

Inthemeantime,SanJuan7&8,whichweresupposedtobedesignatedas“limiteduse”unitsforMATScompliancepurposes,havecontinued tooperateat capacity factorswell abovetheirregulatoryrequirement.Overthelastfiveyears,thesetwounitshavemaintainedanapproximate60percentcapacityfactor.

259CEPR‐JF‐01‐10Attach01.EarlyNoticeofCompliancePlan–MercuryandAirToxicsStandards(MATS)pages9‐10.

260Id.

261WhiletheIRPconsideredscenariosinwhichgasisavailableinthenorth,therewerenostudiestosuppor either the timing, infrastructure requirements, legal or permit requirements, costs, orviability. With respect to this option, PREPA states “Gas to the North presents an appealingproposition of enabling PREPA to build newor convert existing generation close to the demandcenters and reduce the level of dependence on the South‐to‐North electric transmission system.WhilegastotheNorthcouldpotentiallybeachievedviaLNGinfrastructureintheNorthoraSouth‐to‐Northgaspipeline,thefeasibilityofeitheroptionisyettobeevaluated.”PREPA2015IntegratedResourcePlan,Section6.3.1.

262BaseIRP.Section8.10.“Portfolio3Future1(P3F1).P3F1keydecisionsinclude:…6.SanJuan9&10andPaloSeco3&4willbeeitherretiredordesignatedtolimitedusebyDecember31,2020.”AlsoTable7‐5.


Figure 22. Capacity factor at San Juan 7 & 8 FY2012‐2016, relative to “limited use” designation for MATS compliance.263

These inconsistencies in PREPA’s claims and records leave this Commission with anunfortunatesetofuncertainties.ThefollowingquestionsareunresolvedbyPREPA’srecordandremainunclear:

1. WhatPREPAcandotoachievereasonableMATScomplianceatSanJuanPlant,ifatall.

2. WhatstepsPREPAistakingtoachieveMATScomplianceatSanJuan9&10.

3. WhatexpectationshavebeensetwithEPAwithrespecttoMATScomplianceatSanJuan9&10.

4. IfPREPAstillreliesontheassumptionthatSanJuan9&10willbeconvertedtonaturalgaswitha“gastothenorth”scenario.

5. HowtheextremelypooroperationalrecordatSanJuan10comportswithPREPA’sclaimthatthisunitiscriticalforreliabilityinthenorthofPuertoRico.

6. HowPREPAexpectstomeetthelimitedusedesignationforSanJuan7&8.

b. CapitalProjectsatSanJuanSteam

DespitetheextraordinarypoorreliabilityofSanJuan10andthestatedcriticalnatureoftheSanJuanplantaccordingtoPREPA’sclaims,theutilityhaslinedupveryfewcapitalprojectsattheplant.Infact,thelevelofexpendituresaresolowthatunderothercircumstancesitcouldsignalanintenttopotentiallyabandontheplant.

263CEPR‐AH‐03‐07Attach01.Author’scalculations.


Table 11 shows PREPA’s expected spending at San Juan. As shown in the table, PREPAexpectstospend$200,000atallofSanJuanSteaminFY2017,apaltrysumtosupportthereliabilityandoperationsofatheoreticallycriticalplant.At$200,000inFY2017,PREPAiseffectivelyspending$0.50/kW,oraboutanorderandahalfofmagnitudelessthansteamplants of this sizemight typically require. Spending in FY0218 is equally small. Only inFY2019doesPREPA’sexpectedcapitalspendingactuallytranslateintorealprojectspendingat$15millionforimprovementstotheturbinesandboilersatSanJuan9&10.

Table 11. Historic and expected capital spending on individual projects at San Juan steam for projects with spending from

FY2017 to FY2019, millions of dollars.264


FY2018 FY2019 FY202019477 TravelingScreensUnits7to10 0.1 0.0 0.2 0.2 ‐ ‐New AirPreheaterBasketsUnit7 ‐ ‐ ‐ 0.5 1.0 ‐New AirPreheaterSupportTrunionReplacementU‐10 ‐ ‐ ‐ ‐ 2.0 ‐New TurbineImprovementUnit9 ‐ ‐ ‐ ‐ 4.0 ‐New TurbineImprovementUnit10 ‐ ‐ ‐ ‐ 4.0 ‐New InstallHighPressureParts(SH,RH,Econ)Unit9 ‐ ‐ ‐ ‐ 4.5 ‐ Total 0.1 0.0 0.2 0.7 15.5 ‐

ThesespendingexpectationsatSanJuanareproblematicforavarietyofreasons:

First,thelackofspendingintheneartermforSanJuan9&10isinconsistentwiththestatedcriticalnatureof theseunits, the current extendedoutage at San Juan10 and the spottynatureofSanJuan9’sreliability. It isdifficulttoimaginehowPREPAexpectstomaintainunits with an extraordinary failure rate and expect to spend no capital dollars onreplacementpartsatthefailingunitsuntilFY2019.

Second,whileSanJuan7&8havebeendesignatedas“limiteduse”byPREPA,theyarestillabletoprovidepeakingservicesandhaveclearlyplayedaroleinsupportingPREPA’ssystemwhileSanJuan10hasbeendown.PREPA’scapitalbudgetsuggeststhattheseunitswillnotbe maintained at all. Again, like PREPA’s other limited use designated units, if PREPArequiresthecapacityfromtheseunitsduringcriticaloremergencyperiods,theunitsneedtobemaintainedataminimumlevel.IfPREPAabsolutelydoesnotrequiretheseunitsforreliabilitypurposes,theyshouldberetired.TheideathatPREPAcansimplyabandontheseunitsfromacapitalormaintenanceperspectiveandstillhopethattheywillprovideservicesispoorutilitypracticeandendangersPREPA’ssystem.

Third,PREPA’sexpectationsofsubstantialcapitalspendinginFY2019onSanJuan9&10areinconsistentwiththeIRP’sexpectationthattheseunitswilleitherbedeclared“limited

264ScheduleF3‐REV


use”orretiredbyDecember2020.Itisunclearwhythisspendingisdeferredandthenspentjustpriortotheseunitspotentiallycomingoutofservice.

c. RecommendationRegarding Capital Projects at San JuanSteam

We recommend theCommissionapprovePREPA’sFY2017 capital spendingat SanJuanSteam.

Inaddition,theCommissionshouldrequirePREPAtoprovideastrategicplanforSanJuan,includingthefollowingelements,ataminimum:maintenanceplan,MATScomplianceplan,aninvestmentplanformaintaining(ornot)SanJuan7‐10,andareliabilitystudy–i.e.whatstrainsareplacedonthesysteminthepresenceorabsenceoftheSanJuansteamunits.

WerecommendthattheCommissionrequirePREPAtotrackandreportonprojectsatSanJuan7‐10,regardlessofiftheseunitsareconsideredlimiteduse.

6. AguirreandSanJuanCombinedCycleUnits

TheAguirreandSanJuanCombinedCycle(“CC”)UnitsaretheonlycombinedcycleunitsinPREPA’scurrent fleet.LocatedattheAguirreandSanJuanplantsites,respectively, theseunitsarecomprisedofcombustionturbine(“CT”)unitsandheatrecoverysteamgenerators(“HRSG”).Theseunitsaredesignedtocapturewasteheatfromthecombustionturbineandharnessthatheatforthepurposesofcreatingadditionalgeneration.Bothsetsofcombinedcycleunitsburndieselfuel.

ThetwoAguirreCCunitseachhaveanameplatecapacityof296MW,butareconsideredtohave a 260 MWmaximum output by PREPA,265 for a total of 520 MW. Each CC unit iscomprisedof fourcombustionturbineunitsandasingleHRSG.266Theunitswerebuilt in1977andhave a veryhighheat rate (i.e. lowefficiency) for a combined cycleunit (11.1MMBtu/MWh).267

ThetwoSanJuanCCunitseachhaveanameplatecapacityof220MW,butareconsideredtohavea200MWmaximumoutputbyPREPA,268foratotalof400MW.Theunitswerebuilt

265BaseIRP.Table3‐11,footnote3.


267Inconsistentreports:BaseIRP.Table3‐1:~7,700btu/kW.URSJune2013AnnualReport.PREPAExhibit3.02(D)ConsultingEngineersReport.FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Page9:8,253btu/kWh.

268BaseIRP.Table3‐11,footnote3.


very recently, in 2008 and 2009. As such, they have relatively low heat rates (i.e. highefficiency)ataround8.0MMBtu/MWh.269

a. IssuesattheCombinedCycleUnits

i. MATSCompliance

OutsideofthenaturalgasfiredCostaSur5&6,theAguirreandSanJuanCombinedCycleunits are the rare fossil generating units in PREPA’s fleet that are compliantwithMATSbefore any retrofits or refueling options. Falling under the definition of stationarycombustionturbine,270thesecombinedcycleunitsarenotaffectedunitsundertherule.

ii. ForcedOutagesatAguirreCCs

PREPAdidnotprovideforcedoutagerecordsorestimatesfortheAguirreCombinedCycleunits.The2015IRPindicatedanappallinglyhigh20%modeledforcedoutageratefortheseunits.271ConsideringthatareviewoftheIRP’sestimatesagainsthistoricoutageratesoftenappearrelativelylow,andtheitseemslikelythattherecentforcedoutagerateoftheAguirreCombinedCycleunitsarelikelywellabovethatreportedintheIRP.

iii. CommissionIRPRequirementtoPursueRepoweringProjectAguirreCCs

Inthe2015IRP,PREPAdiscussedaplantorepowertheAguirreCCs,replacingtheunit’sturbineto increaseoutputandimprovetheheatrateoftheunit.Presumablythisretrofitprocesswould or could substantially reduce forced outages. The case for improving thisMATS compliantunitwas so compelling that theCommissionordered that “PREPA shallpursuepermittingandstartacompetitivebiddingprocesspursuanttoSection6B(a)(iii)ofAct83andtheJointRegulationapprovedbytheCommissionandPREPAtothateffect,fortherepoweringoftheAguirre1and2CCunitswithnew,dual‐fuelcapableturbines.272TheCommissionnotedthat“turbinereplacementwillallowthisfacilitytocontinueoperating,withmore flexibility.Turbinereplacements for theAguirre1and2combinedcycleunitscome at amoderate cost, but provide a 21% improvement in heat rate and a three‐foldincreaseincapacityfactor,”andfound“thatrepoweringoftheAguirreCCunitsisasoundinvestment.”273

269BaseIRP.Table3‐1.

27077FR9486,§63.10042

271BaseIRP.AppendixB‐1.

272FinalIRPOrder,VII(B)(1)(c)

273FinalIRPOrder,VI(A)(4)


PREPA’smodelingoftheirsystemfortheIRPandasprovidedinthisratecase274

NeithertherepoweringnortheplanningorpermittingcostsfortherepoweringarepartofPREPA’spetitionforarateincrease.

iv. GasConversionofAguirreCC

Inthe2015IRP,PREPAproposestoconverttheAguirreCCtooperateasagasfiredfacilitystartingin2018,withtheconversioncomingonlinewiththeAOGPproject.Intheratecase,PREPAties in thegasconversionof theAguirreCCwith theAOGPproject fromacapitalperspective,seekingtospend$46.6millionattheunitsforconversionwork.Asdiscussedinthatsection,PREPAhasnotyetsignedacontractforthiswork,buthasengagedininformaldiscussionsonthetopicwithGeneralElectric.

ThecostsofthegasconversionattheAguirreCCarenotdiscussedhere.

b. CapitalProjectsattheAguirreandSanJuanCCs

i. General

PREPA’scapitalbudgetsatAguirreandSanJuanCCsarebothverydifferentfromeachotherand tell a storyaboutPREPA’spriorities.AtAguirreCC,PREPAproposes to spend$17.8millionfromFY2017toFY2019,(seeTable12)oranaverageofabout$11.5/kW.Thebulkof these projects comprise scheduled maintenance and “automation” systems. PREPA’sspendingonscheduledmaintenanceinFY2017iscommensuratewiththeamountspentinFY2016,andyetaccordingtoPREPA’srecordstheitappearsthattheoutagesatAguirreCCkeep this unit from contributingmeaningfully to PREPA’s electrical system. In addition,PREPAseekstomakethegasconversionoftheAguirreCCakeycomponentoftheAOGPprojectandgasificationofthesouth.ItisunclearwhyPREPAwouldnotseektoimplementcomprehensiveprojectstoensurethattheplantisavailabletoprovidereliablepowerratherthanjustspendingonabusiness‐as‐usualschedule.

PREPA’sprojectsatAguirreCCappeartobeprimarilydrivenby individual improvementprojects,runbyPREPAstaff.

In contrast, at San Juan CC plant, PREPA seeks to spend $40.8 million from FY2017 toFY2019, (seeTable12)oranaverageofabout$34/kW.Thevastbulkof thespending isdirectedtowardsanexternalmaintenancecontract(PIDs16945&16946),discussedinthenextsection.ItisnotclearwhyPREPAallocatessuchalargerdollarfiguretowardstheSanJuanCCs, aside from the fact that theutilityholds an external contract thatmay requireminimumspendingtomeetcontractqualityobligations.



Overall,thedifferenceinspendingbetweentheCCprojectssuggeststhatPREPAispreparedtohaverelativelypooroperationsatAguirreCC(untiltheunitsarerepoweredin2020orbeyond) while San Juan CC is considered a key critical resource. This spending patternappearsto,again,beareactivemechanismbywhichPREPAreactively–onlywhenacriticalproblememerges–anddefersspendinguntilamajorprojectlikeAOGPcanbejustified.TheCompany’smodelingindicatesthatarepoweredAguirreCCunit,operatingwithalowerheatrateandbetterreliability,isakeymechanismbywhichPREPAcanretirePaloSecoandSanJuan steam units. Nonetheless, the utility’s spending seems potentially low on a unitcharacterizedwitha20percentfailurerate.

Table 12. Historic and expected capital spending on individual projects at Aguirre and San Juan CCs for projects with spending

from FY2017 to FY2019, millions of dollars.275

Unit PID ProjectDescription FY2015 FY2016

FY2017

FY2018 FY2019 FY2020

AguirreCom

bined

Cycle

14672 ScheduledMaintenanceStage1 1.0 2.4 2.1 2.3 2.1 2.07542 CO2CoolingSystemUnitsTurbine 0.0 ‐ ‐ ‐ 0.1 ‐ 16752 SteamTurbineandGeneratorImprovement 0.4 0.8 0.6 ‐ 4.8 4.015763 ValveReplacementandImprovementofHRSG 0.3 ‐ 0.3 0.2 0.3 0.3New FoxboroAutomationSystemforSteamTurbine ‐ ‐ 1.2 1.4 ‐ ‐ 16451 SeawaterCoolingTowerImprovement ‐ 0.5 0.4 0.2 0.2 0.216455 Make‐UpWaterSystemReplacement ‐ ‐ 0.2 0.4 ‐ ‐ 13534 NewTransformer500kVa4160/480 0.1 0.0 0.1 0.1 ‐ ‐ 16017 BatteryBanks ‐ ‐ ‐ 0.2 0.3 ‐

Total(AguirreCombinedCycle) 1.9 3.8 4.9 4.7 7.8 6.5

SanJuan

CombinedCycle

16945 CombinedCycleImprovementU‐5 6.7 8.3 6.0 6.0 6.0 6.016946 CombinedCycleImprovement U‐6 5.4 4.8 6.0 6.0 6.0 6.017039 ScheduledmaintenanceHRSG‐ Unit5 ‐ ‐ 0.3 ‐ 2.0 2.517040 ScheduledmaintenanceHRSG‐ Unit6 ‐ ‐ ‐ 0.3 ‐ 2.517041 SparesandMaintenanceComponents‐ U5&6 0.1 0.7 0.3 0.3 0.3 0.519500 CombustionTurbinesInsulation‐ U5&6 ‐ 0.2 0.3 ‐ 0.5 ‐ New TorqueConverters‐U5&6 ‐ ‐ 0.1 0.3 ‐ ‐ Total(SanJuanCombinedCycle) 12.1 14.0 13.0 13.0 14.8 17.5

ii. MaintenanceContractatSanJuanCC

The primary spending at San Juan CC is amaintenance contract fromMitsubishi‐Hitachi(“MHPS‐PR”)toservicethecombustionturbinesandgenerators(PIDs16945&16946).276Thislong‐termservicesagreement,signedinMarch2016,extendsandexpandsthescopeofservicesprovidedbyMHPS‐PRfromtechnicalsupportandassistance,asagreedtoattheendof2007, toa fullmaintenance contract.TheURS June2013AnnualReportdescribes thecontract:

275ScheduleF3‐REV

276ResponsetoOct202016clarificationcallCEPRRequestNo.4(supplemental).Long‐TermServiceAgreementbetweenPREPAandHMPS‐PR.


TheAuthorityhasalongtermmulti‐yearserviceagreementwiththecombustionturbinevendortoprovidetechnicaladviceandtoperforminspectionsofthecombustionturbinegeneratorsandthesteamturbinegeneratorsthatcompriseSanJuanUnits5&6.TheAuthorityisresponsiblefortheinspectionandmaintenanceofauxiliaryequipmentintheseunits.277

ThefullversionofthiscontractwasonlymadeavailabletotheCommissionaweekbeforetheexpertreportsweredue,andthusitisimpossibleatthisstagetoassessifthescopeofservicesunderthiscontractarereasonablefortheprice,andiftheaggregate$12millionperyearbudgetedbyPREPAisconsistentwiththecontract.

Itisclear,however,thatmost–ifnotall–oftheservicesprovidedunderthecontractareassociatedwithongoingstandardmaintenancecycles.Becausethiscontractisspecificallyassociatedwiththeregularmaintenanceofthegenerator,thecostsofthiscontractshouldlikelybeconsideredanoperationsandmaintenance(O&M)expenseratherthanacapitalcost.

Inspectionsandoverhaulsatpowerplantsaretypicallydividedintotiers,notdissimilartocarcompanies’recommendationsforserviceintervals.Minormaintenanceandinspectionsareperformedmultipletimeseveryyeartolookforoperationalproblems.Moresignificantmaintenancecycles,requiringthetemporaryidlingofaunit,aretypicallyscheduledforlowload seasons. Finally, major maintenance outages occur on multi‐year cycles, and aredesignedtooverhaulandreplacewornequipmentandkeepunitsinagoodstateofrepair.

The San JuanCC contractwithMHPS‐PRprovides for layers of inspections atmulti‐yearintervals,andcanbeexpectedtoprovideongoingserviceforseveralyears.Whilewedonothave evidence that the maintenance contract at San Juan is not currently effective, thecontract lacks any form of performance incentive or metrics. The contract appears toinsulatethecontractorfromperformancefailuresaswell.Whilethetermsofthecontractareconsideredproprietary,thepenaltiesimposedonthecontractorforoutagedelaysdonotappearcommensuratewiththecostofthosedelaysonPREPA.

The 2013 URS Report suggests that in 2011 and 2012 the contractor, who was thepredecessortoHMPS‐PR,failedtokeepthethenbrand‐newunitsonline.Atonlytwoyearssince commissioning, the generators at the San Juan CCs began deteriorating, failingsubstantiallythreetimesbetween2011and2013.

ThefailureofCT6’sgeneratorlateinfiscalyear2011hadcausedtheAuthoritytoputST6inreserveshutdownforeconomy.Infiscalyear2012itwasaccruingdaysinreserveshutdownforeconomywhenUnit5’ssteam

277 URS June 2013 Annual Report. PREPA Exhibit 3.02(D) Consulting Engineers Report. FortiethAnnualReportontheElectricityPropertyofthePuertoRicoElectricPowerAuthority,June2013.Page23


turbinegeneratorrotorfailed.ToreturnUnit5’ssteamturbinetoavailablestatus,theAuthorityinstalledtheUnit6generatorrotorintoST5’sgenerator….ThegeneratorrotorfromST5wassenttoamainlandfacilitytoberefurbished.ItwasreturnedtoPuertoRicoforinstallationinST6inAugust;thesteamturbinewasreadyforserviceinSeptember.Afterlessthan100hoursinoperationtherotorfailedagain.InOctobertheOEMremovedtherotorandsentitforrepairs.TherepairedrotorreturnedtotheislandinDecemberandthesteamturbinewasavailableforserviceinJanuary.278

(Unfortunately,thistrainofeventsasrecordedintheURSstudycannotbeseenintheforcedoutagerecordsprovidedbyPREPA,whichextendbacktoearly2012.)

ItisunclearhowarelativelynewunitliketheSanJuanCCsfailedsomanytimessoshortlyafter commissioning, and if the failures were due to ambient conditions, the quality ofmaterials available, operational problems or failure, staff failures, or poor maintenancecontracts.

c. Recommendation Regarding Capital Projects at theCombinedCycleUnits

WerecommendthatthemaintenancecontractatSanJuanCCberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$12millioninFY2017fromcapitaltoO&M.

WerecommendthattheCommissionapprovetheothercapitalexpensesatSanJuanandAguirreCombinedCycleUnits.

We recommend that the Commission examine the MHPS‐PR contract and theperformance ofMHPS‐PR to determine if the contractor ismeeting performanceexpectationsformaintenanceserviceatSanJuanCC.

Werecommend that theCommissionrequirePREPA to filenoticeofany long‐termcontractwithserviceproviderswithapotentialnetpresentvalueof$25millionvalueorhigher.

7. CambalacheandMayagüez

CambalacheandMayagüezaretwocombustionturbine(CTs,alsoknownas“gasturbines”or GTs) that burn diesel. Cambalache, near Arecibo on the northern central coast, iscomprisedofthreepowerblocksof83MWeach,or249MW.Itwasbuiltfrom1997‐1998.



Mayagüezstation,locatedinMayagüezonthewestcoast,iscomprisedoffourpowerblocksat~50MWeach,or200MW.279TheGTunitsatMayagüezaresomeofPREPA’smostrecentlybuilt,comingintoservicein2009.

PREPA did not provide a comprehensive record of forced outages at Cambalache orMayagüez.The2013URSReportdiscussedacriticalfailureattheCambalacheplantwhenacontrolsystemfaultledtothebuildupofunburntfuelinaturbine,leadingtoanexplosionthatseverelydamagedUnit1in2013.280ThesamereportdiscussesmoreminoroutagesatMayagüez, including an incorrectly installed turbine which required modification underwarrantee.

a. CapitalProjectsatCambalacheandMayagüez

PREPA’scapitalexpectationsforCambalacheandMayagüezarecomprisedentirelyofflatfees for “inspections” at Cambalache (PID 15880) and “improvement” atMayagüez (PID16978).

AtCambalache,theinspectionrepresentsanongoingservicecontractwithAlstom,valuedat$4millionperyear.

AtMayagüez,PREPAsimplyindicatesaflat$600,000peryear“improvement.”

WhilePREPAprovided theAlsommaintenance contract, it isnot clearwhatprojects areincludedundertheMayagüezumbrella,orifitsimplyrepresentsmaintenancespending.Itis also unclear why PREPA anticipates spending $4,000,000 per year at the older, lessefficient249MWCambalacheplant,andonlyone‐sixth(1/6th)ofthatcapitalatthefarmoreefficient, newerMayagüez station. This discrepancy reinforces the concern that PREPA’smaintenancebudgetsarenotwellbalanced,andputtheirmostvaluable,flexibleandreliableresourcesatriskthroughunderinvestment.

i. MaintenanceContractatCambalache

The primary spending at Cambalache is amaintenance contractwith contractor AlstomCaribe (now a division of GE Power). The twelve‐year contract, signed in May 2011, isdesignedtoprovideaninspectionandrefurbishmentofcombustionturbinesandgeneratorseverytwoandahalfyears.281Specifically,thecontractprovidesservicesforwhataretermed

279Conflictingevidence:URSJune2013AnnualReport.Page9:55MWeach.BaseIRP.Table3‐1:50MWeach.


281AlstomCambalache2011Contract,availableasCEPR161020RequestNo.4.Attach01.pdf.TermistoDecember31,2023orsix“Cinspection”cycles,whicheverisearlier.


“C Inspections,”performedat25,000operatinghours282–orapproximatelyevery two tothreeyears.

LiketheSanJuanCCmaintenancecontract,Alstomdividesmaintenanceintocycles,denotedas“A”through“D”inspections.“A”inspectionsoccurapproximatelyeverymonthandahalf(1,000hours)andincludepreventativemaintenance.“B”inspectionsoccureveryyearandhalf(12,500hours)andincludethedisassemblyoftheturbineunit forcloserreview.“C”inspections,everytwoandahalftothreeyears(25,000hours),includetherefurbishmentofthe turbine and combustion chamber. Finally, “D” inspections, every five to seven years(50,000 hours), entail the refurbishment or replacement of anyworn component in thegeneratororturbine.283ThemaintenancecontractatCambalacheisspecificallygearedtothe“C”inspectioncycle.

According to the URS June 2013 Annual Report, “under this contract the OEM [OriginalEquipmentManufacturer]provide [sic]a full time technical assistant (TA)during classCinspections and for the replacement parts needed in the hot gas path during class Cinspectionsofthecombustionturbine.” 284MaintenanceresponsibilitiesunderthecontractaresplitbetweenPREPAandAlstom,whereAlstomprovidesturbinecleaning, inspectionandrefurbishmentservices,butPREPA“employeesareresponsible for the installationofreplacementparts,”285andday‐to‐dayoperationsandsitemaintenance.286

The“C”inspectionsprovidedunderthiscontractwouldappeartofallintostandardongoingstandardmaintenancecycles.Becausethiscontractisspecificallyassociatedwiththeregularmaintenance of the generator, the costs of this contract should likely be considered anoperationsandmaintenance(O&M)expenseratherthanacapitalcost.

While the contract requires that Alstom provide a “permanent on‐site operations andmaintenanceadvisor,”287andprovidesa“technicalfieldadvisor”for“A”and“B”inspections,thecontractdoesnotactuallyspecifytheroleofthetechnicalfieldadvisor,wholeadstheinspectionandrefurbishmentprocess,andmostimportantly,whobearsresponsibilityforcorrectlyexecutedinspections,maintenance,andreplacement.

The contract recognizes that Alstom relies on PREPA staff formuch of the execution ofmaintenance,andspecificallyseekstoreduceAlstom’sliabilityforPREPAstaffnegligenceordeficiencies. Alstom included a contract provision that “excludes any and all liquidated

282URSJune2013AnnualReport.Page6

283URSJune2013AnnualReport.Page6‐7


285URSJune2013AnnualReport.Page26.

286AlstomCambalache2011Contract,Paragraph1.5andAppendix1.

287AlstomCambalache2011Contract,Paragraph1.1(h)


damagesforoutagescheduledelays,unlesssuchdelayisonehundredpercentattributabletoanegligentactoromissionofALSTOM(i.e.ALSTOMfailstodeliveracorrectpartormakeavailable the required personnel and such late delivery/performance causes an outagedelay).”288

SincePREPAdidnotprovidearecordofforcedoutagesatCambalache,includinganyreasonsforoutagesordelays,wecannotthoroughlyevaluatetheperformanceofthemaintenancecontractatCambalache.TheURSJune2013AnnualReportdescribesatwo‐yearoutageatCambalachebasedona control system failure resulting in anexplosion in the turbine.289WhileitisnotclearifthisfailureandthesubsequentoutageareattributabletoAlstom,itoccurredunderthepredecessorcontract.

TheCambalachecontractalsocontainsnoperformanceincentivesorpenaltiestokeeptheunits in operation or in a state of good repair. Alstom’s liabilities are limited to a smallfractionofthecostofthecontract,eveninthemostseverecases.290

b. RecommendationRegardingCapitalProjectsattheCTs

WerecommendthatthemaintenancecontractatCambalacheberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$4millioninFY2017fromcapitaltoO&M.

WehaveinsufficientinformationonthenatureoftheimprovementsatMayagüezstation,andalthoughwesuspectthattheserepresentstandardmaintenancecycles,thisclarificationmustbemadelater.Consideringthisuncertaintyandduetotherelativelysmallimpactofthis specific expense, we recommend that the Commission approve the $600,000FY2017improvementsatMayagüez.

We recommend that the Commission examine the Camabalache contract and theperformance of Alstom to determine if the contractor is meeting performanceexpectationsformaintenanceserviceatCamabalache.

D. AguirreOffshoreGasPortcapitalbudget

1. IntroductionandOverview

TheAguirreOffshoreGasPort(“AOGP”)isare‐gasificationfacilitymeanttoallowtheAguirreSteamandAguirreCombinedCycleunits(collectively“AguirrePlant”)accesstonaturalgasshippedtoPuertoRicoas liquefiednaturalgas(LNG).Thefacility iscurrentlyinthefinal

288AlstomCambalache2011Contract,Paragraph1.1(f)


290AlstomCambalache2011Contract,Paragraph8.3.


stages of permitting and engineering. Per PREPA’s records, the utility has spent limiteddollarson theport facility itself,buthasmadesubstantialmonetarycommitments to theproject and has advanced the process of converting the plants at Aguirre Plant to burnnaturalgas.Thisratecasecontemplates$56.3millioninFY2017and$413.3millioninFY2018capitalspendingforAOGP‐relatedprojects.

AOGP is supported by PREPA on four primary points. Per the Environmental ImpactStatement (EIS) prepared by the Federal Energy Regulatory Commission (FERC), AOGP“would[1]contributetothediversificationofenergysourcesinPuertoRico,[2]allowtheAguirrePlanttomeettherequirementsoftheEPA’sMercuryandAirToxicsStandardrule,[3]reducefueloilbargetrafficinJobosBay,and[4]contributetoenergypricestabilizationintheregion.”

TheAOGPproject iscomprisedof the installationof infrastructuremeanttosupportandtransportnaturalgasfromanoffshoreLNGre‐gasificationshiptotheAguirrePlant.Whencomplete, AOGP’s vendor, Excelerate, would dock a “Floating Storage and RegasificationUnit”(FSRU)attheoffshoreport.ArrivingtankerswouldtransferLNGtotheFRSU,whichwoulddecompresstheLNGonanas‐neededbasis,shippingthedecompressednaturalgasthroughanunderseapipelinetotheAguirrePlant.291

2. AOGPprojectboundaries

AOGP,asaddressedinthisratecase,referstofourdistinctbutinterlinkedprojects,aswellasotherbackofficeinitiativestobringtheprojecttofruition.Thefourprojects:

1. Theoffshoreproject:anLNGberthingplatformandsubmergedpipelinetotheAguirrePlantsite;

2. Theonshoreproject:apipelinefromJobosbaytotheAguirrePlantfacilities;

3. Combinedcycleconversionproject:theinstallationofnaturalgasburnersandcontrolequipmentatAguirreCombinedCycleUnits1and2;

4. Units1&2conversionproject:theinstallationofnaturalgasburners,boilermodifications,andcontrolequipmentatAguirreSteamUnits1and2.

ThedefinitionofAOGPasusedinthisratecasedifferssubstantiallyfromthedefinitioninPREPA’sIntegratedResourcePlan(IRP),docketCEPR‐AP‐2015‐002.IntheIRP,theAOGPprojectrefersexclusivelytotheoffshoreandonshoreprojects,whiletheconversionprojectsareconsideredseparately.Intheratecase,bycontrast,allfourprojectsarebundledunder

291FederalEnergyRegulatoryCommission.February2015.AguirreOffshoreGasPortProject:FinalEnvironmentalImpactStatement(FEIS).DocketNos.CP13‐193‐000andPF12‐4‐000.


oneumbrella.292WhiletherearedifferencesbetweencostsandtimelinesasproposedintheIRPandinthisratecase,careshouldbetakennottooverlookthesubstantialdifferencesbetweenthedefinitionsinbothcases.Forthepurposesofthisreportandcase,AOGPreferstotheentireproject,includingtheoffshoreandonshorecomponents,aswellasthenaturalgasconversionprojectsatAguirrePlant.

This section on AOGP details the Company’s justification for the AOGP project, the duediligenceandanalysesconductedbyPREPA,thereasonablenessoftheconstructioncontractforAOGP,thisCommission’sfindingsintheIRPproceedingwithrespecttoAOGP,andtheroleofthatfindingwithrespecttothisratecase.

3. Cost,contractingandfinancingatAOGP

a. CapitalcostofAOGP

AOGP, as used in this rate case, encompasses four major projects and back officecomponents.Asofthetimethisratecasewasfiled,PREPAanticipatedthecostofthetotalAOGPat$552million,includingfinancingcosts.Thetotalcostoftheprojectalsoincludesfinancing costs. The capital costs that PREPA associates with the AOGP project are acombination of vendor contracts and in‐house efforts.We discuss the contracts held byPREPAinthesectionbelow.

AsofJuly2016,thefilingofthisratecase,PREPA’sbudgetforthewholeAOGPprojectwas$517million,beforefinancingcosts,asshowninTable13,below.

Table 13. Aguirre Offshore Gasport capital budget – 2014, 2016

Project2016BudgetEstimate293

OffshoreProject 339,063,421OnshoreProject 27,345,940CombinedCycleGasConversion 46,637,480SteamUnitGasConversion294 87,490,350BackOfficeComponentsProfessionalServices 6,509,452MOUPermittingCosts 5,000,000PREPAPermits 164,012EnvironmentalJustice 2,000,000ProjectManagement 2,560,744Total 516,771,399 31,014,105

292PREPA’sresponsestoCEPR‐AH‐01‐02andCEPR‐AH‐03‐013.“ThefinancedcostofAOGPasusedin the IRP does not include the cost of performing fuel conversions at the Aguirre steam andcombinedcycleunits,buttheratecasevaluedoesincludetheseconversioncosts.”

293CEPR‐AH‐01‐01,andCEPR‐01‐01_Attach01.

294Includescontingencyat$5.15million.


Thisbudgetisdoesnotappeartobeconsistentovertime.

First,PREPAhasnotcontractedfortheconversionofthecombinedcycleunit,andonlyholdsestimatesfromGE,asdiscussedinthenextsectiononcontracts.Therefore,PREPA’sbudgetforthislineitemisstillunresolved.

Second,whenPREPAexperiencedaconstructionbudgetincreasefortheoffshoreprojectof$20million295(or$24million,dependingonthesource296)italsoreduceditscontingencyholdbackby$2million.Thisispoorengineeringpractice:anincreaseinbudgetshouldsignalthatthecontingencyshouldbemaintained,ifnotincreased–notdecreased.

Finally,PREPAisconductingotherprojectsattheAguirreSteamUnitsthatarespecificallydesigned to support the natural gas conversion, but are not clustered under the AOGPumbrella. Specifically, PREPA identifies three boiler improvements at Aguirre that areprerequisiteandcomplementaryprojectsforthenaturalgasconversion.297Atleastoneoftheseprojects,aheatexchangerreplacementatAguirreSteamUnit1,wasnotcontemplatedinPREPA’s2015majorcapitalprojects,298andaddsanother$12.5milliontotheprospectoftheconversions.299

PREPA’scommitmentsandinvestmentsunderAOGParenotlimitedtothecapitalcostsoftheproject.Aswediscussinthecontractingsectionbelow,thefinalizationoftheAOGPalsotriggers the hiring of the FRSU vessel from Excelerate, a 15‐year commitment at $40.7millionperyear–orapresentvalueof$422million.

Unlike theremainderof itscapitalprojects,PREPAalsoassumes itwillbeable tosecurefinancingformuchofthecostofAOGP.PREPA’sfinancingassumptionsadds$35milliontothecapitalcostsoftheAOGPproject,bringingPREPA’scurrentestimateto$552million.Thefinancingassumptionincludesbothinterestchargedduringconstruction(anequivalentofAllowance for Funds Used During Construction, or AFUDC) and a financing amountassumption.300 In the 2015 IRP, PREPA assumed a different financing rate, with 2%

295CEPR‐01‐01_Attach01,footnote2

296CEPR‐AH‐01‐04(h)(ii).“Theprimarydiscrepancy[betweentheratecaseandIRP]isa$24millionincrease in costs to fund theOffshoreProject,which can be attributedprimarily to the need forHorizontalDirectionalDrilling(HDD)ConstructionWorkandassociatedpermittingcoststoburythegaspipelineundergroundfromtheoffshoregasportthroughthebaytotheshoreline.”

297CEPR‐AH‐06‐09(b)(1‐3).

298CEPR‐AH‐06‐09(d).

299CEPR‐AH‐06‐09(a).

300CEPR‐AH‐01‐04_Attach01.


interest.301Inthisproceeding,PREPAhasassumed5%interestforfinancingAOGP.302ItisnotclearwhyPREPAchangeditsassumptionshere.

b. ContractingatAOGP

PREPAholds,andisplanningon,multiplecontractstosupporttheAOGPproject,includingcontractsforthedevelopmentoftheoffshoregasportitself(“InfrastructureAgreement”),303the gas conversions of the Aguirre Plant units, engineering, the development of theenvironmental impact statement (EIS) materials for FERC, the development andshepherding of permits through Puerto Rico and federal agencies, and legal services. Inaddition, PREPA has signed contracts for the operation andmaintenance of the gasportfacilitiesonceinplace(“TerminalOperationandMaintenanceAgreement”),304andthelong‐termhire of theFRSU facility (“TimeCharterParty andLNGStorage andRegassificationAgreement”,or“TimeCharter”).305

Whiletheauthorsofthisreporthavereviewedthecontracts,wearenotattorneys.Theremay be elements of these contracts that are not clear outside of the legal profession.Therefore,theseareopinionsbasedonalayreviewofthesecontractsanddonotrepresentanattorney’svieworopinion.

AlargemajorityofspendingatAOGPiswillbeincurredbycontractswiththreevendors,andfromPREPA’spersonnel.Thethreemajorvendorsandtheircontractsare:

1. ExcelerateEnergy,aTexasfirmspecializinginoffshoreregassificationofLNG.

a. MemorandumofUnderstanding(MOU),signedMay17,2011.AmendedthroughJanuary30,2014.306DesignedtoprovidepermittingmaterialstoFERC,developengineeringspecificationsforAOGP,andresultinacontractforserviceswithExcelerate(“DefinitiveAgreements”).307ThisMOUexpiredwhentheremainderagreementswithExcelerateweresignedonMarch17,2014.

301BaseIRP,VolumeI,Table5‐1.

302CEPR‐RS‐01‐04(b).

303ProvidedasPREPAResponseROIDRRCEPR‐AH‐03‐09_Attach01.

304ProvidedasPREPAResponseROIDRRCEPR‐AH‐03‐010_Attach01.

305MadeavailableintheIRPProcess(CEPR‐AP‐2015‐0002)inresponsetotheCommission’s3rdROI(March1,2016)

306ProvidedasPREPAResponseROIDRRCEPR‐AH‐03‐010_Attach01throughAttach11.

307PREPAResponseROIDRRCEPR‐AH‐03‐010_Attach11.


b. TimeCharterPartyandLNGStorageandRegassificationAgreement,signedMarch17,2014.ThiscontractisPREPA’sagreementtohireanFRSUfromExceleratefor15yearsatacostofabout$51millionperyear,308plusthecostoffuelandothercharges.

c. InfrastructureAgreement,signedMarch17,2014.ThiscontractisPREPA’scommitmenttopayfortheconstructionoftheoffshoregasportfacility,andrepresentedaminimum$266millioncostatthetimeitwassigned.309

d. TerminalOperationandMaintenanceAgreement,signedMarch17,2014withAguirreOffshoreGasport,LLC,asubsidiaryofExcelerate.Thiscontractprovidesthepersonnelforoperationsandmaintenanceattheoffshoreterminal.

2. AlstomCaribe(nowmergedwithGeneralElectric,orGE).ContracttoconvertAguirresteamunits1&2tonaturalgas.ThecontractforAlstomwassignedonNovember8,2013–wellbeforethecontractsforthegasportfacilitywerefinalizedorPREPA’slimitednoticetoproceedwasexecuted.310

3. GeneralElectric.ConversionoftheAguirreCombinedCycleUnits1&2tonaturalgas.AccordingtoPREPA,thereisnocurrentcontractwithGEfortheseservices,“PREPApersonnelhaveheldinformalconversationswithGE,”311and“noRFPhasbeenissuedfortheAOGPnaturalgasconversionfor[the]AguirreCombinedCycle[gasconversion].”312Nonetheless,onMay28,2014,GEprovideda$31.5millionproposaltoPREPAforalow‐NOxburner.313Atsomelaterpoint,GEprovidedPREPAarevisedproposalincludingcontrolequipment,amountingto$41million.314

PREPA’s disclosure of its proposal and contracting process for AOGP are lacking, and areview of themajor contracts at AOGP reveals substantial concerns with respect to theforward‐goingcommitmentsfacedbyPREPAirrespectiveoftheCommission’sdecisionsonAOGP.Theauthorsaskedforeffectivelyeveryproposal,bidandRFPconsideredbyPREPAwithrespecttoanyaspectofAOGPprojectconstructionoroperation—whetherdisclosedor

308Author’scalculation:TimeCharter,Paragraph9.1(“Hire”)(b)(i)at$111,500perday,andSchedule4Part2(i)at$29,016perday.

309InfrastructureAgreement,Schedule15.

310CEPR‐AH‐03‐08(b)(i).

311CEPR‐AH‐05‐12(a)(i‐ii)

312CEPR‐AH‐05‐12(c)

313CEPR‐AH‐05‐12_Attach02.

314CEPR‐AH‐05‐12_Attach01.Thereisnodate,sourceorothercontextprovidedwiththisdatasheet.


not. PREPA failed to provide many of its contracts, and in some significant cases couldprovidelittleevidencethataformalbidorcontractingprocesshadbeenexecuted.PREPA’stimelines for contract costs and projects often could not be squared against thedocumentationprovided.BelowwedetailsixexamplesofproblematictriggeringeventsandinconsistenciesinthecontractingprocessforAOGP:

i. InfrastructureAgreement

TheInfrastructureAgreement,signedbetweenPREPAandExcelerate’ssubsidiary,AguirreOffshore Gasport, LLC, sets up a series of triggering events that could lead to extensivecommitmentson thepartofPREPA; these commitments could technicallymove forwarddespite the Commission’s order to PREPA to cease spending until such time that theCommission authorizes the project. The Infrastructure Agreement requires that both aLimitedandaFinalNoticetoProceed(LNTPandFNTP,respectively)beexecutedbyPREPAshortlyafterallpermitsarefinalizedandoncefinancinghasbeensecured.OncetheFNTPissigned, the vendor is authorized to immediately beginwork and is authorized to collectdamagesifPREPAthenhaltsconstructionaftertheFNTPissigned.Inaddition,theFNTPtriggers a commitment to hire the FRSU from Excelerate, and PREPA incurs substantialliquidateddamages–$190million–ifthecontractisthenterminated.315TheInfrastructureAgreementdoesprovidePREPAtheopportunitytoavoidsigningtheFNTPifthereispendinggovernmentactionthatseekstorestrainorprohibitthetransaction,316butPREPAmay,atitssolediscretion(perthecontract),waivethatcondition.Forlackofaconditionsprecedentthat specifically requires PREPA to have received regulatory authorization to proceed,PREPAmaysoonbeinapositiontomakefurthercommitmentswithouttheauthorizationoftheCommission.

ii. TimeCharterAgreement

TheTimeCharteragreement,acontracttohireExcelerate’sFRSUforfifteenyears,requiresa daily fixed payment of $111,500,317 or $40.7million per year, and includes substantialliquidateddamagestowithdrawfromtheagreementonceauthorized.Thecontractcontainsa“hellorhighwater”provision,requiringthatPREPApayforthehireoftheFRSU“withoutregardto(i)theamountofLNGdeliveredtoEEforRegasification,(ii)whetherornotLNGdeliveriesareactuallymadebyPREPAor(iii)whetherornotPREPAisabletoreceiveorrequirestheuseofNaturalGasfromorbeyondtheShore‐sideNaturalGasDeliveryPoint.”

315TimeCharter,paragraph62.69(b)

316InfrastructureAgreement3.3(a)Conditionsprecedent…(v)“Thereisnopendingorthreatenedaction, suit or other proceeding brought by any governmental entity or other third party (A)challengingorseekingtorestrainorprohibitthetransactionscontemplatedbythisAgreementor(B)thatmayotherwisehaveanadverseeffectoftheabilityofPREPAtofulfillitsobligationsunderthisAgreementortousetheFacilities,ineachcaseinanymaterialrespect;”

317TimeCharter,paragraph9.1(b)(i)


This contract represents a substantial—and nearly irrevocable—investment that isotherwisenotdisclosedbyPREPAinfilingthisratecase.

iii. AguirreSteamConversion

PREPAidentifiesthatnearlytwo‐thirdsofthecoststoconvertAguirreSteamUnits1&2areforcontractswithAlstomCaribe,asubsidiaryofAlstomPower.318AlstomPower,acquiredby GE to form GE Power in November 2015, is a large multinational generationmanufacturingandenergyservicescompany.

PREPAprovidedone$18.5millionAlstomcontract,whichappearstobe forwhatPREPAcharacterizesas “boilermaterials.”Thiscontractwasexecuted inNovember,2013319andnearly fullypaidby January,2015.320Disconcertingly,however,PREPA identifiesanother$37.1 million in expected Alstom contracts, including for installation and “additionalmaterials,”whichareneitherdiscussednordisclosed.ItisnotclearifPREPAholdsproposalsorcontractswithAlstomforthisadditionalwork,andiftheidentifiedfuturecostshaveanyformofguarantee.

iv. AguirreCombinedCycleConversion

PREPAidentifies$46millioninexpectedcoststoconverttheAguirreCombinedCycleunitsto burn natural gas once AOGP is online.321 These costs begin occurring in FY2017—i.e.,withinthenextninemonths.Nonetheless,PREPAwasunabletoidentifyaformalRFPorbidprocess,indicatinginsteadthatPRPAhashad“informaldiscussionswithGE.”322Itisunclearwhy PREPA has identified GE as the conclusive vendor after only conducting “informaldiscussions.”Itisalsonotclearthattheidentifiedcostshereareinanywayguaranteedorstable.PREPAisexposedtosubstantialcontractpricerisk.Havingmadecommitments—andexpenditures—at other Aguirre units and on the offshore gasport in preparation for theconversion,buthavingsecurednopricecommitmentsatthecombinedcycleunits,PREPAleavesitselfexposedtothepotentialforpricegougingfromcontractors.

v. AECOM

318Author’scalculation.SeeAH‐01‐01Attach01.TotalcostofconversionforUnits1&2at$87.5million.Alstomcontractsaccountfor$55.6million,or64%.

319CEPR‐AH‐03‐08(b)(i).ContractprovidedasPREPAResponseROIDRRCEPR‐AH‐03‐08_Attach14(CONFIDENTIAL).pdf.

320CEPR‐AH‐05‐11(a)(ii).RemainingpaymentsweremadeinOctober2015onceboilermaterialsweredeliveredforAguirreUnit2.CEPR‐AH‐03‐08_Attach10,PDFpage10.


322CEPR‐AH‐05‐12


PREPAidentifiesthatinengagedtheconsultingserviceAECOMtoprovideenvironmentalservices, including theevaluationofPreventionofSignificantDeterioration(PSD) for theAguirre conversion to natural gas. This service, standardwhen a utility is considering apotentialmajormodification to a power plant,would have helped identify, and possiblyprepare,anypermitrequirementsorchangesrequiredbythePuertoRicoEnvironmentalQuality Board (EQB) or U.S. Environmental Protection Agency (EPA). AECOM is anestablished consultancy for these services, and these types of services are critical indeterminingwhattypesofmodificationsmaybedemandedbyenvironmentalregulatorsifalargecapitalprojectisundertaken.Nonetheless,PREPA’sdisclosureofthecontractisstillproblematic. The AECOM contract and two subsequent amendments were provided byPREPA,323 but PREPA failed to provide the attachments that detail the scope ofwork forservices, and PREPA’s contention that “the contract is not currently active”324 does notappeartocomportwiththeexpectedannualspendingscheduleunderthiscontract,whereinmostofthecostisincurredinFY2017andFY2018.325

vi. LTAutomation

PREPA identifiesLTAutomationprovidedcontrol systemupgrades to theAguirreSteamPlant,inacontractsignedNovember18,2011.326Ingeneral,thechanginganticipatedfuelmix, firing patterns and startup procedure for Aguirre steam units would have likelyrequirednewcontrolsystems,andingeneralitisnotunreasonabletoexpectanupgradeofthese systems in older plants to increase efficiencywhen other upgrades are also beingconsidered.However,thecontractprovidedbyPREPAonlyaccountsforonequarteroftheLTAutomationcontractspending‐$6millionoutof$24.5milliontotal.PREPAprovidednodocumentation for the other components listed under the sub‐header of LT Automationcontracts.327PREPAhasnotaccountedforasubstantialamountofcontractdollarsspentatAguirre.

PREPA’sAOGPcontractsrevealseveralsubstantialoverarchingconcernswiththePREPA’scontractingprocessingeneral,andspecificallywithregardstoAOGP.

1. PREPAisunabletoreasonablyaccountfordollarsspentuponrequest,afactthatshouldspeakforitselfwithrespecttoregulatoryoversightrequirements.Underall

323PREPAResponseROIDRRCEPR‐AH‐03‐08_Attach07,Attach08andAttach09.

324AH‐03‐08(a)(ii)

325AH‐01‐01Attach01.

326AH‐03‐08(a)(v).

327Whileitisfeasiblethatsomeofthecontractslistedunderthe“LTAutomationcontracts”headerarenotstrictlywithLTAutomation,andareinsteadonlycoordinatedthroughLTAutomation,wefind PREPA’s response to the requirement to provide the LT Automation contracts less thansatisfactoryorinappropriatelynarrow.


circumstances,PREPAshouldbeabletoprovideanaccurateaccountingofcontractdollarsspent,thecontract–includingscope–underwhichthedollarswerespent,andareasonableaccountingoftheprocessthatledtotheprovisionofthecontract.

2. PREPA’scontractsdonotcontainanyformofConditionsPrecedentrequiringregulatoryauthorization.Toallowforproperregulatoryoversight,PREPA’slargestcontractsshouldbebroughtbeforetheCommissionforauthorization,andshouldcontainconditionsprecedentallowingPREPAtoterminatecontractsthatfailtoreceiveCommissionauthorization.

3. PREPA’sAOGPprocessiscomplexandrequiresmultiplesimultaneousmovingparts,butcontainnomechanismsforsynchronizationandappearpoorlycoordinatedbetweendifferentvendors.ThefactthatPREPAtodayvergesonfinalizingahighcost–andirrevocable–contracttoreceivenaturalgasatAguirrewithintwoyears,buthasnotevensecuredacontracttoconvertakeysetofgenerators(thecombinedcycleunits)tonaturalgasisdeeplyproblematic.Underthesecircumstances,PREPAhasnoguaranteeforprice,onlinedate,orperformanceatoneofthekeygeneratorsthatwillreceivethenaturalgasuponwhichPREPAhasstakedsomuch.EvenattheAguirreSteamunits,PREPAonlyprovidedamodicumofthecontractsforthenaturalgasconversionprocess,andthefewcontractsprovidedhavenoguaranteeofperformanceoncetheconversionoccurs.WhilePREPAhasspentsubstantialeffortonsecuringinfrastructure,operations,andaregassificaitonvessel,theutility’scontractsshowlittlethoughttoensuringthatthegeneratingunitsarepreparedtoreceivethenaturalgas.

c. FinancingAOGP

Inthepreparationofthiscase,PREPAassumedthatitwillbeabletosecureafederalloanguaranteefromtheU.S.DepartmentofEnergy(“DOE”)underSection1703,TitleXVIIoftheEnergyPolicyActof2005.328PREPA’sapplicationtoDOErespondstoa2013solicitationforinnovative(i.e.non‐market)technologies,329andseeksaloanguaranteefor80%oftheAOGPproject costs,330 including offshore, onshore, and gas conversion projects.331 Under thisprogram,theDOEwouldprovideaguarantee,orbackstop,onaloantakenoutbyPREPAfromathirdpartybank.TheDOEloanguaranteeisnotfinalizedyet,andPREPA’slanguage

32842U.S.C.16511–16514

329SolicitationNo.DESOL‐0006303

330PREPAEx.5.0at909‐910.AlsoPREPAresponsetoCEPR‐AH‐01‐04_Attach01.

331FederalEnergyRegulatoryCommission.February2015.AguirreOffshoreGasPortProject:FinalEnvironmentalImpactStatement(FEIS).DocketNos.CP13‐193‐000andPF12‐4‐000.Section1.2.5.


with respect to the status of the loan guarantee application is substantially hedged intestimonyanddiscoveryresponse.

IntheCompany’sinitialfiling,theDOEloanguaranteeisdescribedinasingleclausebyaNavigantpanel,anddescribedas“thepotentialforfinancingofAOGPsupportedbytheDOEloanguaranteeprogram.”332AskedfordetailsabouthowtheCompanyhadcalculatedthefinancedcostofAOGP,PREPAprovidedasimplecalculationandprovidedadisclaimerthat“PREPAisnotassumedtoattemptorbeabletoobtainaconstructionloan.Currently,PREPAis assumed to issue debt under the DOE loan guarantee program at the end of FY2017,beginningofFY2018.”333FromPREPA’sresponse,itisnotclearifPREPAmeansthattheratecasedoesnotassumethataloanisprocured,orthattheloanguaranteeisnotspecificallyearmarkedfortheconstructionofAOGP.Neitheroftheseexplanationsaresound.PREPAis,infact,attemptingtoobtainaloanguaranteeforconstruction,asaffirmedbytheU.S.DOEintheAOGPFEIS,334andtheratecasedoes,infact,assumethataloanisprocured,deducting$413.4million fromFY2018 revenue requirements under the header “Investment CapexFinanced(AOGP).”335

In the Company’s supplemental filing, Mr. Quintana provides slightly more detail withrespect to the loan guarantee application,336 stating that the “fiscal component forqualification…couldbeapotentialissueforPREPA,”andthat“PREPAandtheDOEcontinuetomaintainadialogueaboutpotentialguarantees.”337Further,Mr.Quintanaindicatesthatacritical step inobtaining this loanguarantee is theexecutionof “adiligenceprocess thatwouldbecarriedoutbyathirdpartyadvisorandcouldcostseveralmilliondollars,”butthat“PREPAhasnotconsidereditappropriatetomoveforwardwiththatstepuntilitreceivedan approved IRP.”338 Finally, Mr. Quintana hedges substantially on acquiring the loan

332PREPAEx.5.0at909‐910.

333ResponsetoCEPR‐AH‐01‐04(g)

334FederalEnergyRegulatoryCommission.February2015.AguirreOffshoreGasPortProject:FinalEnvironmentalImpactStatement(FEIS).DocketNos.CP13‐193‐000andPF12‐4‐000.Section1.2.5.“PREPAapplied fora loanguarantee for theconstructionof theAguirreOffshoreGasPortProject(includingthenon‐jurisdictionalfacilitiesdescribedinsection1.4)asubseapipelineconnectingtheOffshoreGasPorttotheAguirrePlant,andconversionofmultipleelectricity‐generatingunitsandothermodificationsattheAguirrePlant.”

335ScheduleF‐3REV

336Exhibit13.00at156‐183

337Exhibit13.00at162‐168

338Exhibit13.00at168‐171


guarantee,statingthat“relyingon[the]DOEloanguaranteeisnottheonlypathtoaffordablefinancingfortheAOGPproject,ifitproceeds.”339

d. ImpactsoffinancingassumptioninRateCase

PREPA is seeking a DOE loan guarantee for 80% of the overall AOGP project cost. TheCompany“assume[sthatitwill]issuedebtundertheDOEloanguaranteeprogramattheendofFY2017[orthe]beginningofFY2018.”340PREPA’srevenuerequirementsareimpactedintwowaysfromthisassumptionset:

1. PREPA’scapitalcostschedule(F‐3REV)inaccuratelyallocatescapitalspendinginallyearstoensurethat80%ofthecostsareincurredinFY2018.

2. PREPAassumesthatitwillincurnocapitalcostsforAOGPinFY2018,despitetherelativelylargespendinganticipatedinthatyear.

i. Capital cost schedule in F‐3 REV inaccuratelyallocatesAOGPcapitalspending

PREPA’scapitalcostschedule(F‐3REV)isorganizedtodisclosePREPA’santicipatedcapitalspendinginFY2017,FY2018andFY2019.Importantly,sincePREPAassumesthatithasnoaccesstocapitalmarkets,capitallistedinthisscheduleisonan“as‐spent”basisratherthanonan“in‐service”basis.Therefore,onewouldgenerallyexpectthatallcapitalprojectslistedin this schedule are shown on an as‐spent basis. Because AOGP represents a very largefractionofPREPA’scapitalbudget–46%fromFY2017throughFY2019,inclusive341–andPREPA proposes to capital on an as‐spent basis into annual rates, it is critical that theproposedspendingbeascloselyalignedaspossibletoPREPA’sspendingschedule.Thisisnot the case for AOGP. Instead, PREPA appears to have allocated funds to FY2017 andFY2018byhowAOGPwillbefinanced,withFY2018fundsrepresentingthe80%projectcostexpectedtobesupportedbytheDOEloanguarantee,andtheFY2017fundsrepresentingtheremainderofprojectcostsnotyetincurredandnotunderthe80%umbrella.

Table14,below, illustrateshowPREPA’sbudgetprocessforAOGPwasdesignedtoplaceexactly 80% of budgeted costs in FY2018 – the exact amount PREPA anticipates beingfinancedthroughtheDOEloanguaranteeprogram.ThetablealsoshowsbothhowPREPA’scurrentspendingtodatedoesnotalignwithPREPA’sreportedbudgetinthisratecase,342

339Exhibit13.00at177‐179

340ResponsetoCEPR‐AH‐01‐04(g)

341ScheduleF‐3REV,sumofline6(“AguirreOffshoreGasPort(AOGP)Investment”)asafractionofthesumofline10(“TotalMaintenance&InvestmentCapex”).

342Note that actual spending fromPre‐FY2015 through FY2016 approximates PREPA’s reportedbudget for FY2016. This simplification indicates that PREPA’s response to CEPR‐AH‐01‐03 wasfactuallyincorrect.ThatrequestaskedthatPREPA“provideadetailedlist,withdescriptions,values


andhowPREPA’sspendinginthefirstquarterofFY2017indicatesthatPREPAisunlikelytomeettheirbudgetassessmentforFY2017.

Table 14. AOGP budget and spending

Pre‐FY2015 FY2015 FY2016 FY2017 FY2018

AOGPBudget343 $47,014,472 $56,339,808 $413,417,119

%ofwhole 9.1% 10.9% 80%

Spendingtodate344 $22,422,611345 $22,546,328 $2,041,360 $4,173 $0

%ofwhole 4.3% 4.4% 0.4% 0% 0%

A reasonable interpretationof these inconsistencies is thatwhatPREPA’sAOGPbudgetexpectationsarenotreliable,aremislabeled,andarelikelyadistortionfromPREPA’sloose financing assumptions. What PREPA labels as the FY2016 budget is, in fact, allspendingincurredpriortoJuly2016.WhatPREPAlabelsastheFY2018budgetistheamountthat theCompanyexpects to finance through theDOE loanguarantee.What remainshasbeenshuffledintotheFY2017budget,andevidencesuggestit isunlikelytobespentthisyear,regardlessoftheCommission’sFinalOrderontheIRP.AdditionalevidenceregardingAOGPspendinginFY2017willbediscussedlater.

ii. PREPAassumesthatnocapitalcostsareincurredinFY2018forAOGP

PREPA’santicipatedcapital revenuerequirements inFY2018arecompletelyoffsetbyanassumption that the Company will successfully obtain a loan guarantee from DOE inFY2018.346InPREPA’saccounting,allAOGPcapitalexpensesincurredinFY2018areoffsetby the financing offer. This assumption is not, and cannot, be considered a “known andmeasurablechange.”Rather,PREPA’scurrentplanis,intheabsenceofaCommissionorderrequiringtheCompanytostopwork,tocontinueintheconstructionandprocurementoftheAOGPproject,regardlessofitsfinancing.Indeed,PREPAhasmadeclearthat,initsopinion,AOGPmustbepursuedasa fundamentalcomponentof theCompany’sMATScompliance

anddates,ofallFY2016spendingassociatedwithAOGP.”PREPArespondedbycitingtoCEPR‐AH‐01‐01Attach01,whichindicates$47millionofspendinginFY2016.However,CEPR‐AH‐05‐11.a.ii‐a.ivindicatethatmostofthisspendingwasincurredinprioryears.

343PREPAResponsetoCEPR‐AH‐01‐01Attach01(FY2016‐FY2018)andF‐3REV(FY2017‐FY2018)

344PREPAResponsetoCEPR‐AH‐05‐11.a.ii,a.iv(SpendingtodateFY2015‐FY2017).AnsweredonSeptember28,2016.

345 Author’s calculation. PREPAResponse to CEPR‐AH‐05‐11.a.i through a.iv (Total spent to dateminusreportedspendinginFY2015‐FY2017)

346ScheduleF‐3REV,line11(InvestmentCapexFinanced(AOGP))


strategy.347PREPA’srevenuerequirementsmodelassumesthatPREPApursuesAOGPandobtainsfinancingfortheAOGPprojectata5%interestratethroughtheDOEloanguarantee.While the assumption of long‐term spending at AOGP is consistent with PREPA’s plans,contractsandtheCommission’s27SeptemberOrder,theassumptionofsuccessfulfinancingforAOGPisnotknownandmeasurable.TheassumptionofDOE‐backedfinancingforAOGPdistortsthecostimplicationsofAOGPonPREPA’sratepayers,particularlyunderaformularate mechanism (FRM) structure. If the Commission approves AOGP’s capital costs aspresentedandallows for anFRMstructure, butPREPAdoesnot obtainAOGP financing,PREPA’scustomerswillbesaddledwithextraordinaryratesinFY2018.

4. CommissionIRPFindingswithrespecttoAOGP

a. FinalIRPOrder

OnSeptember26,2016theCommissionissuedaFinalOrderonPREPA’s2015IRP(“FinalIRPOrder”)determiningthatit“cannotconcludethatAOGPrepresentsaleast‐cost,least‐riskpathforservingcustomers'needsandmeetingPuertoRico'senergypolicygoalsbasedon the facts presented in this proceeding.”348 The Commission approved continuedpermitting,engineeringandplanningactivitiesintheoverallAOGPproject,butsubjecttoarestrictive$15millionspendingcap.349TheCommissionrequiredthatPREPAexpresslyaskforpermissiontoexceedthe$15millionspendingcap,andthatpermissionwouldonlybegranteduponreviewofasoundeconomicanalysisofAOGP.350TheCommissionrequiredthat“PREPAmayrequestpermissionfromtheCommissiontoexceedthe$15millioncap,”but“shallaccompany such requestwith a detailed economic assessment of the AOGP project” (“AOGPEconomicAnalysis”)351

This reportdoesnot revisit the IRP’s assessmentnor theCommission’sdecision, insteadfocusing on the treatment of AOGP in this rate case, and the process used by PREPA todetermineifconstructingAOGPwasareasonablecourseofaction.Thisreportdoesseektoprovide clarity and a pathway for PREPA to complywith the Commission’s Order of 27Septembertore‐filethisratecaseincompliancewithIRPfindings.352

b. 27SeptemberOrder

347PREPA’sMotionforReconsiderationinthe2015IRPFinalOrder.DocketCEPR‐AP‐15‐002.Page34.“AOGPandtheconversionsneedtobeapprovedaspartofPREPA'sMATScomplianceefforts.”

348DocketCEPR‐AP‐2015‐002.FinalOrder.September26,2016.Paragraph255.



351FinalIRPOrderB.1.a.(2)

352DocketCEPR‐AP‐2015‐001.Orderrequiringrevisedtestimony.


OnSeptember27,2016,theCommissionrequiredPREPAtore‐fileexhibitsandtestimonywith respect to revenue requirements (“27 September Order”). The Order specificallyrequired that PREPA’s rate case reflect the Commission’s findings that “spending onpermitting,planningandengineeringrelatedtotheAguirreOffshoreGasPort(“AOGP”)shallnotexceed$15millionfromthedateoftheissuanceoftheIRPOrder,”andfurtherclarifiedthat“thespendinglimitappliestothetotalcombinedspendingassociatedwithAOGPandthegasconversions.”353

TheplainlanguageofthisOrderrequiredthatPREPAre‐fileitsrevenuerequirementwithareflectionofa$15millioncapontheAOGPproject.PREPA’srevisedfiling,receivedOctober15,2016,failedtomakethisadjustment.

The 27 September Order contained what could have been perceived as an inconsistentapproachtothetreatmentofAOGP,aninconsistencyharnessedbyPREPAinexplainingitsfailuretore‐fileinaccordancewiththeCommission’srequirements.First,theCommissionrequiredthatthe“spendingonpermitting,planningandengineeringrelatedtotheAguirreOffshoreGasPort[sic](“AOGP”)shallnotexceed$15millionfromthedateoftheissuanceofthe IRP Order.” However, the Commission’s required that PREPA’s “submissions of [anupdatedfinancialmodel,projectedcapitalspendingandrevenuerequirements]alsoshallassume,forinformationalpurposesonly,thatAOGPwillbecomeoperationalatarealisticallyachievabledate.”The$15millioncapandassumptionofAOGPoperationsata“realisticallyachievabledate”couldbeperceivedasinconsistent.

c. PREPAresponsivetestimony

PREPA’sresponsivetestimonyarguedtwoprimarypointstoexcludethe$15millioncap,anddidnotaddressthecommission’srequirementfortheAOGPEconomicAnalysis.First,thatthe27SeptemberOrderrequirementthatPREPAconformtoa$15millionbudgetcapatAOGPuntilfurtherjustificationisoutweighedbyPREPA’sbelief,asexpressedbyPREPAwitnessMr.Quintana,“thatAOGPis,andwillremain,justifiedandleastcost”andthat“theCommissionwillultimatelyagree.”Second,PREPAarguedthat“evenifAOGPandtheothergenerationprojectsinthenortharenotapproved,PREPA[sic]othersignificantinvestmentswouldhavetobemaderapidly.”PREPAconcludedthatit“hasnoreasontobelievethatthesealternative investments would be any less expensive than the investments currentlyreflectedinitsthreeyearbusinessplan.”354

WhilePREPAcorrectlypointsouttheambiguityofthedirectiveinthe27SeptemberOrder,itsargument isspeciousandunderminesPREPA’ssupportofaFormulaRateMechanism(FRM).

353DocketCEPR‐AP‐2015‐001.Orderrequiringrevisedtestimony.Page2.

354PREPAEx.13.00at106‐116.


i. PREPA’sargumentthatAOGPalternativespendingisequivalenttothecostsofAOGPitselfisunsupported

Mr. Quintana’s assumption that “other significant investments would have to be maderapidly”intheabsenceofAOGPisnotsupportedbyevidenceintheIRPrecordnorinthisratecase,andmakesnoattempttorectifythecostsofthosealternativeactionsagainsttheexpectedcostsofAOGP.TheCommissiondoesnot,andshouldnot,simplyapprovecapitalcostsonaroughandunsupportedassertionthatalternativeinvestmentstoAOGPwouldbeapproximatelythesamecostasAOGP.Mr.QuintanaeffectivelyarguesthattheCommissionshouldhandPREPAanopencheck for$470million355ontheassumptionthateven if theCompanydoesnotpursueAOGPitwouldfindawaytousethecapitalforMATScomplianceatanapproximatelyequalcost.Thisargumentbeliestheresourceplanningprocess,prudentutilitypractice,andreasonableratemakingpractice.

NavigantwitnessMr.Hemphill,testifyingonbehalfofPREPA,testifiesthat“theIRPOrderdoesnotmateriallychange,orchangeinaknownandmeasurableway,PREPA’sadjustedtestyearrevenuerequirement,”356andstatesthatinFY2017revenuerequirements“the$56millionpertainingtoAOGP…shouldnotberemoved.”357Ratherthannotethe$15millioncapitalcap,Mr.Hemphill insteadcitesto“theSeptember27Order’sdirectivethatRPEPAshouldassumethatAOGPandrelatedprojectsarejustifiedtotheCEPR’ssatisfactionandareoperationalatarealisticallyachievabledate.”358

InchoosingtociteonlytotheCommission’srequirementthatcapitalspendingandrevenuerequirements should assume, for informational purposes only, that AOGP will becomeoperational,Mr.Hemphill providesno avenuebywhich theCommission can realisticallyrestrict PREPA’s spending on the AOGP project in a meaningful way. The restriction ofspendingonAOGPisakeydriverbywhich thisCommissionseekstocontrolcostswhiledemandingactionofPREPA.TheFinal IRPOrderrequiredthat toexceedthe$15millionbudgetcapatAOGP,PREPAwouldberequiredtofileanimprovedeconomicassessmentoftheAOGPproject.RatherthanpursuingthisavenuetodemonstratethatAOGPisinthebestinterests of Puerto Rico ratepayers, PREPA has elected to challenge the Commission’sfindingsandsubmitstestimonyindicatingthattheCommission’sexplicitdisapprovaloftheAOGPprojectatthistimehavenobearingontheCompany’scurrentratecase.Intheabsenceof an improved economic assessment, PREPA’s plan to continue spending $56million inFY2017hasnobasisanddirectlycontradictstheCommission’srequirements.

355$470million:sumofFY2017+FY2018AOGPspending.

356PREPAEx.14.00at39‐40.

357PREPAEx.14.00at52‐54.

358PREPAEx.14.00at44‐47.


ii. PREPA’s failure to adjust AOGP‐related revenueRequirementsUnderminesFRM

Mr.Hemphill’sargumentwithrespecttoAOGPalsounderminesPREPA’sproposedFRM.Insupplementaltestimony,Mr.Hemphillarguesthat“theFRMshouldaffordtheCommissionasubstantialandwell‐definedroleinreviewingPREPA’sforward‐lookinginvestmentplansandbudgets…This structure isdesigned tomeshwith the IRPprocess to streamline theplanningandapprovalof capital investment.”359Hecontinues that “theFRMwouldallowPREPAtofileitscapitalinvestmentbudgetinlinewithwhatalreadyhasbeenlitigatedanddecided.”360Unfortunately, the first indicationofhowPREPAwouldrespond toa specificCommissionfindingwithrespecttotheIRPandimpactingtheratecaseisnotindicativeofthissmoothly‐functioningIRPfindingandadjustmentmechanism.Instead,PREPAhasfailedtoaccountinitscapitalbudgetfortheCommission’sfindingsontheIRPinalitigatedanddecidedcase.

Finally, Mr. Hemphill recommends a FRM filing and adjustment schedule beginning inOctober 2017. Mr. Hemphill’s first reconciliation sets rates for FY2019,361 skipping anyadjustmentsordecision‐makingprocess forFY2018. Indeed,hismechanism implies thatPREPA’sprovisionalrateswouldcarrythroughthethirdquarterofFY2017,andthe firstpermanentrateswouldbesetforfivequartersfromQ3FY2017throughtheendofFY2018.Under thisproposal, theCommissionwouldbechargedwithdeciding theentiretyof theAOGPdecisiontodayandsettingratesaccordingly.

5. RecommendationwithrespecttoAOGPcapital

a. OptionsavailabletotheCommissionwithrespecttoAOGPcapital

PREPA’s failure to provide updated revenue requirements for AOGP, as required by theCommissionfromtheSeptember2016IRPFinalOrder,placesaburdenontheCommissiontoenforcesuchaspendingcap.IntheabsenceofafinaldecisiononAOGP,theCommissionhasthreeoptionswithrespecttotheproject:

1. AssumethatPREPA’spre‐existingcommitmentswithrespecttoAOGP,andthehurdleoffindingaMATScompliantalternativeinareasonabletimeframe,aresubstantialenoughthatthereisnorealalternativeoptionforAOGP.Withsuchafinding,theCommissionwouldceaseexaminationofthedecision‐makingprocess

359PREPAEx.16.00at66‐73.

360PREPAEx.16.00at154‐156.

361PREPAEx.16.00at206‐207,alsoEx.16.01.


andfocusinsteadonspending,financing,furthercontracting,andprojectmanagement.

2. MaintainPREPA’srevenuerequirementwithrespecttoAOGPforbothFY2017andFY2018butrequireongoingcompliancefilingstoensurethespendingcapisnotexceededuntiljustifiedundertheAOGPEconomicAnalysis.UnderthismechanismPREPAwouldeffectivelybeentrustedtoself‐governAOGPspending,andtheCommissionwouldrequireatrue‐upifAOGPisfoundimprudent.

3. RevisePREPA’srevenuerequirementwithrespecttoAOGPforbothFY2017andFY2018,reducingFY2017spendingtothe$15millioncapandshuntingremainingfundingintoFY2018.ThisoptionprovidesthemoststringentCommissioncontrolovertheprojectandensuresPREPAresponsivenesstotheCommission’srequirementsfortheAOGPEconomicAnalysis.ItalsorisksincreasingcostsifPREPA’scontractsaredelayedwhiletheCommissiondeliberatesonAOGP.

Intheauthors’opinions,thefirstoption–ceaseexaminationofAOGP–isnotjustifiedbytheevidencepresentedbyPREPAtodate.TheCommission’sorderwithrespecttotheIRPwasbasedonthefindingthattheeconomicsofAOGPasopposedtothealternativespresentedby PREPA were not compelling. The IRP, as prepared, suggested that there were realalternativestoAOGPthatcouldmeetcompliancerequirements,improvePREPA’sflexibilityandabilitytointegraterenewableenergy,andutilizecapitaltoimprovePREPA’sgenerationfleet, rather than shift fuel sources with the same generation fleet. New informationpresentedbyPREPAinthisratecaseindicatethattherearesubstantialcommitmentsthathavebeenmade,andmaysoonbemade,byPREPAthatreducetheCompany’soptions;thesecommitments were not reflected in the IRP’s analysis. Further, in the time that PREPAcommitted internally to the AOGP project it increasingly sidelined viable alternatives,includingMATS‐compliantnewgeneration,environmentalcontrolsonexistinggeneration,increased renewable penetration, a focus on smaller distributed generation, or thecompletionofasouthcoastgaspipeline.PREPAhaspresentedlittleornoevidencethatthoseoptions are not still viable. Further, PREPA’s argument that a failure to move forwardexpedientlyonAOGPexclusively risks immediateandsevere sanction fromEPA362 isnotborneoutbyEPA’spastsettlementswithnon‐compliantutilities.WhileconductingtheAOGPEconomicAnalysismid‐streamisneithertimelynorefficient,itisanecessarycheckonthissubstantialcapitalinvestmentandlong‐termfuelcommitment.

MaintainingPREPA’srevenuerequirementbutrequiringintensivecompliancefilingsplacesanobligationon theCommission toensurePREPA’s spending remains in‐checkuntil theAOGPEconomicAnalysisiscomplete,andmaynotpreventPREPAfrommakingincrementalcommitments even without the Commission’s authority. PREPA’s response to the

362PREPA’sMotionforReconsiderationofProvisionsoftheFinalResolutionandOrder,October13,2016,CEPR‐AP‐2015‐0002,paragraph38.


Commission’sFinalIRPOrdersuggeststhatPREPA’sprimarymodeonAOGPisdefensive,ratherthanseekingtoplacatetheCommission’sconcerns.Inresponsetotheorder,PREPA(a)filedamotionforreconsideration,accusingtheCommissionoffailingtounderstandtheIRPanalysisandseekingtounderminethecredibilityoftheCommission’sassessment,363and(b)declinedtoupdaterevenuerequirementsasrequiredbytheCommission’spost‐orderupdate.364PREPAhasnotfiledanAOGPEconomicAnalysis,hasnotrequestedclarificationontherequirementsforsuchananalysis,noraskedforpermissiontocontinuespending.HadPREPAofferedtoplacatetheCommission’sconcerns,soughttofileaneconomicanalysis– even while contesting the Commission’s findings, or even requested clarification, theCommissionmighthavetakencomfortinallowingPREPAgovernanceoverAOGPspendinginFY2017.GivenPREPA’scontinuingfailuretoprovidefullandcompleteinformation,allowtheCommission theopportunity to audit itsmodels andmodel assumption, andor evenrespond to the Commission’s stated requirements, the Commission should not providePREPAthisgovernancelatitude.

TheonlyquestionremainingisifaCommissionrequirementtorestrictFY2017spendingto$15millionatallAOGP‐relatedprojectswillresultinsignificantunnecessarycostsatAOGP.Stalling projects, cancelling vendors, or missing contractual deadlines could result inincreasedcosts,damages,orpotentiallegalactionsbyvendors.OurlayreviewofPREPA’scontractssuggeststhattheremaybecasesinwhichPREPAisexposedtoincreasingfeesforincurredvendorcosts,andmaydelaysomeprocurementorauthorizations.However,wefind no specificmajor cost increases or damages because of the $15million cap, unlessPREPAsigns theAOGPFNTPwithoutCommissionauthorization.TheCommissionshouldrequirethatPREPAnotproceedtosigntheFNTPuntilitoffersexpressauthorization.

InPREPA’sMotionforReconsiderationontheIRPFinalOrder,theutilitymakesnomentionofanylegalorcontractualbarriersformeetingthe$15millioncap.Inotherutilities,theriskofcontractualliabilitiesisaprimaryconcernforutilitiesfacedwithnear‐termrequirements.

The Commission’s $15 million cap, if applied to FY2017 expenditures, provides over aquarter (27%) of PREPA’s anticipated AOGP budget in that year, and could cover AOGPpermittingcostandpermittingsuccessfees,FY2017paymentforpriorworksconductedbyExcelerate, and professional back office service fees.365 In response to discovery, PREPAindicatesthatithasonlyspent$4,173inthefirstquarterofFY0217,lessthan0.01%ofitsanticipatedbudgetfortheyear.

363Id.,paragraphs54and65.

364PREPA’sPetitionforRateReview,Exhibit13.0,SupplementalDirectTestimonyofJavierQuintanaMendez,lines100‐119.

365CEPR‐AH‐01‐01_Attach01.PermittingCosts=$11.2million,PermittingSuccessFee=$1.2million,EngineeringPriorWorks=$0.7million,ProfessionalServices=$0.5million,PREPApermits=0.02million.Total=$13.6million.


Overall,weexpectthat,barringtheexecutionoftheFNTPatAOGP,PREPAcouldmeetthe$15millioncapinFY2017withminimalcontractrisk.

b. Recommendation

WerecommendthattheCommissionrevisePREPA’srevenuerequirementforFY2017toreflecta$15millionspendingcapatAOGP,reducingFY0217revenuerequirementsby$41,339,807.ToreflecttheCommission’sorderthatPREPA“shallassume,forinformationalpurposes only, that AOGPwill become operational at a realistically achievable date,”werecommend increasing the FY2018 budget by the same $41.3 million increment to$454,756,927.

WerecommendthattheCommissionorderthatPREPAwithholdfromsigningaFinalNoticetoProceeduntilithassubmitted,andtheCommissionhasapproved,theAOGPEconomicAnalysis.PREPAshouldsubmit,ortheCommissionshouldrequirethatPREPAsubmit,theAOGPeconomicanalysisexpediently.

E. TransmissionandDistributioncapitalbudget

1. Overview

PREPA’stransmissionanddistributionsystemservesover1.6millioncustomersovertheislandofPuertoRico.Thetransmissionsystemlinkslargecentralstationgeneratorsinfourlocationsonthenorthandsouthcoaststhroughbothaerialandundergroundwires,runningthroughsomeofthemostdifficultterrainintheUS,includingmountainsandwettropicalforests.Thetransmissionanddistributionsystemrequireconstantupkeep,andfinancialandtalentlosseshaveledtowhatPREPAdescribesasseveredeterioration,andwhatanexternalreviewclearlyindicatesissystemfraughtwithreliabilityconcerns.

WediscussPREPA’sreliabilityissuesinTHESTATEOFPREPA’SSYSTEM,ChapterIII(Page26),notingthesevere–andincreasing–levelofinterruptionsseenbyPREPA’scustomers.

Historically,PREPA’sspendingontransmissionanddistributionhasmatchedbudgetsfairlyclosely,withsomeexceptions(seeFigure23,below).Thebudgetfortheratecaseisroughlyinlinewith2015spendingandbudgets,andincreasessubstantiallythereafter.


Figure 23. Historic transmission and distribution budget and actual spending by fiscal year, 2010 to 2015;366 rate case budget.367

PREPA’shistoric transmissionanddistributionbudgetsandspendingareharder to trackthanotherdirectorates.PREPAallocatesbudgetstoacentralofficeviaeitherthelineitem“transmission anddistribution” (deepartment 65) or to the director of transmission anddistribution(department64).Thesecentralofficesthenallocatebudgetstoregionaloffices.In doing so, the utility creates a disconnect between budget request and the use of thatbudget,relyingonthedivisiontotrackandallocateappropriately.

Figure24,below,showstheeffectofthisbudgetingprocessontheassessmentofbudgetusefrom2010to2015inthetransmissionanddistributiondirectorate.Notably,everyoneofthe regional offices appears to be overbudget on a year‐by‐year basis, while the centraldepartmentsareapproximatelyequallyunderbudget.Acloser lookat line‐itemspendingrevealsthatthosecentraldepartmentsareallocatedlargeblocksofcapitalandusenoneofit.


367ScheduleF‐3REV


Figure 24. Historic over (under) spending in transmission and distribution

Fromoneperspective,theallocationofbudgettoacentraldepartmentmaybeanefficientmechanismofensuringthatbudgetsaremetattheindividualregionaloffices.Ontheotherhand, itblurs theability to (a)provideoversightand (b)ensure strategicdeploymentoffundsacrossthetransmissionanddistributionsystemrelativetootherdemandsonPREPA’ssystem.

2. Transmissionsystemcapitalbudget

a. Overview

Someof themostcomprehensive informationaboutPREPA’s transmissionsystemcomesfrom a 2013 report from URS on PREPA’s assets. 368 The report summarizes PREPA’stransmissionsystemin2013asfollows.

368URSJune2013AnnualReport(“URS2013Report”).PREPAExhibit3.02(D)ConsultingEngineersReport. Fortieth Annual Report on the Electricity Property of the Puerto Rico Electric PowerAuthority,June2013.


TheAuthority’stransmissionsystemisaninterconnectednetworkof230kV,115kV,and38kVpowerlinesthatcarryelectricalpowerfromtheproductionplantstonumerousdistributioncentersfromwhereitisdistributedtoclientsforconsumption.

Atthecloseoffiscalyear2013,thetransmissionsystemwascomprisedof2,478circuitmilesoflines:375circuitmilesof230kVlines,727circuitmilesof115kVlines,and1,376circuitmilesof38kVlines.Includedinthetransmissionsystemtotalsareapproximately35milesofunderground115kVcable,63milesofunderground38kVcableand55milesof38kVsubmarinecable.Inadditiontothehighvoltagelines,thetransmissionsystemincludestransformersatthegeneratingplantsubstations,transmissioncentersforinterconnectionofdifferentvoltagesystemsandswitchyardsandgearforconnectionorseparationofportionsofthetransmissionsystemoperatingatthesamevoltage.HighvoltagetransformersinstalledintheAuthority’stransmissionsystemanditsproductionplantshaveatotaltransformercapacityof19,207MVA.369

Notethatwedidnotrequestanupdateonthismileageforthecurrenttransmissionsystem.Anew230kVlinefromGuayanillatoArecibohaslikelyincreasedthetotalmilage.

PREPA’stransmissionsystem,showninFigure25,below,isanoverlayofa230kVsystem,a115kVsystem,anda38kV.370The230kVsystemessentiallyformsaringaroundtheisland’slesspopulousinterior,connectingmajorcitiesandtowns.Inaddition,PREPAmaintainstwosubstantialnorth‐southcorridorsfromSalinas(nearAguirreandtheAESplant)toSanJuan,andfromGuayanilla(nearEcoEléctricaandCostaSur)toAreciboandManationthenorthcoast.TheprimaryoperationalthermalgenerationinPuertoRicoislocatedonthesouthcoast–EcoEléctrica,CostaSur,Aguirre,andAEScoalplant.Thetransmissionsystemisdesignedtofacilitatetheflowofenergyfromtheseplantstotheprimarypopulationcentersinthenorth.Cambalache,onthenorthcoastnearArecibo,andMayagüez,onthewestcoast,providepeakinggeneration.

369URS2013Report,page3.

370TransmissionmapisfromURS2013Report,page113.Thismapisvintage2013andprovidedforillustrative purposes only. The report andmap are publicly available andwas accessed throughpublic channels on PREPA’s website athttp://www.aeepr.com/INVESTORS/DOCS/Financial%20Information/Annual%20Reports/Consulting%20Engrs%20Annual%20Report%20FY2013.pdf. PREPA provided a 2016 transmissionmapunderconfidentialcoverthat,forillustrativepurposes,canbeconsideredsimilarinmostaspectstothevintage2013map.


Figure 25. PREPA’s Transmission System vintage 2013, from URS 2013 Report. Includes planned transmission through 2018, vintage 2013.


b. Transmissionsystemcapitalbudget

PREPA’s transmissionsystemcapitalbudgetamounts to$81.3million inFY2017, spreadover100separatelineitemprojects..Table15,below,consolidatesPREPA’stransmissioncapitalbudgetbysub‐areaand initiative.371ThemajorityofFY2017spending–68%‐ isconcentratedinthe230kVand115kVsystems,withthevastmajorityofthebudgetsspentonlinerehabilitation.

Table 15. Transmission system capital budget, millions FY2017‐FY2019

FY2017 FY2018 FY2019

230 kV System

230 kV Line Rehabilitation 15.9 17.2 13.8

230/115 kV New Transmission Center & Increase Capacity 2.0 3.0 6.0

230 kV Switchyard New and Expansions 0.5 3.0 0.5

230 kV Total 18.4 23.2 20.3

115 kV System


115 kV New Line 2.4 2.0 0.0

115/38 kV New Transmission Center & Increase Capacity 5.0 8.6 6.0

Capacitor Bank ‐ Transmission 2.0 19.0 13.8

115 kV Switchyard New & Expansions 1.2 0.0 4.1

115 kV Underground System 1.5 14.0 9.3

Transmission Misc. Improv. Plant ‐ Engineering 0.0 0.0 0.5

115 kV Total 37.4 59.7 62.7

38 kV System


38 kV Switchyard New & Expansions 1.0 1.5 5.0

38 kV New Line 0.9 0.6 0.3

38 kV Underground System 0.3 0.3 1.0

Increase Line Capacity 0.6 0.6 0.0

38 kV Total 12.5 10.4 17.4

Other Transmission

Transmission Misc. Improv. Plant ‐ Engineering 7.2 6.3 1.0

Transmission Misc. Improv. Plant ‐ Electric System 2.5 2.9 2.5

Transmission Pole Replacements 1.5 0.8 2.0

Breakers Uprating 0.8 0.5 0.5

Energy Management System 0.5 0.5 0.7

Other Transmission Plant 0.5 0.4 0.8

Grounding Mat Reconstruction 0.1 0.1 0.1

Other Transmission Total 13.0 11.4 7.5

Transmission Total 81.3 104.7 107.9

Overall, PREPA expects to spend on the order of $50,000 per mile of existing 230 kVtransmissionlineonrehabilitation,androughlythesameamountonthe115kVsystem.The

371ScheduleF‐3REV


expected investment in the38kVsystem is far smaller, suggesting thateither the38kVsystemisinbetterrepair,orthePREPAisprioritizingthelong‐distancehighcapacitylines.

i. Transmissionlinecapitalbudget

HalfoftheFY2017dollarsspentintransmission($40.5million)aretargetedtowardstherepairofspecifichighcapacity transmission lines.Specifically,PREPAseeks torepair thefollowingcorridors(FY2017costs):372

‐ Two230kVlines(50900,51000)fromAguirretoAguasBuenas,justsouthofSanJuan(south‐north):$15.7million

‐ 115kVline(37800)fromJobosnearAEStoSanJuan(south‐north):373$13.4million

‐ 115kVline(37400)fromArecibotoSanJuan(east‐west):$4.4million

‐ 115kVline(36100)fromLagoDosBocastowardsPiñas,ontheoutskirtsofSanJuan.(eastwest):$7.0million

AsinallofPREPA’sothercapitalprojects,thesearethe“as‐incurred”costs,ratherthanthefull costof eachproject.Assessing the full costof eachproject, itbecomesapparent thatPREPAexpects tospendabout$600,000permileonthe230kVsystemand,onaverage,about$700,000permileonthe115kVsystem.

PREPA’s projects include the replacement of structures, towers, and foundations, andreplacementofinsulators.Thesecostsappeartobeinlinewithutilityestimatesforsimilarprojects.

Thelargestcomponentofthe38kVcapitalbudgetisallocatedtotherehabilitationofthe38kVsystem.Thereare24specificprojectswithFY2017budgetsthatfallunderthisumbrella,only one of which exceeds one million dollars. We did not request detailed budgetjustificationsfromPREPAforprojectsunderonemillioninFY2017,andthereforeareunabletoprovideinsightonthespecificsoftheseprojects,asidefromtheirgeneraldescriptionsas“improvements”and“increasecapacity.”TheyareconsistentwithPREPA’sindicationofthestateofthesystem.TheonlyprojectinexcessofonemillioninFY2017isareconstructionofsevenmilesofline(PID15610)nearMayaguez.Thisprojectreplacesrottingwoodenpoleswithsteel,effectivelyre‐buildingtheline.Thedollarcostisconsistentwithutilityestimatesforsimilarprojects.

i. Othertransmissionprojects


373ThisistheonlyprojectlistedthatalsoincludesnewtransmissionlinesegmentalongafivemilecorridornearCaguas,southofSanJuan.


MostofPREPA’snon‐line‐specificcostsarescatteredacrossotherinitiatives,mostofwhichare relatively low cost on a per‐item basis, and therefore difficult to assess without anengineering audit. The only substantial additional line items in FY2017 are listed under“TransmissionMisc.Improv.Plant‐Engineering”andareprimarilyallocatedtowardstheconstructionoftwonewcontrolroomsattwomajorswitchyardsinSanJuan:MonacilloandViaducto.Wehavenospecificbasisforcomparisonorworkplanforthesecontrolrooms,butassessthecostsaswithinlineforthisscaleofproject.

3. Distributionsystemcapitalbudget

a. Overview

The 2013 URS Report on PREPA’s assets provides a reasonable overview of PREPA’sdistributionsystem:

AsofJune30,2013,theAuthority’sdistributionsystemconsistedofapproximately31,550circuitmilesofdistributionlines(withoperatingvoltagesrangingfrom4.16to13.2kV)and333substations(withatotalinstalledcapacityof5,018MVA).Thedistributionsystemhasmorethan1,800circuitmilesofundergroundlines.TheAuthorityhas22portabletransformerswithatotalcapacityof349.6MVAtosubstituteforexistingtransformersduringmaintenanceoroutages;similarly,theAuthorityhastwoportablecapacitorbankseachratedat18MVAR.Thereare813privatelyownedsubstations(withatotalinstalledcapacityof3,266MVA).Thedistributionsystemalsoincludesapproximately1,485,200clientmeters.374

Of the 31,550 circuitmiles of distribution lines, PREPA indicated that approximately 24percentofthelinesareat13.2kV,24%ofthelinesareat8.32kV,andtheremainingoverheadlinesareat4.16kV.375 Inadiscovery response,PREPA indicated thatover16,000circuitmilesareprimaryvoltage.376Primaryvoltagesarecarriedtotransformertobesteppeddowntosecondaryvoltagesdeliveredtocustomers.Inmorerecentupdates,PREPAhasindicateditdoesnothave Informationon thepercentageofpolescarryingprimaryandsecondary

374URSJune2013AnnualReport(“URS2013Report”).PREPAExhibit3.02(D)ConsultingEngineersReport. Fortieth Annual Report on the Electricity Property of the Puerto Rico Electric PowerAuthority,June2013.Page17

375Id.Page55.

376CEPR‐AH‐02‐01


lines.377Inaddition,asofPREPAhasapproximately3,100circuitmilesunderground,three‐quartersofwhichis13kV.378

TheCompanyreported333substationsintheURS2013Report,379andlistsapproximatelythesamenumberinresponsetodiscovery.380AccordingtotheURS2013Reportreport:

TheAuthorityhasstandardizedontwosizesofpermanentsubstationsbasedonthetransmissionsystemsupplyvoltage.Thisstandardizationexpeditestheengineering,procurement,andconstructioncycle,increasesflexibilityinpotentiallyutilizingequipmentasspares,andprovidesacosteffectiveinstalledcapacitymarginforloadgrowth.InsituationswheretheAuthorityneedsadditionalsubstationcapacityonaninterimbasisorwithshortleadtimes,theAuthorityinstallstemporarysubstationsthatarestandardizedunitizedmetalcladequipment,whichcanberelocatedasrequired381

The entire PREPA distribution system is divided into seven regions with 26 TechnicalDistricts. 382 PREPA deemed amap of the distribution system as voluminous and criticalenergy infrastructure.383We attempted to review the Company’s geographic informationsystem (GIS), butwere unable to access PREPA’s virtual private network (VPN) despiterequeststoPREPAfortechnicalsupport.Thus,wedonothaveagoodvisualizationoftheregionalanddistrictdetailforthePREPAdistributionsystem.

Likethetransmissionsystem,PREPAhasstatedthatitsdistributionsystemisinapoorstateofrepair:

Ingeneral,PREPADistributionsystemisinoperationalstateorcondition,butwithreliabilityconcerns,duetoagingofthecomponents.Stillahighpercentageofoverhead(OH)distributioncircuitsareattachedtowoodpolesindeterioratedconditions,withOHcablesgagesinappropriatetoservetheelectricalloadinareliableandqualitativemanner.Takingalooktotheunderground(UDG)Distributioncircuits,asignificantnumberofthemhavebeenreplacedin“temporary”waybyOHcircuitsthrustreetlightspoles,toreestablishelectricalservicetocustomerswhenoutagescannotbefixed

377CEPR‐PC‐02‐034

378CEPR‐JF‐02‐06(a).Author’scalculation.

379URS2013Report.Page51.

380CEPR‐PC‐02‐028(a).

381URS(2013)PFD000055

382CEPR‐AH‐02‐06

383CEPR‐PC‐02‐039


repairingtheUDGconductor.MostoftheUDGcircuitsinresidentialcommunities(Urbanizations)withmorethan30yearsofexistencearedirectburial,whichacceleratethecabledeteriorationandpreventfromreplacecablesinreasonabletimeandavoidrepeatedinterruptionsinthefuture.This,alsoincreasesthereconstructionandmaintenancecostswheninterventionisrequired.InadditiontotheDistributionElectricalnetwork(OH&UDG),thesites(substations)wheremainpowerdistributiontransformersresides,areinadvancestateofdeterioration.Mostofthecomponentsneedtobereplacedandperformcontinuousmaintenancetogroundsandbuildings.384

GivenPREPA’sdescriptionofthe“deteriorated”conditionofitsdistributionsystem,itisnosurprisethatPREPA’sreliabilitymetricsarepoorandindicateasystemthatexperiencesfrequentandlongerdurationoutages(see“LowreliabilityforPREPAcustomers,”SectionIII.B,page32).

b. Distributionsystemcapitalbudget

PREPA’sdistributionsystemcapitalbudgetamountsto$74millioninFY2017,spreadover108separatelineitemprojectsinPREPA’sdiversedistributionnetwork..Table16below,consolidatesPREPA’sdistributioncapitalbudgetbysub‐areaandinitiative.385Notethatthisisnotbrokendownbydirectorateandincludesspendingunderboththetransmissionanddistributiondirectorateaswellasthecustomerservicedirectorate(markedwithasterisk).

384CEPR‐AH‐02‐01(f)

385ScheduleF‐3REV


Table 16. Distribution system capital budget, millions FY2017‐FY2019

FY2017 FY2018 FY2019

13.2 kV System

Improvements to the 13.2 kV Aerial Distribution System 1.9 1.9 2.2

Construction and Extension of 13.2 kV Aerial Feeders 4.0 3.8 5.5

Improvements to the 13.2 kV Underground System 2.7 6.4 4.4

Improvements to the 13.2 kV Underground System (Blankets) (00917) 1.3 1.0 1.5

Construction and Extension of 13.2 kV Underground Feeders 0.6 0.6 0.7

13.2 kV Total 10.4 13.5 14.1

8.32 kV System

Improvements to 4.16 kV ‐ 8.32 kV Aerial System (Blankets) (00968) 20.0 15.3 23.0

Improvements to 4.16 kV ‐ 8.32 kV Aerial System (Other) 3.9 3.8 5.4

Construction and Extension of 4.16 kV ‐ 8.32 kV Aerial Feeders 0.5 0.2 0.3

Improvements to the 4.16 KV ‐ 8.32 KV Underground System 6.8 5.3 9.3

Construction and Extension of 4.16 kV ‐ 8.32 kV Underground Feeders 0.3 0.3 0.3

8.32 kV Total 31.5 24.9 38.2

4.16 kV System

Construction and Extension of 4.16 kV ‐ 8.32 kV Aerial Feeders 0.6 0.5 0.9

4.16 kV Total 0.6 0.5 0.9

Substations, Drops, and Meters

Purchase of Meters and Metering Equipment* 12.5 12.5 12.5

New Customer Extensions and Drops (Blankets) (01035) 6.0 3.0 8.0

Street Lighting 5.5 3.3 7.0

Line Extension to Serve New Customers 1.1 1.6 1.6

Rehabilitation of Substations 1.9 1.3 1.3

Miscellaneous Substations Improvements 0.2 0.1 0.2

New Distribution Substations 0.0 0.5 4.0

Purchase and Installation of Breakers and Sectionalizers 0.6 0.4 0.7

Purchase Step ‐ Voltage Regulators 0.2 0.1 0.2

Replacement of Residential and Commercial Services* 0.4 0.4 0.4

New Services Drops Installation ‐ Commercial* 0.9 0.9 0.9

Special Transformers 0.6 0.6 0.7

Distribution Line Voltage Converter 0.4 0.3 0.5

Distribution Lines Capacitors 0.1 0.1 0.2

Distribution System Automation 0.5 0.3 0.7

Distribution System Expansion 0.0 0.0 0.2

Distribution Total 30.8 25.1 39.1

Other

Improvements to 4.16 kV ‐ 8.32 kV Aerial Distribution System 0.3 0.3 0.3

Improvements to the 13.2 kV Underground System 0.5 0.3 0.6

Distribution Total 73.9 64.5 93.1

*FallsunderCustomerServicedirectorate

TherearenotablefeaturesofPREPA’sdistributionbudget.Ingeneral,wecangatherthatthevastmajority of the spending is on the rehabilitationof the existingdistribution system,substations, feeders, and lines. Some of the most substantial spending is allocated to“blankets” which are simply pools that get used on an as‐needed basis for repairs,maintenance, and replacement parts in thewidespread system. Overall, the distributiondepartmentallocates$27.3million,oroverone‐third,ofitsbudgettounspecifiedblanketsestimates. Another $12.5 million, or 17 percent, is allocated to new meters and meter


equipment.ThelargeremainderofPREPA’sdistributionbudgetistargetedtowardsstreetlighting,andbothaboveandbelow‐groundsubstationandfeederrehabilitation.

c. Distributionnetworkcapitalspending

Whilenearly85percentofPREPA’sdistributioncapitalbudget isallocated towardsnon‐meterissues,wewereunabletoassesstheaccuracyorneedforthevastmajorityofthe105non‐meterprojects.Ofthe78distributionprojectsthathadFY2017spendingandwerenoteitherexplicitlyblanketsormeters, theaverageFY2017costwaswellunderonemilliondollarsandthemediancostwasaquarterofamillion.

PREPA did not provide justification for the individual projects or explanations beyondgeneral“requiredimprovements,”andthusitisnearlyimpossibletoassesstheprudence,value,orreasonablenessoftheseprojects.

Itisdifficulttoevenjudgethevalidityofthedistributionbudgetagainsthistoricspendingbecause in prior years as PREPA’s records of past capital spending do not appear to bebrokenintothesamecategoriesasinforward‐lookingspending.

PREPA’s distribution staff did thoroughly answer questions during the October 20ConferenceCallregardingthenatureoftheexpenditurestheyseektomake,andindicatedthat(a)thecapitalwasabsolutelyrequiredsimplytobeginaservicerestorationprogramand(b)thatthecapitalbudgetswerelikelylowerthanwererequiredtoestablishreasonableservice.

d. Metercapitalspending

In FY2013, PREPA indicated that its system contained approximately 1,485,200 clientmeters.386 In response to discovery, PREPAnoted that its number ofmeters currently is1,620,401installedmetersandincludesbothactiveandinactivemeters.387

In its FY2017 through 2019 budgets, PREPA has requested $10.6 million per year topurchasenewmeters.PREPAdescribesitsbudgetrequestforfuturemeteracquisitionsas:

PID16677‐$10.6M‐AcquisitionofMetersSecondaryLines.TheamountbudgetedinPID16677isforacquisitionofmetersforresidentialcustomers.Theseamountsdonotincludemaintenanceoroperatingexpenses.PREPAplanstomaintainitsreplacementplanwithanemphasisinequipmentupgrades.TheestimatedamountofmeterstobeacquiredforFY2017is29,000meters.Theestimatedcostis$200/meterpluscommunication

386URS2013Report,page17

387CEPR‐JF‐02‐05


infrastructure,ifweconsiderthecostoftheinstallationofnewtechnology(AMI)theadditionalcostisaround$198/meter388

Forcomparison,PREPAreplaced47,296metersinFY2014,40,111metersinFY2015,and21,133metersinFY2016.389InFY2017,PREPAanticipatesreplacing29,000meters.Ifthereplacement cost formeters is $200/meter and the number ofmeters being replaced is29,000;thentheFY2017meteracquisitioncostwouldbe$5.8millionor$4.8millionlessthanthebudgetrequestmadebyPREPA.ForFY2018andFY2019,PREPAanticipatesreplacing20,000meters,whichat$200permeterwouldcost$4millionperyearor$6.6millionlessthanthedollarsbudgetedforeachfiscalyear.390

Theapparentdifferencebetweenthebudgetrequestandthecalculatedamountappearstobe the continued implementation of smartmeters (“AMI”) by PREPA. Thesemeters aresubstantially more expensive than PREPA’s current generation of meters and requiresubstantialinfrastructuretobeuseful.AccordingtoPREPA,thesenewmeterswouldcostapproximately $398 per meter, a substantial increase on a per‐meter basis and aquestionableuseofPREPAfundsatthistime.

i. Moving from Advanced Meter Readers to SmartMeters

Electricmeterscanbegenerallycategorizedintothreetypes:radialmeters,AMRmeters,andsmartmeters.

Electromechanicalmeterscontainametaldiscthatspinswithenergyuse.Theamountthatthedialturnsindicatestheenergyconsumed.Thesemetersrequirehumanmeterreaderstovieweachindividualmeteratitslocationtoobtainbillingdata.

AutomatedMeterReading(AMR)metershaveonewaycommunicationscapabilitiestotransferenergyconsumptiondatatoreadersordatabasesremotely,eitherviawireorradiofrequencies.Thesemetersobviateaneed forameterreader tomanuallyattendameter.UtilitiescanobtaininformationfromAMRwitheitherhand‐helddevicesorvehicledrivenreaders.PREPAindicatesthatitsAMRmeterstransmitdatabackthroughthedistributionsystem.

Thesystembeinginstalledutilizesaproprietarytechnologythatcommunicatesbetweenmetersandremotecontrollersbysuperimposinga

388CEPR‐AH‐02‐06

389CEPR‐JF‐02‐05

390URS2013Report.Page48.


frequencymodulatedsignalontheAuthority’sexistingdistributionlinesbetweentheclientmeterandtheAuthority’ssubstation.391

Inaddition,PREPAnotesthatitsAMRmetersalso:

Beginninginfiscalyear2011theAuthorityhasbeeninstallingmeterswithanintegraldisconnect/connectfeature.Thesemetershavegreatlyreducedthetimerequiredforcustomerservicedisconnectionsandreconnectsforshorttermclientsorinproblematiclocations.392

AdvancedMeteringInfrastructure(AMI,or“smartmeters”):Smartmetersarecapableof two‐way communicates between individual meters and the utility. The two‐waycommunications have the capability of providing interval data at a level of time scalegranularityunavailabletoradialandAMRmeters.However,smartmetersrequireboththecommunication,datastorage,anddataanalysisinfrastructuretosupportthevastamountsofdatarequired.

Todate,PREPAhasindicatedthatitstartedAMRmetersinstalledin2000andcompleteditsAMRinstallationin2013.393AMRmetershaveatypicallifespanof15‐20years.394PREPAhasnotprovidedameterreplacementprogramforitsexistingmeterexceptforintroducingAMImetersintoitssystem.

ii. PREPA’sjustificationforsmartmeterinfrastructure

PREPAindicatesthatitssystemcontainsapproximately13,000smartmetersbasedontheabilitytoconducttwo‐waycommunicationswithbetweenthemetersandPREPA.395In2015,PREPA initiated a pilot program to install 30,000 smartmeters in twophases thatweresummarizedintwoPowerPointpresentations,ProyectoPilotodeMetrosInteligentes(July2015)andProyectodeMetrosInteligentes(September2016).396TheJuly2015presentationprovidedanoverviewofPREPA’sgoals(translated):

Developthenecessaryspecificationstodevelopasystemthat:

GathersthenecessarydatatoraisethespecificationsoftheoptimalsystemforESA

391CEPR‐JF‐02‐05

392URS(2013)PFD000063

393URS(2013)PFD000062

394https://www.vectren.com/assets/cms/livesmart/pdfs/amrfactsheet.pdf

395CEPR‐PC‐08‐03parte.

396CEPR161020RequestNo.8Attachment1andAttachment2


Improvetheefficiencyoftheelectricalsystem

Significantlyimprovetheserviceofferedtocustomers,suchasSAIDI,SAIFI,CAIDIandthusreducetheneedformanagementinthecommercialofficesandCustomerServiceCenter(telephones)

Automaticremotedisconnectanddisconnect,toimproveserviceandtoallowservicecut‐off,improvingcashflow397

PREPA’s goals in pursuing smart‐grid technology are similar to those of other utilitiesembarkingonSmartGridimplementation.IntheJuly2015presentation,PREPAcomparestheirpilotprogramtostateofHawaii,398butdoesnotprovidecontextforincludingthestate–asopposedtoanyotherstatethathasalsotestedsmartgridtechnology.

Utilities have generally found it difficult to justify smart grid technology on a costeffectivenessbasisandinthiscaseHawaiimayhaveservedasafaultyexampleforPREPA.OnMarch31,2016,theHawaiianElectricCompaniesfiledasmartmeterimplementationplaninDocket2016‐0087.TheHawaiiplanwouldinvolve467,000metersacrossfiveoftheHawaiian Islands. The filed analysis indicated that the deployment cost of $413millionwouldlikelyexceedthe$345millioninestimatedbenefitsfortheproject.399ItisunknownifHawaii will proceed with full smart grid implementation since the Hawaiian ElectricCompanies own analyses indicate that the project as currently envisioned are not cost‐effective.

Yetmoresubstantially,theslidedeckrevealsadeeplyfaultyjustificationfortheproject:aperceptionof excess cash flow in the customer servicedepartment.The slide states that“PREPA annually allocates $10 million to the meter replacement program. This plancontemplatesusingthisfundtoimplementthereplacementofthenetwork.”400

ItisinconceivablethatatapointwherePREPA’ssystemisactivelyfailing,anydepartmentwouldlookattheirbudgetasinviolableunlesstheinitiativeisabsolutelynecessaryforthesafe,reliableandefficientoperationofthesystem.PREPA’ssecondpresentation,fromJuly2016,clearlyindicatesthatthesmartmeterprogramwouldnotbeexpectedtoprovidecostsavings,andisadistractionfromPREPA’simmediaterequirements.

iii. SmartMeterPilot

397CEPR161020RequestNo.8Attachment1:Slide3


399HawaiianElectricCompanies.Application.Table2,Page.8.March31,2016.Docket2016‐0087.



In theSeptember2016draftpresentation,PREPA indicated that it spendsapproximately$10millionper year on replacementmeters and thatPREPAwanted touse someof thefundingforsmartmeters.401

Thedraftpresentation indicatesthatstarting in June2015,PREPAordered15,000smartmetersfromacompany,CaribbeanMeteringSystems(CMS).402,403PREPAreportsthatithadreceived10,216meters,data collectordevices, andnetwork support.404 In total,PREPA’sestimatedtotalcostforthePhaseIpilotwas$7.5millionorapproximately$500permeter.

Thenextphaseof theSmartMeterPilotproject is toexpandtheexistingpilotprojectbyanother15,000smartmeters.For thisphase,PREPAhasestimatedthat thecostswillbe$4.675millionor$311permeter.405PREPAanticipatesthatmetercostswilldecreasefrom$223.62permeterinthefirstphaseto$193permeterforthesecondphase.Inaddition,PREPAhasestimatedthatitwillnotneedadditionaldataconcentratorsthatwereinstalledduringPhaseI.

TheresultsofPREPA’spilotindicatethatithadnotachievedsomeofitsgoals.


402CEPR‐JF‐02‐05

403CEPR161020RequestNo.8Attachment2:Slide8.PREPAhasincludedthecostofthe4,784smartmetersstillpendinginstallation.




Table 17. Findings from PREPA CMS Smart Meter Pilot: September 2016

Goal FindingsImprovemeterreadingeffectiveness 99%target,93%observed406Improveprocesstoresolvenon‐payment Reducedtimestoprocessdelinquentaccounts

throughautomaticdisconnectionReductioninenergytheft Observedtheftin 169outof6,844metersinstalled.

Basedondeploymentof250,000meters,PREPAdeterminedthatsmartmeterswouldnoteconomicallyviable.407

Durabilityofmeters PREPAneededtoreturn88metersandseveralmetershadminorissues.408

Qualityofmetersreceivedfromvendor PREPAnotedthat5,840metersor57percent ofthe10,216metersreceivedwererejectedandreturnedforrecalibration.PREPAnotedthatthemanufacturerisworkingonimprovingqualitycontrol.409

MeasuringServiceQuality PREPA noted that it needed to improve serverinfrastructure and technical support from thesupplier.410

ThesummarizedfindingsindicatethatPREPAisnotreadyforamoresignificantdeploymentofsmartmeters.First,PREPA’sestimatesoftheCMSprojectforyearstwo(2)througheleven(11)wouldcostapproximately$18‐20millionperyearfortheinstallationof250,000smartmetersora replacementof17percentofPREPA’s1.4million customermeters.411 If oneincludesthe$12.175millionofspending(spentandplanned)ontheinitialsmartmeterpilot,thissuggestsacumulativeprojectcostof$192to$212millionfor250,000meters.Onapermeterbasis, the250,000meterswouldcost$768to$848permeter.Withrespect to theestimatedannualcostsof$18to20millionforyears2through11,PREPAhasnotprovidedany specificdetailsof the scopeand justificationof theadditionalwork toexpandsmartmeters.Furthermore,PREPAhasnotprovideasummaryoftheplanorcostsofthe250,000smartmeterdeploymentoranyotherfuturedeploymentplans.








ThereisnoevidencethatPREPA’ssystemwillactuallybebenefitedinthenearormid‐termby the deployment of smartmeters, and smartmeters do little tomitigate a generation,transmission,anddistributionsystemthatsimplyfailstooperatereliably.

AnotherconcernhighlightedinthefindingsfromtheinitialpilotprojectisthequalityofthemetersdeliveredtoPREPA.Ofthe10,216metersdeliveredtoPREPAasofSeptember2016,overhalfofthemetersneededtobereturnedforre‐calibration.Evenifoneassumesthatnomoreofthe15,000Phase1meterswillfail, itstillresultsinaninitialrejectionrateof39percent. These rejection rates should raise serious questions about the contracts andexperienceofboththesmartmetervendorandsupplier.

PREPAitselfhasnotedthatitneedstoevaluatecomplimentarytechnologiestoaddressthetopologicalchallengesoftheisland.Acellularcommunicationnetworkwillnotbeabletocommunicateitmountainousterrainsincethecommunicationrequiresalineofsightpath.As part of the evaluation, PREPA is considering the need to pilot a 10,000 smartmeterdeploymentwithradio‐frequencycommunicationstechnology.412

iv. SmartMeterRecommendations

Basedontheresultsoftheinitialpilotproject,werecommendthefollowing:

1. PREPAdiscontinueanyadditionalworkbeyondthecontracted30,000smartmeters.

2. PREPAprovideadetailedbusinesscaseincludingcostsandbenefitstojustifytheneedforsmartmeterdeployment.

3. ThebusinesscaseshouldprovidedetailofthescopeandscaleofdeploymentincludingbutnotlimitedtotechnologicalissuesthatwillneedtobeaddressedbyPREPA.

4. ThebusinesscasealsoneedstoaddressbudgetallocationssinceasmartmeterprojectwilldivertfundingandmanagementfocusfromotherpressingneedsfacedbyPREPA.

4. RecommendationswithrespecttoTransmissionandDistributionCapital

Wemake the following recommendations with respect to transmission and distributioncapital:



1. WerecommendthattheCommissionapprovethefullFY2017transmissionsystemcapitalbudgetasrequestedbyPREPA.

2. WerecommendthattheCommissionrequirePREPAtoceasefurthersmartmeterorAMIpurchasesuntilastrategicplanjustifyingtheneedforAMIisproducedbyPREPAandapprovedbytheCommission.

3. WerecommendthattheCommissionreducePREPA’smeterbudget(PID16677)from$10.6to$5.8millionforFY2017toreflecttheanticipatedcostofcontinuingPREPA’snormalAMRacquisitions.

4. WerecommendthattheCommissionapprovethefullFY2017distributionsystemcapitalbudgetforallitemsexceptsecondarylinemeters(PID16677).

F. Othercapital–TransportationandComputerEquipment

1. Overview

ThelastcomponentofPREPA’scapitalbudgetareallocationsforvehicles($19.4million)andcomputerequipment($13.1million).Inthisassessment,wefocusedonthestateofPREPA’sgeneration fleet and transmission, and have relatively little information about theseanticipated expenditures. This section summarizes the information that we do haveavailable.

2. Vehicles

a. Autofleet

PREPArequestsabudgetof$16millioninFY2017forvehiclereplacement(PID13713).WerequesteddetailonPREPA’santicipatedspending,andwereprovidedalistof132vehiclesthatPREPAisseekingtoacquire.413Mostofthereplacementvehiclefleetisorientedtowardsthepurchaseofbucketorboomtrucks– the truckscommonlyused for lineworkwithabucketandboom.Abouthalfthetrucksareearmarkedforthetransmissionanddistributiondirectorate ($12.5million).Another40 trucks are earmarked formeter readers in clientservices ($2 million), and another 24 pickup and vans are targeted for the generationdirectorate ($946,500). The executive directorate acquires the remaining eight vehiclesincludingoneboat,whichweassumeisforoffshorework(total$467,000).

PREPAmaintainsa fleetof3,593vehicles,ofwhichapproximately1/3areSUVsorbasicpickuptrucks.Anotherthirdofthevehiclesareeithertrailersorhighlyspecializedvehicles(suchasbulldozers,cats,trenchersandloaders),theremainingvehiclesareeitherbucket

413CEPR161020RequestNo10Attach01


trucks or other line vehicles.While the number of SUVs, Jeeps and pickup trucks seemssomewhatextraordinaryforthenumberofemployeesatPREPA,thebudgetfortheseclassesofvehiclesarerelativelysmall.

b. Helicopter

PREPArequests$3millionforareplacementhelicopterinFY2017.

The National Transportation Safety Board (“NTSB”) reports that on July 24, 2015 ahelicopterownedbyPREPAcrashedinshallowwaternearVegaBaja.TheNTSBreportsthatthe helicopter was engaged in electrical power distribution construction, and had amechanicalfailuremid‐flight.Thepilotsuccessfullydescendedandnoneofthethreecrewwereinjured,buttheNTSBreportedthatthe“mainrotorstruckthehelicopter’stailboom,separatingitfromthefuselage.”414Thecrashedhelicopter,anMDHelicopter369isalightutilitycraft.Acursoryreviewofsimilarcraftsuggeststhata$3millionpricetagiswithinreason,butwehavenotmadeathoroughreviewofthevalidityofthiscost.

3. Computerequipment

PREPArequests$13.1millionforvariouscomputerequipment,datamanagementsystems,andnewnetworkequipment.PREPAhasdividedtheirrequestinto18sub‐categoriesundertheexecutivedirectorate,allbuttwoareunder$1million.Themostsubstantialrequestisanear‐term(FY2017)datacentermigrationtoPREPANetworkfor$8million($6.3millioninFY2017).

We have no specific knowledge of these initiatives or the basis of the cost. The primaryconcerninthiscasecouldbetheinappropriateuseofanaffiliate(PREPANetwork).However,wehavenoreasontobelievethatthisinitiativeisinappropriate.Maintaininganeffectivedatamanagementsystemandcomputernetworkwillbecriticaltocontinueddatacollection,statisticalanalysis,andreporting.

4. Recommendationswithrespect to transportationandcomputerequipment

We recommend that the Commission approve the full FY2017 transportation andcomputerequipmentbudget.

G. Recommendation

414NTSBReportERA15LA280.http://www.ntsb.gov/_layouts/ntsb.aviation/brief.aspx?ev_id=20150724X55322&key=1


WeprovidebelowasummaryofourrecommendationsonPREPA’scapitalbudgetrequest.

1. WerecommendtheCommissionapprovePREPA’sFY2017capitalspendingatAguirre,PaloSeco,andSanJuanSteamPlants.

2. WerecommendthatthemaintenancecontractatSanJuanCombinedCycleberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$12millioninFY2017fromcapitaltoO&M.

3. WerecommendthattheCommissionapprovetheothercapitalexpensesatSanJuanandAguirreCombinedCycleUnits.

4. WerecommendthattheCommissionexaminetheMHPS‐PRcontractandtheperformanceofMHPS‐PRtodetermineifthecontractorismeetingperformanceexpectationsformaintenanceserviceatSanJuanCC.

5. WerecommendthattheCommissionrequirePREPAtofilenoticeofanylong‐termcontractwithserviceproviderswithapotentialnetpresentvalueof$25millionvalueorhigher.

6. WerecommendthatthemaintenancecontractatCambalacheberemovedfromthecapitalbudgetandreassignedasanannualmaintenanceexpense,areassignmentof$4millioninFY2017fromcapitaltoO&M.

7. WerecommendthattheCommissionexaminetheCamabalachecontractandtheperformanceofAlstomtodetermineifthecontractorismeetingperformanceexpectationsformaintenanceserviceatCamabalache.

8. WerecommendthattheCommissionrequirePREPAtoprovidestrategicplansforSanJuanandPaloSecosteamplants,includingthefollowingelements,ataminimum:maintenanceplan,MATScomplianceplan,aninvestmentplanformaintaining(ornot)SanJuan7‐10andPaloSeco1&2,andareliabilitystudy–i.e.whatstrainsareplacedonthesysteminthepresenceorabsenceoftheSanJuanand/orthePaloSecosteamunits.

9. WerecommendthattheCommissionrequirePREPAtotrackandreportonprojectsatSanJuan7‐10andPaloSecoUnits1&2,regardlessofiftheseunitsareconsideredlimiteduse.

10. PREPA’snextlong‐termplanningexerciseshouldevaluatetheextenttowhichthe“limiteduse”unitsshouldbemaintainedandcontributetopeakrequirementsonaneconomicbasis.

11. WerecommendthatPREPAtrackcapitaldollarsspentateach“limiteduse”facilityseparatelythentheunitsthatareexpectedtobemaintained.ForCostaSur,PREPAshouldtrackandreportonprojectsatUnits3&4.Further,PREPA’slong‐term


modelingshouldconsistentlyassessiftheselimitedusefacilitiesareavailableforpeakpurposes,andifnot,assessthevalueofmaintainingunitsthatneithercontributetopeakpurposesorprovideenergytothesystem.PREPA’slackofmaintenancespendingattheseunitsisinconsistentwithitscontinuedpeakuseofthefacilities.

12. WerecommendthattheCommissionrevisePREPA’srevenuerequirementforFY2017toreflecta$15millionspendingcapatAOGP,reducingFY0217revenuerequirementsby$41,339,807.ToreflecttheCommission’sorderthatPREPA“shallassume,forinformationalpurposesonly,thatAOGPwillbecomeoperationalatarealisticallyachievabledate,”werecommendincreasingtheFY2018budgetbythesame$41.3millionincrementto$454,756,927.

13. WerecommendthattheCommissionorderthatPREPAwithholdfromsigningaFinalNoticetoProceeduntilithassubmitted,andtheCommissionhasapproved,theAOGPEconomicAnalysis.PREPAshouldsubmit,ortheCommissionshouldrequirethatPREPAsubmit,theAOGPeconomicanalysisexpediently.

14. WerecommendthattheCommissionapprovethefullFY2017transmissionsystemcapitalbudgetasrequestedbyPREPA.

15. WerecommendthattheCommissionrequirePREPAtoceasefurthersmartmeterorAMIpurchasesuntilastrategicplanjustifyingtheneedforAMIisproducedbyPREPAandapprovedbytheCommission.

16. WerecommendthattheCommissionreducePREPA’smeterbudget(PID16677)from$10.6to$5.8millionforFY2017toreflecttheanticipatedcostofcontinuingPREPA’snormalAMRacquisitions.

17. WerecommendthattheCommissionapprovethefullFY2017distributionsystemcapitalbudgetforallitemsexceptsecondarylinemeters(PID16677).

18. WerecommendthattheCommissionapprovethefullFY2017transportationandcomputerequipmentbudget.

19. WerecommendtheCommissionrequirePREPAtopresenttheresultsofitsinvestigationintotheAguirreSeptember21st2016outagetothisCommissionanddiscloseitsbudgetrequirementsfromthatevent.


VI. FUELANDPURCHASEDPOWERBUDGETS


Inthischapter,wediscussPREPA’sbudgetsforfuelandpurchasedpower.Together,thesebudgetsrepresent$1.47billioninFY2017–approximatelyhalfofPREPA’sFY2017revenuerequirement.415 At present, both of these components of PREPA’s system costs arecompensatedthroughtheuseofadjustableriders,meaningthatPREPA’srevenuesadjusttoensuresufficientcollectionforthesecostsevenasfuelpriceschangethroughouttheyear.Assuch,thevaluespresentedintheratecasearenotstrict“budgets”inthesamesenseastheotherlineitemsofPREPA’srevenuerequirement,butratherareexpectedspendinglevels.PREPApresentsthesebudgetswiththeexplicitunderstandingthatitsspendingisexpectedtodiffer,potentiallybyalargemargin,asfuelcostschange.

In the interest of setting realistic rates and appropriately managing the expectations ofratepayers,theCommissionmustdecideifPREPA’sfuelandpurchasedpowerbudgetsarereasonable towithin an acceptablemargin. If theCommissiondecides thesebudgets arereasonable,nochangewillbe required. If theCommissiondecides thatPREPA’s fuelandpurchasedpowerbudgetsarenotreasonable,itmustthendecidewhetherornottoorderadjustmentstothesebudgetsand,ifso,inwhatamounts.

Below,wediscussthegroundsofPREPA’sfuelandpurchasedpowerbudgets,describehowPREPAformulatedthesebudgets,andcritiqueboththemethodologyandtheresults.OurconclusionisthatPREPA’suseofproductioncostmodelingwithPROMODisappropriatefordetermining fuel and purchased power budgets in the short term, and that PREPA isgenerally successful at forecasting its spending on purchased power from thermalgeneratorsbut is subject to significantuncertainty in theonlinedatesofnew renewableenergy projects. In addition, we conclude that PREPA’s fuel price forecasts wereunrealistically low and therefore an adjustment to the fuel budget is necessary. Werecommendanadditiontothefuelbudgetof$461,305,000.

B. PREPA’suseoffuel

PREPA’sowngenerationfleetconsistsalmostexclusivelyoffossil‐firedgenerators.416Apartfrom two units at the Costa Sur plant, all of PREPA’s thermal generators burn either

415PREPA’sPetitionforRateReview,ScheduleA‐1REV.Author’scalculation.

416Base IRP,Volume I, Table3‐1. In addition to its approximate5GWof operational fossil‐firedcapacity,PREPAoperates60MWofhydropowergeneration.


distillate417orresidualfueloil.418Ingeneral,PREPA’scombinedcycleandgasturbineunitsburndistillateoilwhileitssteamunitsburnresidualfueloil.CostaSurunits5and6burnablendofnaturalgasandresidualoil.

AsdiscussedintheCommission’sFinalOrderontheIRP,419anddemonstratedinScheduleA‐1 REV, PREPA’s costs are dominated by fuel expenses. Historically, PREPA has spentbetweenoneandtwobilliondollarsayearonfuelalone.420Overtime,thecompositionofPREPA’sfuelmixhasshifted.InFY2011,#6fueloilrepresentedninetypercentofthefuelburnedbyPREPAatitsunits(onthebasisofMMBTUoffuel),withtheremaindermadeupsolelyofdistillateoil.421PREPAstartedburningnatural gasatCostaSurunits5and6 inFY2012,andthefuelhassincecometorepresent27percentofPREPA’stotalfueluse.422Useofdistillatehasalsoincreasedtoapproximatelyone‐fifthofPREPA’sfuelmix,whilePREPA’srelianceonresidualfueloilhasdeclinedandnowrepresentsonlyhalfofthefuelburnedbyPREPA.423Thesechangesareshowninthefigurebelow.

417DistillatefueloilisonecommonnamefortheNo.2gradeoffueloil.ItisalsosometimesreferredtoasdieselfuelbyPREPAandothers.

418ResidualfueloilisonecommonnamefortheNo.6gradeoffueloil.IthasahigherviscositythanNo.2fueloilandmaycontainhigherlevelsofimpurities.

419IRPFinalOrder,SectionII(E).

420PREPAEx.14.02.xslm,“Inputs”tab,line25(“FuelOilAdjustment”).

421Author’scalculation,basedonvaluesprovidedinPREPA’sresponsestoCEPR‐AH‐06‐05(b)(v)oftheCommission’sFourteenthROI(November7,2016;refertoCEPR‐AH‐06‐05Attach01.xlsx)andCEPR‐AH‐06‐01(d)of theCommission’sFourteenthROI (November3,2016).PREPAconsumedatotalofapproximately147,700,000MMBTUoffuelinFY2011,ofwhichapproximately133,00,000wasNo.6fueloilwithNo.2fueloilmakinguptheremainder.

422Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐06‐05Attach01.xlsxandCEPR‐AH‐06‐01(d).PREPAconsumedatotalofapproximately139,600,000MMBTUoffuelinFY2011,ofwhichapproximately73,600,000MMBTUwasNo.6 fueloil,28,900,000MMBTUwasNo.2 fueloil,and37,100,000MMBTUwasnaturalgasburnedatCostaSur.

423Id.


Figure 26. Composition of PREPA's fuel mix over time, on a heat content basis.

Notably,PREPA’s total fuel consumptionhasalsodeclinedbyapproximately fivepercentsinceFY2011, inaccordancewithPREPA’s falling sales.424 IfPREPAconstructsnewunitswhichhavelowerheatratesandburnprimarilydistillateornaturalgas,wewouldexpectitsfuel mix to shift yet further. However, neither PREPA nor the Commission expect thecompositionofPREPA’sgenerationfleettochangesubstantiallyinFY2017andassuchwewouldexpecttotalfuelconsumptiontofollowaroughlysimilarpattern,intermsofamountsofindividualfuelsandamountoffuelasawhole,asseeninFY2016.

C. PREPA’sPowerPurchaseandOperatingAgreements(PPOAs)

Inadditiontoitsownunits,PREPAbuyselectricityfromtwofossil‐firedplantsandavarietyofrenewableunitsownedbythird‐parties.PREPAhascontractualagreementswiththesethird‐parties, known as Power Purchase and Operating Agreements, or PPOAs. PREPA’scostsforpurchasedpowerarelargelydependentonthestructuresofitscontractswithitsPPOAcounterparties,whichthemselvesvarywidely.

PREPA’stwofossilPPOAsarewithEcoEléctrica,whichoperatesa507MW,naturalgas‐firedcombinedcycleplant,andAES,whichoperatesa454MWcoal‐firedsteamplant.425PREPA’scontractualtermswithAESarestraightforward.Asdescribedbytheutility,PREPApaysAESfortheenergyitproducesandthedependablecapacityitprovides,aswellascompensating

424Id.

425BaseIRP,VolumeI,Table3‐1.


AES for its startup‐related costs after any shutdown of the unit that was requested byPREPA.426Allofthesecostsaresetonayearlybasis.427PREPA’senergypaymenttoAESiscomprised of two components: a modified fuel pass‐through and a charge for variableoperationsandmaintenancecosts.428AES’per‐kWhenergypriceisfixedeveryyear,subjecttoaguaranteefromPREPAthattheunitwillbedispatchedatacapacityfactorofatleast50percent.429Similarly,thecapacitypricerepresentsbothcapital(debt)andfixedoperationsandmaintenancecostsatAES.430

PREPA’s contract with EcoEléctrica is somewhat more complex. It includes a capacitypayment and a base energy charge (both of which are structured similarly to the AEScontract, although unlike the AES energy charge, the EcoEléctrica base energy charge isadjustedbasedontheunit’sheatrateatdifferentlevelsofoutput).431LikeinitsagreementwithAES,PREPAhasalsoagreedtopaychargesrelatedtounitstart‐upassumingthatPREPArequested the preceding shut‐down. However, unlike the AES contract, the EcoEléctricacontractcontainsaprovisionthatPREPAwillpayan“excessenergypayment”forgenerationabove a 76 percent capacity factor.432 On a Conference Call, PREPA clarified that the 76percent capacity factor limit is set on amonthly basis, that EcoEléctrica sets the excessenergyrateunilaterally,and that theexcessenergyratecanchangeonaweeklybasis.433These provisions make it impossible for PREPA to accurately predict its spending atEcoEléctricaduringayearlybudgetingprocessunlessPREPAcommits todispatchingtheunitatacapacityfactoroflessthan76percentinanygivenmonth–which,aswediscussbelow,PREPAdoesnotdo.

InadditiontoitsthermalPPOAs,PREPAhasactivecontractswithseveralrenewableenergyproviders.Thisincludestwowindfarms,alandfillgas‐firedgenerator,andfoursolarfarms,totalingapproximately157MWofcapacity.434PREPA’scontractswithrenewablegeneratorsgenerally contain a base energy price with a yearly escalator and a REC price.435 The

426PREPA’sresponsetoCEPR‐AH‐06‐03(c)oftheCommission’sFourteenthROI(October26,2016).

427Id.

428Id.

429Id.

430Id.

431PREPA’sresponsetoCEPR‐AH‐06‐02(b)oftheCommission’sFourteenthROI(October26,2016).

432Id.


434 PREPA’s response to CEPR‐AH‐03‐02(a) (PUBLIC) of Commission’s Seventh ROI (August 31,2016).RefertoCEPR‐AH‐03‐02Attach01.xlsx

435Id.


problemsassociatedwithPREPA’sexistingrenewableenergycontractswerediscussedatlengthintheCommission’sFinalOrderontheIRP436andwillnotberehashedhere.

Wenote,however,thatitshouldbeveryeasytoPREPAtoforecastitsspendingonrenewablePPOAswith a high degree of accuracy, as the contract prices are known in advance andPREPAmustpayforalloftheenergygeneratedattheseunitsregardlessofwhetherornotit isused to serve load.437 Indeed, thishigh levelof certainty isoneof themain financialbenefitsofrenewableenergycontracts,ascomparedtorelianceonunitswhosevariablecostofgenerationmaychangesignificantlyandabruptlygivenchangesinfuelprices.ImportantlyforPREPA,sincethepricesof liquidfuelsaregenerallyvolatile,suchchangesoccurquitefrequently,exposingratepayerstothevagariesoftheglobaloilmarket.

D. PREPA’sfuelandpurchasedpowerbudgetingprocess

ThereareseveralkeyinputsrequiredforandstepsinvolvedinPREPA’sfuelandpurchasedpowerbudgetingprocess:

First,PREPAmustprepare,orhavepreparedonitsbehalf,aforecastofloadandaforecastoffuelpricesovertheperiodoftimecoveredbythebudget.ThesepricesarethenusedtopredictbothhowmuchofeachfueltypePREPAexpectstoburnandwhatPREPAexpectsthetotalcosttobeofthefuelthatituses.

Second, PREPA must assemble relevant cost data related to its PPOA contracts andperformanceandoperational costdata related to itsunits (asdiscussed further inLOADFORECAST,ChapterIV,Page41).

Finally,PREPAcanthenusethesecost inputsinamodelthatdeterminesanoptimal(i.e.,lowestcostwhilestillsatisfyingdemand)dispatchpatternforallunitsonPREPA’ssystem,includingPPOAunits.Thismodeliscalleda“productioncostmodel”asitforecaststhecostsofproductionindetail.

InorderforPREPA’sfuelandpurchasedpowerbudgetstobereasonableandrealistic,eachofthesestepsmustbeconductedrigorously.Weexamineeachstepinthesectionsbelow.

1. Fuelforecasts

Forthepurposesofthisratecase,PREPAreliedonfuelpriceforecastspreparedbySiemensinFebruaryof2016.438ThesesameforecastswerealsousedintheIRPcasetopreparethe

436IRPFinalOrder,SectionF.

437Id.,SectionF(3).

438PREPA’sresponsetoCEPR‐AH‐03‐03(a)(i)oftheCommission’sSeventhROI(August26,2016).


UpdatedFuelIRP.439PREPAaskedSiemenstopreparetheseforecasts“duetothedownturnoftheoilandotherfuelpricesinthesecondhalfofyear2015.”440AttheTechnicalHearingontheIRP,Siemenssaidthattheseforecastsrepresenteda“lowerbound”ontheexpectedtrajectoryoffuelprices.441However,PREPA’suseoftheseforecaststoprepareitsbudgetsinthisratecasebeliesthatassertion.

IntheFinalResolutionandOrderontheIRP,theCommissioncommentedontheseforecasts,andfoundtheyweresignificantlybelowcontemporaneous forecasts fromcrediblepublicforecasts.442Whencomparedwithseveralmonthsofactualdata,wenotethatthisconclusionfromtheFinalOrderhasbeenborneoutovertherecentpast.WecompareherePREPA’spredictions of prices for natural gas traded at Henry Hub and prices of West TexasIntermediatecrudeoiltoactualvalues.PREPAusesthese“fundamental”pricesasindicestopredict and determine its PREPA’s actual fuel prices;443 indices relate to prices at globaltradinghubsandarenotimpactedbyfactorsspecifictoPuertoRico.444

Asshowninthefigurebelow,actualfuelpricesstartedtodivergefromPREPA’sBaseIRPforecast in Julyof2015.However,natural gasprices started to rebound in Juneof2016,whichwasbeforethestartofFY2017butaftertheinitialfilingofPREPA’sratecase.WTIprices,meanwhile,increasedstartinginFebruaryof2016–thesamemonthinwhichthelower‐bound forecastwasprepared,andseveralmonthsbeforePREPA filed its ratecasewiththisCommission.Asofthiswriting,actualWTIpricesareapproximatelytwicewhatPREPAhadpredicted.ConsideringthatfueloilisthedominantfueltypeburnedbyPREPA,thisobservationraisesconcernsthatPREPA’srelianceonanunrealisticallylowfuelpriceforecasthasledittoseverelyunderbudgetforfuelinthisratecase.

439

440CEPR‐AH‐03‐01(a)(ii).

441CEPR‐AP‐2015‐0002,TechnicalHearing,April6,2016,NelsonBacalao,00:14:15ofpart5ofthehearingrecording.

442IRPFinalOrder,Section4(B)(1).

443BaseIRP,VolumeIII,Section2.3.3.

444IRPFinalOrder,Section4(B)(1).


Figure 27. PREPA's fuel price forecasts as compared to actual values.445

2. PPOAcontractterms

InordertorepresentPREPA’scontracttermswithitsPPOAprovidersinPROMOD,PREPAmakes several adjustments and assumptions. PREPA’s representation of renewablecontractsisstraightforward:PREPAfoldsboththeenergyandRECchargesassociatedwiththesecontractsintoasinglecostforenergy.446PREPAappearstomodelprojectswithasetcapacityfactor—forexample,themodeledcapacityfactorofallexistingsolarprojectsis21percent.447

PREPA’srepresentationofitsfossilPPOAsissomewhatmorecomplex.WhatPREPApaysforasan“energycharge”atAES,theymodelwithtwocomponents:afuelcostandavariable

445BaseIRPvaluesfromBaseIRP,VolumeIII,AppendixG.RatecasevaluesfromCEPR‐AH‐03‐03.ActualvaluesfromtheEnergyInformationAdministration(EIA).HenryHubandWTIactualmonthlyclearing prices are available at https://www.eia.gov/dnav/ng/hist/rngwhhdm.htm andhttp://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=rwtc&f=Drespectively.WTIpricesin$/bblwere converted to$/MMBTUusinganassumedheat contentof5.719MMBTU/bbl, alsosourcedfromEIA(https://www.eia.gov/forecasts/aeo/pdf/appg.pdf).

446PREPA’sresponsetoCEPR‐JF‐01‐22oftheCommission’sSixthROI(PUBLIC;August23,2016).RefertoCEPR‐JF‐01‐22Attach01.xlsx.

447Id.;author’scalculation.


operationsandmaintenance(O&M)cost.448PREPAmodelsthecapacitychargeatAESasacombinationofafixedO&Mcostandacapitalcostthatisaddedinapost‐processingstep.449

PREPA’s representation of EcoEléctrica is similar. According to PREPA’s description, itcaptures the shift in pricing from the base to the excess energy charge by modelingEcoEléctricaashavinglimitedaccesstoasetquantityoffuelpricedatthebaseenergychargelevel,with the totalenergycontentof that fuelbeingequivalent to theamountofenergyproducedbygenerationatEcoEléctricauptoacapacityfactorof76percent.450Afterthatsupply of fuel is expended, the unit has access only to amore‐expensive fuel, priced torepresent the expected level of the excess energy charge.451 We consider this to be areasonablemethodologyforrepresentingPREPA’sactualcontracttermswithEcoEléctrica.However,becausetheexcessenergychargeisnotsetinadvance,weconsiderthemodeledcostofthissecondfueltobe,atbest,aneducatedguessonPREPA’spart.

Altogether,wefindnocauseforconcerninthemethodologyoutlinedabove.WecanconfirmthatthebudgetedenergychargesatAESandEcoEléctricaareequaltothetotalFY2017fueland variable O&M charges at these units, adjusted for inflation, and that the budgetedcapacity charges are equal to the total fixed O&M and capital charges, adjusted forinflation.452

3. UseofPROMODandconcernsassociatedtherewith

AsmentionedinSECTION,above,PREPAusesaproductioncostmodelcalledPROMODtoforecastsitsdispatchandcostsoverthecomingyear.PREPA’sfuelandPPOAbudgetsare,inessence, takendirectly fromPROMOD,asconfirmedbyPREPAontheOct.20ConferenceCall.453

ThereareseveraldifficultiesinherentinmodelingdispatchofPREPA’ssystem,whichhasrelativelyfewunitsandrelativelyhighforcedoutagerates.Inparticular,PREPAassertsthatitdispatchessteamunitsonamust‐runbasisbasedonunitavailabilitywhileunitsfiredondistillatearedispatchedinmeritorderbasedonthevariablecostsofdifferentunits.454ThisrequiresPREPAtousehighminimumruntimesinmodelingitssteamunits,asidentifiedby

448Id.

449Id.


451Id.

452CEPR‐JF‐01‐22Attach01.xlsx


454PREPA’sresponsetoCEPR‐JF‐01‐12(c)oftheCommission’sSixthROI(PUBLIC;August23,2016).


several intervenors in the IRP proceeding,455 which may cause strange or distortionaryresults(suchasmodeledunderutilizationofPREPA’smoreeconomicandflexibleunitsdueto the need to avoid decommitment of inflexible steam units). These difficultiesnotwithstanding,PROMODisanindustry‐standardmodelandwefindittobeanappropriatetoolforPREPAtouseforthepurposeofforecastingspendingintheyearahead.

RegardlessofouracceptanceofPREPA’sgeneralmethodology,wehaveobservedseveralinconsistencies betweenPREPA’sPROMODoutputs anddata or attestations providedbyPREPA elsewhere. We are concerned that these inconsistencies indicate inaccuracy inPREPA’s PROMOD outputs, in assertions PREPA hasmade to the Commission and otherauthorities,orboth.Wedescribetheseinconsistenciesbelow.

a. Assumed fuel blend at Costa Sur does not comportwithhistoricalpatterns

Asdescribedabove,units5and6attheCostaSurplantburnablendofNo.6fueloilandnaturalgas.IntheIRP,PREPAassertedthatthisblendwasnaturalgas‐dominant:

CostaSur5&6STunitsburnnaturalgasandNo.6fueloilinadualfuelfiringscenario.CostaSur5burns80percentofnaturalgasand20percentofNo.6fueloilandCostaSur6burns75percentofnaturalgasand25percentofNo.6fueloil.456

This assumptionwas used in PROMOD: PREPAmodeled Costa Sur unit 5 asmodeled asburningexactly80percentnaturalgas(onaheatbasis)ineverymonth,whileCostaSurunit6 burns exactly 75 percent natural gas in every month.457 Depending on the relativegenerationofthetwounits, thisequatestoapproximatelya76to78percentnaturalgascontentinthefuelburnedattheplantasawholeoverthecourseofFY2017.

WhilePREPAmaybecorrectaboutthefuelmixthatwillbeburnedatCostaSurinthefuture,theactualfuelmixburnedattheplantisheavieronfueloilthanthelevelsclaimedbyPREPAabove.Onamonthlybasis,theplanthasonlyexceeded75percentnaturalgasinitsfuelmixinbriefexcursionssinceearlyincalendaryear2015.Apartfromthesespikesingasusage,CostaSur’sfuelmixhashoveredaround70percentnaturalgas,significantlybelowthelevelsassumedbyPREPAinthemodelingusedforbudgetpurposes(Figure28).Indeed,ifthefuelmixattheplantinFY2017actuallyconformedtothevaluesofferedbyPREPA,itwouldresultinthehighestpercentageofnaturalgasintheplant’smixsinceFY2013(thefirstfullyearinwhichtheplanthaddual‐fuelcapability).

455IRPFinalOrder,AppendixB.

456BaseIRP,VolumeI,Table3‐1,note(2).

457Author’sanalysisofdatacontainedinCEPR‐JF‐01‐22Attach02.xlsx.


Figure 28. Historical and modeled fuel mix (% natural gas by heat content) burned at Costa Sur units 5 and 6.458

WefindthatthefuelmixburnedatCostaSurishighlyuncertainandislikelytobemorefueloil‐heavy thanPREPA’smodelingshows.Wehavenoevidencewithwhich toexplain thevolatilityofthefuelmixattheplant.However,wealsohavenoevidencewithwhichtoadjustthetotalamountsof fuelburnedattheplantthatwereoutputbyPROMOD.Ashift intheassumed fuelmixat theplantwould change its assumedvariable costof generationandpotentially itseconomics relative to the restofPREPA’s fleet.Therefore, anentirelynewPROMOD runwould be necessary to predict total fuel consumption at the plant given adifferentfuelmixassumption.Althoughthecurrentmodelingmayunder‐predictfuelcostsslightly,webelievesuchchangeswouldbesmallandeasilyhandledthroughthefuelchargeadjusteralone.Therefore,werecommendnoadjustmenttofuelmixassumptionspertainingtoCostaSurvaluesatthistime.

b. PREPA plans to significantly exceed the “base energycharge”generationlevelatEcoEléctrica

Asdescribedabove,PREPA isexceptionallyvulnerable tovariations in theexcessenergypricechargedbyEcoEléctrica.ThemainwaythatPREPAcanavoidpotentiallyhighchargesis by restricting EcoEléctrica’s generation to a 76 percent capacity factor or below on amonthlybasis.Doingsoappearstomakegoodsense:inFY2016,theaverageexcessenergy

458 Historical values from PREPA’s response to CEPR‐AH‐06‐01(d)(iii) of the Commission’sFourteenth ROI (November 3, 2016). Forecasted values are author’s calculation based on valuesprovidedinCEPR‐JF‐01‐22Attach02.xslx.


pricechargedbyEcoEléctricawashigherthanPREPA’saveragecostofgenerationatallofitsunitsexcepttheAguirreCCs,gasturbines,andCambalache.459WhilewerecognizethattherelativeeconomicsofPREPA’sunitscanshiftonamonthlyorevenaweeklybasisgiventhestructureofPREPA’scontractwithEcoEléctrica,thiscomparisonisstillindicativeoftherisksofrelyingonEcoEléctricaabovetheexcessenergythreshold.Ifthispatternholds,itwouldbeprudentforPREPAtoresistdispatchingEcoEléctricaabovea76percentcapacityfactor except as a close‐to‐last resort to meet unusually high demand. Indeed, PREPAassertedduringtheOct.20ConferenceCallthatthiswouldbeitsstrategyandthat ithasbeensuccessfulindoingsointherecentpast.460

This assertion does not appear to be entirely factual. Over the 84 months of FYs 2010through2016, therewere58months inwhichEcoEléctricawasdispatchedata capacityfactorofhigherthan76percent.461DispatchwasnoticeablylowerinFY2016,however,withEcoEléctrica’sgenerationmeetingorexceedingtheexcessenergythresholdinonlyhalfofthe year. PREPA’s PROMOD outputs show that PREPA’s budget is based dispatch ofEcoEléctrica at veryhigh levels:EcoEléctrica ismodeledasbeingdispatchedbelowa76percentcapacityfactorinonlyonemonthandisdispatchedatacapacityfactorof95percentoraboveformostoftheyear.462Table18,below,summarizestheseresults.

Table 18. Historical and modeled EcoEléctrica capacity factors by month.463

2010 2011 2012 2013 2014 2015 2016 2017*July 97% 90% 72% 85% 78% 93% 80% 88%August 92% 94% 82% 87% 91% 88% 72% 95%September 89% 93% 85% 88% 92% 90% 49% 95%October 92% 97% 80% 86% 91% 96% 60% 99%November 88% 91% 82% 84% 83% 89% 79% 59%December 90% 66% 81% 87% 85% 77% 84% 97%January 68% 81% 43% 72% 53% 76% 76% 96%February 27% 60% 79% 11% 41% 62% 79% 95%March 73% 56% 80% 90% 69% 33% 76% 83%April 85% 75% 75% 87% 98% 49% 77% 96%May 96% 74% 81% 90% 97% 67% 76% 99%June 90% 88% 86% 90% 97% 81% 65% 99%*FY2017valuesareforecasted

459Author’scomparisonofaverageproductioncostsasprovidedinPREPA’sresponsetoCEPR‐AH‐03‐07(e)oftheCommission’sSeventhROI(PUBLIC;August31,2016)andaverageFY2016excessenergycostasprovidedinCEPR‐AH‐06‐02(e)(iv).

460October20ConferenceCall,2:30:00.

461Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐02(e)(i).


463 Shaded cells denote values above 76%. Historical values from CEPR‐AH‐06‐02(e)(i). FY2017forecasteddispatchfromCEPR‐JF‐01‐22Attach02.xlsx


We assume that this modeling reflects PREPA’s belief that EcoEléctrica’s excess energychargesinFY2017willbelower,relativetoPREPA’svariablecostofgeneration,thantheywere inFY2016.However,wehavenoevidencesupportingthisbelief.Moreimportantly,PREPA (to our knowledge) has no evidence supporting this belief, as EcoEléctrica canapparentlychargePREPAwhateverexcessenergypriceitdeemsfit.

Wehavetwomainconcernsproceedingfromthisobservation:first,wenotethatPREPA’smodeled reliance on EcoEléctrica above the excess energy threshold increases theuncertaintyofPREPA’spurchasedpowerandfuelbudgets.PREPAcannotknowinadvancewhatthecostofthatexcessenergywillbe.Indeed,PREPA’sassertionthattheexcessenergychargewouldbe“around$9/MMBTU”464wasbasedonPREPA’srelativelylowdispatchofEcoEléctrica.Atahigherdispatchlevel,weseenoreasonthatthatpriceexpectationwouldhold true. If PREPA’s actual dispatch shifts away fromEcoEléctrica and towards its ownunits,wewouldexpectthePREPA’sfuelspendingtobehigherthanthebudgetsetforthintheratecaseanditspurchasedpowerspendingtobelower.IfPREPAdoesactuallydispatchEcoEléctrica at such high levels, PREPA’s purchased power spending could increasesubstantiallyascomparedtoitsstatedbudget.

Second,consideringthelowavailabilityandinflexibilityofmanyofPREPA’sunits,weareconcerned thatPREPAwouldnotbeable to rapidly commit sufficient capacitywithonlyseveraldays’noticeifEcoEléctricadecidedtochargeveryhighspotprices.IfEcoEléctricaobservesthatPREPAdoesnothaveenoughunitsingoodworkingconditiontoreplaceitsexcess energy, it might decide to charge PREPA extremely high prices. PREPA, and theratepayers of PuertoRico,would be leftwith little recourse. This concern highlights theextenttowhichPREPA’sPPOAwithEcoEléctricaisunfavorabletoPREPA.

c. Some renewable energy contracts appear to bemisrepresented

DuetothestructuresofPREPA’srenewablePPOAsandthesimplicityofrepresentingthesestructuresinPROMOD,itshouldbestraightforwardtoreconcileaveragecostofgeneration(total cost divided by total generation) of PREPA’s renewable PPOAs with the reportedcontractualcostsoftheseunits.Ingeneral,ourexpectationinthismatterholdstrue:mostunitsmodeledbyPREPAdohavemodeledcosts that are consistentwith thecontractualcostsreportedbyPREPA.However,therearethreeexceptions,asnotedinthetablebelow.

464October20ConferenceCall


Table 19. Modeled and reported renewable energy contract costs.

Facility

ReportedEnergyCost*($/MWh)

ReportedRECCost*($/MWh)

ReportedTotalCost**($/MWh)

ModeledTotalCost†($/MWh) Note

AESIlumina 159 35 194 194 BlueBeettleIII 143 20 163 163 CIROOneSalinas(CIROGroup 137 35 172 172 DesarrollosdelNorteInc.dbaAtenasSolarFarm 145 15 160 160 FonrocheEnergy 150 25 175 175 Xzerta‐TecSolarI 150 15 165 165 GCLSolarEnergy(CaribePlanetaSolarLLC;GuayamaSolarEnergy) 141 30 171 171

HorizonEnergy 143 35 178 178

Energycostnotescalatedfromonlineyear(CY2015)

IrradiaEnergyUSA 144 23 167 167 OrianaEnergyLLC(Yarotek) 150 30 180 180 PatternSantaIsabel 131 26 157 157

PuntaLima(GoGreenPR) 131 13 143 156

Modeledcostnotconsistentwithstatedcontractcost

SanFermínSolarFarm 150 35 185 185

Energycostnotescalatedfromonlineyear(CY2014)

WindmarRenewableEnergy 162 35 197 197

*CEPR‐AH‐03‐02Attach01.xlsx**SumofreportedenergyandRECcosts.‡Author’scalculationbasedonvaluesinCEPR‐JF‐01‐22Attach02.xlsx

Fortwooftheprojects(HorizonEnergyandSanFermínSolarFarm),PREPAappearstohavemodeledtheenergycostasifithadnotchangedsincetheunitscameonlineincalendaryears2015and2014,respectively.Thisisunrealistic,consideringthatPREPAassertedthatbothunitshaveacontractual2percentperyearescalationofenergycosts.465Forthethirdproject,the Punta Lima wind farm, we are simply unable to reconcile the costs provided indiscovery466withthemodeledvalue.467

Wehavenoexplanation for thesedeviations.We recommend that in lightof this lackofclarity,theCommissionrequirePREPAtoimmediatelyupdate,andkeepupdatedwheneverchangesoccur,aversionofthespreadsheetprovidedinresponsetoCEPR‐AH‐03‐02.468This

465CEPR‐AH‐03‐02Attach01.xlsx466Id.

467AscalculatedbytheauthorfromvaluesinCEPR‐JF‐01‐22Attach02.xlsx.

468CEPR‐AH‐03‐02Attach01.xlsx;specifically,“PPOAList”tab.


spreadsheet should tabulate the names, owners, contract numbers, initial energy costs,currentenergycosts,initialRECcosts,currentRECcosts,anyrelevantescalators,andthefirstoperationaldate(actualoranticipated)oftheproject.

AsthesedeviationswouldhaveonlyaminorimpactonPREPA’spurchasedpowerbudget,wedonotrecommendthatPREPAberequiredtocorrectthematthistime.

E. EvaluationofPREPA’spurchasedpowerandfuelbudgets

1. Purchasedpower

a. Historically,PREPAhasbeenabletopredictitspurchasedpowerspendingwithacceptableaccuracy

We examined PREPA’s actual and budgeted spending on fossil PPOAs since FY2010 toevaluatewhetherPREPAhasbeenabletosuccessfullypredictitspurchasedpowerspendinginpastyears. Ingeneral,wefindthatPREPA’sfossilpurchasedpowerspendinghasbeencommensuratewithitsbudgets.PREPAisabletoforecastitsspendingatAESwithahigherdegreeofaccuracythanitsspendingatEcoEléctrica,whichiswithinourexpectationsgiventhedifferencesincontractstructuresbetweenthetwoagreements.Assuch,whenPREPAover‐orunder‐spendsitsPPOAbudget,thedeviationsarelargelyduetospending(orlackthereof)atEcoEléctrica.ThispatternisshownclearlyinFigure29.469

469NotethatbudgetdeviationsinFY2016aregenerallynotinformativeasPREPA’sFY2016budgetwas identical to its FY2015 budget (PREPA’s response to CEPR‐AH‐05‐01 of the Commission’sEleventhROI,October19,2016).


Figure 29. Budgeted and actual spending on PREPA's PPOAs with AES and EcoEléctrica, FYs 2010 through 2016.470

Historically, PREPA’s spending at AES has been within four percent of its budgetexpectations,withPREPAover‐budgetingasoftenasitunder‐budgets.471PREPA’sspendingatEcoEléctrica,bycontrast,deviatesbyitsbudgetsby12percentonaverageandasmuchas19percentinthepastfiveFYs.472Over‐orunder‐spendingatEcoElectricis,aswewouldexpect,drivenprimarilybytheenergybudget:PREPA’sspendingonenergyatEcoEléctricahasdeviatedfromitsbudgetsbyoveragesofasmuchas54percentandbyunderspendsofasmuch as 24 percent since FY2010.473 These are large deviations, representing tens ofmillionsofdollarsofdifferencebetweenPREPA’sbudgetsanditsactualspending.474Whilethese fluctuations in spending are balanced by PREPA’s relative certainty regarding itscapacity budget at EcoEléctrica and its total budget atAES, they also highlight again thevulnerablepositioninwhichPREPAfindsitselfduetoitscontracttermswithEcoEléctrica.

470EcoEléctricavaluesfromCEPR‐AH‐06‐02(e)(ii),(e)(iii),(e)(iv),and(f).AESvaluesfromCEPR‐AH‐06‐03(e)(i),(e)(ii),and(f).TotalsaresumofvaluesforAESandEcoEléctrica.

471Author’scalculation,basedonvaluespresentedinCEPR‐AH‐06‐03(e)(i),(e)(ii),and(f).AveragesareirrespectiveofFY2016.

472Author’scalculation,basedonvaluespresentedinCEPR‐AH‐06‐02(e)(ii),(e)(iii),(e)(iv),and(f).AveragesandmaximumvalueareirrespectiveofFY2016.

473Id.

474Id.


In total,PREPA’sspendingon fossilPPOAshashewntowithintenpercentof itsbudgetsthroughouttherecentpast,withtheexceptionofFY2013(whenPREPAoverspentitsbudgetby13percent)andFY2016(inwhichPREPAdidnotprepareanewbudgetandsimplyusedvaluesfromFY2015).475Intheaggregate,wefindthistobeanacceptablelevelofcertaintyandwouldnotexpectsignificantlymoreaccuracyfromPREPA’sforecastsgiventhecontracttermsmentionedabove.

With regards to PREPA’s renewable PPOAs, we again note that the main uncertaintyregardingPREPA’sspendingonoperationalprojectspertainstotheactualoutputoftheseunits rather than the costs thereof. As above, we find that PREPA’s representation ofrenewablecontracttermsisgenerallyaccurate.Therefore,ouroverallconclusionregardingPREPA’s PPOA budgets is that PREPA’s actual spending is expected to be withinapproximately tenpercentof itsbudgetsand that this is anacceptable levelofdeviationconsideringthewayinwhichPREPAcollectsrevenuesrelatedtoPPOAspending.

b. PREPA’s purchased power budget is close to actualspendinginFY2017sofar

OurexpectationinthismatterisborneoutbyanexaminationofPREPA’sactualspendingonPPOAssofarinFY2017.PREPAhasreleasedtwomonthlyreportsforFY2017,coveringthemonthsofJulyandAugust(ofcalendaryear2016).476PREPA’sspendingonPPOAsinthesemonthshasbeenunderitsbudgets,assummarizedinthetablebelow.

Table 20. PREPA's budgeted versus actual spending on purchased power in FY2017.477

Budget($000) Spent($000) %Over(+)/Under(‐)BudgetJuly 70,165 59,024 ‐16%August 72,239 60,206 ‐17%

While these deviations are somewhat larger than theyhave beenhistorically,we seenocauseforconcern,fortworeasons:first,thedeviationsweanalyzedaboveareonanannualbasis andwewould expectmonthly values to be somewhatmore volatile. Second, theseresults indicate that PREPA may have overbudgeted somewhat for PPOAs. Since thedeviationisnotextreme,themoreconservativeresponseistoallowanyoverpaymentfromratepayerstobecompensatedthroughadjustmentsof thePPOAriderratherthantoriskunderpaymentbyadjustingthebudgetdownwards.

475Author’scalculation,basedonvaluespresentedinCEPR‐AH‐06‐02(e)(ii),(e)(iii),(e)(iv),and(f)andCEPR‐AH‐06‐03(e)(i),(e)(ii),and(f).

476ExhibitFisherandHorowitz,Exhibit03.

477ValuesfromPREPA’sJulyandAugust2016monthlyreports.


Recently,PREPAalsopreparedanupdatedbudgetforPPOAspending(andfuel,asdiscussedfurther below) in response to a question from Commission advisors.478 This budgetoriginates fromanewPROMODrunbasedondifferent fuelprice forecasts (asdiscussedfurtherbelow)andincorporatesthreemonths’worthofactualvalues(inotherwords,oneadditionalmonth as compared to the table above). PREPA’s total expected spending onPPOAs in this updated budget differs by only one percent from the value presented inPREPA’soriginalfilinginthisratecase.Thisisaninsignificantdifferencethatiseasilywithintherangeofexpectationsandwouldnotbeexpectedtoleadtoanyformofrateshock.Assuch,webelieveitwouldbemorestraightforwardtosimplymaintainPREPA’sPPOAbudgetintheratecaseasfiled.

Table 21. PREPA's as‐filed and most recent PPOA budget expectations.

RateCase* ResponsetoOct.31CallQ9** %Increase(+)/Decrease(‐)EcoEléctrica $387,433,152 $403,363,647 4%

AES $305,139,587 $309,015,491 1%Renewables $135,820,274 $107,094,044 ‐21%

Total $828,393,012 $819,473,181 ‐1%*PREPA’sPetitionforRateReview,ScheduleA‐6REV.**CEPR161031RequestNo.9Attach01.xslx.

c. PREPA faces significant uncertainty regarding the onlinedatesofnewrenewableenergyprojects

WhenpressedonthedifferencesbetweenthePPOAbudgetoriginallypresentedintheratecaseandthatpreparedinresponsetothequeryfromCommissionadvisors(asshowninthetableabove),PREPArespondedthattherearetwofactorsexplainingthedeviations:first,the“realitiesofdispatch”ledtosmallchangesinexpectedspendingonfossilPPOAs.479Second,andmoreconcerningly,notalloftherenewableenergyprojectsfromwhichPREPAexpectedtopurchasepower inFY2017haveactuallycomeonline.480Weunderstandthis tobe themaincauseofthe21percentdropinPREPA’sexpectationsregardingspendingonrenewablePPOAs.481Althoughwerequestedmoreinformationonthistopic,itwasnotprovidedintimetobeincorporatedintothisreport.

PREPA’s uncertainty with regards to the status of its contracts with pre‐operationalrenewableenergyprojectswasamajorpointofcontentionintheIRPproceedingandwas

478 PREPA’s response to the Request 9 of the Commission’s October 31 Conference Call ROI(November14,2016).RefertoCEPR161031RequestNo.9Attach01.xslx.

479CEPR‐AP‐2015‐0001,November15,2016ConferenceCall.

480Id.

481Author’scalculation,basedonvaluesinScheduleA‐6REVandCEPR161031RequestNo.9Attach01.xlsx.


discussedatsomelengthintheCommission’sFinalOrderontheIRP.482WearedismayedthatPREPAfacessuchahigh levelofuncertaintywithregards to thesecontracts that itsexpectations regarding spending on renewable energy can change by over a fifth,representingalmost$30millioninspending,inlessthansixmonths.Thisoccurrenceonlyunderscores the Commission’s concerns with regards to PREPA’s large number ofoutstandingcontractsforunrealizedrenewableenergyprojects.

d. Noadjustmentofthepurchasedpowerbudgetisnecessaryatthistime

Consideringtheevidenceanalyzedabove,wefindnocompellingcausetoadjustPREPA’stotalPPOAbudgetatthistime.WerecommendthattheCommissionmaintainPREPA’sPPOAbudgetasoriginallyfiled,withtheunderstandingthattherelativeproportionsofPREPA’sactualspendingthatoccursatEcoEléctrica,AES,andrenewableprojectsmaydeviatefromthevaluespresentedinScheduleA‐6REV.

We also recommend that the Commission order PREPA to prepare an updated,comprehensive database (in spreadsheet form) of its renewable energy contracts, asdescribedabove.TheCommissionshouldalsorequirePREPAtoupdateitatregularintervalswith regards to itsmost current expectedonlinedates of each contractedproject. Thesereporting requirements should be considered supplemental, rather than replacing therequirementssetforthintheFinalOrderontheIRP.483

2. Fuel

a. PREPAhassignificantlyunderbudgetedforfuel

i. Evidenceforunderbudgeting

ThedeviationbetweenPREPA’s fuel forecastsandactual fuelprices suggested tous thatPREPAhadlikelyunderbudgetedforfuel.Assuch,wesoughttocomparePREPA’sactualfuelspendinginFY2017thusfartoitsfuelbudget.Asportendedbythegapbetweenforecastedandactualfuelprices,wefoundasignificantdeviationbetweenPREPA’sbudgetedandactualspending.WeexaminedPREPA’sbudgetedandactual fuel spending in thesame JulyandAugustmonthlyreportsmentionedabove.WefoundthatPREPA’sfuelspendinginthefirsttwomonthsofFY2017wasdouble itsexpectedoutlay.WesummarizethesedeviationsinTABLE below. We therefore conclude that PREPA’s FY2017 fuel budget significantlyunderstatesitslikelyspendingonfuelinFY2017.UnlikePREPA’sspendingonPPOAs,thesedeviationsarenotableandindicateaneedforadjustment.

482IRPFinalOrder,§IV(F)(4).

483IRPFinalOrder,§VII(B)(1)(i).


Table 22. PREPA's budgeted versus actual spending on fuel in FY2017.484

Actualspending

%DeviationoverBudgetBudget($000)

No.2($000)

No.6($000)

NG($000)

Total($000)

July 53,062 54,190 38,905 20,106 113,201 113%August 56,455 43,615 43,213 23,755 110,583 96%July+August 109,517 97,805 82,118 43,861 223,784 104%

ii. Motivationforadjustment

AlthoughPREPAhastakenpainsinrebuttaltestimonytopointoutthatitisguaranteedtorecover itsactual costson fueldue to thestructureof the fueladjuster clause,485westillbelievethistobeaworthwhileexercise.PREPA’sattestationthatincreasesinfuelcostshave“noeffectonrates”486iscertainlyfalsefromthepointofviewofPREPA’sconsumers.PREPAhasunderbudgetedforfuelbyapproximately60percent,indicatingthattheactualcostofthefueladjusterclauseislikelytobesignificantabovewhatPREPAhaspurportedinitsratecasepetition.Weareconcernedthatsuchasignificantdeviationwouldleadtosurprisesforratepayers,whichliescountertotheratemakingprinciplesofstabilityandpredictability.

Moreover,PREPAuseditsunrealistically‐lowfuelvaluesthroughoutitsratecasefiling.AsdiscussedbyCommissionadvisorPaulChernick,thesevaluesimpactPREPA’scalculationofmarginal costs and calculation of its cost of service. The significant and wide‐reachingimpactsofthedeviationbetweenPREPA’sfuelbudgetandarealisticexpectationofPREPA’sspendingonfuelmeanthatacorrectionisofparamountimportanceforthiscase.Therefore,werecommendthattheCommissionorderanadjustmenttoPREPA’sfuelbudgettobringitsfiledratesmoreinlinewithrealisticspendingexpectations.

b. CalculationofamorerealisticFY2017fuelbudget

Inlightofthesedeviations,weusedseveralmethodstocalculateamorerealisticexpectationforPREPA’sspendingonfuelinFY2017.Wewillnotehereexplicitlythatallofthesemethodsaremeanttoprovideback‐of‐the‐estimatesestimatesonlyandwerecalculatedwiththebest

484ValuesfromPREPA’sJulyandAugust2016monthlyreports(ExhibitFisherandHorowitz04).WeassumedthatPREPA’s listedspendingon“Gas” fuelreferstoNo.2;“Oil” fuelreferstoNo.6;and“NaturalGas”referstonaturalgasburnedatCostaSur.

485PREPA’sPetitionforRateReview,Exhibit23,RebuttalTestimonyofFrancisX.Pampush,LucasD.Porter,andDanT.Stathos.lines728‐731.

486Id.,line730.


data available to us. We do not endorse any of these methods for PREPA’s budgetingprocessesinthefuture.

ThethreemethodsweusedtoestimaterealisticFY2017fuelspending,487andtheresultsofthoseestimates,areelaboratedbelow.

i. Method1:Prorationbyindividualfuels

Inthefirstmethod,weexaminedwhatpercentageofPREPA’sspendingoneachfuelituseshasoccurredinthemonthsofJulyandAugustinFYs2010through2016.Foreachfuel,wefoundtheaveragepercentageofspendingthatoccurredinJulyandAugustoftheseyears.We then divided spending by fuel by these percentages to arrive at a by‐fuel proratedexpectation of total spending in FY2017. Mathematically, this proration is justified byobservingthat,foranygivenportionofayear:

Spart/Sfull=fpart

whereSpartisspendinginagivenpartoftheyear,Sfullisspendingoverthecourseofthefullyear, and fpart is the fractionof spending that occurred in the identifiedpart of the year.Therefore:

Spart/fpart=Sfull

The percentages of annual fuel spending that occur in July and August by fuel and forPREPA’ssystemasawholearesummarizedinthetablebelow.

487 Regarding this analysis as a whole, we will note here the assumption that PREPA providedhistoricalvaluesinnominaldollars(i.e.,thesevalueswerenotcorrectedforinflation).Becauseouranalysesinthissectionrestonratiosorpercentagesofdollarsspentinthesameyearasoneanother,correctingforinflationwouldnotimpactourresults.


Table 23. Historical percentages of annual spending on fuel occurring in July and August.488

PercentofAnnualSpendingOccurringinJulyandAugustFY No.2 No.6 NG* Total2010 21% 14% ‐ 15%2011 13% 15% ‐ 15%2012 15% 19% 0% 18%2013 19% 20% 11% 19%2014 23% 17% 18% 18%2015 25% 21% 23% 22%2016 15% 26% 19% 21%Average 19% 19% 18% 18%

*NG=naturalgas.PREPAbeganburningnaturalgasatCostaSur inFY2012.Theaveragepercentage fornaturalgas isthereforeforfiscalyears2013‐2016only.

Usingthismethodasappliedtotheby‐fuelJulyandAugustspendingshowninTABLEabove,weestimatedexpectedspendingbyfueltobeapproximately$527millionforNo.2fueloil;approximately$435millionforNo.6fueloil;and$249millionfornaturalgas,suggestingatotalfuelbudgetof$1.211billionbeforeperformance‐relatedsavingsaretakenintoaccount.

ii. Method2:Prorationbytotal‐systemfuelspending

Inthismethod,weproratedPREPA’stotalsystemspendingonfuelratherthanproratingbyindividual fuel.Thismethod returns slightlydifferent results thanMethod1because thepercentagesofspendinginJulyandAugustbyfuelarenotidenticaltooneanotherortothepercentage of total system spending on fuel in July and August. This method does notproduce an estimateof total spendingby fuel and allows forpossibility that the relativeratiosofPREPA’sspendingondifferentfuelsmayshiftthroughouttheyear.

Usingthismethod,weestimatePREPA’stotalFY2017spendingonfueltobe$1.225billionbeforeperformance‐relatedsavingsaretakenintoaccount.

iii. Method3:Expectationofspendingbasedonshort‐termpriceforecastsandPREPA’seffectivemultiplieronfuelindices

Method 3 uses a different conceptual framework than Methods 1 and 2. Rather thanpredictingPREPA’sfull‐yearspendingfromspendinginFY2017thusfar,wepredictitbasedon the historical relationship between fuel indices (Henry Hub and West TexasIntermediate)andPREPA’sactualfuelpricesas incurred.WedividedPREPA’sactual fuelpricesonamonthlybasisbyhistoricalsettledHenryHub(HH)andWestTexasIntermediate

488PREPA’sresponsestoCEPR‐AH‐06‐01(d)andCEPR‐AH‐06‐04(a)oftheCommission’sFourteenthROI(November3,2016andOctober25,2016,respectively).


(WTI)pricesasrecordedbytheEnergyInformationAdministration.489Weaveragedthesevalues for FYs 2015 and 2016, finding that on average PREPA pays 3.49xHH prices fornaturalgasona$/MMBTUbasis,andpays1.7xand1.2xWTIpricesfordistillateandresidualfueloils,respectively,ona$/bblbasis.490

Wethenmultiplied these inflatorsbyexpectedHHandWTIprices fromthe8NovembereditionofEIA’sShort‐TermEnergyOutlook(STEO),which ispublishedmonthly.491STEOpricesare short‐term, look‐ahead forecastswithvalues foroneyearpast thepublicationdate.Althoughtheycannotbeusedforlong‐termplanningpurposes,STEOforecastsareareliablesourceofforecastedfuelpricesforthenear‐term.BymultiplyingSTEOforecastsbyourestimateofPREPA’saverageinflators,wearrivedatashort‐termforecastofPREPA’slikelypricesfordeliveredfuelsoverthecourseofFY2017.

Finally, we multiplied these expected fuel prices by PREPA’s monthly forecasted fuelconsumptionasreportedinitsPROMODoutputs492toarriveatexpectedmonthlyspendingon a by‐fuel basis.We emphasize here that inputting different fuel prices into PROMODwouldcertainlyresultinchangestothedispatchlevelsofPREPA’sunits,therebyalteringexpectedfuelconsumption.However,wenoteagainthatthismethodismeanttoprovideaback‐of‐the‐envelope estimate of PREPA’s total spending on fuel rather than an exactpredictionofspendingbyfuel.PREPA’sPROMODoutputsprovidenaturalgasconsumptionincubicfeet.Weassumedaheatcontentof1050BTU/ft3fornaturalgastoallowuseofafuelforecastin$/MMBTU,whileweusedtotalbarrelsoffueloilconsumptionandaper‐barrelfuelpriceforecastforfueloil.493

Through use of this method, we arrived at estimated FY2017 spending levels ofapproximately$257milliononNo.2fueloil,approximately$650milliononNo.6fueloil,andapproximately$341milliononnaturalgas,resultinginatotalestimatedFY2017fuel

489MonthlyHenryHubpricesareavailableonlineat:https://www.eia.gov/dnav/ng/hist/rngwhhdm.htm.MonthlyWTIpricesareavailableonlineat:http://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=rwtc&f=D. 490Author’scalculation,basedonHHandWTIpricesfromEIAaswellasdatafromCEPR‐AH‐06‐05Attach01.xslxandCEPR‐AH‐06‐01(d).

491STEOdataisavailableonlineat:http://www.eia.gov/beta/steo/.


493WeassumedthisvaluebecauseitappearstobewhatPREPAusedinmodeling:wewereabletorecreatePREPA’snaturalgasfuelpriceforecast(whichisin$/MMBTU)exactlybyfirstmultiplyingmonthlynaturalgasconsumptionincubicfeetbythisvalueandthendividingmonthlyexpendituresonnaturalgasbytheresult.However,wenotethatitcontradictsPREPA’sassumptionintheIRPofa1,000BTU/ft3heatcontentfornaturalgas,assetforthinTable5‐2ofVolumeIoftheBaseIRP.Weareunabletoexplainthediscrepancy.


budgetofapproximately$1.248billion,beforeperformance‐relatedsavingsaretakenintoaccount.

Ourresultsfromallthreemethods,aswellasPREPA’soriginalfuelbudget,areshowninthetable below. To avoid false precision, we have rounded by‐fuel values to the nearestthousandandtotalfuelbudgetstothenearestmillion.

Table 24. PREPA's as‐filed FY2017 fuel budget and author’s calculations of estimates of expected spending.

No.2 No.6 NG Handling/Others Total BudgetasFiled* 168,314,322 387,448,919 202,010,094 5,921,743 763,695,078Expected Method1 527,219,000 434,726,000 249,079,000 ‐ 1,211,000,000Method2 ‐ ‐ ‐ ‐ 1,225,000,000Method3 257,468,000 650,429,000 340,520,000 ‐ 1,248,000,000

*ScheduleA‐6REV.

iv. Otherparties’calculationsofrevisedfuelbudgets

BothPREPA494andthepanelofTomSanzilloandCatherineKunkel(onbehalfofISCE‐PR)495havepresentedupdatedestimatesofwhatthesepartiesbelievePREPAwillspendonfuelinFY2017.PREPA’sestimatewaspresentedinresponsetoadirectrequestfromCommissionadvisors,496whileSanzilloandKunkeldescribedtheirestimateofPREPA’sbudgetshortfallintestimony.497Forsimplicity,weinterpretedSanzilloandKunkel’sassertionthat“PREPA’sactual[fuel]costfor2017[may]beasmuchas$400‐$500millionmorethantheproposedbudget”498aspropositionofa$450millionincreaseinPREPA’sproposedbudget.

Meanwhile,PREPApresentedanewtotalfuelbudgetasoutputfromanewPROMODrun,asdiscussedabove.PREPA’snewestimationofitsFY2017fuelbudgetis$1,222,696,925.499WenotethatthisPROMODrunwasbasednotonanewandupdatedsetoffuelpriceforecastbutratheronasetofolderfuelpriceforecaststhanthoseusedinPREPA’soriginalpetition.500TheseforecastswerealsopreparedbySiemens,inDecemberof2015.501

494CEPR161031RequestNo.9Attach01.xlsx

495CEPR‐AP‐2015‐0001,testimonyofTomSanzilloandCathyKunkelonbehalfofICSE‐PR.


497TestimonyofTomSanzilloandCathyKunkelonbehalfofICSE‐PRatlines305‐307.

498Id.atline306.


500November15ConferenceCall.

501Id.


We find PREPA’s choice to go back to a prior forecast rather than preparing (or havingpreparedonitsbehalf)anewforecasttobenonsensical.AsthedeviationbetweenPREPA’sBaseIRPandUpdatedFuelIRP/ratecaseforecastsandrealvalues(showninFIGUREabove)shouldmakequiteclear,fuelpricesarevolatileanddifficulttopredictwithaccuracy.Themostrecentandup‐to‐datedatashouldbeusedwheneverpossible.Theproblemwiththeforecastsoriginallyused inthisproceeding isnot thedateonwhichtheywereprepared.Rather,Siemensintentionallypreparedthoseforecastsasa“lowerbound”onexpectedfuelprices,502anditwasclearlyinappropriateforPREPAtousethemforbudgetingpurposes.Lower and upper bound sensitivity forecasts are of crucial importance for evaluating autility’s risk to fluctuations in fuelprices, but theyarenot suitable foruse in short‐termplanningandbudgeting.Instead,PREPAshouldhaveuseditsmostrecentexpectationofthebasecase—themostlikely—trajectoryforfuelprices.

The fact that ithasbeenalmostayearsincePREPA’s lastbasecase fuelprice forecast isequallyconcerning.BecausePREPA’scostsaredominatedbyfuel,itbehoovestheutilitytomaintainanup‐to‐dateunderstandingofthelikelyfuturepricesofthatfuel,regardlessofhowtheutilityiscompensatedforitsfuelspending.WerecommendthattheCommissionrequirePREPAtopreparefuelpriceforecasts,orhaveforecastspreparedonitsbehalf,atleastbiannually.Wefurtherrecommendthat theCommissionrequirePREPAtosubmitacomprehensivedescriptionoftheforecastingmethodologyanddatasourcesusedbyitselforitsadvisorsfortheCommission’sreviewaspartofitsinitialfilinginthenextmajorrate‐orplanning‐relatedcaseitbringsbeforethisCommission.

v. Suggestedadjustment

WesummarizetheresultsofallthevariousbudgetestimationmethodsdescribedaboveinTABLE.Weobservethat,ingeneral,thedifferencebetweenthesevariousmethodsissmall:thereisgeneralagreementamongallparties(includingPREPA)thatPREPA’sfuelbudgetforFY2017wastoolowbyapproximately60percentorbetween$400and$500million.

Table 25. Comparison of proposed adjustments to PREPA's FY2017 fuel budget.503

IncreasefromFiledBudgetParty EstimatedBudget*($) $ %PREPA(As‐Filed)** 763,695,078 ‐ ‐FisherandHorowitz Method1 1,211,000,000 447,304,922 59%Method2 1,225,000,000 461,304,922 60%Method3 1,248,000,000 484,304,922 63%

PREPA(New)† 1,222,696,925 459,001,846 60%SanzilloandKunkel‡ 1,213,695,078 450,000,000 59%

*Estimatedbudgetsarepriortoexpectedsavingsfromperformanceimprovements.**ScheduleA‐6REV

502IRPTechnicalHearing,April6,2016,NelsonBacalao.

503Suggestedadjustmentisbolded.


†CEPR161031RequestNo.9Attach01.xlsx‡Author’scalculationbasedontestimonyofTomSanzilloandCathyKunkelonbehalfofICSE‐PRatlines305‐307.

WerecommendthattheCommissionadoptoursecondproposedmethodasitresultsinthemedianestimateofthethreemethodsdescribedabove.Thismethodsuggestsanadjustedfuelbudget,beforeperformancesavings,of$1,225,000,000intotalforNo.2fueloil,No.6fueloil,andnaturalgasburnedatCostaSur.WethereforerecommendthattheCommissionadjustPREPA’sfuelbudgetupwardsby$461,305,000.

WedonotrecommendthattheCommissionadoptPREPA’srevisedbudgetfortworeasons:first, our calculation, although back‐of‐the‐envelope, is based on current short‐termforecasts,whilePREPA’sisbasedonaforecastthatisalmostayearoutofdate.Wecannotendorse the use of such stale data. Second, because we are not recommending that theCommissionupdateanyoftheothervaluesthatmayarisefromanewPROMODrun,thenewfuelbudgetwouldproceedfromadatasetthatisinconsistentwiththerestoftheproposedrevenuerequirement.504

vi. Status of fuel‐related performance improvementsavings

Throughout its rate case petition, one of the forms of “known andmeasurable changes”describedbyPREPA is savingsexpected from improvements inPREPA’sperformance. Inotherwords,PREPAhaspresentedbudgetsthatare,intheory,predicatedontheassumptionthatPREPA’soperationsareexactlyasefficientastheywereinthetestyear.TheseconceptsaredescribedinmoredepthbyotheroftheCommission’sadvisors.505Forourpurposes,itisimportantonlytonotethatPREPAhaspresentedafuelbudgetthatiscomparabletoitspastspendingandclaimedthatitwillinactualityspendlessthanthatbudgetduetoimprovedcorporate performance. PREPA’s expected savings due to fuel‐related performanceimprovementsamountto$107,726,711inFY2017.506

During the October 20 Conference Call, PREPA described these savings as “run‐rate,”meaning that theyhavealreadybeenpartiallyrealized,andstatedthat itexpects the fullvalue of the savings to be realized over the course of the year. In discovery from theCommission,PREPAconfirmedthatitexpectstheslightlymorethanthefullvalueofthefuel‐related performance savings claimed in the revenue requirement to be achieved in

504WerecognizethedifficultyinclaimingboththatthefuelpriceforecastusedinPREPA’soriginalPROMODrunisunsuitableandinrecommendingthattheCommissioncontinuetorelyontheoutputsofthisrunregardless.Simplystated,PREPAhasnotprovidedaPROMODrunbasedonfuelpriceforecastsofanacceptablevintageandquality.Leftwithonlyunsatisfactoryalternatives,wehavechosenthesimplest.

505RefertothereportofCommissionadvisorsRalphSmithandMarkDady.

506PREPAEx.14.02.xlsm,“Inputs”tab,cellH27.


FY2017.507WethereforeseenoreasontoadjustPREPA’sclaimedperformancesavingsandhavemaintainedthesesavingsinourcalculationofanadjustedFY2017fuelbudget.

F. Recommendation

1. Fuelbudget

WerecommendthattheCommissionadjustPREPA’sFY2017upwardsby$461,305,000,foratotalFY2017fuelbudgetof$1,117,273,289aftersavingsfromperformanceimprovements.

2. Fuelpriceforecasts

WerecommendthattheCommissionrequirePREPAtopreparefuelpriceforecasts,orhaveforecasts prepared on its behalf, at least biannually. We also recommend that theCommission audit PREPA’s fuel price forecasting methodology, which the CommissionshouldrequirePREPAtosubmitaspartofitsinitialfilinginthenextmajorrate‐orplanning‐relatedcasePREPAbringsbeforethisCommission.

3. Purchasedpowerbudget

WerecommendnochangetotheFY2017budgetforpurchasedpower.

4. Renewableenergycontractstructuresandstatusupdates

WerecommendthattheCommissionrequirePREPAtoestablishadatabase,inspreadsheetform, of its contracts with renewable energy providers as described above. We furtherrecommendthattheCommissionrequirePREPAtoupdatethisdatabaseinaregularandtimelyfashion.

507PREPA’sresponsetoCEPR‐RS‐05‐17(d)oftheCommission’sTwelfthROI(October11,2016).


VII. OPERATINGEXPENSEBUDGET


Inthischapter,weaddressPREPA’sbudgetforspendingonoperationsexpensesotherthanfuelandpurchasedpower.PREPA’soperationalbudgetincludesthemajorityofitsplannedspendingonlabor,508aswellasitsplannedspendingonthematerialsandservicesneededtooperate its plants, maintain its transmission and distribution system, and function as acompany. In its revenue requirement, PREPA presents both the labor and non‐laboroperationalbudget,aswellasabreakdownoftotaloperationalspendingbyfunctionarea.509

TheCommissionmustdecide if thePREPA’s total operationsbudget, and theallocationsthereof into labor and non‐labor and functional area expense budgets, is reasonable. Inparticular,theCommissionmustascertainwhetherPREPA’soperationsbudgetissufficientbut not excessive to allow PREPA to operate a safe and reliable electric system. If theCommission decides that PREPA’s operations budget and the breakdowns thereof arereasonable,nochangewillberequired.IftheCommissiondecidesthatPREPA’soperationsbudgetisnotreasonable,itmustthendecidewhetherornottomakeadjustmentstothatbudget.

Below, we discuss the significance of the operations budget, describe the evidencesupporting that budget (or lack thereof), and comment on several concerning patternsobserved in PREPA’s historical spending and FY2017 budget. We come to severalconclusionsregardingPREPA’soperationsbudget:

1. First,andmostimportantly,weconcludethatPREPA’soperationalspendinghasnotbeenconsistentwithoperationofasafeandreliablesystemsinceatleastFY2014.Thispointisofcrucialimportanceforthisratecaseasawholeandweemphasizeitssignificancebelow.PREPA’soperationalspendinghasbeenrestrictedinrecentyearsandtheperformanceofboththegenerationandtransmissionsystemshavedeclinedaccordingly,accordingtobothourownanalysisandexplicitstatementsfromPREPAtothiseffect.Weareconcernedthat,evenbeforeFY2014,PREPA’s

508Approximately14percentofPREPA’shistoriclaborspendinghasbeenrelatedtocapitalprojects(author’s calculation based on values provided in PREPA’s response to CEPR‐JF‐01‐24(a) of theCommission’sFourthROI,August23,2016,PUBLIC;refertoCEPR‐JF‐01‐24Attach02.xlsxthroughCEPR‐JF‐01‐24Attach05.xlsx).However,PREPA’scapital‐relatedlaborbudgetisconvolvedwithitsoverall capital budget and difficult to audit. Our analysis does not address capital‐related laborspendingexplicitlyorseparatelyfromoverallcapitalspending.

509PREPA’sfunctionalareasincludegeneration,customerbilling,administrativeandgeneral,andtransmissionanddistribution.Transmissionanddistribution listedindependently inthefinancialmodelbutaregroupedtogetherinhistoricalrecordsprovidedbyPREPA,soitisnotclearwhetherornotthesedirectoratesareconsideredtobeonesingleortwoseparatefunctionalareasbyPREPA.


budgetingprocessencouragedausteritytothepointthatPREPAmustspendsufficientlyonitsoperationssuchthatitcanoperateasafeandreliablesystemonagoing‐forwardbasis.

2. Second,wefindthatPREPAmadesolittledocumentationoftheprocessandvaluesunderlyingthebudgetavailabletoCommissionadvisorsthatwehavenochoicebuttoconsiderthebudgetandtheallocationthereoftobe,ineffect,unsupported.Mostespecially,wefindthattheallocationofoperationalexpensesbyfunctionalareaislikelytobearlittle,ifany,relationshiptothebreakdownofPREPA’sactualFY2017operationalspending.510

3. Third,weobservethatamajorcomponentofPREPA’soperationalspendingoccursinitsAdministrativeandGeneral(A&G)functionalarea,andthatspendinginthisareahasincreasedinrecentyearsforunexplainedcauses.WerecommendthattheCommissioninvestigatethisissuefurtherinitspendinginvestigationregardingPREPA’sperformance.

4. Finally,throughcomparisonwithhistoricalspendingandoperationalpatterns,weproposeaseriesofreallocations,recategorizations,andcorrectiveadjustmentstobetteralignPREPA’soperationalbudgetwithitsapparentneedsforoperatingasafeandreliablesystem.

B. ConstituentelementsofPREPA’soperatingexpenses

PREPA’snon‐fueloperatingexpensebudget511forFY2017totals$559,752,076.512Thisvalueiscomprisedofspendingonlaborandotheroperationalandmaintenanceexpenses.PREPAprovidedthelaborandnon‐laborcomponentsoftheoperationsbudgetinScheduleA‐2REV.Asisthecasewithfuel,PREPApreparedbothabasebudgetandaproposedactualbudgetaftertakingintoaccountexpectedsavingsfromperformanceimprovementinitiatives.WecalculatethatPREPAwouldexpect itstotaloperationsspendingtobe$662,502,076ifno

510 This conclusion hasmajor implications for PREPA’s cost of service study, as itmay call intoquestiononeofthebasesofPREPA’scostallocation.ThissubjectisdiscussedbyCommissionadvisorPaulChernick.

511PREPApresentsvaluesforbothits“TotalLabor&OtherOperatingExpenses”andits“TotalNon‐FuelO&MExpense”inScheduleA‐1REV.Thedifferencebetweenthetwoisasinglelineitementitled“CILTSubsidyRecoveryRequiredinBaseRate.”CILTandsubsidiesarethepurviewofCommissionadvisorPaulChernickandwedonotaddresstheminthisreport.Instead,ouranalysisherefocusesonthevaluedescribedas“TotalLabor&OtherOperatingExpenses.”

512SumofFY2016Non‐FuelO&MExpenses,ScheduleE‐6REV.


performance‐relatedsavingswereachieved.513PREPAestimateditsFY2017performancesavingsatavalueof$102,750,000.514

Below, we describe the constituent elements of PREPA’s operating expenses, which cangenerallybecategorizedaseitherlabor‐ornon‐labor‐related.WealsoreviewhowPREPA’soperationalspendinghaschangedoverthecourseoftherecentpast.

1. Laborexpenses

As presented in PREPA’s financialmodel, labor expenses consist of salaries515 aswell asspending on pensions516 and benefits.517 Salaries and wages have historically made uproughly two‐thirdsofPREPA’soperational laborspending.518PREPAexpectsspendingonlabor in FY 2017 to be $434,937,821 before savings expected from performanceimprovements,519ofwhich64percentisexpectedtobespentonsalaries.520

PREPA’soperationallaborbudgetcomprisesitscompensationforalloftheworkersthatareinvolvedinPREPA’sday‐to‐dayoperations.CompensationforPREPA’sadministrativeandexecutivestaff,aswellasallof itscustomerserviceemployees, iscontainedinthisvalue.PREPA’soperationallaborbudgetalsoincludescompensationforlinesmen,meterreaders,andtheplantworkersthatoperateandmaintainPREPA’sgeneratingunits.521Indeed,ofthe6,694employeesincludedinPREPA’sFY2017operational laborbudget,63percent522areworkersintheGenerationandT&Ddirectorates.523Withoutthelaboroftheseemployees


514Id.

515Weunderstandthisheadingtorefertocompensationofbothsalariedandwageworkers.

516Weunderstandthisheadingtorefertopensionfunddepositsonbehalfofcurrentemployees,ratherthanpaymentstocurrentretirees.

517PREPAEx.14.02.xlsm,“Inputs”tab,lines58‐59.

518 Id., author’s calculation (FY2010‐2016 average of “Salaries” values as a percentage of “TotalOperatingExpenseLaborCosts”is64%).

519ScheduleA‐2REV.

520Author’scalculationbasedonvaluesprovidedinPREPAEx.14.02.xlsm,“Inputs”tab,lines58‐59.

521 Author’s inference based on values provided in PREPA’s response to CEPR‐AH‐06‐13 of theCommission’sFourteenthROI(November4,2016).RefertoCEPR‐AH‐06‐13Attach01.xlsx.

522Author’scalculationbasedonvaluesprovidedinPREPA’sresponsetoCEPR‐MC‐01‐03(a)oftheCommission’sFourteenthROI(November1,2016).

523Wenoteherethatbecauseemployeecompensationratesarenotidenticalforallpositions,thisstatisticdoesnotindicatethat63percentofPREPA’slaborspendingisdevotedtotheGenerationandT&Ddirectorates.


(whichmust certainly be compensated), PREPA’s systemwould simply cease to operatecorrectly or at all—aswe shall see below. Although PREPAmay be able to improve theefficiencyofitsuseofstaff,thereisnowayforPREPAtooperateasafeandreliableelectricsystemwithoutsufficientspendingonemployeelabor.

2. Non‐laborexpenses

PREPA’spre‐savingsnon‐laboroperationsbudget,whichitlabelsas“Unclassifieddivisionexpenses” in Schedule A‐2 REV, is $227,564,256.524 This budget is further elaborated inPREPA’s financial model, where it is revealed as comprising ten different categories ofspending. The table below summarizes these categories and their budgeted amounts forFY2017.

Table 26. PREPA's proposed spending on non‐labor expenses by category of spending.525

CategoryofSpendingAmountBudgetedin

FY2017PercentofFY2017Non‐Labor

OperationsBudgetAdditionalSafetyUpgrades $4,432,500 2%Materials $32,863,721 14%Transportation,PerDiem,andMileage $33,753,326 15%Property&CasualtyInsurance $20,200,000 9%Restructuring $28,000,000 12%RetireeMedicalBenefits $22,220,000 10%Security $21,210,000 9%BankingServices $14,140,000 6%MaintenanceandUtilities $17,957,800 8%OtherMiscellaneousExpenses* $32,786,908 14%

Total $227,564,255 100%*AccordingtoPREPA’sbusinessplan,miscellaneousexpensesconsistprimarilyofrent,legalfees,utilities,andprofessionalservices.526

Apartfromonlyafewparticularexceptions,wewillnotdiscussthesecategoriesindepth.527We note, however, that this budget includes some costs that are essential to PREPA’soperations.528Forexample,allofthecostofallthephysicalgoodsneededfortheroutine,

524ScheduleA‐2REV.

525PREPAEx.14.02.xslm,“Inputs”tab,lines69‐79.

526PREPAEx.3.02,p23.

527Asanote,weobservethatPREPA’snon‐fueloperationsbudgetdoesnotincludeanyspendingonfinesorpenalties,forexampleonpenaltiesrelatedtonon‐compliancewithMATSorthePuertoRicoRenewablePortfolioStandard.PREPAconfirmedduringtheOctober20ConferenceCallthatnosuchfineswere included in the revenue requirementdespite the fact that PREPA is currently innon‐compliancewithmultipleapplicablepolicies.

528 This statement does not imply, however, thatall of the spending included in this category isessentialtoPREPA’soperations.


day‐to‐daymaintenance of PREPA’s plants andwires is contained in this budget.529 ThisbudgetalsocoversthecostofthetransportationneededforPREPA’semployeestotraveltoits lines530 (which are sometimes located in remote or difficult‐to‐access areas531) formaintenanceandrepairpurposes.Asabove,althoughPREPAmaybeabletoimprovetheefficiencyofitsuseofmaterialsandservices,sufficientoperationalspendingisnecessaryforPREPAtooperateasafeandreliableelectricsystem.

3. Expensesbyfunctionalarea

Inadditiontothebreakdownintolaborandnon‐laborexpensesprovidedbyPREPAinthefinancialmodelandScheduleA‐2REV,PREPA’soperationalbudgetisalsobrokendownbyfunctionalarea inSchedulesA‐1REVandE‐6REV.Asdescribedbelow, themethodologyusedbyPREPA (or its advisors) to calculate theseby‐areavalueswasadhoc at best.Asstated,PREPA’sfunctionalareaoperationalbudgetsforFY2017aresummarizedinthetablebelow.

Table 27. PREPA's proposed allocation of operational expenses by functional area.532

FunctionalAreaAmountBudgetedin

FY2017PercentofFY2017Operations

BudgetGeneration $122,410,515 22%Transmission $34,222,179 6%Distribution $169,277,149 30%CustomerService/Billing $84,944,941 15%AdministrativeandGeneral $148,897,292 27%

Total $559,752,076 100%

A sound presentation of anticipated operations spending by functional area is of greatimportancetothisCommission,asitisthroughthesevaluesthattheCommissioncanjudgetheextenttowhichPREPAisplanningforadequatelevelsofmaintenanceactivitiesatitsgeneratingunitsandT&Dinfrastructure.Ideally,generationandT&Doperationalbudgetswould be supported by PREPA with a strategic plan and clear documentation detailingplannedmaintenanceandoperationalactivities,expectedstaffingrequirements,expectedspendinglevels,andanybufferamountsheldinreserveincaseofunplannedoccurrences.Generation expenses, in particular, should be straightforward for the Commission toreconcile with PREPA’s planned generation levels by plant. Customer service and A&Gbudgetsshouldbesimilarlysupportedwithaclearelaborationofstaffandfundsthatmustbe committed for both ongoing activities, newprograms, and special internal initiatives.

529Author’sinferencebasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xlsx.

530Id.

531PREPA’sresponsetoCEPR‐AH‐02‐01(e)oftheCommission’sSixthROI(PUBLIC;August29,2016).

532ScheduleE‐6REV.


Unfortunately,thisisidealisfarfromwhatPREPApresentedtotheCommissioninitsratecasepetition.

C. PREPA’sFY2017operatingbudgetiseffectivelyunsupported

Asdescribedabove,whilethedocumentationprovidedbyPREPAinsupportofitscapitalbudgetisthin,wecanatleastobservethatPREPA’stotalcapitalbudgetisthesumofalargenumberofindividualprojectsorbudgetitems.PREPAenumeratedtheseprojectsonanitem‐by‐itembasisinScheduleF‐3REV.Furthermore,wecantracetheoriginsofPREPA’sfuelandpurchased power budgets directly to outputs from PROMOD. For PREPA’s operationalbudgets, we have no such support. In this section, we discuss the informationwewereprovided by PREPA with regards to its FY2017 operational budget. We describe thecalculationoftheoperationalbudgetwithinthefinancialmodel,thecalculationoftheby‐functional‐area sub‐budgets tabulated above, and the unsatisfactory nature of PREPA’sresponsestorequestsforadditionalsupport.

1. PresentationofPREPA’soperatingbudgetinPREPA’sratecase

Theoperationalbudget,aspresentedinSchedulesA‐1REV,A‐2REV,andE‐6REV, istheresultofacalculationbasedonfoursetsofinputsinthefinancialmodel.PREPAusedtheseinputs to calculate (1) expected labor spending; (2) expected non‐labor spending; (3)expected savings from performance improvements; and (4) the breakdown of its laborbudgetbyfunctionalarea.

PREPA calculated its labor budget simply as the sum of two hardcoded values, labeled“Salaries” and “Pension&Benefits.”533The sumof these lines is labeled “TotalOperatingExpenseLaborCosts.”534TheSalariesandPension&BenefitsvaluesforFY2017arenottheresultofanysupportingcalculationswithinthefinancialmodel.535InScheduleF‐4,PREPAdescribeshowitcalculateditsbudgetsforthesecategories:

Salariesareforecastbasedonprojectednumberofemployeesandhourlywages,andassumingtheaverageemployeeworks1,950hoursayear.Headcountfor2016isestimatedtobe6,710employees,decliningto6,395

533PREPAEx.14.02.xlsm,“Inputs”tab,lines58and59.

534Id.,line60.

535PREPAsuppliedsomeadditionaldetailsregardingthesevaluesinitsresponsetoCEPR‐RS‐06‐07of the Commission’s FourteenthROI (October 21, 2016). Refer to CEPR‐RS‐06‐07Attach 01.xslx.However,thesedetailsarescant.Forexample,ratherthancontainingasinglevalueforsalaries,thisdocumentcontainsacalculationoftotalpayrollexpensesasthesumofhardcodedvaluesforsalariesandovertime.ItisnotclearfromthiscalculationhowPREPAcalculateditsbudgetforeithersalariesorovertime.Moreover,thisdocumentcontainsnofurtherinformationregardingthebreakdownofPREPA’splannedspendingbyfunctionalarea.


employeesby2019andstayingconstantforthebalanceoftheforecastperiod.Retirementsareassumedbasedonemployeeswithmorethan24yearsofservice,withnoadditionalattritionassumed.SalariesarebasedonhourlysalariesofPREPAinMarch2015;andgrowby1%peryearfrom2017onwards.

Employeebenefitsarecalculatedas63.19%ofBaseSalarybeginninginFY2016.Benefitsincrease1%annuallyfromFY2017throughFY2019andareheldconstantat66.19%thereafter.Benefitscalculatedforovertimearecalculatedinthesamemannerastheyarecalculatedforsalaries.PREPA’spensionfundingisforecasttostayconstantfromthehistoricaltotheforecastperiod.536

PREPA did not provide any supporting workpapers showing these calculations, nor didPREPAprovideabreakdownofemployeesandlaborspendingbyfunctionalarea.

PREPAcalculated its laborbudgetas thesumofvalues in thecategories listed inTABLEabove.Withtheexceptionofthevalueforthecategorylabeled“AdditionalSafetyUpgrades,”the by‐category values are the product of hardcoded values for FY2016 and hardcodedinflators.Theby‐categoryvaluesforFY2016arenottheresultofanysupportingcalculationswithinthefinancialmodelandwerepresentedbyPREPAwithoutfurtherdocumentation.The inflators for FY2017 are all equal to one percent, except the inflator for the“RestructuringFees”category,whichistwelvepercent.Theseinflatorsarenottheresultofanysupportingcalculations.PREPAmentioneditsassumptionthatitsnon‐laboroperationalexpenses(apartfromrestructuringfees)wouldincreaseatarateofonepercentperyearinScheduleF‐4537butitdidnotattempttojustifythisassumption.

PREPAcalculateditstotalexpectedsavingsfromperformanceimprovementsasthesumofhardcoded values for several different performance initiatives. Those performanceinitiativeswithforecastedsavingsreflectedintheFY2017revenuerequirementarerelatedto increasing disconnection costs; theft recoveries and reduced T&D losses; PREPA’sautomotive fleet and shops; procurement and inventory management; medical benefitsavings;andheadcountreduction.538Theexpectedsavingsvaluesarenottheresultofanysupporting calculations within the financial model. Although PREPA describes theseperformance initiatives in Schedule F‐4,539 it does not explain therein how it calculatedexpectedsavingsvalues.

536ScheduleF‐4,IV(A)and(B).

537Id.,IV(D).

538PREPAEx.14.02.xlsm,“Inputs”tab,lines65and92

539ScheduleF‐4,VI(B)‐(D).


ThetotaloperationalexpensebudgetforFY2017isthesumof laborexpenses,non‐laborexpenses,andperformanceimprovementsavings.540

2. Financialmodelbreakdownofexpensesbyareaisadhoc

PREPAdidnotpresentanyhardcodedbudgets forby‐areaoperationalexpenses. Instead,PREPAcalculatedfunctionalareabudgetsinamannerthatitdescribedthusly:“functionalexpenseisbasedonauditedFY14allocations.”541Fromanexecutionalstandpoint,whatthisstatementmeansisthatPREPAfoundthepercentageoftotal(laborandnon‐labor)spendingforwhicheach functionalareawasresponsible inFY2014andthensimplymultiplied itstotal FY2017 operational budget by these percentages to arrive at by‐area operationalbudgets.542

Wefindthismethodtobeunsatisfactory.Aselaborateduponbelow,thesevalueshavenoinherent relationship to theoperational activitiesplanned to occur in FY2017 in eachofPREPA’sfunctionalareas.ItisnotinfactcleartouswhetherornotPREPAconsidersthesebudgetstobebindingonorevenapplicabletoitsfunctionalareas,orwhethertheseby‐areavaluesaresimplyaconstructforthepurposesoftherevenuerequirementcalculationwithinthefinancialmodel.

StatementsfromPREPAhaveonlyincreasedouruncertainty.WhenaskedwhetherornotPREPAexpectstheproportionsbyfunctionalareaofitstotaloperationalexpensespendingto remain constant over time at FY2014 levels, PREPA’s advisor replied that, at present,PREPA“doesnothaveanexpectationforhowtheseproportionswillchange”543WedonotknowwhetherPREPAmeantthisstatementtoapplytoFY2017oronlytofutureyears.

On a clarification call, PREPA asserted that it budgets for operations spending on a by‐departmentbasis.544PREPA’sadvisorthenadmittedthatthemechanismitusedtoallocatetotaloperationalexpensesbyfunctionalareainitsrevenuerequirementmaynotrelateinanywaytoitsactualbudgetingprocess.545PREPA’sadvisorassertedthatPREPAdidthebestthatitcouldtoprovideoperationalexpensesbrokendownbyfunctionalarea,implyingthat

540 PREPA Ex. 14.02.xlsm, “RF_Schedules” tab, lines 489‐493. PREPA does not present this valueexplicitlybutincludeditimplicitlyintheformulaeunderlyingtheby‐areavaluesinScheduleE‐6.

541ScheduleE‐6REV.

542 PREPA Ex. 14.02.xlsm, “RF_Schedules” tab, lines 489‐493. Hardcoded FY2014 operationalspendingbyfunctionalareaisincolumnE;thepercentageofexpendituresspentbyeachfunctionalarea is in column J. Column H contains the product of these percentages and PREPA’s totaloperationalbudget.

543PREPA’sresponsetoCEPR‐AH‐02‐09(c)oftheCommission’sSixthROI(August25,2016).

544October20ConferenceCall,ErnestoRamos,3:41:00.

545October20ConferenceCall,LucasPorter,3:45:40.


PREPAhadnot,infact,everpreparedaby‐departmentoperationalbudgetforFY2017—orthat,ifithad,ithadnotprovidedthisbudgettoitsadvisors.

Moreover,wesimplydisagreewiththeassertionthatPREPA’scalculationisthebestpossiblemechanismforallocatingFY2017operationalexpensesamongitsfunctionalareasgiventheapparentlackofabottom‐up,by‐areabudget.Ifnothingelse,wefinditimpropertomixdatafromtwoseparateyears in thewaythatPREPAdid:PREPAcalculated laborexpensesasinflated from FY21016 values but allocated these expenses according to FY2014proportions. Thismethod implicitly asserts that therewereno “knownandmeasurable”changesinthedistributionofoperationalspendingbyfunctionalareabetweenFY2014andFY2016—anassertionthatweknowtobefalse,basedonthehistoricalspendingpatternsexploredbelow.Alternatively,wecaninterpretPREPA’smethodologyasassertingthatanychangesinthebreakdownofoperationalexpensesbyfunctionalareathatdidoccurbetweenFY2014andFY2016willberevertedforFY2017;thiswouldbeanunusualclaimandPREPAhascertainlynotpresentedevidence to thiseffect.Themost likely,but leastsatisfactory,explanationisthatPREPAmeanstoassertneitherofthesestatements,butratheremployedanadhoc,poorly‐justifiedcalculationinthepreparationofoneofthemostmajorelementsofitsrevenuerequirement.

There are at least two more serious deficiencies with this approach: PREPA’s methodinherently assumes that the same proportions are valid for both labor and non‐laborspending and that that they are valid for PREPA’s planned savings from performanceimprovements.Wefindthatneitherassumptionistrue,andthatPREPAcouldhaveeasilycorrectedtheseassumptionsgiventheinformationweareconfidentwasavailabletoit.Wediscussthesefaultyassumptionsbelow.

WealsodescribeanunexplaineddeviationbetweentheFY2014valuesusedbyPREPAinthefinancialmodelandthoseprovidedinhistoricalrecords.Weareunabletojustifyorreconcilethisdeviation.

a. Proportionof laborspendingbyareadiffers from thatofnon‐laborspendingbyarea

PREPA prepared separate and independent labor and non‐labor expense budgets, asdescribedabove.However,itsallocationofoperationalexpensesbyfunctionalareaisbasedonasinglesetofpercentagevaluesasappliedtototalspending.Thismethodisonlyvalidifeither:(1)thepercentageofoperationallaborspendingineachfunctionalareaisthesameasthepercentageofnon‐laboroperationalspendingineachfunctionalarea;or(2)theratiooflaborexpensestonon‐laborexpenseshasremainedconstantsinceFY2014,suchthatthetotalproportionalbreakdownoflaborandnon‐laborspendingineachfunctionalareaalsoremainsconstant.

We found that neither of these conditions is true. An examination of the breakdown ofFY2014 operational spending by functional area and expense type (labor or non‐labor)


demonstratesthatthefunctionalareasclaimverydifferentproportionsoflaborandnon‐laborspending.

Table 28. Percentages of FY2014 labor and non‐labor operational expenditures by functional area.546

FunctionalArea%ofFY2014Labor

Spending%ofFY2014Non‐Labor

Spending%ofFY2014Total

OperationalSpendingGeneration 22% 23% 22%Transmission 7% 4% 6%Distribution 34% 20% 30%CustomerService 17% 10% 15%A&G 20% 45% 27%

In particular, we note that while the proportions of labor and non‐labor spendingrepresentedbygenerationexpensesaresimilartooneanother,T&Dspendingtakesupamuch greater proportion of labor spending than non‐labor spending. Conversely, A&Gexpenses dominate non‐labor spending but are a relatively small proportion of laborspending.ThesedifferenceshighlighttherisksinPREPAhavingreliedonasinglefactortoallocateatotalbudgetcomprisedofbothlaborandnon‐laborelements.

If,however,PREPA’s laborandnon‐laborspendinghadinflated(ordeflated)atthesameratesinceFY2014,thesedifferenceswouldbeimmaterialastheratiooflabortonon‐laborspendingwouldbeidenticalinFY2017toitsvalueinFY2014.Inreality,however,PREPA’slabor and non‐labor budgets have shifted in opposite directions since FY2014, as wedemonstrateinthetablebelow.

Table 29. Comparison of PREPA's FY2014 actual operational spending and FY2017 proposed budgets.

FY2014Actual* FY2017Budgeted** %Change,FY2014toFY2017Labor 530,797,419 434,937,821 ‐18%Non‐Labor 203,319,764 227,564,256 12%Total 734,117,183 662,502,076 ‐10%

*CorrectedvaluesasprovidedbyPREPAinCEPR‐AH‐02‐08(b).**PREPAEx.14.02.xlsx,“Inputs”tab,lines60and79.

WenotethatPREPA’slaborandnon‐laborbudgetshavenotjustchangedatdifferentrates,butratherthelaborbudgethasdecreasedascomparedtoFY2014spending,whilethenon‐laborbudgethasincreased.Thishasresultedinasignificantchangeinthefractionoftotaloperationalspendingrepresentedinbylaborexpenses,whichwas72percentinFY2014547butisbudgetedatonly66percentinFY2017.548

546PREPA’sresponsetoCEPR‐AH‐02‐08(b)oftheCommission’sSixthROI(August25,2016).

547Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐02‐08(b).

548Author’scalculationbasedonvaluesprovidedinPREPAEx.14.02.xslm,“Inputs”tab,lines60and79.


Because the proportions of spending by area are not the same of labor and non‐laborexpenses,andthebreakdownoftotalspendinginto laborandnon‐laborcomponentshaschangedsinceFY2014,PREPA’sproposedallocationofspendingbyfunctionalareaislikelyinvalid.WeareunsureastowhyPREPAusedonlyasinglepercentagefortotalspendingtoallocateitsoperationalbudgetbyfunctionalarea,consideringthatitclearlyhadaccesstototallaborandnon‐laborspendingamountsinhistoricalyears.Basedontheseobservations,we propose a reallocation of PREPA’s operational budget by functional area in SECTIONbelow.

b. Allocationofperformanceimprovement‐relatedsavingsbyareaisnotconceptuallysound

PREPA calculated its allocation of operational funds by functional area using a totaloperationalbudgetthatincludedexpectedsavingsfromperformanceimprovements.549Indoing so, PREPA implicitly assumed that its savings fromperformance improvements (ifrealized550)wouldbespreadacrossitsfunctionalareasinthesameproportionsasitslaborandnon‐laborspending.Thisassumption isnonsensical. Inmanycases, theperformanceimprovements PREPA listed in the financial model and Schedule F‐4 are likely to haveimpacts that disproportionately affect spending in one particular functional area. Forexample,PREPAcategorizesoverhalfof itsannualspendingonmaterialsasagenerationexpense.551However,PREPA’sallocationmethodinherentlyandwithoutbasisassumesthatonly 22 percent of its expected $37,500,000 of savings related to “Procurement andInventory”552 will be realized as a reduction in generation expenses.553 Similarly, PREPAclaims an expectation of $10 million in savings related to (additional) headcountreductions.554Withouta summaryofwherewithinPREPA’sorganization it expects theseheadcountreductionstooccur,wecannotputcredenceinPREPA’sassumptionthatthese

549 PREPA Ex. 14.02.xlsm, “RF_Schedules” tab, lines 489‐493. The value for performanceimprovementsavingsisoneofthreetermsthatareaddedtogetherbeforebeingmultipliedbytheby‐areaallocationfactorsdescribedabove.

550WedonotcommentonthestatusofPREPA’sexpectedoperationalperformanceimprovementsavings inthisreport.However, thattopic iscovered inthereportofCommissionadvisorsRalphSmithandMarkDady.

551Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xlsx.BetweenFY2010andFY2016,generationexpensesaccountedfor,onaverage,59percentofPREPA’sspendingunderthedescription“Materialsandsuppliesgen(warehouse)”and55percentofPREPA’sspendingunderthedescription“MaterialsBlankets.”

552PREPAEx.14.02.xlsm,“Inputs”tab,line98.

553Separately,wequestiontherealismofPREPA’sassumptionthatitsprocurementandinventory‐relatedsavingswillexceedtheentiretyofitsspendingonmaterialsinFY2017.



savings can be apportioned among its functional areas in exactly the proportions it hasassumedfortotalspending.

Wealsonote that, forsomeof theperformance improvementsavingsclaimedbyPREPA,casting these initiatives as reductions to expenses may be inappropriate. Most notably,PREPAconvolvessavingsfrom“TheftrecoveriesandreducedT&Dloss.”IfPREPAweretoincrease itsrelianceondistributedenergyresourcesor improvethepowerqualityon itstransmissionlines,itmayverywellseereductionsintransmission‐anddistribution‐relatedlosses,whichwewouldexpecttoseemainlyasareductioningeneration,fuel,andpurchasedpowerexpenses(as itwouldreducetheoverallquantityofenergyPREPAwouldneedtogeneratetoservedemand).However,inScheduleF‐4,PREPAdescribesthisinitiativenotasanimprovementtotheactualefficiencyofdeliveringpowertocustomers,butasareductionin“non‐technical losses,”atermthatPREPAusestodescribetheftofelectricity.555Ratherthanadecrease inspending,PREPAdescribes this initiativeasresulting inan increase incollections.556WethereforebelievethatPREPAmis‐categorizedthislineitem.

Because we have scant data with which to adjust PREPA’s allocation of performanceimprovementsavings,wedonotattempttodoso.Rather,webaseouranalysisofPREPA’soperationalbudgetbelowsolelyonPREPA’spre‐savingsbudgetvalues.

c. FY2014valuesarenotconsistentwithhistoricalrecords

InseekingtocomparePREPA’soperationalbudgetswithitshistoricalspendingpatterns,wecomparedFY2017valuestohistoricalrecordsprovidedbyPREPAindiscovery.Indoingso,we found an unexplained inconsistency in the by‐functional‐area breakdown of FY2014values.Asdescribedabove,PREPAincludedahardcodedbreakdownofFY2014operationalspending by functional area in the financial model.557 PREPA provided a table furtherdividing these values into labor and non‐labor FY2014 spending by functional area inresponse to discovery.558 PREPA also provided two separateworkbookswith records ofhistoricaloperationalspendinginmoregranulardetail,includinglistingexpensesbybothfunctionalarea559(ordirectorate)andtype(laborornon‐labor)orkindofexpense.

Wesummedbothlaborandnon‐laborexpensesbyareafromthesehistoricalrecordsandfoundinexplicabledeviationsfromthevaluesPREPAusedinthefinancialmodeltocalculateitsFY2017allocations.Whilethetotalvaluesforlaborandnon‐laborspendingarerelativelyclose,thefunctionalareatotalsdifferbymillionsortensofmillionsofdollars.Weshowthese

555ScheduleF‐4,VI(B)(3).

556Id.

557PREPAEx.14.02.xlsm,“RF_Schedules”tab,lines489‐493,columnE.

558CEPR‐AH‐02‐08(b).



deviationsinthetablebelow.PREPAprovidedhistoricalvaluesintheattachmenttoCEPR‐AH‐06‐13.560 The values in PREPA’s response to CEPR‐AH‐02‐08561 match those in thefinancialmodel.562

Table 30. Deviations between PREPA's FY2014 operational spending as reported in different ROIs.563

FunctionalArea Labor

CEPR‐AH‐06‐13 CEPR‐AH‐02‐08 DeviationA&G $101,879,925.16 $104,705,353.46 ‐$2,825,428.30CustomerService $102,254,632.64 $91,969,357.40 $10,285,275.24Generation $139,665,830.89 $114,439,783.74 $25,226,047.15T&D $186,997,009.81 $219,682,924.47 ‐$32,685,914.66

Total $530,797,398.50 $530,797,419.07 ‐$20.57 Non‐LaborFunctionalArea CEPR‐AH‐06‐13 CEPR‐AH‐02‐08 DeviationA&G $116,444,612.75 $90,574,065.76 $25,870,546.99CustomerService $13,474,708.09 $19,436,287.49 ‐$5,961,579.40Generation $39,551,514.63 $46,102,118.00 ‐$6,550,603.37T&D $32,414,853.27 $47,207,293.15 ‐$14,792,439.88

Total $201,885,688.74 $203,319,764.40 ‐$1,434,075.66 TotalFunctionalArea CEPR‐AH‐06‐13 CEPR‐AH‐02‐08 DeviationA&G $218,324,537.91 $195,279,419.22 $23,045,118.69CustomerService $115,729,340.73 $111,405,644.89 $4,323,695.84Generation $179,217,345.52 $160,541,901.74 $18,675,443.78T&D $219,411,863.08 $266,890,217.62 ‐$47,478,354.54

Total $732,683,087.24 $734,117,183.47 ‐$1,434,096.23

These deviations call PREPA’s allocation of operational budgets by functional area evenfurther into question,564 as the percentages of total spendingby functional area are verydifferentbetweenthetworecords.Wetabulatethosepercentagesbelow;weemphasizethatduetothesizeofPREPA’soperationalbudget,aonepercentdifferenceinallocationisequaltoseveralmilliondollars.

560Id.;thesevaluesmatchthosefoundinCEPR‐JF‐01‐24Attach02.xslxthrough…Attach05.xslxforyearsFY2012throughFY2015.

561CEPR‐AH‐02‐08(b).

562WiththeexceptionthatPREPA’svaluefortotalnon‐laborspendinginFY2014isincorrectinthe“Inputs” tabof the financialmodel, as explained inPREPA’s response toCEPR‐AH‐02‐08(a). ThiserrordoesnotimpactthecalculationsofexpensesbyareainScheduleE‐6.

563ValuesinPREPA’sresponsetoCEPR‐AH‐02‐08matchthoseinthefinancialmodel.

564WenotealsothatthesedeviationsraiseadditionalconcernsregardingthequalityandconsistencyofPREPA’srecordkeeping.


Table 31. Allocation of FY2014 operational spending by functional area as reported in different ROIs.

%ofTotalSpendingFunctionalArea CEPR‐AH‐06‐13 CEPR‐AH‐02‐08A&G 30% 27%CustomerService 16% 15%Generation 24% 22%T&D 30% 36%

Total 100% 100%

Inouranalysisbelow,werelyonlyontheFY2014valuesfromdetailedhistoricalrecords,ratherthanusingthevaluessetforthbyPREPAinthefinancialmodel.

3. PREPA failed to provide supporting documentation for theoperationsbudget,evenafterseveralrequests

Thelaborandnon‐laborcomponentsoftheoperationalbudgetvaluespresentedbyPREPAinthefinancialmodelarecitedtothe“AlixPartnersbusinessplan”and“MillimanActuarialStudy.”565PREPAwasaskedbyCommissionadvisorstoprovideadditionalsupportoramorecomplete budget several times during this proceeding. In response to a request for theFY2017operational budget,566 PREPA simply provided aPDF version of the same valuesfoundinthefinancialmodel,withnoadditionalsupportinginformation—andthenPREPAprovidedthissamedocumentagaintwomonthslater, inresponsetoasecondROIforitsoperationalbudget.567

Commissionadvisorsalsoaskedfora“detailedbreakdown”ofthevaluesusedbyPREPAinitscalculationofoperationallaborexpenses,“includingtheworkbook(s)thatgeneratedallhard‐coded values.”568 In response, PREPA provided detailed workbooks showinganticipatedandactualspendinglevels,bydirectorateanddivision,withdescriptions,foritsentireoperationalbudget(includingfuel,purchasedpower,labor,andnon‐laborexpenses).Theseworkbookswouldhavebeensatisfactoryforourpurposes—exceptthatPREPAonlyprovidedworkbooksforFYs2012through2015,569despitethefactthatthefinancialmodelhashardcodedlaborexpensevaluesforeveryyearthroughFY2035.Becausetheypertain

565PREPAEx.14.02.xlsm,“Inputs”tab,lines94‐103.

566PREPA’sresponsetoCEPR‐RS‐03‐05oftheCommission’sSixthROI(August15,2016).RefertoCEPR‐RS‐03‐05_Attach01.pdf.

567PREPA’s response toCEPR‐RS‐06‐12of theCommission’sFourteenthROI (October21,2016).RefertoCEPR‐RS‐06‐12Attach01.pdf.

568CEPR‐JF‐01‐24;emphasisadded.

569WenotethatPREPAdidnotlabeltheworkbooksprovidedinthisresponsewiththeyeartowhichtheypertain.Wewereonlyable to identify theapplicableyearsbyreconcilingvalues fromtheseworkbookswithvaluescontainedinCEPR‐AH‐06‐13Attach01.xslx.


neithertoFY2017nortoFY2016(theyearonwhichPREPA’sbaseditsnon‐laboroperationalexpensevalues),noneof theworkbooksprovided inresponse to thisROIcanbeusedassupportforPREPA’sFY2017operationalbudget.

Finally,CommissionadvisorsaskedPREPAdirectly fordocumentationsupportinghow itdevelopedandcappeditsoperationalbudgetinFY2017.570PREPArespondedwithatwo‐pagedescription,inprose,ofmultipleelementsofitsrevenuerequirement.TheentiretyofthedescriptionofhowPREPA’snon‐laboroperationalbudgetwasdevelopedreads:

BudgetwasdevelopedbydirectoratebasedonFY16actualswithadjustmentsmadeforknownandmeasurablechanges.AdjustmentsweremadeforknownperformanceimprovementinitiativesforFY17.571

PREPAdidnotprovideanysupportingworkpapersinresponsetothisquery.

We are very concerned by this lack of documentation. Given the lack of supportinginformationsuppliedbyPREPAthusfar,weareunabletoascertainwhetherornotPREPAactuallydevelopedbudgetsbyfunctionalareaforFY2017.Weconsideritentirelypossiblethat the entirety of PREPA’s operational budgeting process for non‐labor costs was, aspresentedinthefinancialmodel,totakeactualspendinglevelsfromFY2016andinflatethembyonepercent.This issimplynotasatisfactorymethodology.We find itproblematic forPREPAtohavebaseditsbudgetedvaluesonspendingfromayearinwhichPREPAhadnoapprovedbudget,572aswehavenoassurancesthatthelackofanapprovedbudgetdidnotcause distortionary effects in PREPA’s actual spending.We are unable to concludewithcertainty that PREPA’s budget as presented reflects its actual planned operational andmaintenance activities in its various functional areas, because we have absolutely noevidencetothateffectpertainingtoeitherFY2017orFY2016.

Moreover, as we discuss below, however, PREPA’s operational spending in FY2016wassimplynotsufficienttoenableittooperateasafeandreliablesystem.Instead,PREPAhasbeen operating on a skeleton crew focused on reactive, rather than preventative,maintenance.WhileweunderstandthatPREPA’sbudgetingeneralisbasedonatestyearwith “known andmeasurable changes,”wewould have hoped that operating a safe andreliablesystemwasoneknownandmeasurablechangethatPREPAintendedtomakefromFY2016toFY2017.

570CEPR161020ClarificationCallRequestNo.9.

571Id.

572 CEPR‐AH‐05‐01. Because PREPA’s board did not approve a new budget prior to the start ofFY2016,PREPAautomaticallyadoptedtheFY2015budget.


D. PREPA’soperatingbudget isnotconsistentwithhistoricalpatternsorwithoperationofasafeandreliablesystem

GiventhelackofinformationprovidedtousbyPREPAregardingthebasisforitsFY2017operational budget, our main avenue of analysis for evaluating this component of therevenue requirement is through comparison with PREPA’s historical spending onoperations. In this section, we discuss our analysis of PREPA’s operational budgets andexpendituresforFYs2010through2016andcommentonthepatternsweobservetherein.

1. PREPA’soperationalspending fellsharplybetweenFY2013andFY2016

a. Declinesinoperationalspending

PREPA’soperationalspendinghasdeclinedinrecentyears,573fromahighofapproximately$829million inFY2012574 toonly$667million inFY2016.575PREPA’soveralloperationalspendingpatternisshowninFIGURE.

573Regardingallhistoricalvaluesinthischapter,andouranalysisbasedthereon,wewillnoteherethatweassumedPREPAprovidedhistoricalvalues innominaldollars(i.e., thesevalueswerenotcorrected for inflation). For the sake of simplicity, we did not convert historical dollars into aconsistent dollar‐year. However, readers should note thatwhenwe identify declines in PREPA’soperationalspending,thesedeclinesarelikelytobegreaterinrealterms.Wewillcontinuetonotethispointinfootnotesbelowasappropriate.

574CEPR‐AH‐06‐13Attach01.xslx,“FY2012Variance”tab.

575Id.,“FY2016Variance”tab.


Figure 30. PREPA's total operational spending by functional area, FYs 2010‐2016.576

As can be seen from the figure above, spending on customer service, transmission anddistribution(T&D),andgenerationhaveallfallenbysignificantamounts,withanespeciallysharp drop from FY2014 to FY2015. Meanwhile, A&G spending has actually increasedsomewhat.WedonotknowwhatcausedthelargespikeinA&GspendinginFY2012,butofferbelowsomecommentaryontheincreaseinA&GspendingfromFY2014tothepresent.

PREPA’s historical labor spending has followed a similar pattern to its total operationalspending, as shown in FIGURE. Labor spending was approximately constant for thegeneration,customerservice,andT&DareasfromFY2010toFY2014577Thereafter,PREPAcut its labor spending ineveryarea.However, themost severecutswereexperienced ingeneration:theFY2016generationbudgetis23percentbelowthearea’sbudgetinFY2010,representingcutbacksofalmost$33millioningeneration‐relatedlaborspending.578


577Effectively,declininginrealterms.

578Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.


Figure 31. PREPA's operational labor spending by functional area, FYs 2010‐2016.579

PREPAalsocut thenon‐laborbudgetof itsgenerationarea,byapproximately27percentbetweenFYs2010and2016(amountingtoadecreaseinbudgetofnearly$13million).580Intheaggregate,PREPA’snon‐laborspendingoneverythingexcepttheA&Gareahasdeclinedby 28 percent since FY2010, as we show in FIGURE below. Meanwhile, A&G non‐laborspendingincreasedby44percent—orasumof$50million—overthesameperiod.581


580Id.

581Id.


Figure 32. PREPA's operational non‐labor spending by functional area, FYs 2010‐2016.582

ThesepatternsindicateachangeinprioritiesatPREPA.WhilePREPAhasspentlessandlesson the materials and staff needed to operate its generating units and transmission anddistribution infrastructure, it has chosen to devote more resources to its own internalcorporateoperations.ThischoicehashadsevererepercussionsforthesafetyandreliabilityofPREPA’ssystem,aswediscussfurtherbelow.

b. Relationshipbetweenoperationalbudgetsandspending

TherelationshipbetweenPREPA’soperationalbudgetsanditsactualoperationalspendinghas also shifted over time. Prior to FY2014, PREPA routinely overspent its operationalbudgets.PREPA’sunderbudgetingwasnotevenlydistributedamongitsfunctionalareas:ingeneral, PREPAmet or slightly underspent its generation and customer service budgetswhile PREPA overspent its T&D budgets by approximately five percent on average andoverspentitsA&Gbudgetsby24percent.583

582Id.

583Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.AveragesomitFY2013,whichappearstobeanoutlier:PREPAexceededeveryoneofitsoperationalsub‐budgetsexceptcustomerservice‐andgeneration‐relatedlaborexpensesinthatyear,forunknownreasons.Inparticular,theextenttowhichPREPAoverspentitsgenerationnon‐laborbudgetisunusual:thebudgetwasexceededby146percent,primarilyduetoa$16millionoverageonmaterialsspending.


InFY2014andthereafter,PREPAbegantounderspenditsalready‐reducedbudgets.584TheaverageproportionofPREPA’sgenerationexpensebudgetthatwentunspentdoubledfromfourtoeightpercent;theproportionofunspentcustomerservicebudgetmorethantripledover thisperiod.585Fromhavingayearlyoverrun, theT&Dareabegan tounderspend itsbudgetsbyfivepercentonaverage.586OnlytheA&Gareacontinuedtooverspend,atarateofeightpercentperyearonaverage.587Theseshiftsarereflectedinthefigurebelow.

Figure 33. Percent by which PREPA under‐ or overspent its operational budgets in FYs 2010‐2012 and FYs 2014‐2016, by type of

expense and functional area.588

Thechanges inPREPA’sspendingpatternsareparticularlynotablewithregards to laborspending.PREPAhasbeenupfrontaboutthefactthatitisseekingtotrimitslaborforceforthesakeofgreaterefficiency.589However,wenotethatwhiletheseeffortswouldgosome

584Id.NotethatthefigureswepresentedaboveshowforPREPA’sactualspendingratherthanitsbudgeted spending. However, spending cutbacks are reflected in PREPA’s budgets as well: thecustomer service, T&D, and generation budgets were cut by three, six, and twenty‐five percentrespectivelyinnominaltermsfromFY2010toFY2016.

585Id.;averagescoverFYs2014through2016.

586Id.

587Id.


589AsdiscussedfurtherbyCommissionadvisorsRalphSmithandMarkDady.


waytoexplainingdecreasesinPREPA’sbudgetedlaborspending, it isnotclearhowtheyrelate to the deviation between PREPA’s budgets and its actual expenditures. We areconcernedbythepossibilitythatthispatternindicatesthateitherPREPA’sbudgetsdonotreflectitsactualplansorthatPREPAhasbeenunabletoretaindesiredemployeesoverthepastseveralyears.

2. PREPAdescribeditsownbudgetformationprocessasonebasedonprinciplesofausterity

Thedeclines inPREPA’soperationalbudgetover timeareconcerning,butnot surprisinggiven PREPA’s constrained financial situation and the manner in which it prepares itsbudgets.Weemphasizeherethattheconceptualbasisoftherevenuerequirementisthatitrepresentsall of the revenueneededby theutility tooperate a safeand reliable electricsystem.Fullreturnofavertically‐integratedutility’sprudently‐incurredcostsisoneofthebasicpremisesofutilityoperations.However,asiswell‐knownbytheparticipantsinthisproceeding,PREPA’srevenueshavebeeninsufficienttocoveritscostsforsometime.

Perhapsasaresultofsituation,PREPAhasadoptedanattitudeofausteritytowardsitsownspending.WhenaskedwhetherornotPREPAhadhistoricallycappeditsoperationalbudgetsin thesamemanneras itscapitalbudgets,590PREPArespondedthat“ofcourse,” ithad.591PREPA’s Director of Finance proceeded to explain that PREPA’s policy is to cap its’departments’ operational budgets at the lower of the previous years’ budget or actualspending,unlessdepartmentheadsmakeacompellingcaseforanincreaseinbudgetandpresentevidencethattheincreaseisinlinewithPREPA’sorganizationalgoals.592

On the same call, PREPA claimed that this years’ process of setting operational budgetsdifferedsomewhatfromthepastdueandthatbudgetcapsforFY2017weresettoagreewith“therevenuerequirement.”Theintendedmeaningbehindthisstatementisnotclear,giventhattheFY2017operationalbudgetisbasedonthe(restricted)FY2016operationalactualspendingasdiscussedabove.Weareconfident,however,thatthisstatementdoesnotimplythattheFY2017operationalbudgetisinfactsufficienttocoverPREPA’sactualneedsfortheoperationofasafeandreliablesystem.

3. IncreasesinAdministrativeandGeneralspending

Instarkcontrasttotheausteritydescribedabove,wenotedabovethatPREPA’sspendingonAdministrativeandGeneraloperationshasactually increasedoverthepastseveralyears,evenasspendingonitscoregenerationandT&Doperationshascratered.Weinvestigatedthe sourceof this increase and found it to originate entirelydue to additional non‐labor

590Thatis,atalevelthathasnoapriorirelationshiptoitsactualneeds

591October20ConferenceCall,4:03:00.

592Id.,4:06:00.


spendingonthepartoftheA&Gfunctionalarea.Asmentionedabove,PREPA’snon‐laborA&Gspendinghasincreasedbyapproximately$50millionsinceFY2010.593

Concerningly,thebulkofthisincreaseinspendingisinanareadescribedas“GenMiscExpControlled by Resp.”594 We interpret this heading as “General Miscellaneous ExpensesControlled by Responsibility.” PREPA’s A&G spending in this area has increased fromapproximately $80million in FY2010 to $122million in FY2015 and a staggering $134millioninFY2016.Inotherwords,PREPA’sspendinglastyearonmiscellaneousA&G‐relatedexpenseswasmorethanitsentireproposedbudgetforgenerationexpensesinFY2017.Itisdifficultforustooverstatehowconcerningthisis.Moreover,wehaveabsolutelynofurtherinformation about what, exactly, PREPA spent these funds on, although we can makeinferences. A separate discovery response indicates that they were likely spent by the“Corporate Responsibilities” directorate595 but we do not find this to be particularlyinformativeastothenatureoftheexpenses.WerecommendthattheCommissionaskformore documentation from PREPA on this matter and require PREPA to justify theseexpenditures.

4. PREPAhasexplicitlystatedthatstaffandbudgetrestrictionshaveledtodeclinesinsafetyandreliability

Aswe clearly establish in SECTIONabove, the reliability of PREPA’s generationunit andtransmissionanddistributioninfrastructurehaspalpablydeclinedasPREPAhascutbackonits operational spending. PREPA itself has made a number of explicit statementsacknowledgingtheeffectsofthesecuts.Inparticular,PREPA’sreducedstaffinglevelsappearto be leading to severe consequences for system safety, reliability, and operability. Forexample, with regards to the “deteriorated” state of its transmission and distributioninfrastructure,PREPAsays:

Duetothelackofadequatequalifiedpersonnel,theDistribution,liketheTransmissionSystemisdeterioratedbothfromastructural/mechanicalperspectiveandfromanelectricalperspective…Thisdeteriorationisthecauseofcontinuousinterruptions,nomatterofweatherconditions.596

Inaddition,PREPAcitesthat:

…thelackoftransportationorspecialequipmentandthenonavailabilityofqualifiedhumanresources,arethemajorreasonsthatleadtolongservice

593Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.


595 CEPR‐JF‐01‐24_Attach 05.xlsx. All spending in “Gen misc exp controlled by resp” expensesoccurredunderthedirectoratelabeled“A14‐ResponsabilidadesCorporativas”.

596CEPR‐AH‐02‐01(e).


interruptions(affectingSAIDI,SAIFI&CAIDIindexes)andimpactedreliability.597

Finally,PREPAstatesbaldlythat:

…theeffectsofT&Dworkforcereduction(~22%fromJan2014toJul2016)havebeensignificantontransmissionanddistributionsystemmaintenance.Theseeffectshavebeenexacerbatedbyashortageoffundsnecessarytoexecuteawell‐plannedpreventativemaintenanceprogram.Onedirecteffectoftheworkforcereductionhasbeenthereassignmentofmultipleconstructioncrewstofocusonreactivemaintenanceinsteadofpreventativemaintenanceandnewconstruction…theseeffectshaveledtothesteadydeclineintransmissionanddistributionsystemmaintenanceandperformanceacrosskeyperformanceindicatorsincludingCAIDIsince2014.598

WenoteherethattheyearidentifiedbyPREPA,2014,isthesameyearweidentifiedaboveastheturningpointinspendingpatterns.

On a similar note, PREPA says the following with respect to the impact of reducedoperationalspendingonitsgenerationunits:

Theeffectofgenerationworkforcereduction(~23%fromJan2014toJul2016)ongenerationmaintenancehasbeensignificantparticularlywhencombinedwithshortageoffundstousethirdpartylabortoperformmaintenanceactivities.Theworkforcereductionhasaffectednotonlytheresourcesthatperformedmaintenanceplanning/monitoringbutalsothosethatexecutedthemaintenance.Thisresultedinamigrationfrompreventivefocusedmaintenancetoamorebasic/correctivefocusedmaintenancetomaintainunitsoperatingwithlimitedresourcesandfunding.Thatcombinedwiththedepartureofcriticaloperationalexperiencefromthepowerplants(notjustinmaintenance)increasestheseverityoftheproblem.PreventivemaintenanceinolderunitslikePREPA’sisparticularlycriticalandifrelaxedoveraperiodoftimeleadstoincreasedfrequencyanddurationofforcedoutageeventsashavebeennoticedinrecenttimes.599

Asdemonstratedabove,wecanconfirmPREPA’sassertionthatforcedoutageeventshaveoccurred at increasing rates in the past two years. PREPA itself also cited “Loss of a

597CEPR‐AH‐02‐01(g).

598PREPA’sresponsetoCEPR‐RS‐03‐03oftheCommission’sSixthROI(August22,2016).

599Id.


significantnumberofexperiencedpersonnel,”“insufficientstafftoperformallthenecessarymaintenancework,”and“defermentsonpayingvendorleadtodelayinreceivingmaterials”askeycausesofitsforcedoutagesinaninternalpresentation.600

We note this presentation in particular as it was provided to the Commission only indiscoveryandnotasadirectjustificationforPREPA’srevenuerequirement.Onemighteasilyinterpret the quotes above as an attempt on PREPA’s part to justify higher costs than itactuallyneedstooperateasafeandreliablesystem.Wedonotbelievethistobeanaccurateinterpretation, for several reasons. First, PREPA’s rates of forced outages and systeminterruptions have undeniably risen in the recent past—this is not a fiction or amisrepresentationonPREPA’spart.Second,PREPA’sproposedoperationalbudgetisnotinfactsubstantiallyhigherthanthebudgetsoftherecentpast;infact,itissubstantiallylowerinsomeareas.Finally,PREPAwouldhavenoreasontodissembleonthistopicinaninternalpresentationfocusedonanalyzingitsownforcedoutagesandthecausesoftheseoutages.Explicitly,wedonotbelievethatPREPAismisrepresentingthedirenessofitssituationortheimpactsofitsreducedoperationalspendingthereoninanattempttoartificiallyinflatetherevenuerequirement.

5. Implicationsforbenchmarkingstudy

As part of its filing, PREPA witness Larry Kaufmann presented the results of a studycomparingPREPA’soperationalexpenses(lessfuelandpurchasedpowerspending)tothoseofseveral“peer”utilities.Dr.Kaufmannusedthisstudytoconcludethefollowing:

Overall,thesefindingsleadmetoconcludethatPREPA'sexpensesarenotbeing artificiallyinflatedbecauseofinefficientoperationsorexcessivewagepaymentsto PREPAemployees.TheevidencesuggeststhatPREPA'sinternalcostmanagementis nottheprimaryfactorinPREPA'sfinancialdifficulties.ThePREPAfigures,however, likelyreflectdownwardpressuresonspendingduetoitsfinancialdifficulties.601

Inlightoftheabove,wefindthisstudytobecompletelyimmaterialtothetaskfacingtheCommission: deciding whether or not PREPA’s revenue requirement and accompanyingratesarejustandreasonable,andwhetherornottheyaresufficientbutnotexcessiveforthepurposesofoperatingasafeandreliableelectricsystem.

Dr.Kaufmann’sanalysis,basedondatafromyears2008through2014,ispredicatedontheassumptionthatPREPA’soperationalcostsintheseyearsareinfactfullyrepresentativeofPREPA’sneedsfortheoperationofasafeandreliablesystem.Dr.Kaufmann’sownstatement

600PREPA’sresponsetoCEPR‐JF‐01‐16oftheCommission’sSixthROI(PUBLIC;August23,2016).RefertoCEPR‐JF‐01‐16Attach01(PUBLIC).pdf

601PREPA’sPetitionforRateReview,Exhibit6.0,DirectTestimonyofLarryKaufmann,lines532‐536.


above, thatPREPA’soperational costs “likely reflectdownwardpressureonspending,”602contradictthisbaselineassumption.ThemanystatementswequotedinthesectionabovealsobelieDr.Kaufmann’spremise.EvenifweassumethatPREPA’sspendingthrough2014wasfullysufficientforitsneeds(anassertionofwhichwehavesomedoubt),itisclearlythecasethatPREPA’scostsoverthepasttwoyearsandaspresentedinthisratecasearenotrepresentativeofthelevelofspendingcommensuratewithoperationofasafeandreliablesystem.

Moreover,itissimplynotusefultocomparePREPA’scoststothoseofotherutilitieswhenthose other utilities manage to spend comparable amounts to PREPA and yet operatesignificantlymorereliablesystems,withsuperiorreliabilityindicesandlowerforcedoutagerates.Forexample,DukeFlorida—oneofthe“peer”utilitiesidentifiedbyDr.Kaufmann603—reportsanadjustedSAIDIgoalofapproximately80minutesperyearfor2015.604PREPA’sSAIDIin2013(ayearcomfortablywithinDr.Kaufmann’speriodofanalysis)was,onaverage,51minutespermonth.605Inotherwords,PREPA’ssystemisapproximatelyeighttimeslessreliable(basedonthismetricalone)thanDukeFlorida’s.

Due to the austerity built into PREPA’s budgeting process, the Commission cannot beconfident that PREPA’s past spending or current revenue requirement represent the fullamountthatPREPAwouldrequiretomaintainitssysteminassafeandreliableastateaspracticable.Similarly,PREPAhasneverpresentedthisCommissionwithanestimateofwhatPREPA’scostswouldbewereittoimproveitssystemtothepointofhavingsimilarreliabilitytothemainlandpeerutilitiesdiscussedbyDr.Kaufmann.Assuch,wecannotconsiderthisstudytobeofuseandrecommendthattheCommissiondisregardit.

E. CalculationofadjustmentstoPREPA’soperatingbudget

Inlightoftheweightofevidencepresentedabove,wefounditusefultocomparePREPA’soperational laborandnon‐laborbudgetstoitsaveragespendingintheyearsinwhichitssystemwasmaintainedinapassable(ifnotexcellent)stateofrepair.606Indoingso,wehavearrivedatseveralrecommendedadjustmentsthatweassertwillbringPREPA’soperationalbudgetsmoreinlinewithhistoricalpatternsandclosertowhatPREPArequirestorunas

602Id.,line536.

603Id.,line191.

604SeeReliabilityTrends,ChapterIII.B.1,Page35above.

605Author’scalculation(simpleaverageofvaluesassumedtobemonthly),basedonvaluesprovidedinPREPA’sresponsetoCEPR‐MC‐01‐11oftheCommission’sFourteenthROI(November4,2016).RefertoCEPR‐MC‐01‐011Attach01.pdf.

606WebasethisstatementontheunitavailabilityanalysispresentedaboveandthestatementfromPREPA’s external Consulting Engineers that its systemwas in “good repair and sound operatingcondition”asoftheendofFY2013.ScheduleI‐1,ExecutiveSummaryp3.


safeand reliable a systemas it ran inFY2014andyearsprior.Asmentionedabove, ouranalysis here focuses on PREPA’s budget before performance improvement savings aretaken intoaccount.Wedonot recommendanyspecificadjustments toPREPA’sassumedperformancesavings.

Wesummarizeouradjustments,andthemethodologyweusedtoarriveatthem,below.

1. Reallocationofbudgetsbasedonexpected laborandnon‐laborproportionsofspendingbyarea

Asdiscussedabove,wehaveobservedthatthepercentagesofPREPA’slaborspendingthatfallundereachfunctionalareahavehistoricallybeenverydifferentfromthepercentagesofnon‐laborspendingforwhicheachfunctionalareaisresponsible.Ourfirststepincalculatingan adjusted FY2017 operational budget was to reallocate the total labor and non‐laborbudgetsproposedbyPREPAtobetterrealignthemwiththisobservation.

Our adjustment is based on the average percentages of labor and non‐labor spendingaccountedforbyeachfunctionalareafortheperiodFY2010‐FY2014.Wechosethisintervalas it incorporatesthe largestamountofhistoricaldataavailabletousbeforethepointatwhich PREPA’s spending patterns shifted due to its financial constraints. Moreover, weobservedthatthebreakdownofPREPA’slaborandnon‐laborspendingbyfunctionalareawasfairlyconsistentduringtheseyears.Thefiguresbelowshowhistoricalpercentagesoflaborandnon‐laborspendingbyfunctionalarea,theFY2010‐FY2014averagesweusedtofor budget reallocation purposes, and PREPA’s proposed FY2017 allocation factors. WebelievethatourproposedreallocationfactorsmorecloselymatchPREPA’sactualhistoricalspending patterns during years in which it was running a nominally functional system,especiallyintheareaofnon‐laborspending.


Figure 34. Percentage of labor spending by functional area, with proposed allocation factors.607

Figure 35. Percentage of non‐labor spending by functional area, with proposed allocation factors608

607Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.Opensymbolsshowhistoricalvalues;closedsymbolsshowPREPA’sproposedallocation factors for theFY2017revenue requirement; dashed lines show FY2010‐FY2014 average percentages, as used for ourreallocationanalysis.

608Author’scalculationbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.Opensymbolsshowhistoricalvalues;closedsymbolsshowPREPA’sproposedallocation factors for theFY2017revenue requirement; dashed lines show FY2010‐FY2014 average percentages, as used for ourreallocationanalysis.


WeobservethatPREPA’sproposedallocationfactorsarelikelytobemisleadingifPREPA’sactual spending on labor and non‐labor expenses comports with historical patterns. Inparticular,PREPA’sallocationlikelyoverstatesA&GlaborexpensesbutvastlyunderstatesA&Gnon‐laborexpenses,whileoverstatingnon‐laborspendinginallotherfunctionalareas.PREPA’smethodalsolikelygivesshortshrifttogeneration‐relatedlaborexpenses.

WedescribeourreallocationsofPREPA’s laborandnon‐laborexpensebudgets,basedonFY2010‐FY2014averageallocations,below.WeemphasizeherethatthesereallocationsarenotreflectiveofwhatwethinkPREPAshouldbespendingoneachfunctionalarea,basedonabottom‐upevaluationofPREPA’sactualneeds.Rather,wepresenthereourexpectationofwhatPREPAwouldspendbyfunctionalareainFY2017ifitsoperationalspendingpatternconformedtohistoricaltrends.

a. Labor

WecalculatedareallocatedlaborbudgetbymultiplyingourproposedreallocationfactorsbyPREPA’stotallabor,aspresentedinthefinancialmodel.Wealsocomparedthesevaluesto an assumed laborbudgetby functional area,whichwe calculatedusing the allocationfactorsproposedbyPREPA,todeterminetheadjustments(i.e.,additionsorsubtractionstoPREPA’sproposedbudgets)indicatedbyourreallocation.

The tablebelowshowsabreakdownof laborspendingby functionalareausingPREPA’sallocationfactors,areallocatedbreakdownusingourproposedallocationfactors,andtheadjustmentsthatproceedfromthisreallocation.

Table 32. Reallocation of PREPA's FY2017 operational labor expense budget.

FunctionalArea

PREPAAllocationFactors*

AssumedFY2017LaborExpense

Allocation**ReallocationFactors†

ReallocatedFY2017LaborExpense

Adjustment(Rounded)

A&G 27% $115,697,810 19% $82,180,802 ‐$33,517,000CustomerService 15% $66,001,814 19% $83,777,704 $17,776,000Generation 22% $95,116,552 27% $115,430,151 $20,314,000T&D 36% $158,121,645 35% $153,549,163 ‐$4,572,000

Total 100% $434,937,821 100% $434,937,821 ‐*PREPAEx.14.02.xlsm,“RF_Schedules”tab,lines488‐493,headed“FY2014Allocation.”**ProductofPREPAallocationfactorsandtotalFY2017laborbudget,PREPAEx.14.02.xslm,“Inputs”tab,line60.†Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xlsx.AveragepercentagebyfunctionalareaoftotallaborspendingforyearsFY2010‐FY2014.

As expected, this adjustment removes a significant amount of the funds assumed to beallocatedtotheA&Gareaforlaborexpensesandreallocatesthesefundstothegenerationandcustomerserviceareas.ThisreallocationalsoadjustsT&Dlaborexpensesdownwardsslightly.

b. Non‐Labor


WhilePREPAhadlaborexpensesineveryfunctionalareasinceFY2010,thesetofcategoriesinwhichitexpectstospendnon‐laboroperationalfundshasexpandedsomewhatovertime.Inparticular,therearetwocategoriesofspendinginPREPA’snon‐laboroperationalbudgetthatarenewsinceFY2014:“RestructuringFees”and“AdditionalSafetyUpgrades.”PREPA’sbudget for Restructuring Fees consists primarily of its payments to its restructuringconsultant, AlixPartners, and other advisors in the restructuring process.609 PREPA firstengagedAlixPartnersinFY2015.610PREPA’sbudgetforAdditionalSafetyUpgradescoversanumberofsafetyinitiativesthatwereidentifiedbyPREPA’sconsultantDuPontasaresultofDuPont’sauditofPREPA’ssafetyrecordandpractices.611AdditionalSafetyUpgradesisanewcategoryforFY2017,asindicatedinthefinancialmodel.612

Inordertoperformafair,apples‐to‐applescomparisonofPREPA’snon‐laboroperationalbudgetwithhistoricalvaluesfromFYs2010through2014,wefirstsubtractedthefundsinthesetwocategories.ThisallowsustocomparePREPA’splannedversushistoricalspendingoverthesamesetofspendingcategories.Thetotalamountofplannedspendinginthetwonew categories is $32,432,500,613 leaving a total legacy‐category non‐labor operationalbudgetof$195,131,755.Ourreallocationofthisbudgetsub‐totalaccordingtotheaverageFY2010‐FY2014allocationofnon‐laboroperationalspendingisshowninthetablebelow.

Table 33. Reallocation of PREPA's FY2017 operational non‐labor expense budget.

FunctionalArea

PREPAAllocationFactors*

AssumedFY2017Non‐LaborExpense

Allocation**ReallocationFactors†

ReallocatedFY2017Non‐LaborExpense

RoundedAdjustment

A&G 27% $51,906,998 57% $110,462,390 $58,555,000CustomerService 15% $29,611,244 8% $14,836,592 ‐$14,775,000Generation 22% $42,673,364 21% $41,468,897 ‐$1,204,000T&D 36% $70,940,150 15% $28,363,877 ‐$42,576,000

Total(legacy) 100% $195,131,755 100% $195,131,755 0*PREPAEx.14.02.xlsm,“RF_Schedules”tab,lines488‐493,headed“FY2014Allocation.”**Product of PREPA allocation factors and FY2017non‐labor budget (PREPAEx. 14.02.xslm, “Inputs” tab, line 79) lessbudgetsforRestructuringFeesandAdditionalSafetyUpgrades(Id.,lines69and73).†Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xlsx.Averagepercentagebyfunctionalareaoftotalnon‐laborspendingforyearsFY2010‐FY2014.

AsexpectedfromFIGURE,theimpactofthisreallocationisprimarilytoshiftfundstotheA&G area from all other functional areas.We repeat here the recommendation that the

609PREPA’sresponsetoCEPR‐RS‐07‐11(b)oftheCommission’sSixteenthROI(November4,2016).

610PREPAenteredintoaprofessionalservicesagreementwithAlixPartnersonSeptember15,2014,according to its fifth contract amendment:http://www.aeepr.com/Documentos/Ley57/CONTRATOS%20GENERAL/2015‐P00036E%20Alixpartners.pdf.

611ScheduleF‐4,IV(C).


613Id.,sumoflines69and73.


CommissioninvestigatePREPA’snon‐laboradministrativeandgeneralspending,potentiallyaspartofitspendinginvestigationonPREPA’sperformance.

2. Correctiveadjustments

The reallocation of PREPA’s operational budget we performed allows for greatercomparabilitybetweenPREPA’sFY2017budgetlevelsanditsactualpastspending.Weusedhistoricalvaluestobenchmarkthereallocatedbudgetsandasindicators(byinspection)ofaneedforcorrectiveadjustments.614OuraimwiththeseadjustmentsaretobringPREPA’sgeneration‐andT&D‐relatedoperational spendingmore in linewithwhat itappearedtoactually require to operate a nominally safe and reliable system, as it did in the periodbetweenFY2010andFY2014.WebalancethisaimwithadesiretoavoidundulyincreasingPREPA’srevenuerequirement.

a. Labor

OurbenchmarkingofPREPA’sFY2017laborbudgetsissummarizedbelow.

Table 34. Benchmarking and correction of PREPA's FY2017 operational labor expense budget.

FunctionalArea

ReallocatedFY2017LaborExpense

BenchmarkingValue

DifferencefromBenchmarking

ValueCorrectiveAdjustment

CorrectedFY2017LaborExpense

A&G $82,180,802 $65,124,014* $17,056,788 ‐$17,057,000 $65,123,802CustomerService $83,777,704 $87,184,112* ‐$3,406,408 0 $83,777,704Generation $115,430,151 $144,469,557** ‐$29,039,405 $9,680,000 $125,110,151T&D $153,549,163 $192,438,980** ‐$38,889,816 $19,445,000 $172,994,163

Total $434,937,821 ‐ ‐ $12,068,000 $447,005,821*MinimumspendinglevelinyearsFY2010‐FY2016,asbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.**AverageofFY2010‐FY2014spendinglevels,asbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.

WeusedaveragespendingongenerationandT&DlaborexpensesinyearsFY2010‐FY2014to benchmark our estimates of PREPA’s FY2017 anticipated spending levels. ThiscomparisonrevealsthatPREPA’sFY2017budgetsaresignificantlybelowitsactualspendingin prior years. Generation labor expenses are approximately $29 million below theirhistoricallevels,whileT&Dlaborexpensesarealmost$39millionshortcomparedtopastspending.Inlightofourobservationsaboveregardingtheimpactofreducedspendingonlaborintheseareas,werecommendincreasestothelaborbudgetsforgenerationandT&D.Ourrecommendedcorrectiveadjustmentforgeneration‐relatedlaborspendingisonethirdof the shortfall between theFY2017valueand thehistorical average.Our recommendedcorrective adjustment for T&D‐related labor spending is half the shortfall between theFY2017value.Bothvalueswereroundedtothenearestthousandtoavoidfalseprecision.

614Asmentioned above, all calculations, comparisons, andbenchmarking are performedwithoutcorrectinghistoricalvaluesforinflation.


We set these values on the basis of our judgment as they reduce the deviation betweenPREPA’sFY2017budgetsandhistoricalspendinglevelstoapproximately$20millionineacharea,whichwebelievetobeanappropriatecompromiselevelgivenPREPA’sconstrainedfinancialenvironment.We chose to benchmark PREPA’s A&G and Customer Service labor budgets to PREPA’sminimumspendinglevelsontheseexpensesintheperiodcoveredbyhistoricalrecordsinourpossession(FY2010throughFY2016).Webenchmarkedthesevaluesdifferentlyaswehave seen no evidence that labor shortages in A&G or customer service have negativelyimpactedthesafetyorreliabilityofPREPA’ssystemandthereforebelieveitisappropriateforPREPAtorestrictspendingintheseareasconsideringitsfinancialposition.TheFY2017customerservicelaborallocationisalreadyslightlybelowthebenchmarkedvalue(whichcorrespondstoPREPA’sspendingoncustomerservicelaborexpensesinFY2015615)andwetherefore suggestno furtheradjustment to thisvalue.TheFY2017A&G laborbudget,bycontrast, is approximately $17 million above the benchmarking value of FY2016 actualspending.Werecommendthatthisbudgetbereducedtolastyear’sactualspendinglevel.Thiscut isnotamatterofausterityperse;rather,wehaveseennoindicationthata$17million increase in PREPA’s spending on its administration will increase the safety andreliabilityofPREPA’ssysteminanyway,andsafetyandreliabilitymustbePREPA’s firstpriorities.

b. Non‐Labor

Weperformed a similar benchmarking analysis on PREPA’s non‐labor budgets in legacycategories.However,becauseweinferthatPREPAhaslessdirectcontroloveritsnon‐laborcoststhanoveritslaborcosts,webenchmarkedallcategoriestoPREPA’saveragespendinginFYs2010through2014.616Wetabulatetheresultsbelow.


616ThisresultsinamoregenerousbenchmarkingforA&Gandcustomerservicenon‐laborexpensesthanforlaborexpensesinthoseareas.TheFY2017non‐laborbudgetforA&Gisapproximately$6millionabovePREPA’sminimumspendingonthoseexpensesinthepastseveralFYs(whichoccurredinFY2011),andthenon‐laborbudgetforcustomerserviceisapproximately$4.8millionPREPA’sminimumspendinginthatarea(whichoccurredinFY2015).Author’scalculation,basedonvaluesinCEPR‐AH‐06‐13Attach01.xslx.


Table 35. Benchmarking and correction of PREPA's FY2017 operational non‐labor expense budget.

FunctionalArea

ReallocatedFY2017Non‐LaborExpense

BenchmarkingValue*

DifferencefromBenchmarking

ValueCorrectiveAdjustment

CorrectedFY2017LaborExpense

A&G $110,462,390 $123,012,301 ‐$12,549,911 0 $110,462,390CustomerService $14,836,592 $16,328,449 ‐$1,491,858 0 $14,836,592Generation $41,468,897 $45,963,710 ‐$4,494,814 $4,495,000 $45,963,897T&D $28,363,877 $31,214,509 ‐$2,850,633 $2,851,000 $31,214,877

Total(legacy) $195,131,755 ‐ ‐ $7,346,000 $202,477,755*AverageofFY2010‐FY2014spendinglevels,asbasedonvaluesprovidedinCEPR‐AH‐06‐13Attach01.xslx.

Weobserve that every functional area has a shortfall as compared toPREPA’s historicalspending levels. However, as above, we prioritize additions to the generation and T&Dbudgets.Becausethenon‐laboroperationalbudgetismuchsmallerthanthelaborbudget,correcting the entire budget shortfalls in these areas has a relatively small impact onPREPA’s overall revenue requirement. Therefore, we recommend additions in the totalamountofthedeviationsbetweenPREPA’sFY2017budgetsandhistoricalactualnon‐laborspendingongenerationandT&Dexpenses—$4.495and$2.851million, respectively.WerecommendnochangestothecustomerserviceandA&Gnon‐laborbudgets.

3. OtheradjustmentstoPREPA’snon‐laborexpensesbudget

ApartfromareallocationandcorrectionofPREPA’sFY2017operationalbudgettobettermatchhistoricalspendingpatternsfromyears inwhichPREPA’ssystemwas inastateofrelatively good repair, we make two other sets of adjustments to PREPA’s non‐laboroperationalexpensebudget.Thefirstsetofadjustmentsaddsbackinplannedspendingonnon‐laborcategoriesthatdidnotexistinFY2014orbefore.ThesecondsetofadjustmentsaddsinthevalueofseveralmaintenancecontractsthatPREPAappearstohaveimproperlycategorizedascapitalexpenses.Wepresenthereatablesummarizingtheseadjustments,withdescriptionsbelow.

Table 36. Other adjustments to PREPA's FY2017 non‐labor operational expense budget.

FunctionalArea

CorrectedFY2017Non‐LaborExpense

AssignmentofNewNon‐LaborExpenses*

RecategorizationofMaintenanceContracts**

Adjusted FY2017Non‐LaborExpense

A&G $110,462,390 $28,000,000 $138,462,390CustomerService $14,836,592 $14,836,592Generation $45,963,897 $2,216,250 $16,000,000 $64,180,147T&D $31,214,877 $2,216,250 $33,431,127

Total $202,477,755 $32,432,500 $250,910,255*PREPAEx.14.02.xslm,“Inputs”tab,lines69and73.**ScheduleF‐3REV,PIDs15880,16945,and16946.

a. Assignment of non‐labor spending in categories that arenewsincethetestyear


InordertocompleteouradjustmentsofPREPA’sFY2017non‐laboroperationalbudget,wewerefacedwiththenecessityofmakinganassumptionregardingtheappropriateallocationofPREPA’splannedspendingincategoriesthatarenewsinceFY2014.Asdiscussedabove,thesecategoriesare“RestructuringFees”and“AdditionalSafetyUpgrades.”617

With regards to the “Additional Safety Upgrades” category, we consulted the safetyassessment presentation prepared by DuPont.618 This presentation describes necessarysafety improvements both at PREPA’s generating units and during its transmission anddistributionmaintenanceactivities.Therefore,wemadeasimpleassumptionthatspendinginthiscategorywouldfallevenlybetweenthegenerationandT&Dfunctionalareas.

WeassumedthattheentiretyofPREPA’sspendingonrestructuringfeesistheresponsibilityofitsadministrativeandgeneralarea.

b. Recategorizationofmaintenancecontracts

As described in SECTIONS above, we have identified several line items that PREPAcategorized as capital expenses but that we believe should actually be consideredoperational expenses. These line items are PREPA’s planned spending on contracts forroutinemaintenanceatitsSanJuanandCambalacheunits.619Wehaverecategorizedthesecontractsasgeneration‐relatednon‐laborexpensesandaddedthemintotheFY2017budgetestimateaccordingly.

4. Separationoftransmissionanddistributionbudgets

PREPA’shistoricalrecordsgrouptransmissionanddistributionintoonefunctionalarea,butPREPA’sfilingtreatstheseareasasseparate.Therefore,itwasnecessaryforustoperformthecalculationsaboveintermsofT&DcollectivelyandthenseparateouttransmissionanddistributionbudgetsindividuallyinordertomatchtheformofPREPA’sfiling.Ananalysisofhistorical values broken down into transmission and distribution labor and non‐laborspending from a variety of sources620 shows that transmission spending has historicallyaccounted for approximately 17 percent of both labor and non‐labor spending (withdistributionspendingaccountingfortheremaining83percent).Weusedseparateaveragevaluesforlaborandnon‐laborspending(toallowgreaterprecisionthananassumedintegerpercentagevalue)toallocatetotalT&Dlaborandnon‐laborbudgetsintotransmissionanddistributionexpensesseparately.Wesummarizetheseallocationsinthetablebelow.

617PREPAEx.14.02.xlsm,“Inputs”tab,lines69and73.

618PREPA’sresponsetoCEPR‐TW‐01‐09oftheCommission’sTenthROI(October5,2016).RefertoCEPR‐TW‐01‐09Attach01.pdf.

619PREPA’sPetitionforRateReview,ScheduleF‐3REV,PIDs15880,16945,and16946.

620ScheduleI‐1AppendixIII;ScheduleE‐6REV;CEPR‐AH‐02‐08(b);CEPR‐AH‐02‐09(b).


Table 37. Allocation of T&D operational budget into transmission and distribution budgets.

FunctionalArea

PREPAAssumedPre‐SavingsOperationalExpenseBudget

Labor Non‐LaborTotalAdjusted

Budget†%* Adjusted

Budget%** Adjusted

BudgetT&Dtotal $240,852,630 100% $172,994,163 100% $33,431,127 $206,425,630Transmission $40,503,828 17.1% $29,622,092 16.8% $5,615,067 $35,236,828Distribution $200,348,802 82.9% $143,372,071 83.2% $27,816,059 $171,187,802

*Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐02‐08(b).**Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐02‐09(b).†Totalbudgetsbasedonroundedtotaladjustmentvalues.

5. Summaryofadjustments

ThefollowingtablesummarizesPREPA’spre‐savingsoperationalbudgetsbyfunctionalareaasimpliedoriginallyinthefinancialmodel,thesumtotaloftheadjustmentsdescribedabove,and our final recommended operational budgets by functional area. Adjustments andrecommendedbudgetshavebeenroundedtothenearestthousand.

Table 38. Summary of FY2017 operational budget adjustments.

FunctionalAreaPREPAAssumedPre‐SavingsOperationalExpenseBudget

TotalRoundedAdjustment

RecommendedFY2017OperationalExpenseBudget

A&G $176,232,177 $27,354,000 $203,586,177CustomerService $100,534,690 ‐$1,920,000 $98,614,690Generation $144,882,579 $44,408,000 $189,290,579Transmission $40,503,828 ‐$5,267,000 $35,236,828Distribution $200,348,802 ‐$29,161,000 $171,187,802

Total $662,502,076 $35,414,000 $697,916,076

Although increased from PREPA’s proposal, our adjusted budget remains $63.7 millionbelowPREPA’saverageoperationalspendinginFY2010‐FY2014.621OuradjustmentsresultinsignificantincreasesingenerationexpensesaswellasA&Gexpenses(largelyduetothereallocationofnon‐laborspendingintotheA&Garea).PREPA’sallocationof36percentoftotaloperationalspendingtoT&Dwas,inourobservation,unprecedentedinrecenthistory.Assuch,ourreallocation(beingbasedonhistoricalvalues)resultsincutstothetransmissionanddistributionexpensebudgetsascomparedtothevaluespresentedbyPREPA.However,giventheadhocnatureofPREPA’soriginalbreakdown,wearenotconcernedbythisresult.

We note here explicitly that we do not endorse this methodology for PREPA’s futurebudgetingpractices.Aswithourfueladjustment,thesecalculationsaremeanttoprovideaback‐of‐the‐envelopecalculationofwhatreasonablespendinglevelsatPREPAshouldlooklike.Inthefuture,theCommissionshouldexpecttoseewell‐documentedand‐supported,

621Author’scalculation,basedonvaluesprovidedinCEPR‐AH‐06‐13.xlsx.


bottom‐upbudgetsforeachfunctionalarea,detailingexpectedplansandspending.Forthepurposesofthisratecase,however,webelievetheadjustedbudgetswepresentherewillsuffice.

6. Achievabilityofadjustedspending

StatementsfromPREPAandouranalysisofPREPA’sunderspendingonoperations inthepastthreeyearsbothsuggestthatevenifallocatedadditionalfunds,PREPAmaynotbeabletoactuallyspendtheminFY2017.Indeed,withregardstoitsbaselaborbudgetasdiscussedabove,PREPAstated:

ItisimportanttonotethatPREPAdoesnotbelievethatitwill(orshould)achieveaworkforceof6,694622bytheendofFY2017.ThisnumberwasprojectedinFY2015whentheworkforcewassignificantlylargerinsize.623

AsPREPAexpectsitsactualemployeecounttobelowerthanthenumberonwhichitspre‐savingslaborbudgetisbased,wecaninferthatPREPAdoesnotbelieveitwillactuallyspenditsentirelaborbudgetinFY2017.PREPAdoesnotappeartoviewthisasproblematic—infact, PREPAbelieves it is sustainable tohave several hundred fewer employees than areprojected in the financial model.624 PREPA believes this because of “the significantoperationalimprovementsthathavebeenputinplaceaswellastheworkforceefficiencygainsthathavebeenachievedoverthepasttwofiscalyears.”625

WefindthisstatementoddinthefaceofPREPA’smanyexplicitclaimsthatithasinsufficientemployeesavailabletoperformbasicmaintenance.WhileitisentirelypossiblethatPREPAhasachievedimprovedworkforceefficiency,wefinditimpossibletodeconvoluteanygainsinworkforceefficiencyfromtheseverecutsinstaffthathaveallowedPREPA’sforcedoutageand service interruption rates to rise. As we have established clearly, PREPA’s systemrequires significantmaintenance.Therefore, unlikePREPA,wedo find it concerning thatPREPA believes it will not have—or need—the number of employees it set forth in thefinancialmodel.

Ultimately, the Commissionmust judgewhether or not PREPA’s revenue requirement issufficienttoallowittooperateitssystemsafelyandreliably.Webelievethat,asstatedbyPREPA,itsoperationalbudgetisnotsufficientforthesepurposes.RegardlessofwhetherornotPREPAachievesthis levelofspendinginFY2017, therecordmustshowwhatPREPAactuallyneedstooperateitssystem,ortheclosestavailableestimateofthatvalue.IfPREPA

622Matchesvaluefor“Headcount”inthefinancialmodel.PREPAEx.14.02.xlsm,“Inputs”tab,line61.

623PREPA’sresponsetoCEPR‐RS‐07‐07(b)oftheCommission’sSixteenthROI(November10,2016).

624PREPA’sresponsetoCEPR‐RS‐07‐02(d)oftheCommission’sSixteenthROI(November10,2016).

625Id.


underspendsitsbudgets,excessvaluewill likelybereturnedtoratepayersthroughsomeformofreconciliationmechanism.

InlightofPREPA’sstatementsonthisissue,werecommendthattheCommissionmaintainactive oversight of PREPA’s budget and spending. We further recommend that thesequestions,regardingPREPA’splansanditsactualneeds,beconsideredintheCommission’spendinginvestigationonPREPA’sperformance.

F. Recommendation

1. Additionoffundstooperationsbudget

Asdescribedabove,werecommendaseriesofadjustmentstoPREPA’soperationalexpensebudget. These adjustments result in an addition of $35,414,000 to PREPA’s budget foroperational expenses, of which $16million is a recategorization of funds from PREPA’scapitalbudget.Assuch,oursuggestedadjustmentsaddatotalof$19,414,000toPREPA’srevenuerequirement.

2. Increasedbudgetoversight

In addition to the adjustments we describe above, our analysis of PREPA’s operationalexpensebudgethasemphasizedtheimportanceofincreasedoversightofPREPA’sbudgetingprocessandactualspending.Toenablegreatervisibilityonthesematters,werecommendthattheCommissionrequirePREPAtoadjustitsmonthlyreportformattolistmonthlyandyear‐to‐dateactualspendingandbudgetedvaluesbylaborandnon‐laborexpensesinthesamefunctionalareasusedherein.WerecommendthattheCommissionalsorequirePREPAtoincludeitstotalannualbudgetsandpercentofbudgetsspentinthepastmonthandyear‐to‐datetoenableeasiertrackingofPREPA’sspend‐downsofbudgets.


VIII. CERTIFICATION

Byfilingthisreport,Icertifiedthattheinformation,facts,schedules,exhibitsandanalysisprovidedhereismydirecttestimonyand,tothebestofmyknowledge,trueandcorrect.

JeremyI.Fisher

November21,2016

Byfilingthisreport,Icertifiedthattheinformation,facts,schedules,exhibitsandanalysisprovidedhereismydirecttestimonyand,tothebestofmyknowledge,trueandcorrect.

ArielI.Horowitz

November21,2016

commonwealth of puerto rico puerto...

Documents