energy system risk assessmentenergy system risk assessment james d. mccalley, [email protected] iowa...

Energy System Risk AssessmentEnergy System Risk Assessment

James D. McCalley, [email protected] State University

Workshop on Overarching Issues in Risk Analysis

October 28, 2005

Ames, Iowa

mailto:[email protected]

Five Infrastructure ProblemsFive Infrastructure Problems

1. Transmission control center economy/security system maneuvering

2. Responding to low probability, high consequence events with blackout potential

3. Maintenance: maximizing cumulative risk reduction with limited resources

4. Investing in capital-intensive infrastructure under uncertainty

5. Reliability/economy of national energy transportation system: electric, gas, coal.

Objective: Present five energy infrastructure problems involving risk, with some level of approach to each.

Energy control centers

Security assessmentSecurity assessment• Security: ability of the power system to withstand any of a defined set of next contingencies• Contingencies:

• faults followed by removal of faulted element(s)• “N-k”: k is # of removed elements. Prob ↓ as k ↑• Industry plans for, prepares for: N-1, some N-2.

• Consequences:• Circuit overload• Low voltage

• Voltage instability• Cascading

• Uncontrolled load interruption Blackouts

All control centers analyze these; they are precursors to next level consequences.

Reducing risk?Reducing risk?• Action must be taken if any one prescribed contingency violates performance criteria• Action: Redispatch generation (involves $$)

Prob 1: Optimal Power Flow (OPF)Minimize GenCosts

subject to1. Power flow equations2. Normal condition constraints

Prob 2: Security-constrained OPFMinimize GenCosts

subject to1. Power flow equations2. Normal condition constraints3. Security constraints

But risk is not quantified in these formulations:• Contingency probability varies• Probability of voltage instability/cascading varies

A Stressed SystemA Stressed System

•System is heavily stressed, but there are many companies making lots of $$ for their shareholders !

• August 12, 1999, Thursday, 2 pm• Ambient temperature is ~103 degrees F and still rising• Large city control center• Loading above that in 1999 summer peak planning case

• Bus1 500/230 kV bank has Bus1 500/230 kV bank has been over 100% since noonbeen over 100% since noon

•• It’s loss will result in collapseIt’s loss will result in collapse•• Now, at 2:10 pm, it is 110%Now, at 2:10 pm, it is 110%

Bus 1 500

Bus 1 230

Bus 1 115

Bus 2 500Bus 2 500

Bus2 230Bus2 230

Bus 2 115Bus 2 115

110%

Imports Operator’s view Operator’s view at 2:10 pm, at 2:10 pm, 8/12/998/12/99

200 MW flow

94%

93%

0.95 pu volts

Simulation Simulation Results of a Results of a Preventive ActionPreventive Action

Bus 1 500Bus 1 500

Bus 1 230Bus 1 230

Bus 1 115Bus 1 115

Bus 2 500Bus 2 500

Bus2 230Bus2 230

Bus 2 115Bus 2 115

103%

Imports

0 MW flow

101%

104%

0.91 pu volts

Risk Calculation

Evaluated using variants of power flow calculation (Newton-Raphson iterative solution to high dimensional nonlinear algebraic equations)

Forecasted operatingcondition

Low voltage

Cascading

Overload

Voltage instability::

p(1)

p(2)

p(3)

p(n)

∑∑=

jtype

problemj

kiescontingenc

kXSevkpXRisk )|()()(

Severity functions...• Risk:Expected severity in next hour. • Severity=1.0 atimposed limit.• Risk=1.0 isequivalent to one violation.

0.90

1

Sev

erity

Voltage magnitude (p.u.) 0.95

Fig 6.4: Low voltage severity function

90

1

Sev

erity

100 Percent loading

Fig 6.7: Overload severity function

0 5 Fig 6.9: Voltage instability severity function

1

%margin

Sev

erity

1E6

This value assigned to provide Risk=1.0 if lowest probability contingency (p=1E-6) results in voltage instability.

This amount of risk equates to what industry has indicated is unacceptable.

Contingency Probability EstimationDistinguish between contingency probabilities based on

• Historical outage data

• Physical attributes: length, voltage level, geography

• Temporal attributes: weather

1. Separate outage data into 24 pools (8 zones x 3 kV levels)

2. For each pool,

• separate outage data into “weather blocks” and compute failure rate/mile for each block, resulting in about 30 values

• Use linear regression to determine dependence of failure rate on weather

3. On-line, evaluate failure rates/mile for each pool, multiply by line length for each circuit.

Contingency Probability Estimation

1 2 3 4 5 6 7 8 9 1011

1 2 3 4 5 6 7 8 9101112

0

2

4

6

8

Windspeed(miles/hour)/3Temperature(F)/10

Num

ber o

f out

ages

(200

1~20

04)

North central data

24

68

10

05101520

-14

-13

-12

-11

-10

-9

x2:Wind speed (miles/h)/3x1:Temperature (F)/10

y Lo

g(Fa

ilure

rate

)

0 5 10 15 20 253

3.5

4

4.5

5

5.5

6

6.5

7x 10-5

Time(one day)

failu

re ra

te

Visualization over time

…and operating conditions

and the ability to drill-down to identify nature & cause of high risk

Top 30 High-Risk Contingencies (from 714)Showing resulting OL+LV risk

…and space

Prob 3: Risk-Objective SCOPFMinimize GenCosts +β×Risk


• This formulation requires sensitivity of risk to each generator injection. • Why is this formulation an improvement?

Deterministic security limits (normal and contingency) are enforced, on targeted basis

Overall risk is reduced.The balance between cost and risk reduction

may be observed (and controlled).

Risk reduction using “targeting” redispatchProb 2: Security-constrained OPFMinimize GenCosts


Risk reduction using “targeting” redispatch

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

6.65 6.66 6.67 6.68 6.69 6.7 6.71 6.72 6.73

β=10000β=6000

β=5500

β=5200

β=5000

β=4800

Risk

Cost of redispatch

NN--k contingenciesk contingencies

What to say, operationally, about high-order (N-k, k>1) contingencies?

• If probability is high, put it in contingency list and treat it as any other N-1 contingency (e.g., take preventive actions as needed)

• If probability is low, monitor it, perhaps identify corrective actions for it should it occur, but do not spend money in preventive actions.

How to identify high-probability N-k contingencies?

Probability OrderProbability OrderDefinition: the probability order of a contingency is the number of independent events necessary for occurrence of that contingency.

Probability order is a rough way of comparing probabilities of different contingencies.

Assume the probability of any single event (faulted line, protection failure, etc) is 10-4.

Then, for independent events, occurrence of • two events is P(A)*P(B)=10-8 (order 2), • three events P(A)*P(B)*P(C)=10-12 (order 3), etc.

But probability of 2 dependent events is P(A)P(B|A), and P(B|A) can be 1.0, so the probability of the two dependent events is P(A).

Classification of NClassification of N--k contingenciesk contingenciesProtection system failures (NERC category C or D):Protection system failures (NERC category C or D):

Inadvertent operation, failure is exposed after a firstInadvertent operation, failure is exposed after a first--faultfaultFailure to operate when needed.Failure to operate when needed.

Both cases require Both cases require fault+existencefault+existence of protection failure: order=2.of protection failure: order=2.Are there single events that cause NAre there single events that cause N--k outages? Yes,…k outages? Yes,…

1.1. Common mode outage (NERC class D) such as hurricane, earthquake,Common mode outage (NERC class D) such as hurricane, earthquake, airplaneairplane2.2. Breaker fault (NERC class D)Breaker fault (NERC class D)3.3. Substation topology problem (maintenance or careless switching)Substation topology problem (maintenance or careless switching)4.4. Cascading following a firstCascading following a first--faultfault

But common mode and breaker faults have probabilities closer to But common mode and breaker faults have probabilities closer to order 2. This leaves #3 and#4.order 2. This leaves #3 and#4.Since #3 requires only a single fault, it has order 1.Since #3 requires only a single fault, it has order 1.Since #4 is dependent, it can have order close to 1.Since #4 is dependent, it can have order close to 1.

Example of topological weakness

BR

EA

KE

R-1

BR

EA

KE

R-2

BR

EA

KE

R-3

SWIT

CH

-1

SWIT

CH

-2

SWIT

CH

-3 BUS-1

BUS-2

Line-1 Line-2 Line-3

SWIT

CH

-1

SWIT

CH

-2

SWIT

CH

-3

BUS-1

BUS-2

Line-1 Line-2 Line-3

There are thousands of such substations in a model.

We perform a topological graph search using the breaker/switch and connectivity data to identify these high-probability N-k contingencies.

Remove bus 1 from service, and a single fault on any line results in N-3 contingency.

CascadingCascading risk depends on:• Occurrence probability of a first contingency k (Level 0).• Probability of all possible Level 1 trips, computed as a function of

post-contingency loading on remaining circuits• Severity of cascading sequence following each Level 1 trip,

computed as a function of number of circuits lost, or if no convergence, as a function of voltage collapse severity.

Level: 0 1 2 3 4 5 6 7 8 9 10 11 12…

NL1=5NL2=8

NL4=Voltage instab severity

NL5=12

NL3=1

P1

P2

P3

P4

P5

∑=

=m

iLii NPkPkRisk

1*)()(

P(k)

Graphic comparison of different probability models for N-k contingencies

110 -10

10 -8

10 -6

10 -4

10 -2

10 0

2 3 4 5 6 7 8

Number of lines lost in each contingency (log scale)

Log-Probability

Poisson Model (Concave)

Cluster Model (Concave)

Power Law Model (Straight line) Observed Prob.

( ) ( )

1

1

3.1151 1

1 1

0.12657 1

3.115 2Pr( 3.115, 1.12657)

1

1.1266 1 3.115 for cluster model 1.1266 1 3.115 1.1266 1 3.115

Pr 0.12657 0.12657 1 ! for poisson

x

x

xX x

x

X x e x

α μ

λ

−

−

− −

− −

− −

⎛ ⎞+ −= = = = ⎜ ⎟

−⎝ ⎠

⎛ ⎞−⎛ ⎞× × ⎜ ⎟⎜ ⎟− + − +⎝ ⎠ ⎝ ⎠

= = = −

3.78

model

Pr(X 3.78 ) 0.9098 for pow er law model -x p x

⎧⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪ = = =⎪⎩

Final result for maximum likelihood estimation of parameters for the three probability models

WHAT HAPPENED ON WHAT HAPPENED ON AUGUST 14, 2003???AUGUST 14, 2003???

1. 12:05 Conesville Unit 5 (rating 375 MW)2. 1:14 Greenwood Unit 1 (rating 785 MW)3. 1:31 Eastlake Unit 5 (rating: 597 MW)

INITIATING EVENT

4. 2:02 Stuart – Atlanta 345 kV5. 3:05 Harding-Chamberlain 345 kV6. 3:32 Hanna-Juniper 345 kV7. 3:41 Star-South Canton 345 kV8. 3:45 Canton Central-Tidd 345 kV9. 4:06 Sammis-Star 345 kV

SLOW PROGRESSION

10. 4:08:58 Galion-Ohio Central-Muskingum 345 kV11. 4:09:06 East Lima-Fostoria Central 345 kV12. 4:09:23-4:10:27 Kinder Morgan (rating: 500 MW; loaded to 200 MW)13. 4:10 Harding-Fox 345 kV14. 4:10:04 – 4:10:45 20 generators along Lake Erie in north Ohio, 2174 MW15. 4:10:37 West-East Michigan 345 kV16. 4:10:38 Midland Cogeneration Venture, 1265 MW17. 4:10:38 Transmission system separates northwest of Detroit18. 4:10:38 Perry-Ashtabula-Erie West 345 kV19. 4:10:40 – 4:10:44 4 lines disconnect between Pennsylvania & New York20. 4:10:41 2 lines disconnect and 2 gens trip in north Ohio,1868MW21. 4:10:42 – 4:10:45 3 lines disconnect in north Ontario, New Jersey, isolates NE part

of Eastern Interconnection, 1 unit trips, 820 mw22. 4:10:46 – 4:10:55 New York splits east-to-west. New England and Maritimes

separate from New York and remain intact.23. 4:10:50 – 4:11:57 Ontario separates from NY w. of Niagara Falls & w. of St. Law.

SW Connecticut separates from New York, blacks out.

FAST PROGRESSION

Location Date Scale in term of MW or Population

Collapse time

US-NE 10/9/65 20GW, 30M people 13 mins New York 7/13/77 6GW, 9M people 1 hour

France 1978 29GW 26 mins Japan 1987 8.2GW 20mins

US-West 1/17/94 7.5GW 1 min US-West 12/14/94 9.3GW US-West 7/2/96 11.7GW 36 secondsUS-West 7/3/96 1.2GW > 1 min US-West 8/10/96 30.5GW > 6 mins

Brazil 3/11/99 25GW 30 secs US-NE 8/14/03 62GW, 50M people > 1 hour London 8/28/03 724 MW, 476K people 8 secs

Denmark & Sweden 9/23/03 4.85M people 7mins Italy 9/28/03 27.7GW, 57M people 27mins

Larger Blackouts in Last 40 years

An Analogy to Air Traffic ControlAn Analogy to Air Traffic Control

Unfolding Cascading Event

Remedial action by the operator

Normal Stage

Emergency Stage

Catastrophic Outcome

Large area blackout avoided

Emergency Response System

Without action

time

Airplanes getting too close to each other Avoidance Action

by the TCAS

Normal Stage Emergency

Stage

Collision

Collision avoided

Traffic Alert and Collision Avoidance System

Without action

Rapid Response to Unfolding EventsRapid Response to Unfolding Events

Bus voltage

Contingency-1: Fau lt +N-3 outage from stuck breaker

Action-1: insert shunt cap

Behavior-2:Fast voltage co llapse due to lack of reactive power

Behavior-3:Slow voltage collapse due to LTC action

Action-2: block LTCs

10 sec 5 min

Behavior-1: System is normal. Everything is within lim it.

Time 0

1.0

B1 C1

S1 S2 S3

S4

S5 S7

S6

A1

B2

A2

B3

A Dynamic Decision Event TreeA Dynamic Decision Event Tree

Power System Maintenance (actually an interesting subject)

Deterioration level 4

(Failed)



Deterioration level 1 New

Haz

ard

Rat

e

Time

Δlambda

Δt

Before maintenance (Level 3)

After maintenance (Level 1)

When maintenance needs require resources that exceed When maintenance needs require resources that exceed available resources, how to strategically allocate available available resources, how to strategically allocate available resources to maximize benefit gained from them?resources to maximize benefit gained from them?

Fault from power line to trees Tree trimming

Examples of failure modes & maintenanceExamples of failure modes & maintenance

Transformer oil degradation and insulation failure

Oil Reconditioning

Cumulative risk calculation

Year long, hourly risk variation for each contingency

Contingencyprobabilities

Power system simulator

Mid-term load forecast

Benefit: Maintenance reduces contingency probabilities which reduces cumulative-over-time risk

Obtaining failure rate reduction

ajk

deterioration

function g(c(T+1))

Historical data c(t) t=1,…,T

Statistical

Processing

Most recent observation

c(T+1)

a23 a12 a34 Level 1 (new)

Level 2 (minor)

Level 3 (major)

Level 4 (failed)

⎭⎬⎫

⎩⎨⎧

=Δ

=Δ ∑8760

00

00

),(),()(

),,(),,(t

tkRtkCumRiskkp

tkmptkmCumRisk

Developed for each failure mode of each component

Optimization∑ ∑∑

component task time time) task,ent,ion(componRiskReduct

subject to:budget limits, crew constraints for different maintenance categories, timing constraints for maintenance tasks requiring outages

maximize

0

20

40

60

80

100

120

140

160

180

200

Quarterly allocated CRR and resource allocation for different maintenance categories -Case B

Tree Trimming

Trans Line

Xfmr Minor

Xfmr Major

CB Maint.

Quarter 1Quarter 2 Quarter 3

Quarter 4Fiscal Year

ECRR (k$)

Labor (100Hrs)Cost (k$)

The National Electric Energy SystemThe National Electric Energy System

Fig. 1: Gas, Rail, and Electric Trans portation Systems

NEES:NEES: iintegrated infrastructure associated with ntegrated infrastructure associated with production, transportation, storage, endproduction, transportation, storage, end--use of four use of four energy forms: electricity, gas, coal, and water.energy forms: electricity, gas, coal, and water.NEES integrity depends on NEES integrity depends on

electric generation and transmission subsystems electric generation and transmission subsystems ability to produce and transport the fuelability to produce and transport the fuel

Vulnerability to disruptions is due to natural causes, Vulnerability to disruptions is due to natural causes, equipment failure, labor unavailability, equipment failure, labor unavailability, communication failures, terrorist attacks.communication failures, terrorist attacks.

The National Electric Energy SystemThe National Electric Energy System

The NEESRainfall & Snow

RunoffGas Wells Coal Mines

… … …

Raw Energy Supplies

WaterHydraulic System

GasPipelines

CoalRailroads, Barge

… ……

Gas StorageReservoirs Coal PilesStorage & Transport. Systems

… … …

… … …

Generation System

… … … … … … … … … … …Electric Energy Demand

ElectricityElectric Transmission System

Electric Transm. System :

Nuclear Plants

Other Plants

(wind, solar, etc)

Examples of DisruptionsExamples of Disruptions

Lightning strike

Labor strikesPekin, IL: 13 345kV transmission lines destroyed by a tornado in May 2003

El Paso, NM, 2000: Gas pipeline rupture

Ellet Valley, VA, 2003: Norfolk Southern coal train derailed

Examples of DisruptionsExamples of Disruptions

Black Thunder, WY, 2005: Coal train derailment

1993 Flood Stops Barge Traffic

Disruption to Gulf Coast Gas Production from Katrina/Rita

Examples of Disruptions Examples of Disruptions On August 22, 2000, a 30 inch pipeline ruptured in New On August 22, 2000, a 30 inch pipeline ruptured in New Mexico, and was forced out of service, taking with it two Mexico, and was forced out of service, taking with it two parallel lines that together form a major artery into Californiaparallel lines that together form a major artery into California. . This decreased availability of gas in California, significantly This decreased availability of gas in California, significantly driving up price as seen by owners of gasdriving up price as seen by owners of gas--fired electricity fired electricity suppliers as well as residential and commercial gas endsuppliers as well as residential and commercial gas end--users. users. At the same time, California was experiencing a decrease in At the same time, California was experiencing a decrease in precipitation, forced outage of several large coalprecipitation, forced outage of several large coal--fired units, fired units, and a weakened transmission system. These factors and a weakened transmission system. These factors contributed to what is now well known as the California contributed to what is now well known as the California energy crisis, characterized by electricity shortages and high energy crisis, characterized by electricity shortages and high prices. Yet, electricity endprices. Yet, electricity end--users were insulated from the users were insulated from the electricity price increases because of regulatory price caps. electricity price increases because of regulatory price caps. Therefore, as gas prices rose, and electricity prices did not, Therefore, as gas prices rose, and electricity prices did not, many consumers quite naturally switched from gas heat to many consumers quite naturally switched from gas heat to electric heat, further exacerbating the electricity shortage. electric heat, further exacerbating the electricity shortage.

Examples of Disruptions Examples of Disruptions

The 1993 Mississippi River flood caused major disruptions The 1993 Mississippi River flood caused major disruptions in the U.S. energy supply. The Mississippi River itself is a in the U.S. energy supply. The Mississippi River itself is a crucial part of the Midwest’s economic infrastructure. crucial part of the Midwest’s economic infrastructure. Barges carry 20% of the nation’s coal, a third of its Barges carry 20% of the nation’s coal, a third of its petroleum, and half of its exported grain. Barge traffic was petroleum, and half of its exported grain. Barge traffic was halted for two months; carriers lost an estimated $1 million halted for two months; carriers lost an estimated $1 million per day. Some power plants along the river saw their coal per day. Some power plants along the river saw their coal stocks dwindle from a twostocks dwindle from a two--month supply to enough to last month supply to enough to last just for a few days.just for a few days.

Examples of Disruptions Examples of Disruptions Ten giant coal mines in Wyoming 's Powder River Basin produce neTen giant coal mines in Wyoming 's Powder River Basin produce nearly arly 40% of the U.S. supply. And coal powers more than half of U.S. 40% of the U.S. supply. And coal powers more than half of U.S. electricity generation. A heavy snowstorm blanketed Wyoming on Melectricity generation. A heavy snowstorm blanketed Wyoming on May ay 11 2005, just as the ice in the surrounding mountains had begun 11 2005, just as the ice in the surrounding mountains had begun to thaw. to thaw. Icy water and coal dust merged into a thick, dirty slurry and ooIcy water and coal dust merged into a thick, dirty slurry and oozed across zed across a 100a 100--mile section of railroad freight track, causing two derailments mile section of railroad freight track, causing two derailments with with major track damage. Spotmajor track damage. Spot--market prices for the basin's coal are up nearly market prices for the basin's coal are up nearly 70% year to date. The hot summer weather left power plants with 70% year to date. The hot summer weather left power plants with especially low stockpiles exiting the summer, so utilities may nespecially low stockpiles exiting the summer, so utilities may not be able ot be able to rebuild stockpiles until after next year. As a result, electrto rebuild stockpiles until after next year. As a result, electric utilities ic utilities have been relying more on natural gashave been relying more on natural gas--fired plants to satisfy demand. fired plants to satisfy demand. Then the full effects of Katrina and Rita on coal (Mississippi bThen the full effects of Katrina and Rita on coal (Mississippi barge arge traffic) and on gas (Gulf wells) are not yet known. traffic) and on gas (Gulf wells) are not yet known.

How will prices and availability of electric energy evolve in thHow will prices and availability of electric energy evolve in the next year?e next year?

Reliability of the NEESReliability of the NEES

Identify conditions that significantly impact price Identify conditions that significantly impact price and availability of electrical energy.and availability of electrical energy.Assess the overall reliability of the energy system Assess the overall reliability of the energy system in order to elaborate preventive and corrective in order to elaborate preventive and corrective plans to avoid massive energy shortages.plans to avoid massive energy shortages.Present an network flow optimization model for Present an network flow optimization model for reliability assessment of the NEES, where the reliability assessment of the NEES, where the subsystems are analyzed together in a single subsystems are analyzed together in a single integrated mathematical framework for the energy integrated mathematical framework for the energy production, transportation, storage, generation, production, transportation, storage, generation, and transmission. and transmission.

HighHigh--level Representationlevel Representation

Period 1 for coal

Period 1for

electricity

Period 1 for gas

Period 2for

electricity

Period 3for

electricity

Period 4for

electricity

Period 2 for gas

… … … …

… … … …Period 2 for coal

Period 5for

electricity

Period 3 for gas

Period 6for

electricity

Period 7for

electricity

Period 8for

electricity

Period 4 for gas

… … … …

… … … …

……… …

Example of network representationExample of network representationS

CG1 CG2

CP2CP1

CS1

GG1 GG2

GP1

GS1

WG1

WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CG1 CG2

CP2CP1

CS1

GG1 GG2

GP1

GS1

WG1

WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

Raw-Energy time step 1


Electric time step 1 Electric time step 2 Electric time step 3 Electric time step 1 Electric time step 2 Electric time step 3

b1(CA1,1) b1(CA2,1) b1(CA1,2) b1(CA2,2) b1(CA1,3) b1(CA2,3) b2(CA1,1) b2(CA2,1) b2(CA1,2) b2(CA2,2) b2(CA1,3) b2(CA2,3)

b2(WS1)b2(GS1)b1(WS1)

b1(CS1)

b1(GS1)

b2(CS1)

SS

CG1 CG2

CP2CP1

CS1

GG1 GG2

GP1

GS1

WG1

WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CG1 CG2

CP2CP1

CS1

CG1CG1 CG2CG2

CP2CP2CP1CP1

CS1CS1

GG1 GG2

GP1

GS1

GG1GG1 GG2GG2

GP1GP1

GS1GS1

WG1

WS1

WG1WG1

WS1WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,1CA1,1 CA2,1CA2,1

CG1,1CG1,1 CG2,1CG2,1 GG1,1GG1,1 GG2,1GG2,1 WG1,1WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,2CA1,2 CA2,2CA2,2


CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CA1,3CA1,3 CA2,3CA2,3


CG1 CG2

CP2CP1

CS1

GG1 GG2

GP1

GS1

WG1

WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CG1 CG2

CP2CP1

CS1

CG1CG1 CG2CG2

CP2CP2CP1CP1

CS1CS1

GG1 GG2

GP1

GS1

GG1GG1 GG2GG2

GP1GP1

GS1GS1

WG1

WS1

WG1WG1

WS1WS1

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CA1,1 CA2,1

CG1,1 CG2,1 GG1,1 GG2,1 WG1,1

CA1,1CA1,1 CA2,1CA2,1


CA1,2 CA2,2

CG1,2 CG2,2 GG1,2 GG2,2 WG1,2

CA1,2CA1,2 CA2,2CA2,2


CA1,3 CA2,3

CG1,3 CG2,3 GG1,3 GG2,3 WG1,3

CA1,3CA1,3 CA2,3CA2,3




Electric time step 1 Electric time step 2 Electric time step 3 Electric time step 1 Electric time step 2 Electric time step 3

b1(CA1,1) b1(CA2,1) b1(CA1,2) b1(CA2,2) b1(CA1,3) b1(CA2,3) b2(CA1,1) b2(CA2,1) b2(CA1,2) b2(CA2,2) b2(CA1,3) b2(CA2,3)

b2(WS1)b2(GS1)b1(WS1)

b1(CS1)

b1(GS1)

b2(CS1)

NodesNodes: actual facilities : actual facilities of the NEES of the NEES

ArcsArcs: transportation : transportation routes and modesroutes and modes

Energy system facilities can be represented Energy system facilities can be represented adequately by arcs and nodes. adequately by arcs and nodes. A network model allows representation of A network model allows representation of capacities, costs, and efficiencies.capacities, costs, and efficiencies.Bulk energy movements can be Bulk energy movements can be represented as flows.represented as flows.Take advantage of fast and existent Take advantage of fast and existent network optimization algorithms:network optimization algorithms:

Generalized minimum cost (GMC)Generalized minimum cost (GMC)Generalized maximum flow (GMF)Generalized maximum flow (GMF)

Capacities: uncertainty due to disruptionsCapacities: uncertainty due to disruptionsCosts/prices: market uncertaintyCosts/prices: market uncertaintyDemand: end user uncertaintyDemand: end user uncertainty

Where is the Where is the Randomness in the Randomness in the

Network Model?Network Model?

Why a Network Model?Why a Network Model?

Stochastic Network Model

Research OutlineResearch OutlineBuild an operational model, which should integrate and assess reBuild an operational model, which should integrate and assess reliability of liability of the NEES.the NEES.Gather & organize the necessary data of the different subsystem Gather & organize the necessary data of the different subsystem networks.networks.Predict and represent uncertainty associated with extreme continPredict and represent uncertainty associated with extreme contingencies: It is gencies: It is necessary to build a model to represent the uncertainty associatnecessary to build a model to represent the uncertainty associated to ed to catastrophic events and their effects in the energy grid.catastrophic events and their effects in the energy grid.Define plausible and credible multiple contingency scenarios, usDefine plausible and credible multiple contingency scenarios, using 2 different ing 2 different criteria: Cascading events and common mode events.criteria: Cascading events and common mode events.Evaluate the impact on the NEES of the contingency's scenariosEvaluate the impact on the NEES of the contingency's scenariosEvaluate how the effects of those contingencies propagate geograEvaluate how the effects of those contingencies propagate geographically and phically and in time.in time.

Conclusions

Transmission system risk assessment and related decision Transmission system risk assessment and related decision depends on:depends on:

Identification of contingencies and their probabilitiesIdentification of contingencies and their probabilitiesRigorous analysis of contingency consequencesRigorous analysis of contingency consequencesAppropriate operator decisionAppropriate operator decision--support tools for both preventive support tools for both preventive and corrective controland corrective controlLongLong--term implementation of strategic resource allocation for term implementation of strategic resource allocation for maintenancemaintenanceRiskRisk--informed decisions in facility investmentinformed decisions in facility investment

Delivering reliable and economic electric energy also Delivering reliable and economic electric energy also depends on understanding the entire, integrated energy depends on understanding the entire, integrated energy system from fuel source to electric distribution substation.system from fuel source to electric distribution substation.

energy system risk assessmentenergy system risk assessment james d. mccalley, [email protected] iowa...

Documents