“.

TITLE.

AUTHOR(S).

SUBMITTED TO

LA-UF+-92-1373

DE92 013431

WASTE MEASUREMENTS AT A PLUTONIUFI FACILITY

Joseph R. ~achter

ESARi)A

DISCLAIMER

Thic ropl wm prcporcd Mman mxounl O( work sponoored by an s#cncy d Ibe IJrmcd Malec

Oovernmen! Ncllhcr the Uni!ai SmIes Govcrnmeni nor ●ny agency Ihcreof, nor my of their

employecc, makes nny wmrraniy, enprcu or lmpliaJ, or mcumai any legal Iiablluy or remprrrti.

bility for the ●xurnuy, complelcnoos, w uwfulnm Of W infurm~lion, nPW~lul, producf, or

pnxem dirdd. of reprc*n\n Ihal its we would not mffin~e prwmely ownal r}~hirn. Refer-

ence herein 10 uny qmofic commercial pnx!ucl, prcwxac, or Krwu hy Irnde name, Irsdemark,

mwrufscturer, or olherww dm nol nocamtily cormlilule o.’ Imply IIS cndurmomcnl, rc.com-

mendailon, or (svmmg hy (he (Irukl WICS (iovcrnmcnt or my qency thereof The VICWS

and opmmrru of sulhum e~preocod herein do nor nwcwnrily maIa or rellod Ibu O( !he

Unilal Slntoc (iow nment or wry mgcncy Ihorcd

n

Los Akmms,

MASTERLos Alamos National Laboratorylms Alamos,New Mexico 87545

4* ● I (1,( i

WASTE MEASUREMENTS AT A PLUTONIUM SCRAP RECOVERY FACILITY

Joseph R. WachterLos Alamos National Laboratory

Los Alamos, NM U.S.A.

solid plutonium contaminated wastea are often highlyheterogeneous, span a wide range of chemical compositions andmatrix types, and arm packaged in a variety of container sizes. NDAanalysis of this waste depends on operator knowledge of theseparameters so that proper segregation, instrument -election,quality assurance, and uncertainty estimation can take place.. Thisreport dascribea current waate measurement practices anduncertainty estimates at a U.S. plutonium scrap recovery facilityand presents a program for doterminin~ reproducibility and bias inNDA measurements. Following thim, an operator’s perspective ondesirable NDA upgrades is offered.

Los Alamos National Laboratory (ML) is a U.S. nationaldefense facility involved in th~ r-covery and processing ofplutonium. Wastes and residues uro routinoly gonorated hero frommany ●tage~ of plutonium metal fabrication including pyrochomicaland aqueous processing of plutonium scrap. Thoso processing stopsproduca a wido variety of leaner scrap and wasto fores ●uch aaplutonium oxida from burned realduas, Pu-bearing salts fromproduction/reduction and metal purification procossos, impuroplutonium metal, metal reduction slags, ash, undisnolvod oxidohaalm, coramicm, cleaning rags, plastics, HEPA filters, and otherromnanta and apparatus qonaratod from processing and cleanuptasknl’z. If th- procasnad raaiduo ham a concentration such thatplutonium rocovary is economically justified, it is classified asscrap and rotainod for later traatm.nt. If it is below ●conomicalrocovary limits, it in dirnpoaed of am radioactive waste. Bothliquid and solid wastes r~sult from those procosaing stapm. Liquidwaatas inclu~s ●ffluontm from ion-axchanga column-, oxaletofiltratom, and cauetic solutions qonara~ad by various head-ond andpurification oparationo. Tho solutionm aro filtarod, traatod in ar,●vaporator, and chemically ●amplad to dstormino di~card criterion.No Nondamtructivo AnalyBes (NDA) aro performad on the liquld

1

wastes. Upon meeting the criterion, the filtrates and sludges arefixed in cement and gent to a waste storage complex. Solid wastescontaminated with plutonium are divided into two categories:Transuranic (TRU; waste containing greater than 3700 Bq/g (100nCi/g) of radioactivity and Low hvel Waste (LLW) containing lessthan this amount. TRU waste is obtained from various processing andcleanup steps within the processing areas while the LLW is derivedfrom locations where contamination is expected to be light. Bothsolid waste categories receive NDA analyses to determineradioactive cGntent. Presently, neither waste form is immobilizedprior to disposal. The purpome of this report is to review thecurrent status of solid waste measurement operations here, examinethe reliability of those measurements, and consider desirableupgrades in support of future waste analysis. The first section ofthe report will briefly review current measurement practices atthis facility. In this section, ~egregation of the waste prior toanalysis, packaging, instrument selection, and quality assuranceproceds~res will be discussed. Proper management of these functionsoften contributes as much to measurement reliability as NDAtechnique. Next, the report will discuae reproducibility and biasostimatcs for wasto maasurementa performed hero and how these●stimatoe wer, justified. Finally, an oparator’s perspectlvo on themos~ important needs in current NDA methodology will be prooanted.Limtud hers ars software, ha~dware, instrument ●valuation, and●tandards roquiromentu needed to m-at anticipated safeguards,quality assurance, and regulatory damanda.

Aftar tho various processing ●tapn, the ro-ulting rosidua Ic●egragat~d according to matrix typo, inspectad for hazardouamaterials (flammabloa, ●xplosivas , carcinogano, ●tc) , and packag~dinto contain~rs ranging from 10 cm diam x 10 cm tall to 208 1drum. It is than cov~rad in plastic shooting, romovod from thaprocoesinq glov,box, and dalivor~d to the NDA laboratory foranalysia. Matarials ●uch as !}apor, plaatics, collulosicm, ash,powdarm, and othar low d.nsity matarialn contaminated withplutonium nra maamured with Sogmontod Gamma Scannars (SGS) usinghigh purity Go d.tsctorc. High and intarmodiata dansity TRU wast~,including metals, loaded qlovaa, tools, motore, and ●oma laan saltro~idums are analyzed by pamaiva neutron coincidanco counting (NCC)mathdm. If tha maasuromant ●-tmbliahoa that an item in below●conomic discard limit- ant can b. disposod of am TRU wasta, it is

2

loaded into a drum with other items of similar matrix composition.Finally, the filled drums are sealed and given an NDA confirmationmeasurement in order to validate the initial measurements andverify t!~at ‘,!vers~on has not taken place. If the confirmationmeasurement is within established limits, the drum is delivered toa waste holding area to await permanent disposal.

For Low Level Waste, compatible materials are first separatedfrom noncompactibles. The two waste forms are then packaged intocardboard boxes, measured to detemine that they meet LLWcriterion, and removed from the processing facility for shallowland burial. LLW is analyzed with a multienergy detection schemeusing a NaI detector to monitor the x-ray region for items whichmeet LLW criterion, and subsequently crossing over to Frogrepsivelyhigher energy gamma rays to measure elevated activity levels.Oversize waste (i.e., waato that in too lhrge to fit into standardLLW containers) is presently analyzad with four slab counters whicheurround the sample and measure the uncorrelated neutron signal.

At present, there are 15 gamma-ray and neutron as~ayinstruments used to measure radioactive wauta at this facility.They are operated and maintained by seven calibration andinstrument repair personnel, ten measurement ●mployees, and fivetechnical support and ov.rsight staff.

A wide variety of tests and measurements are perfomed toassure instrument performance and proper accounting of specialnuclear materials. When a new NDA instrument arrivoe at thisfacility, it i= subjacted to a rigorous ●eries of qualliicatior,checks and maasurementn under ●xpoctad operating conditions. Thoseinclude hardwara, softwar., and algorithm chack-out; short and longterm Instrument stability ●valuation; as~oy range determination;asmassmant of chemicnl forms and matrix typflswhich can b. reliablyasmayod with the naw inmtrumant; raaGlution of uoasuramant controlparameters; e~tablirnhmant of calibration standards and frequency;documentation devolopmant; and oparator training. New instr~unentqualification typically rquire~ about ●ix months of ●ffort and ioreviewed by safeguards, waste management, and quality aoouranc~personnel. Upon completion of tho qualification raquiramonts, aninstrument i- put into regular narvico. ?’haraafter, instrumentparformanca in monitored throuqh daily and weakly stability tests,monthly revlcw of stability dato by oporators and safeguardsparsonnel, periodic calibration chocks, and annual training andoperatil~g procedur. updates. Instruments which are found to beoperating out-ids the stability tc~t limits arm immadiataly pulledfrom ●ervice until the reasonm for tho failura hava bean diagnosed

3

and a new series of stability measurements have been successfullycompleted. Other elements in the medley of measurements ar,dchecksthat assure instrument performance and plutonium accountability arethe confirmation waste measurement protocol mentioned above, ad hocstudies of instrument performance, near real time processaccountability, wudits by outside agencies, and the MeasurementVerification Program (MVP)o

Despite the instrument qualification assessment, measurementcontrol procedures, and waste segregation, nondestructive assays ofthe waste forms indicated above are often troublesome and may poseformidable challenges to the measurement :;pecialist.Reasons fordifficulties include:

- TRU waste comes in a variety of chemical compounds, physicalsizes, isotopic proportions, and matrix compositions. Each ofthese may present complications for different measurementmethodologies. For example, the presence of plutonium metalshot in the waste may impede gamma-ray measurements whereasmultiplication and (ajn) effects can limit the credibility ofneutron counter assays. Knowledge and quantification of these●ffects ar6 often difficult for some waste forms.

- At this facility, most of the T’RU and LLW waste is of aheterogeneous nature. Geometric variations in the spatiallo-cationofradiating materials affect instrument response andcan limit assay reliability.

- Comparison measurements of SGS and NCC assays with highlyaccurate techniques such as calorimetry or spectrochemicalanalyses aro a useful tool for ●stablishing measurementefficacy. However, these are often not possible because of thediverse nature of the vast. and the axpanao of the altarnativoanalyses. Thus, an important supplementary tool for biasevaluation is not availablo for measuromonte of some wastoforms c

- Reprasantativa standards for many kinds of waste matsrials donot ●xist and may prova impossible to fabricate. Indeed, thenature of many hateroganoous waste ●amplos defy attampts tomatch thornwith raprasantative standards.

- Oversize wasto maasuraments are often hampered by backgroundand geometry ●ffects. Thosa may limit aasay sensitivity andintroduco unacceptably largo uncartaintics into the maasuredvalues.

To maintain additional confidence in the assays performed at

4

LANL, studies are routinely conducted to monitor the reliability ofthe NDA measurements of the plutonium product materials, scrap,residues, and waste generated here. During the past three years,for example, studies have been carried out on neutron and gamma-raymeasurements of hydrofluorination residues, matrix effects inpassive neutron counters, comparative NDA analyses of molten saltextraction residues, evaluations of three gamma-ray isotopic codesfor high 241Am materials, appraisal of a self attenuation correctionfor SGS analysis, and NCC measurements of direct oxide reductionsalts, among others. In addition to these studies, there is anongoing program for evaluating NDA measurements over the broadspectrum of process residue and TRU waste forms generated at thisfacility. This Measurement Verification Program was originallyIntended as a mechanism for resolving Inventory differences and foruncovering unreliable instrument performance. But the datacollected over the five year span of the program, to date, havealso revealed information on random and @ystematlc uncertainties Inthe measurements that would be difficult to detemine by othermean=. Although the program generally concentrates on scrap andresidue materials, that is, on materials which have a higherconcentration of plutonium than waste; their chemical composition,matrix fore, instrument selection, and packaging are generallyequivalent to TRU waste here. Therefore, examinations of instrumentperformance for scrap and residueu can be used to provide insightinto the performance of the Instruments for measuring wa~te. Thereare several segments to the program; however only the two thatpertain to waste measurements will be Included in thi~ report. Onepertinunt ●egment comparas SGS and NCC analyses with praviousmeasurements made by the same instrument, while a second segmentcompares the SGS and NCC measurements with referenco valuasobtained from calorimetry + isotopics (CI) analysis. A discussionof the rasults obtainod from those ●egmanta of tha program is givenbelow.

The MVP is applied to tho ●ntire range of NDA instruments thatmeauura TRU wast-. Process materiala containing plutonium as anoxide, a ●alt, or metal; and .mbedd@d in the wide variety ofmatrices typical of a scrap recovery facility, are measured withthese instrumantu. That im, rags, tools, crucible parta, plastics,rubber and lcadod gloves, furnace parts, non-plutonium matal, etcare ~11 analyzad with those instruments. For the program, the

5

materials are segregated and instrument method is selected in themanner indicated above. Plutonium content varies according to

chemical and matrix form but ranges from 1 gram to several hundredgrams for oxides and salts and less than 200 g for metals. The SGSand NCC assays for both segments of the program are singlemeasurements of 15 minutes, or less, duration. During the course ofthe program reported here, all instruments were under themeasurement control procedures cited above. For that segment inwhich the game item is assayed on the same instrument, measurementintervals ranged from one week to as much as seven years. Or?average however, the two measurements were separated by about 18months. Providing the instrument used for the assays has beenoperating in a stable fashion c -r the time period separating the

measurements, as guaranteed by tne measurement control procedures,this comparison gives an estimate of the long tem reproducibilityof the measurement for the NDA technique and type of processm~terial under review. For that megment of the KVP which comparesNCC or SGS assays to reference CI ar~alysesto evaluate ‘measurementbias, the two measurements are generally separated by about 3.5years, although some differ by as much as ten years. Previousstudies have indicated that the reference measurements are accurateto within t 1 % of the nominal plutonium value in the sample.

Table I summarizes the raw results from analysis of these twosegments of the KVP data. Ratio Rl in the table is the average ofthe mass weighted ratios of a single SGS or NCC measurement to asecond measurement made on the same instrument at : later date. Theratio R2 in the average of the mass weighted ratios of a single SGSor NCC measurement to that of a reference Cl measurement made at alater time. Results from 3 SGSS and 6 Nt7Cswere folded together toobtain the reported ratios. The~e in~cruments have been used tomeasure over 95 # of the waste generated here. The uncertaintiesassociated with the ratios are the standard deviation (lo) of thecombined data. Plutonium content for the SGS meaeuraments rangedfrom 1 g to 350 g, whereas for NCC measurements, tha range was 1 gto 370 g.

The table indicates that RI is 1.04 for SGSS; that is, SCSremeasurement averaged 4 % lower than the initial measurement.This ratio is consifitent with settling of heavier plutoniumparticles inside the container over time. Settling has beenobserve< to occur for some of the salts measured by this technique.When the denser plutonium pa~-ticl~tifall to the bottom of thecontainer between the first and second measuramentn, the rasultingeelf attenuation and and effect counting lousas tand to bias the

6

later assay low. The ratio R2 for SGSS is 0.94, indicating thatthese measurements are lower by 6 t, on average, than the referencemeasurements. Because most sources of SGS error, such as endeffects and self attenuation, tend to bias these assays low, thisresult is not surprising. Note that the standard deviationsattached to both SGS ratios are greater than 20 % of the ratiovalues, indicative of the large variability in measurements ofprocess residues and waste.

For NCCs, RI equals 0.99, indicating that remeasurements usingthese instruments averaged 1 t higher than the original values.Here, the magnitude of the remeasurement difference is about onefourth the magnitude of its SGS counterpart. For neutron counters,in contrast to SGSS, a high remeasurement value may result fromsettling because it could lead to closer packing of the plutoniumwithin the containers. This would tend to increase neutronmultiplication and bias the second assay high with respect to thefirst. Although it is known that rnettlingdoes take place withinsome of these containers, this interpretation is speculative sincethere is no complementary evidence, at present, to suggest that theplutonium particles are more closely coupled in these containers.The ratio R2 for NCCS is 1.03, indicating an average high bias of3 % in these measurements. Again, this result Is not surprisingsince multiplication and (a,n)●ffects tend to overestimate neutroncounter measurements. While the standard deviations for both thereproducibility and bias data are again large, they aresignificantly smaller than their SGS analogues.

Two additional points should be mentioned h regards to theaverages proeanted In Table 1. First, each instrument of the sametype shows similar bias and reproducibility tendencies. That is,for all the SGSS used in the study, the later measurements areconsistently smaller than the earlier maasuremonts. Thus, inaddl.tionto the overall RI average being greater than unity, thesame is true for ●ach of the individual SGSS. Also, all the SGSSused in the study are biased low compared to the reference CImeasurements. Tha same is true for NCC measurements. Bias andreproducibility tendencies for the individual instruments all trendin tho aama direction as the overall avarages. The ratios,therefore, reflect consistent instrument susceptibilities. Thesecond point is that, although three different gamma-ray imotopicscodes were used to establish reference values in this study,additional data was also taken to assure that ●ach code gaveessentially the oame isotopic percentages for the materialsreported here. Thus, the biases ara due entirely to inadequacies in

7

the SGS or NCC measurements, not to error in the CI analyses.The data summarized in Table I incorporate all the

measurements performed over the duration of the WP program, todate. However, during the course of the program, certain biastrends were noted and corrections were applied to future assays ofthose material types. In some cases, the ?fVPdata indicated thatcertain process residues could not be reli~’>lyassayed by SGSS, sothe efficacy of NCC measurements was investigated. In other cases,bias correction factors had to be applied to NCC assays to obtainunbiased results. Yet again, the use of an instrument to measurecertain matrix forms had to be limited to no more than 50 g ofplutonium in order to assure credible assays. Therefore, some ofthe raw data summarized above represent measurements which weresubsequently determined to be unreliable and the averages in TableI include results which were later ‘discarded for improvedmeasurement methodologies. For this reason, the summarized datawere retabulated, but with those measurements kncwn to be biasedomitted. The revised results are shown in Table II. These resultsinclude the measurements using the improved methodologies.

The revised table indicates that the ratio RI for SGSS has nowbeen reduced to 1.91; that is, later SGS measurements are lowerthan the original assays by 1 %, on average. This is approximatelyone fourth the difference seen in Table I. The revised value mayintimate that some settling is still occurring In a fraction of thematerials receiving SGS measurements, but that those with thegreatest proclivity for settling are no longer receiving thismeasurement. ?!oreovor,because the average difference Is only 1 *,whereas the dispersion in RI remains high at 21 %, then anyinterpretations on the causes for the disparity are highlyspeculative. Substantial Improvement is also seen in R2, theaverage ratio of the SGS measurements to ~“eferanceCI values. Anaverage measurement is now biased low by about 2 % which is lessthan half of the bias seen in Table I. The variability in this’rat~n has also improved markedly.

For NCCS, the revised reproducibility ratio RI indicates adifference of 1 % between the two measurements, about the same asin Table I. Apparently, those materials which were omitted from theoriginal data because of unsuitable chemistries or matrices werenot the cause of the differences in the two NCC analyses. Again, itshould be noted in this regard, however, that the 1 t difference issmall compared to the variability (14 t) in this data. The biasratio R2 for NCCS, on the other hand, does show substantialimprovement. An average measurement is now biaiaed 1 3, down

8

substantially from the 3 % average seen in Table I. There has beenlittle change in the standard deviation for this ratio over theprevious data.

Several conclusions can be drawn from a comparison of the twotables. First, it is apparent that the MVP program has beensuccessful in determining unreliable instrument performance. Fromthe initial data in Table I, five different material types weredissociated from their original measurement method and reassayedwith another technique. The resulting improvements inreproducibility and bias seen in Table II validate theirdissociation and markedly improve overall measurement credibility.Because those materials which most tended to shift the bias andreproducibility ratios away from unity have now been identified andalternative techniques have been selected, future improvements willbe more difficult. This is compounded” by the large standarddeviations found in the ratios which will make identification ofsmall bia~es within this dispersion increasingly troublesome. Thelarge variability seen in the standard deviations is not isolatedto just a few material types, but instead is spread across theentire spectrum of materials that established the revised values inTable II. It is also true that for some materials, no technique isentirely appropriate. That is, there is a high probability for asmall bias no matter which NDA technique is used. Certain saltsgenerated at this facility fit this category. The salts containproduct slag mixed with magnesium sand and crucible remantsresulting from PuF4 reduction. Residue and waste from this processcontain CaF2 salt, unknown amounts of PuF4, and plutonium metalshot. This matrix presents difficulties for NCC analysis because oflarge (a,n) effects resulting from the presence of fluorine atoms.Likewise, SGS assays are hindered by attenuation of the plutoniumgamma rays in the shot. Identification and treatment of thesemeasurement kinds of measurement problems must be done on anindividual basis. A second conclusion to be drawn from thjs”program is that the averages and standard deviations determinedhere provide a better assessment of measurement uncertainty forprocess residues and waste than most established estimates.Generally, precision and bias estimates are made on the basis ofrepeated measurements of standards (reference materials) or arebased on an educated guess. Reference materials are usuallydesigned to uuit the NDA techr,ique to which they are applied. Thatis, the chemical and matrix form of the standard, the stability ofthe plutonium within the container, and its homogeneity are welladapted to the ❑easurement technique which uses them. Precision and

9

bias values for these refarence materials provide useful estimatesof optimal instrument performance under the facility’s operatingconditions, but do not reflect the measurement uncertainties thatshould be applied to process material assays. The latter depend onthe matrix and chemistry of the material, packaging, location ofthe plutonium within the sample, the presence of radioactive

impurities 6uch as 241m, calibration variability, etc. That is,they depend on a wide asso~tment of properties whose effects on themeasurements are not clarified by precision and bias valuesassociated with analysis of reference materials. Often these

properties are beyond the observation and control of the NDAinstrument operator, such as the presence of interferants or thehomogeneity of the plutonium particles within a container. In thesecases, estimates of measurement uncertainty based on referencematerials measurements are particularly unsuitable ~nd NDA

instrument operators must resort to opinions based on theirexperience or process knowledge, that is~ an educated guess. Theselatter estimates are sometimes useful, but all too frequently, arelater proven to be false. Moreover, they require verification to beof value for safeguards purposes. Generally, this is difficult toprovide.

With these difficulties in estimating uncertainties in mind,Table III was composed. This table lists several differentassessments of bias and reproducibility that are either calculatedat the Los Alamos plutonium scrap recovery facility or are ingeneral use among U.S. TRU waste measurement agencies3. The firstrow of entries under Reproducibility in the table are the percentstandard deviations (% SD) in 15 sequentialplutonium standard containing 56 g of 239Pu.precision values taken randomly from ameasurements on instruments used to assay TRU

measurements of aThese are typicalseries of weeklywaste here and are

often used as indicators of measurement precision among safeguardspersonnel at this facility. The second entries under the same”heading are the average t SDS of the daily standards measurementstaken aver the five year span of the MVP program. These numbersrepresent one estimate of the long term precision in aninstrumentis measurements. Conventional Reference Values (CRV), thethird listings under this heading and the second row of entriesunder the Bias heading are taken from Reference 3. Thereproducibility value is defined as the standard deviation in aseries of repeated measurements; i.e., the same estimate that wasused to establish the second row of entries under thisiheading. Thespread in values for this CRV is determined from statistical

counting errors resulting primarily from counting time limitationsand plutonium loading. Under the Bias heading, the ConventionalReference Value is defined as the closen,=ss of a measured value toits lxue value and is estjmated by the diffc:rence between ameasured average and its accepted reference valuf~.Factors such asmatrix type, chemical form, homogeneity, packaging, etc determinethe spread in these values. The CRV entries presently serve as

precision and bias guidelines for measurements of TRU wastedestined for the WIPP radioactive waste disposal rritein the U.S.The fourth Reproducibility entries in Table III are the % SDassociated with the R1 ratios in Table 11. Under Bias in Table III,the average percent relative differences (% F~) resulting from thedaily measurements of a 56 g plutonium refer~encestandard over thefive year period are shown in the first row. Fina21y, the third

~tries under the Bias heading list the % J?D values excerpted fromthe R2 ratios in Table II.

The first observation that can be drawn from Table III is thatthe criterion presently used here to determine SGS and NCCmeasurement reproducibility has limited practical value. Thereproducibility 8 SDS from 15 successive measurements of a standardshow only how stably the instrument performed over a short timespan. They provide little insight into measurement reproducibilitytaken over an extended period of time or for process materialmeasurements. The second row of entries under Reproducibility are,:hrough comparison with the first row, useful for pointing outdifferences between the short term and long term dispersions in areference standard’s measurements. The two SGS values are seen tohave a particularly large difference, indicative primarily of theirsensitivity to random elec~ronic and microphonics noise in theenvironment. It is also useful to contrast the long termreproducabilities with the Conventional Reference Values. Bfjcauseof the measurement times and plutonium content used to detelminethe second row of entries in Table III, the minimum CRV valueE;should correspond closely to those values. At this facility, theiong term reproducibility of the SGS and NCC measurements are bothslightly below the CRVS, which demonstrates that these instrumentscompare favorably with other instruments using the same technique.The final entries under Reproducibility display the valuesdetermined from residue and waste measurements at this facility.The very large differences between these values and the other NCCand SGS reproducibility values in the table reveal that the otherestimates are clearly inadequate for determining uncertainties inresidues and wa6te. These last reproducibility values are about 5

11

to 10 times larger than the other estimates. The large differencesare important for NDA Laboratory operators and safeguards personnelwho are required to assess the degree to which the measurement ofwaste and residues can be repeated. The large uncertainties shownin the fourth row of the table more realistically reflect thisrepeatability whereas the other estimates are more closely relatedto optimal instrument performance.

Under the Bias heading In Table III, the average SGS and NCCbiases in measuring reference standards (the first row of entries)are seen to have small positive values. These values serve asuseful indicators that the instruments at this facility are beingoperated under a reasonably effective measurement control program,but should not be interpreted as ●stimates of the bias to beexpected in waste and residue measurements. The C!RVvalues in thesecond row under this heading are larger than the values in thefirst row, by comparison. Again, this indicates that theinstruments used here ccmpare favorably with othf!rNCCS and SGSS,although the comparison is somewhat strained since the values inthe first raw were determined from repeated measurements while theCRV definition is an estimate of sxpected bias in a singlemeasurement. In any case, a better estimate of bias in waste andresidue assays can be obtained from the third row of entries inthis category. These indicate that, at thin facility, SGS aasaysare biased low by about 2 t whereas the NCC assays have been biased1 * high. These values provide the most usefu.iestimate of bias inwaste measurements here because they take into account matrix andchemical variations, inhomogeneities, packaging differences, thepresence of impurities, ●tc. In addition, they can be umed tocalculate the ●xpected plutonium inventory difference due toInaccuracies In TRU waste maaaurements. Since the corrections inassay methodology indicated above have been implemented, theplutonium content in the waste ham been underestimated by less than15 g/year, on average.

Comparable bias and reproducibility ●stimatam for low levelwaote measurements have not yet been obtained. These ●stimateswill be diffjcult to acquira bacauso of the labor involved indetermining roferonce values for this waste form. Only chemicalanalysas can provide the sensitivity to achieve useful referencevalues for tha small quantities of plutonium found in this waste,and the ●xpenso and r-source allocation required to prepare thenumber of chemical samples that mimic tho range of LLW matrixcategories is prohibitive.

12

From an operator’s perspective, no report is complete withouta list of what he believes to be the most pressing NDA needs.Although these types of lists often reflect a local perspective,that is, the needs at the operator~s own facility, I have attemptedto incorporate impressions gathered from visiting other U.S. wastemeasurement facilities. In general, the needs do not express adesire for radical improvements in NDA hardware design. Instead,they concentrate on methods for improving measurement qualitythrough upgrades to existing designs. The needs are driven by NDAworkers’ desires for dependable instrument operation, requirementsfor improved and verifiable non destructive measurement anduncertainty analysis, and in response to anticipated demands forincreased automation. A brief description of these needs is givenbelaw:

- NDA instrument electronics should be more robust to withstandtho difficult work ●nvironments ●ncountered at many processingfacilities. Frequently, these ●nvlronments are dirty, noisy(both RF and microphonics noisa may be present), and connoctedto fluctuating ACpower supplies. fn addition, some instrumentdesigns incorporate noisy cabling and stepping motorarrangements. Those degrade measurement quality and lead toincreased instrument failure rates. The effects areparticularly acuto for gamma-ray instruments. At thisfacility, SGS failure rates have bean as high as 3 -/month forcome instruments. NDA instrument developers should considermore robust desigl~ or improvod methods for isolatingdetectors, amplifiers, cabling, and computars from noise andpower fluctuation-.

- Assays of ovmrsize TRU and LLW waste are often highlyproblematic at wasta measurement facilities. Typically this’wasto is dense, consisting of metals or compactod materials,so neutron counting methods offer the best prospect. Thesemethods ara susceptible to measurement ●rror resulting fromfluctuating backgrounds, unknown spatial location of neutronsources within the waste, and matrix ●ffects. In most wastemaacurement facilities w~thin the U.S. weapons complex,inatrumants to assay ovorsize wasto are ●ithar ●ntirelylacking or perform only marginally. Tha resultlngunreliability In the ●stimates of plutonium content isbtginning to come under incroaning regulatory scrutiny.

13

Instrument developers should consider increased developmentalefforts in this area.

- Several improvements in SGS analysis are needed. Peak fittingand peak stripping routines should become a regular part ofthe SGS assay. In the U.S., several attempts have been made toincorporate these capabilities into SGS software, however fewunits have been fully tested and Intogratod into a productionenvironment. This capability is of major importance for olderwaste forms that are inadequately documented or whose originsare poorly understood. For this waste, operators often haveonly a limited understanding of the nature of the spectralinterferences they may encounter. Peak fitting and strippingalgorithms can mitigata ●rror amaoclated with thesemeasurements and enhance the operator’s ability to performreliable assays of these waste forms. In addition, continuedrasearch Into corrections for ●elf-attenuation and end effectsmust be sustained. Because waatc in oftan het~rogenoous andsubject to settling inside containers, improved measurementalgorithms which incorporate these r~rrections will proveespecially usaful.

- Standards, otandards, standards. All U.S. waata measurementfacility operators complain of the inadequacy of their ownstandard raferenco materials and of thair Inability to procuronaw ones. Obstaclas to procurement includo lack of fabricationfacilities for new etandards, ●xpanne, location of purestandards sourca matarialm, shipping problams, and lack of on-site storage capability. Bocaus@ auditors and regulatorsincroaringly raquiro verification of NDA maasuramontm throughcomparisons to standards maaauramants, this naed will continuoto ●acalate in tho future.

- Safeguard and wllat9 dimpoual regulators aro becomLngincreasingly concarnod with NDA measuromont uncertainty.Presantly, most instruments includo statistical countingvariations am tha only component in tho moasurod uncertainty.Howover, thin in usually tho minimum unc~rtainty in thomeasurement and does not reflect variability in calibration,background, containar ●ffmctm, the influence of ●patialvariation on datoctor ●fficiancy, matrix ●ffoct~, ●tc .Al~orithms should bc dcvalopad to include all significantrandom and bias ●ourcas a- part of tho t~tal uncertaintyroportad with an NDA moasuromont,

- Hoiqhtoned attention to ●xpert ●valuation of now ●oftware andhardware dmmigna im n.adad. Our facility ham ●xparionc~d

14

numerous ●rrors in measurement algorithms, hidden softwareglitches, and poor or noisy hardware design in recent

instrument purchases. In addition, 8ome commercial vendorstend to over~ell the measurement capabilities of their

instruments. Discussions with oporators at othar U.S.facilities have reinforced these impressions. To corroct thiscondition, a center where now instrument designs can beindependently ●valuated is required. The center would testuoftwarm and hardwara components under common operatingconditions and appraise measurement algorithms for a widevariaty of materials, including waste. Thair assoasmont of aninstrument’s perfonance will laad t!> improved confidenceamong operators that their measurement needs ara mat, alertdesignars and manufacturara of flaws and limitations in theirproaucts, and assure auditors and regulators of tha ~arlts ofmeasuromenta perfozmed with thama inotrumentm.

- Several davolopmantc aro underway in U.S. wasta managamontfacilities which may affect NDA instrumentation design andanalyuiu. Them Includo automatod waste handling, wastestabilization, cnd roquirom~nta for maasuramants of non-nuclear proportion of waste. Driving tha davolopmant forautomatod wasta handling is a doairo to roduco tho riuka fromradiation ●posuro and accidants that come about from humancontact. In tho futura, automctod equipment will b. us-d withgraat~r froquoncy to movo, load, moasuro, and dieposo ofwamtm. NDA inatrumont dovolopors should be ●waro of the-edev~lopments and alert to their implications. Naw instru.montsmay hava to b~ dasignad that aro compatible with a~tomatodconveyor ●ymtama, can be loadod with robot-, and intarfaco●asily with computors that control othar maasur-mont andloading quipmont. Inetrumant dwmlopars may b. asked toaddress ~uch issues as standardization of softwaro protocolr,roliabi*ity of non NDA .quipmant in automat-d systems, andintegrated ●ystam dauign. Stabilization of a qrsatar varietyand numbar of TRU wast~ form is baing motivatad by raqulatoryconcarns and delays in the opaning of a pannanont disposalrepository in tih, U*S. Ccmontation, bitumanization,polymerization, and vitrification ara all procas-an which havo●ithar bean •~~ccassfully implcmant~d or show promino for wastaimmobilization. NDA maasuramant ●quipmont may ba uaaful foranaly-is of stabilized waato, howavor ●xiating rasserch onamsay r-liability is spars., Incroasod attantion shauld b.givan to tho ●ff.ct of diffarant ●tabilization ●chemmm on

15

maasuremellt accuracy and to development of improvedinstn.umntation for its analysis. In addition to NDAmoaauraments, most TRW waste in the U.S. will also besubjoctod to radiography analysis, alpha particle monitoring,waight maaaurmmento, ultrasound tasting (for drum integrity),and head-pace analysis by gas chromatography (for VOCS andhydrogen). Integration of these five tasks into a ●inglemcasur~ment wtation would reduce moving and loadingrequirements and could mesh well with improved automat-d wastehandling systems discussed above.

Roportad above are a review of wasta maasuroment operations ata U.S. plutonium scrap racovory facility, uncertainty estimates forthose moasuromonts, and ●oms rocommondationa for future NDAdevolopmant ●fforts. In tha first ●action, waato packaging andhandling activitiam are briefly dimcumrnod, than instrumentsolaction and quality assuranca procaduros arc raviawed. The nextsection conaidera the raaults of a program to ●stim~toreproducibility and bias in maamuromants of rasidues and wa~tehers. SGS moasurananta of thosa matarials wcra found to b.r-producible to within 21 t and to b. ●ubjact a biar that avaragod-2 4 ovar tho courso of tha program. Neutron countor maasuramontswara rsproduciblo to within 14 t and had a +1 ~ bias wh~n comparedto r-faranca values. Raamonm for tha difformncos in assay ra~ultsand varioua Qstimatas of maasuram~nt uncertainty wora almodiacussod. Finally, this report closad with an identification ofNDA nooda that, from an operator’s parap.ctiv~, roquiro futurodovaloprnent ●ffortm. The noods prosont ●oftwar., hardware,Inntrumont ●valuation, and ntandarda rqulramantn for improvod●afoguarda, quality amauranca, and diapoaal of radioactive waste.

16

5. References

1. D.C. Christensen, B.F. Bowersox, B.J. HcXerley, and R.L. Nance,“Wastes from Plutonium Conversion and Scrap Recovery Operations”,Los Alamos National Laboratory Report, Mar 1988.

2. D.C. Christensen and L J. Xuliins, “Pros@nt Status of Plutonium?latalProduction and Purification at Los Alamos-1982”, Los AlamosNational Laboratory Report, Jun 1983.

3. F.J. Schultz and J*T. Caldwell, ‘DOE Assay Methods Used forCertification of Contact-Handled Transuranic Waste”, Oak RidgeNational Laboratory Report for the U.S. Department of Energy, 1989.

c:\wP51\sPAIN

17

Table L Results from analysis of all SGS and NCC MVP data are summarizedbelow. RI is the ratio of a single SGS or NCC measurement to a secmdmeasurement made on the same instrument at a later time. F?z is the ratioof the SGS or NCC measurement to the Cl result. The numbers inparentheses are the number of measurements used to determine each of theratios.

RI R2

SGS 1.04 +/- 0.21 (333) 0.94 +1- 0.20 (75)

NCC 0.99 +/- 0.14 (277) 1.03 +/- 0.12 (93)

Table Il. The revised MVP :esults are summarized below. These resultsexclude all assays known to be biased from the original data.

R, ‘2

SGS

Ncc

1.01 +/- 0.21 (248) 0.98 +/- 0.14 (56)

0.99 +1- 0.14 (249) 1.01 +/- 0.12 (45)

Table Ill. Variou~ e8tirnates of reproducibility and bias in NDA analvses. % SD isthe percent standard deviation of the measured data whereas % R6 is thepercent relative difference between the measured and reference values.

REPRODUCIBILITY

% s!) of 15Sequential Measurements

% SD of DailyMeasurements

Conventional ReferenceValues

% SD trom Table II

BIAS

Average % RD ofDaily Measurements

Conventional ReferenceWdues

Average % RD fromTable II

SGS Ncc

I 1.0 %I 2.6 %

I

I 2.1 % 1 2.8 % II 3%-loo% I 3%-50% I

I + 0.4 % + 0.4 %

I +/- 0.5 -10 % II

‘/- 0.5- 10%

-2% +1%