Clint Miller
Pacific Gas and Electric
ABSTRACT
At Waste Management '96 we presented a paper on the derivation of constant scaling factors at Diablo Canyon Power Plant.(1) These scaling factors were base on nine years of operating data from each of the two units at the site. In 1994, a fuel defect in Unit 2 was identified. This was only the second cycle in which either unit had operated with a fuel defect.
This paper will discuss the affect on our scaling factors due to the fuel defect. Samples of cleanup resin and cartridge filters were obtained from Unit 2 to determine if the scaling factors had to be altered. The results, as expected, indicated that the constant scaling factors previously derived were still valid.
The use of the Sample Analysis Program software made this evaluation quick and simple.
In addition, operation with a fuel defect offered an opportunity to obtain actual conservative analytical results for Tc 99 and I 129. Historically, analysis by commercial laboratories had never identified Tc 99 or I 129 in any sample from Diablo Canyon. The use of LLDs values for these nuclides in waste characterization imposed gross conservatism which can adversely affect disposal site performance evaluations.
Samples of resin and cartridge filters from Unit 2 were sent to Battelle PNNL for Tc 99 and I 129 analysis using mass absorption spectrometry (MAS). These results have enable us to derive constant scaling factors for Tc 99 and I 129 which are orders of magnitude below such factors previously derived from LLD values. The sampling and derivation of these scaling factors was less expensive than the cost of commercial computer codes which derive Tc 99 and I 129 from reactor coolant samples.
INTRODUCTION
For radioactive waste classification purposes, a waste stream is any waste product or collection of waste products where the concentration of non-gamma emitting nuclides can be determined by the same set of scaling factors. Constant scaling factors for corrosion products have been derived at Diablo Canyon Power Plant. The constant corrosion product scaling factors (e.g., Fe 55/Co 60, Ni 63/Co 60, Ni 59/Co 60) are waste stream specific at Diablo Canyon. There are two spent resin waste streams, two spent filter waste streams and one dry active waste stream at Diablo Canyon. Once the quantity of Co 60 in any waste stream is determined, the quantity of each non gamma emitting corrosion product can be calculated using the appropriate scaling factor.
In addition, transuranic (TRU) nuclides have each been found to have a constant scaling factor in relation to Pu 239. This relationship is independent of waste stream. If the quantity of Pu 239 is determined, all other transuranic nuclides can be calculated using the appropriate scaling factor.
In practice, annual waste samples from Diablo Canyon need only be analyzed by off-site laboratories for Sr, C 14, Pu 239, Tc 99 and I 129 in addition to a gamma spectrum. Tritium analysis is performed on-site. Use of the constant scaling factors enable all the other nuclides for waste classification at Diablo Canyon to be calculated.
DISCUSSION
The radioactive waste classification program at Diablo Canyon requires new off-site analysis of radwaste when ever conditions exist which could substantially alter waste classification assumptions. One such condition is operation with a fuel defect. The radwaste nuclide spectrum includes activated corrosion products (e.g., Co 60, Fe 55, Ni 63) fission products (e.g., Cs 137, Sr 90, TRU, Tc 99, I 129) and activation products (e.g., C 14, H 3). Operation with a fuel defect is likely to increase only the concentration of fission products in radwaste and thereby, possibly alter the waste classification. Industry experience has been that small fuel defects do not affect the generation of TRU which is dominated by tramp uranium from the exterior of new fuel.
Spent resin at Diablo Canyon is sampled and analyzed on a batch basis. The affect, if any, on waste classification from a fuel defect will be evident from the routine resin analysis.
For cartridge filters and dry active waste (DAW), an annual off-site analysis is obtained and dose-to-curie calculations are performed on each subsequent filter or DAW container to enable classification. The affect of a fuel defect, if any, cannot be ascertained from a single annual sample. In order to assess the affect of the Unit 2 fuel defect, several filter samples associated with the Unit 2 fuel defect cycle were obtained for full spectrum off-site analysis.
The operation with a fuel defect offered an opportunity to obtain actual analytical results for Tc 99 and I 129. Historically, analysis by commercial laboratories had never identified Tc 99 or I 129 in any sample from Diablo Canyon. The use of LLD values in lue of actual results for these two nuclides impose gross conservatism for waste classification. This conservatism can adversely affect disposal site performance evaluations.
Samples from the Unit 2 filters discussed above and primary resin samples from both units were sent to Battelle PNNL. At Battelle these samples were analyzed for Co 60 and Cs 137 by gamma spectrometry and Tc 99 and I 129 using mass absorption spectrometry (MAS).
Three Unit 2 filter samples were obtained during the fuel defect cycle. A letdown filter sample was obtained early in the cycle (about one month after startup) upon learning that a fuel defect existed. A full spectrum off-site analysis of this filter was performed. Based upon the corrosion product, Sr, TRU and C14 results from this filter no waste stream affects were noted. During the seventh refueling of Unit 2 a letdown filter and a refueling water purification filter sample were obtained. These samples were split. One sample portion was sent to the commercial off-site lab for analysis of all nuclides except Tc 99 and I 129. The other portion of the samples and a sample from the letdown filter removed early in the fuel cycle were sent to Battelle PNNL for MAS analysis of Tc 99 and I 129.
The analytical results from these three filters were loaded into the Sample Analysis Program, version 1.4.(2) The program produces plots of scaling factors for the waste analysis data selected. The waste analysis data selected is specified by the user but, usually will be a waste stream, all waste streams or all waste from a unit. The program calculates a median scaling factor and plots the factor-of-ten limits above and below the median scaling factor line for visual comparison of the waste data selected. In addition, the program calculates the dispersion of the data to quantify the variation. In this program the dispersion is defined as the antilog of the standard deviation calculated from the log ratios.
Plots of the various scaling factors were generated using the Sample Analysis Program to compare these three "fuel defect" filters versus historical filters from Diablo Canyon. The evaluation examined the scaling factor trend plots for corrosion products, activation products, fission products and TRU.
As expected the fuel defect did increase the cesium concentration in reactor coolant. Cesium is soluble, however, and not removed in large quantities by cartridge filters. The results from the "fuel defect" filters did not show an increased cesium concentration. Therefore, this major fission product was not and need not be trended. Cesium is not waste classification controlling for filters at Diablo Canyon.
Of the non-gamma emitting corrosion products, Ni 63 is waste classification controlling. Fe 55 and Ni 59 have no impact on waste classification. The trend plot of the Ni63/Co60 scaling factor for filters showed that the three fuel defect filter results were well within the allotted factor of 10. Therefore, the fuel defect did not alter the filter waste stream due to corrosion products. Furthermore, the constant corrosion product scaling factors derived previously were still valid.
Between H3 and C14, C14 is waste classification controlling. No constant C14/Co60 scaling factor has been derived due to the dispersion in the historical data. The trend plot of C14/Co60 scaling factor for all filters, however, showed the three fuel defect filter results were well within the allotted factor of 10. The fuel defect did not, therefore, alter the concentration of C14. Thus, the annual C14/Co60 filter scaling factor was still valid.
The concentration of all transuranics is proportional to Pu239. Constant TRU/Pu239 scaling factors have been derived for Diablo Canyon based upon historical data. No constant Pu239/Co60 scaling factor has been derived due to the dispersion in the historical data. A trend plot of the Pu239/Co60 scaling factor for all filters showed that the three fuel defect filter results were well within the allotted factor of 10. The Diablo Canyon fuel defect did not alter the cartridge filter waste stream due to TRU. Thus, the constant TRU/Pu239 scaling factors derived previously and the annual Pu239/Co60 filter scaling factor were still valid.
Although, there was only one defect in the fuel cycle at Diablo Canyon it occurred in a high power pin. This single defect was equal to four to seven low power pin defects. The TRU results confirm that small fuel defects are insignificant contributors of TRU material versus tramp uranium.
Strontium 90 is included in Table II of 10 CFR 61 for waste classification. Strontium is a fission product but, for filters it cannot be scaled to Cs137 because filters do not contain sufficient quantities of cesium. Strontium 90 is not waste classification controlling for filters at Diablo Canyon. Carbon 14 is the waste classification controlling nuclide for filters at Diablo Canyon. Strontium ranks far down the list for classification controlling nuclides. Both Ni63 and Cs137, if present, are more important than Sr90 for waste classification control. Increasing the historical Sr90 concentration by a factor of 10 has no affect on the classification of a filter at Diablo Canyon.
No constant Sr90/Co60 scaling factor has been derived due to the dispersion in the historical data. The trend plot of Sr90/Co60 scaling factor for filters shows two of the fuel defect filter results at or above the factor of 10 line. When the fuel defect filters are plotted separately, the median scaling factor result is within a factor of 10 of historical filter data, though just barely. The filter from early in the Unit 2 fuel defect cycle is not within a factor of 10 of the two "outage" filters and serves to lower the "fuel defect" median scaling factor.
The fuel defect did increase the concentration of Sr90 in filters, perhaps by as much as a factor of 10. However, since even a factor of 10 increase in Sr90 does not affect filter waste classification at Diablo Canyon, no new waste stream was created by the fuel defect. The annual Sr90/Co60 scaling factor was still used as a valid figure.
For filters at Diablo Canyon Tc99 and I 129 are not waste classification controlling. The concentration of these nuclides in power plant radwaste is extremely low. These nuclides are generated from the fissioning of tramp fuel during a normal fuel cycle. Activation of Mo99 corrosion products is also a source of Tc99. A fuel failure will increase the concentration of these nuclides in radwaste versus a normal fuel cycle. The fuel defect results from Battelle were orders of magnitude lower than LLDs reported by commercial labs on past Diablo Canyon waste. Therefore, the fuel defect did not alter the cartridge filter waste stream due to Tc99 and I 129.
The results from Battelle were loaded into the Sample Analysis Program. Once again, due to the lack of Cs137 in all of the fuel defect filters, Tc99 and I 129 had to be scaled to Co60. The median scaling factor derived for Tc99/Co60 was 8.36E-7 and I 129/Co60 was 1.96E-10. Although the dispersion of these scaling factor results were relatively large (eg, 7.04 and 7.65) they were below a factor of 10. Since these scaling factors were derived from fuel defect data, they are conservative. These conservative constant scaling factors will be used in the future for all filters at Diablo Canyon. These conservative scaling factors will also be used for DAW since that waste stream is so similar to the filter waste stream at Diablo Canyon.
In the fall of 1995, Diablo Canyon Unit 1 had a refueling outage. During this outage samples of resin from a letdown bed and shutdown cleanup bed were obtained. The letdown bed was in service for two 18 month fuel cycles with no fuel defect. During the Unit 2 outage a sample of resin from the letdown bed was obtained. This bed was also in service for two 18 month fuel cycles. The second cycle this bed was in service had the fuel defect.
These three resin samples were split like the Unit 2 filters. A portion of each sample was sent to Battelle for Tc99 and I 129 analysis using MAS. These three results were loaded in to the Sample Analysis Program. Plots of Tc99 and I 129 were made versus Co60 and Cs137. The correlation of Tc99/Co60 was extremely good with a dispersion of only 2.06. The correlation of I 129/Cs137 was extremely good with a dispersion of only 2.63. These low dispersion figures prove that the fuel defect had no affect on these scaling factors. The scaling factor for Tc99/Co60 was 1.68E-8 and I 129/Cs137 was 9.02E-8. These conservative constant scaling factors will be used in the future for all resin at Diablo Canyon.
An evaluation of the Unit 2 letdown resin scaling factors was also performed. Similar to the results for filters corrosion products, C14 and TRU concentrations were not affected by the fuel defect. The concentrations of strontium and cesium
did increase significantly.
CONCLUSION
An evaluation of the affect a fuel defect had on scaling factors was performed. Physical samples of filters and resin were obtained specifically to perform this evaluation. The Sample Analysis Program was used to trend and compare the fuel defect data with historical data. Use of this program made the comparison quick and easy.
The fuel defect had no affect on the corrosion products, C14 or TRU. Strontium concentrations did increase significantly as did cesium. These increases had no affect on the waste classification of filters at Diablo Canyon. In addition, actual results for T99 and I 129 were obtained for the first time. This resulted in the derivation of constant scaling factors for these nuclides. Even though these scaling factors are conservative they are orders of magnitude lower than those previously derived based upon LLD values. The use of such scaling factors will reduce the gross conservatism and adverse impact on disposal site performance assessments.
In order to derive the conservative constant scaling factors for Tc 99 and I 129, only six samples were required. The cost of the Battelle MAS analysis for these six samples was $20,901. This is less than the cost of available commercial computer software to derive T99 and I 129 concentrations from reactor coolant.
REFERENCES