HNF-3233

Analysis of SX Farm Leak Histories --Historical Leak Model (HLM) August 1998

I. Background:

Single-Shell Tanks (SST’s) that have leaked various amounts of high level waste into the soil column have been a fairly common past occurrence at the Hanford Site. All told, some 67 of the 149 SST’s have reportedly leaked (45%). These waste tanks at Hanford had been used for a variety of waste storage missions and of all of these missions, none were more stressful to the tanks than that of containing boiling or aging wastes. The combination of high temperatures, sudden steam "bumps", and caustic high nitrate, nitrite, and aluminate wastes all contributed to the degradation of these tanks.

The set of 25 tanks that contained aging waste (15 SX Farm, 6 A Farm and 4 AX Farm tanks) comprise a particular problem with respect to estimating leaks for two reasons. First, these tanks were cooled by evaporation of water as the waste boiled. Upwards of 200 kgal/mo of condensate was evaporated, cooled, and replaced for some of these tanks during their peak operations. Along with such large turnovers in volume, differentiating between a normal loss of water (i.e. makeup) in the process from that volume waste lost as leaks was extremely difficult. Of this set of 25 boiling waste tanks, 16 (64%) are reported leakers. Of the nine tanks in SX Farm that were heavily used for boiling waste, all have leaked.

As these wastes concentrated, precipitates of sodium nitrate, nitrite, and aluminate would produce cement-like scale on the cooler surfaces within each tank. Solid monoliths of salt have been reported in many tanks with concentrated waste. As a result, this scale would often seal leaks as the tank cooled and therefore further complicate any effective leak detection by material accountability, especially as the tank cooled.

As a result of these and other difficulties, all of the tank leaks in SX Farm were first detected in the soil and not from material accountability. That is, these leaks were only indicated after the fact as radioactive contamination appeared in vertical or lateral drywells around and under each tank. Thus, the period of time for peak heat load for each tank was the time it was most likely to leak and was exactly the most difficult period to deduce a leak by material lost. The HLM is an attempt to better define leak estimates for four SX Farm tanks by reconciling their fill histories with the volume evaporated by their heat loads.

II. Approach:

The Historical Leak Model (HLM) uses limited information for tank waste volumes and waste transactions that is now available on a quarterly basis (from 1956-60 on a monthly basis, from 1960-65 only semi-annually) over these tanks’ histories (WSTRS). The HLM then reconciles this volume information with the evaporation rate that is calculated based on the total heat load of each tank. The historical tank heat loads are based on the radionuclides processed on a month by month basis (by Watrous and Wootan) with a calculation of the short-lived radionuclide heating based on the Cs-137 and Sr-90 fission products by use of effective decay heat curves for each batch processed.

Table 1 shows the radionuclide heating that is used. An effective cooling curve was derived by fitting an ORIGEN2 calculation of spent fuel heat generation decay and scaled relative to Cs-137 and Sr-90 fission products. We found that four surrogate radionuclides adequately represented the cooling rate within ±5% from 100 days to six years. After six years, the majority of the decay heat is due to Cs-137 and Sr-90 decay. This approximation allows us to recreate the heat generation rate of each load of fuel that was processed and decay those rates appropriately.

Table 1. Short-lived radionuclide heating.

All of this information is placed into an Microsoft Excel workbook on a month-by-month basis and is termed the HLM spreadsheet. Within the HLM, account is not only made for short-lived radionuclides but also for tank-to-tank transactions, which can move significant amounts of radionuclide inventory from one tank to another. A further assumption is made about the soluble versus the insoluble fraction of radionuclide heating in order to allow moving fractions of the tank heat load among the tanks. All Cs-137 is assumed to reside in the supernatant and all Sr-90 is assumed to reside in the sludge. As a result of moving tank inventory, more tanks than just these four have been analyzed but are not included in this report.

Once unaccounted volume losses and dates for those losses are established, the HDW model provides estimates for leak compositions by assigning unaccounted volume losses to tank inventory leaked to ground. If more detailed transaction and volume information becomes available, the HLM leak estimates can be further refined.

The radionuclide data is expressed as activity in Ci decayed to 1-1-94 for each batch of fuel processed for the history of Hanford and each fuel batch includes a cooling time in days. We have taken that data and regrown the Cs-137 and Sr-90 to the waste addition date and binned the batches appropriately (monthly, quarterly, or semiannually) depending the transaction records reporting frequency. If there were waste additions to more than one tank in a given period, radionuclides were simply partitioned in proportion to the transaction volumes associated with each tank. This fraction is specified in the fr.rads. column in the spreadsheets in App. A.

Given the amount of Cs-137 and Sr-90 regrown for each batch of fuel processed, each of the four short-lived surrogates are calculated and correspondingly binned into the transaction period. Each surrogate radionuclide is decayed on a month-by-month basis in the spreadsheet according to its half-life. Liquid waste transfers from each tank as well as leaks to the soil column result in removal of only the soluble Cs-137 surrogate set while the insoluble Sr-90 surrogate set always remains in the tank into which it was first placed.

Continue to next section