HNF-3233

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

Stephen F. Agnew and Robert A. Corbin
Chemical Science and Technology Division
Los Alamos National Laboratory

August 1998

Summary:

This report uses readily available historical information to better define the volume, chemical composition, and Cs-137/Sr-90 amounts for leaks that have occurred in the past for tanks SX-108, SX-109, SX-111, and SX-112. In particular a Historical Leak Model (HLM) is developed that is a month by month reconciliation of tank levels, fill records, and calculated boil-off rates for these tanks. The HLM analysis is an independent leak estimate that reconstructs the tank thermal histories thereby deriving each tank’s evaporative volume loss and by difference, its unaccounted losses as well. The HLM analysis was meant to demonstrate the viability of its approach, not necessarily to establish the HLM leak estimates as being definitive. Past leak estimates for these tanks have invariably resorted to soil wetting arguments but the extent of soil contaminated by each leak has always been highly uncertain. There is also a great deal of uncertainty with the HLM that was not quantified in this report, but will be addressed later.

These four tanks (among others) were used from 1956 to 1975 for storage of high-level waste from the Redox process at Hanford. During their operation, tank waste temperatures were often as high as 150°C (300°F), but were more typically around 130°C. The primary tank cooling was by evaporation of tank waste and therefore periodic replacement of lost volume with water was necessary to maintain each tank’s inventory. This active "reflux" of waste resulted in very substantial turnovers in tank inventory as well as significant structural degradation of these tanks. As a result of the loss of structural integrity, each of these tanks leaked during their active periods of operation. Unfortunately, the large turnover in tank volume associated with their reflux cooling has made a determination of leak volumes very difficult.

During much of these tanks’ operational histories, inventory losses because of evaporative cooling could have effectively masked any volume loss due to leak. However, careful comparison with reported tank levels during certain periods clearly show unaccounted volume losses for many tanks. As a result of the HLM analysis, SX-108, SX-109, SX-111, and SX-112 all show clear evidence of unaccounted volume losses during the period 1958 to 1975. Likewise, the HLM does not show similar unaccounted volume losses for tank SX-105, a tank with no reported leak history, verifying that the HLM is consistent with SX-105 not leaking.

These unaccounted volume losses establish the leak start date and rate, and when propagated over time show that SX-108 lost 203 kgal followed by SX-109 at 111, SX-111 at 55, and SX-112 at 44 kgal. These leak volumes represent maximum or upper bounds estimates of each leak and are in total volume about six times the previous leak estimates. Minimum leak estimates are about 50% of these values based on judgments about the heat and leak rate uncertainties. Except for tank SX-112, these results all assume that leaks are continuous from start to end and depend only on three parameters: an elevation in kgal, a rate in kgal per month per kgal above that elevation, and a start date for the leak.

Acknowledgments:

This work has been performed under the auspices of the Department of Energy under contract to the University of California.

Scope:

To use best information currently (as of October 1996) available to estimate the leak volumes for four SX Farm tanks, SX-108, SX-109, SX-111, and SX-112. Once leak volumes and dates have been established, derive curies of Cs-137 and Sr-90 as well as leak chemical composition for major analytes (Na⁺, OH^-, CrO₄^2-, NO₃^-, NO₂^-, AlO₂^-).

One must recognize that there are multiple unstated assumptions and uncertainties with all existing leak estimates. The HLM is simply an attempt at deriving an independent estimate for a particular set of waste tanks, boiling waste tanks. The HLM analysis was meant to demonstrate the viability of this approach, not necessarily to establish the HLM leak estimates as being definitive.

Continue to next section