(18d) Development of Waste Acceptance Criteria for Modular Caustic-Side Solvent Extraction Unit | AIChE

(18d) Development of Waste Acceptance Criteria for Modular Caustic-Side Solvent Extraction Unit

Authors 

Adu-Wusu, K. - Presenter, Savannah River National Laboratory
Walker, D. D. - Presenter, Savannah River National Laboratory
Edwards, T. B. - Presenter, Savannah River National Laboratory
Fink, S. D. - Presenter, Savannah River National Laboratory


A Modular Caustic-Side Solvent Extraction Unit (MCU) is being designed and built at the Department of Energy's (DOE) Savannah River Site (SRS) in Aiken, South Carolina. The Caustic-Side Solvent Extraction (CSSX) process removes cesium from filtered highly alkaline radioactive waste solutions containing high concentrations of sodium and potassium using an organic solvent. In the CSSX process, the solvent is contacted with the cesium-rich aqueous waste solution in an extraction stage to remove the cesium. Following separation of the aqueous and organic phases, the aqueous waste solution exits the MCU as a cesium-depleted effluent stream to be made into a grout. The solvent goes to a scrub stage where it is contacted with 0.05 M nitric acid solution to remove sodium, potassium, and other soluble salts including any trace levels of aluminum, iron, or calcium. Neutralization of any hydroxide carryover from the waste also occurs in the scrub stage. The metals removal and hydroxide neutralization are essential for a robust stripping stage operation. The scrubbed cesium-laden solvent proceeds to a stripping stage, where it is contacted with 0.001 M nitric acid solution to remove the cesium. The nitric acid solution leaves the MCU as a cesium-rich effluent to be made into glass. The solvent from the stripping stage moves to a wash stage, where it is contacted with 0.01 M sodium hydroxide solution to remove residual impurities and any solvent-degradation products before recycling to the extraction stage. The organic solvent comprises four components: (1) an extractant, a calixarene crown ether (Calix[4]arene-bis(tert-octylbenzo-crown-6)) known as BOBCalixC6, that has superior cesium selectivity over sodium and potassium; (2) a modifier, a fluorinated alcohol solvating agent (1-(2,2,3,3-tetrafluoropropoxy)-3-(4-sec-butylphenoxy)-2-propanol) known as Cs-7SB, that improves extraction and helps prevent third-phase formation by increasing the solubility of the extractant; (3) a suppressor, tri-n-octylamine (TOA), that inhibits impurity effects, and improves and stabilizes stripping; and (4) a diluent, Isopar®L, which is a blend of C10 to C12 branched alkanes, that promotes good hydraulics due to its low viscosity and density. The MCU will process waste with varying compositions. The variations in waste solution composition will impact the process performance by changing the final cesium concentration. To ensure adequate process performance, limits must be placed on the composition of the waste feed solution. The objective of the study was to develop waste feed acceptance criteria necessary to meet the target final cesium concentration in the effluent product stream. The study involved proposing ranges for 12 waste feed components (i.e., Na+, K+, Cs+, OH-, NO3-, NO2-, Cl-, F-, SO42-, PO43-, and CO32-, and AlO2-) through a compilation of SRS waste data. Statistical design methods were used to generate numerous wastes with varying compositions from the proposed ranges. An Oak Ridge National Laboratory model called SXFIT was used to predict the cesium extraction distribution coefficients (D-values) between the organic (solvent) phase and the aqueous waste phase using the waste component concentrations as inputs. The D-values from the SXFIT model were used as input along with MCU base case conditions or process parameters to a SASSE (Spreadsheet Algorithm for Stagewise Solvent Extraction) model to calculate the final cesium concentrations for the MCU. The SASSE model was developed at Argonne National Laboratory. The SXFIT D-value and waste component concentration data were used to develop a mathematically simpler model (neural network model) to use in lieu of the SXFIT model that predicts D-values within 15% of the SXFIT D-values. Both the SXFIT and the neural network model revealed the following. The solvent extractant concentration ratios are approximately equal to the corresponding D-value ratios; a useful feature that could be used to predict extraction D-values when the extractant concentration in the solvent changes in the MCU operation. Also, potassium is the only waste component out of the 12 that shows a distinct relationship with the cesium extraction D-values; an indication of potassium's competition with cesium in the CSSX process. A waste feed acceptance model suitable for assessing wastes within relatively wide ranges of D-values (0.6 ? 40) and initial cesium-137 concentrations (0.2 ? 12.8 Ci/gal) has been developed from the SASSE outputs. The waste feed acceptance model is an equation involving initial cesium-137 concentration and D-value that results in a final cesium-137 concentration of 0.1 Ci/gal, the target concentration for the MCU. For example, the waste feed acceptance model shows the minimum acceptable extraction D-value based on MCU base conditions is 5.73. The waste feed acceptance model is defined by a simple linear relationship for extraction D-values > 7. This tool facilitates quicker calculations.