Comparison of simulants to actual neutralized current acid waste: process and product testing of three NCAW core samples from Tanks 101-AZ and 102-AZ [report]

E.V. Morrey, J.M. Tingey, M.L. Elliott
1996 unpublished
Reprint of historical document PVIPC9Sm.03E. dated September 1995. Data. formaaing. and ocher mnvcntiom refla standards at thc original dak of printing. Technical p a r review and cdilorii reviews may m have been pufamcd. Oriice o i k i e n t i k and Technicai iniormation, p . 0 . Sox 12, Oi'h KidgL, T S 3X3:; prices-available from (615) 576-8401. Available lo the public from the National Technical Information Service, US. Department of Commerce, 5285 Port Royal Rd., Springfield, VA 221 61 @ f
more » ... eld, VA 221 61 @ f i e document was printed on recycled paper. t I - Acknowledgements The process and product testing of the core. samples required contributions horn a large group o1 scientists, engineers, and technicians. Garry Richardson was responsible for processing the waste samples through frit addition and drying; he also performed physical and rheological characterization along each step of the process. Ron Holeman, Jim Dunn, and Dewayne Smith were responsible for ' calcining and vitrifying the waste/f'rit mixtures. Ron Holeman, Jim Dunn, Dewayne Smith, and Tim Reining performed sample preparation and durability testing of the resulting glasses. Chemical and ' radiochemical characterizations were provided by Dale Archer, A vitrification plant is planned to process the high-level waste (HLW) solids from Hanford Site tanks into canistered glass logs for disposk in a national repository. Programs have been established within the Pacific Northwest Laboratory Vitrification Technology Development (PVTD) Project to test and model simulated waste to support design, feed processibility, operations, permitting, safety, and waste-form qualification. Parallel testing with actual radioactive waste is being performed on a laboratory-scale to c o n f i r m the validity of using simulants and. glass property . models developed from simulants. Laboratory-scale testing has been completed on three radioactive core samples from tanks 101-AZ and 102-AZ containing neutralized.current waste (NCAW), which is one of the first waSte types to be.processed in the high-level waste vitrification plant under a privatization scenario. Properties of the radioactive waste measured during process and product testing were compared to simulant properties and model predictions to donfirm 'the validity of simulant and glass property models work. This repoh includes,re&s from the three NCAW core samples, comparable results from slurry and glass simulants, and comparisons to glass property model predictions. Experimental Approach ' The three NCAW samples were retrieved from the tanks in cylindrical segments 1 .inchin ' ' diameter and 19 inches long. Several segments representing a complete vertical sample of the settled solids in the tank were combined and blended to make up a core sample. Solids from each core sample were pretreated using a water wash/settle/decant process, including a ferric-nitrate flocculent additive, settle/decant, and two water washes (3 volunies deionized water to 1 volume solids). The washed solids were then characterized chemically, radiochemically, physically, and rheologically (101-AZ Core 1 only). After adjusting the samples to 125 g waste oxide& the waste was trea'ted with formic acid to adjust the feed rheology and to reduce the redox-sensitive species for introduction into the melter,. Off-gas analysis during formic acid addition was performed on 102-AZ Core 2 and . is described in a'separate report.") The formated slurry samples were characterized chemically, physically, and rheologically. Frit was added to each of the foryted s h i e s and samples were characterized chemically, radiochemically, physically, and rheologically. The frit/slurry mixtures were dried and melted at 115OOC in crucibles; resulting glass was characterized with respect to . chemical and radiochemical composition, durability (Product Consistency Test IpcT) and Materials Characterization Center MCC-l]), crystallinity, redox, and density. 1 Two types of simulants were prepared and tested for comparison with the actual waste glass. Process-based Slurry simulants were used to develop and tes't hot-cell'procedures and to provide a direct comparison with the core sample feed chemistry and rheology. Major and minor insoluble components were co-precipitated with NaOH from nitrate solutions and washed to remove the sodium (a) Langowski, M.H., E.V. Morrey, J.M. Tingey, and M.R. Beckette. 1993. w g a s Characteri~*onfrorn the Radioactive NCAW Core -le (102-AZ-Cl) and Simlant During Hwvp Feed Preparation Testing. Letter Report for US. Department of Energy. Pacific Northwest Laboratory. Richland, Washington. iii and nitrate. Soluble minor components were added separately, following washing. Glass simulants, similar to those used to develop glass property models, were prepared to provide a direct comparison with the actual waste-glass product quality. Three simulant glasses were prepared to match the chemical composition of the three actual waste glasses by batching and melting appropriate amounts of dry chemicals. The simulants were characterized the same as the actual waste samples, excluding radiochemical, to provide a direct comparison of simulant and actual waste. Process and product behaviors of the actual waste were compared to simulant behavior, glass property models, slurry property correlations, and simulant behavior from other studies and larger scale tests. Statistical comparisons of simulant waste-glass durabilities and model predictions to actual waste-glass durabilities were made based on 95 % confidence t-tests. Other comparisons are primarily nonstatistical. Results and Conclusions Slurry Chemical Characterization. The chemical composition of the three NCAW core samples and simulants were similar. The major components in all three samples are Fe, Al, and Na 8s OH, C q -, NO;, and NO;. The pH of the washed solids were approximately 12.7 for the core samples from tank 101-AZ, 10.2 for tank 102-AZ, and 10.4 for the simulants. High washing efficiencies of the major cations as measured by the percentage of the analyte remaining in the washed solids slurries compared to the prewashed solids were only observed for Na (30%) and Cr (60%), but significant quantities of Al, Cr, K, Na, F, C1', NO;, NO;, and SO: were removed from the sludge in the washing steps. Phosphate is the only measured anion in which p significant percentage remained in the sample. Comparison of the concentration of Na in the washed solids, the sludge prior to washing, and the reference nominal value for the previously planned HWVP indicates that acceptable washing efficiencies are being achieved on the laboratory-scale processes. A comparison of chemical composition of simulant 102-AZ Core 1 and the corresponding core sample indicates that accurate chemical simulants can be prepared. During the formating process CO?, NO;, and NO; react to produce gas, and the concentrations of these anions in the sample decrease. Slurry chemistry and offgas generation reactions are similar between the core sample and simulants. Observed offgas differences between simulant and core sample could be explained by differences in testing conditions and slurry chemical composition. Formated slurry was combined with frit to achieve melter feeds with waste oxide loadings of 25 to 28 percent; therefore, the major constituents in the melter feeds are the frit components. These major components include Si, Na, B, and Li. The frit components were added as the oxides; therefore, the majority of the elements in the melter feeds are as oxides. Other anions which are present in significant quantities are NO;, NO;, C1-, F, and SO: . The supekatant from the melter feeds contained only three cations in significant quantities (Na, IC, and Li). The major anions associated with the Supernatant are similar to those observed in the slurries with the exception of the oxides which are not soluble. Slurry Radiochemical Charactex4zation. Handling and disposal of chemical simulants is much more cost effective than radiochemical simulants; therefore, no radiochemical simulants were prepared in these studies. The major radionuclides present in the core samples are lr'Cs, 'OSr, '"Ce, and lMRu. All of the slurry sampks are transuranic (contain > 100 nCi/g transuranic isotopcs). The iv majority of the transuranic activity is due to americium and plutonium. None of the supernatant or wash solutions were transuranic, and 13' Cs is the primary radionuclide in the supematant. The only radionuclides affected by the washing steps were l3'Cs and *%b. The only radionuclide which may have been affected by the formating steps was lq. The data is not definitive, but it appears that some of the iodine may have been lost during the formating process. Measured radionuclide activities were within the previous HWW specifications with the exception of '3, 90Sr, and %o. and simulants ranged from 1.04 to 1.14 g/mI. The density of the fomtod slurries was similar to that of the washed solids. As expected the density of the samples increased with increasing solids concentrations. A correlation between the density of the formated slurries and the solids concentration is observed, and simulants are representative of actual waste with respect to this correlation. The density of the melter feeds (1.28 to 1.47 glml) increases significantly from the formated slurry and washed solids density. This trend is also observed for the centrifuged solids density. The simulants have a significantly lower centrifuged solids density than is observed in the core samples. The density of the centrifuged liquid (1.02 f 0.03) was similar for all of the slurries and is comparable to the density of water. V and nitrate. Soluble minor components were added separately, following washing. Glass simulants, -~~_ ~-Vitrification Plant (HWVP). After accounting for differences in concentration,' the simulated formated waste exhibited yield stresses and apparent viscosities roughly two times greater than those for actual formated waste. Act& f o h t e d waste samplesexhibited greater initial settling rates, greater degrees ofsettling, and deker &ntrifuged solids than simulant formated samples, indicating a difference in microstructure. A. comparison with rheological data fiom full-scale formated simulant tests showed essentially identical results with laboratory-scale formated simulant data from this study. A comparison with historical NCAW formated simulant data dating back to 1985 showed actual formated waste results to be equal to or lower than the weakest (i.e., lowest shear stress and apparent viscosity for given concentrations) simulants reported. For melter feed samples, actual waste exhibited lower yield stresses and apparent viscosities than simulatgl waste, which again i'i; attributed to differences in microstructure. Rheological behhior at the radioactive @d simulant melter feed slurries was best represented as thixotropic, yield pseudoplastic with varying degrees of hysteresis. Yield stresses of the radioactive melter feed ranged from 1.4 Pa to 10.3 Pa compared to simulant melter feed yield stresses of 2.2 Pa and 12.4 Pa. Apparexit'viscosities*of the actual waste samples at 50s' ranged from 38 CP to 260 CP compared to simulant viscosities of 58 CP and 365 cP. GIass Characterization. Initial chemical characterizations of the glass were inadequate, using the standard KOH/Na202 preparation methods for inductively coupled argon plasmalatomk emission spectrometry (ICP/AES) analysis. Analysis' of the first 'two radioactive glasses accounted for only 91 9%-93 % of the mass of &e glass. Additional analysis using an HF preparation and comparable standard glasses were required to arrive at a reasonable glass composition. Subsequent procedural improvements to the KOH/Na202 preparation methods resulted in satisfactory results for the third core-sample glass. For all three radioactive glasses, the measured major analytes were generally within 10% of calculated values, which were determined from washed solids composition, frit compositions, and assumed waste loadings. Achieved waste loadings were slightly greater than targeted (i.e., 2% to 5%), because of accuracy'limitations on slurry sampIing or total oxides analysis. Glass redox for the actual waste glasses as measured by Fe+2/Fe+3 ratio ranged between 0.026 to 0.085.compared to a simulant redox of 0.005. Each of the glasses measured was within the design limit for the prior planned Hwvp plant. Glass redox of the actual waste compared well with historical simulant data correlating glass redox to formic acid added and initial nitrite and nitrate Compositions. Radioactive glass samples were analyzed by X-ray diffraction to determine the extent and composition of crystalline phases. As predicted by models and simulant experience, no substantial crystallinity was found (Le., likely under 1%). Density of the three radioactive glasses was measured to be 2.56 glcc for 101-AZ Core 1, 2.67 g/cc for 101-AZ Core 2, and 2.54 g/cc for 102-Az Core 1 at room temperature, values typical of simulant glass densities. GIass Durability. Each of the three radioactive and simulant glass formulations produced highly durable glasses in all cases at least 20 tim& more durable than the Savannah River Environmental Assessment (EA) glass as measured by the PCT. Sevenday PCT boron releases for . the radioactive glasses ranged from 0.13 to 0.21 g/m2 compared to simulant boron releases of 0.20 to 0.34 g/mz. The magnitude of the increase from radioactive to simulant releases ranged from 28 to vi 67%. In each case the differences in boron release were found to be statistically significant to a 95 % confidence. Model predictions for each of the three glass formulations were greater than both actual and simulant waste glass releases. Over the limited amount of tests performed; the achal and simulant waste glass releases fell within the 95% prediction interval for the model 56% and 89% of the time, respectively. Twenty-eightday MCC-1 results for actual and simulant glasses showed similar results, however the differen& were greater. An indeterminate portion of the difference was attributed to differences in leach containers used for these tests. Radiation dose has been shown to have a significant effect on glass corrosion in aqueous leach tests; however, the prediction and explanation of the radiolytic effects are complex. The durability differences between dctual and simulated waste glasses reported in this study were equal to or lower than differences observed by others, which was consistent with removing part of the radiolytic effect (Le., tests performed in Ar atmosphere). To the extent Ar backfilling of the leach containers was effective, the effect of radiolytic generation of nitric acid was 'eliminated. Based on the type of test performed and the relative durability of the glasses in this study, the dominant corrosion mechanism is expected to be network hydrolysk, which is favored under higher pHs. Had=the leach containers contained air, one would have predicted the radiolytic effect to be decreased pH and glass corrosion. With the absence of air in the system, it was not clear whether radiolytic affects shofld increase or decrease corrosion. Sevenday PCT and 28day MCC-1 radionuclide releases were measured, calculated, and compared to results from prior studies. As with prior studies, Am had normalized releases significantly lower than B, ranging from 0.1% to 6% of B. Also consistent with prior studies, Tc, U, Np, and Cs were generally more soluble than Am (Le., 2 10% of B-normalized release). Not consistent with prior studies, Pu exhibited significant normalized releases near B. Strontium was relatively soluble in MCC-1 tests and insolubleh P c . L1
doi:10.2172/403946 fatcat:jxqnoyaj5jfzlf3ttr5jdcfn2y