L. A. Fort, S. L. Lambert and G. E. Stegen
SGN Eurisys Services Corporation
Richland, Washington

Glass volume and composition models have been used to successfully optimize the composition of various high-level waste glasses resulting from the treatment of Hanford’s single-shell and double-shell tank wastes. In addition, various thermodynamic models, such as the Environmental Simulation Program, have been developed as an important tool for predicting the solubility properties of high-level waste components in the tank wastes. These models can now be used in tandem to identify the potential impact of modifying chemical separations processes, such as TRUEX and SREX, that are being considered for high-level waste treatment and disposal at the Idaho National Engineering and Environmental Laboratory (INEEL). For example, preliminary studies suggest that the volume of high level waste glass could be reduced by 22 percent through precipitating phosphate from the TRUEX and SREX strip solutions prior to vitrification. Through the use of such tools, it has been shown that even more substantial volume reductions may be possible if the actinides are precipitated as phosphates or oxalates. These precipitates would produce plutonium-laden glasses, such as those being developed at Hanford by linking the current glass models with solution solubility models during the feasibility study of the various flowsheet scenarios proposed for the INEEL waste treatment facility.

The Idaho National Engineering and Environmental Laboratory (INEEL) reprocessed spent nuclear fuel (SNF) from 1953 through the 1980s. Resulting from this action, a substantial amount of waste remains that needs to be treated and disposed of in a geological repository. Most of the high-level waste (HLW) raffinates have been solidified and stored as granular calcine solids. Also stored at INEEL are tanks of high-activity transuranic (TRU) liquid waste referred to as sodium bearing waste (SBW), that are relatively high in sodium content.

A major feasibility study was performed in 1997 on the treatment of INEEL wastes¹. The study identified chemical separation processes to partition the SBW and calcine wastes into two product streams: a high-activity waste (HAW) stream to be vitrified and disposed in a geological repository; and a low-activity waste (LAW) stream to be immobilized in grout for near surface disposal. The proposed waste separation process consists of ion exchange to adsorb cesium and two solvent extraction processes; the TRUEX process to remove radionuclides, such as Am, Pu, U, Np, Cm, Tc, and lanthanides; and the SREX process to remove strontium. The concentrated strip solutions that contain the cesium, TRU, and strontium from these processes make up the HAW portion that is the feed to the vitrification process. By minimizing the quantities of key components in the strip solutions that affect the volume of the vitrified glass, these process solutions can be used to increase the amount of waste per unit volume within the glass matrix, the waste loading. Optimizing the waste loading, minimizes the volume of glass produced, thus reduceing the cost and size of equipment and facilities needed to process the waste.

A primary purpose of the feasibility study was to develop a baseline case design concept detailed enough to support preparation of a project cost estimate and overall process inputs and outputs. This work included development of a process flowsheet and material balance. The baseline case process includes the following elements important to the current discussion:

The baseline case flowsheet described above was used to develop the feasibility study process and facility description and associated estimates of project cost and process performance. However, results of the baseline case calculations suggest that alternative approaches for processing strip solutions from the solvent extraction may significantly improve process performance and reduce cost. In particular, it appeared that there might be opportunities to further reduce ultimate HLW glass volume, which is important because of the high estimated costs for storage, transportation, and repository disposal of the glass.

In order to further evaluate some of these options, two computer modeling tools were used to predict behavior of the strip solutions and impacts of this behavior on the downstream vitrification of the high activity wastes.

The ESP computer model is a chemical process simulator licensed from OLI Systems, Inc.. This model provides the ability to determine the solubility limits and the composition of solids, which will precipitate from aqueous solutions after evaporation and concentration². The ESP model uses global regression techniques to minimize the total free energy of the aqueous system. Large databases are used to construct the activity-composition-pressure-temperature relationships used to estimate the solubility limits and composition of solids in complex multicomponent solutions. The ESP model has been used extensively at Hanford and as a valuable tool for predicting behavior of tank wastes, during storage and processing. These wastes are typically complex multicomponent solutions with relatively high concentrations of dissolved salts.

The ESP model was used to investigate several of the strip solutions that are being considered to determine the possibility of precipitating undesirable components and chemical species of criticality importance. This model was also used to estimate the maximum volume reduction that could be obtained before the precipitation of pertinent elements and components from solution.

Models that correlate the variation glass properties as a function of glass composition have been developed and used extensively in the Hanford High-Level Waste Program³. These models have been developed by researchers at the Pacific Northwest National Laboratories based on a large amount of glass property data developed for the glass composition ranges expected for nuclear waste glasses⁴. The models provide information on properties that are important for product quality and the ability to process the glass in a melter, including melting temperature (or viscosity as a function of temperature), durability (or leachability), electrical conductivity, and component solubility in the molten glass as a function of temperature. The models mainly consider major components in the glass, such as SiO₂, B₂O₃, Na₂O, Li₂O, CaO, MgO, Fe₂O₃, Al₂O₃, and ZrO₂. Other components included in the models are Cr₂O₃, MnO₂, NiO, P₂O₃, and K₂O. Given a waste feed composition, the models allow glass formulation to be optimized for maximum waste loading consistent with the applicable processing and product quality constraints. This allows the flowsheet and feed composition to be directly assessed in terms of the potential impact on the glass volume.

The composition of the high-activity stream from processing the SBW and dissolved calcine waste is expected to have a significant effect on the design and operating characteristics of the melter. Based on the chemistries of the separation process, the TRUEX strip solution contains a significant concentration of 1-hydroxyethane 1,1-diphosphonic acid (HEDPA), together with a small concentration of tri-butyl phosphate (TBP) and other residual phosphates⁵. If the glass is formulated to produce the minimum number of canisters (or maximum waste loading), the glass models indicate that P₂O₅ (or phosphate) will be the limiting component in the glass. This composition constraint is important because (1) borosilicate glasses with more than 10 wt.% P₂O₅ may be susceptible to phase separation in the melter, and (2) phosphate-rich glasses are potentially corrosive in metal wall and ceramic-lined melters, depending on the residual contaminates in the waste. The preferred option appears to be the selection of a melter, such as the cold crucible melter, that is not susceptible to phosphate-induced corrosion.

The SBW and dissolved calcine feeds for this facility will be concentrated to about 2.4 M HNO₃ in the high-activity waste evaporator and stored for at least one year before being processed though the vitrification facility. Since the volume of SBW feed is relatively small, this feed will be stored on an interim basis for two years. These wastes will be initially transferred to a feed staging tank. Each batch of feed will be analyzed and adjusted if necessary with adjusting chemicals (sodium nitrate) before transfer by air-lift to the melter feed tanks. The feed will be transferred by a double-stage air-lift to a constant level head-pot and metering wheel, which provides feed at a constant rate to the seal pot and melter (60 liters/hour).

Based on current material balance projections, about 307,250 liters of concentrated SBW feed must be processed through the vitrification facility in the first two years of operation. The SBW feed will be delivered in five batches per year from the HAW concentrate tank, with each batch consisting of 30,200 liters of feed (962 kg of nonvolatile oxides per batch). Because the melter is limited to 60 l/hr of feed, it should take about 438 hours to process each batch through the vitrification process, depending on the water content of the feed. This feed rate corresponds to an overall operating efficiency of 25 percent, based on 8,760 hours per year. The melter will produce 31,350 kg of SBW glass (19 canisters) over two years, with an average production rate of 7.16 kg per hour of glass (about 24 percent of the nominal capacity of the slurry-fed cold crucible melter). Spent cesium ion exchange resin will be transferred in batch lots from the HAW solids surge tank. Since the settling characteristics of this resin are not well understood, the resin will be segregated from the SBW feed and collected in a separate solids feed tank. For purposes of this study, it is assumed grinding or reducing the size of the spent ion-exchange resin is necessary. These solids will be transferred as a slurry to the melter at a rate that is proportional to the glass formulation estimates for the SBW glass.

Dissolved calcine feed will also be provided as a 2.4 M HNO₃ solution from the HAW concentrate tanks. About 3,020,000 liters of this feed will be produced over the 20-year operating life of this project (150,000 l/yr.). Approximately five batches of dissolved calcine feed (30,000 liters each) will be delivered to the vitrification facility each year. Under these conditions, the vitrification plant is expected to operate 2,190 hours per year, based on the feed requirements for the melter (60 l/hr. of water). This feed rate corresponds to an overall operating efficiency of 25%. According to the glass models, the melter should produce 53,200 kg/yr. of dissolved calcine glass (24.3 kg/hr.). Assuming that the 3-meter (10 ft. by 2 ft. diameter) canisters are used, about 32 canisters of such glass should be produced each year (70 percent of the nominal capacity of a slurry-fed cold crucible melter).

For the SBW waste, the main streams being sent to vitrification are the cesium ion exchange effluent, TRUEX strip, SREX strip effluents, and the spent ion exchange resin. The SESC glass models were used to predict the composition and volume of SBW glass. The SESC models indicate that 16 canisters of SBW glass will be produced in the vitrification plant, with the glass volume being limited by the amount of P₂O₅ in the glass (9.91 percent). This glass has an equivalent waste loading of 37.62 percent.

According to these models, the total calcine waste is expected to produce 645 canisters of glass, with average waste oxide loading of 30.48 percent. The glass volume is limited once again by the amount of P₂O₅ in the glass (9.76 percent). The feed staging requirements for this waste are also described, including evaporation and sugar denitration of the feed. The volume and composition of off-gas from the evaporator, condenser, nitric acid absorber, and SCR are provided.

The INEEL glass composition estimates for aluminum calcine predict that the produced 34 canisters of glass will have a mean composition of 11.69% P₂O₅ or 6,480 kg of P₂O₅ in the glass. The SESC glass models produced a total different result that being 310 canisters of glass with 51,200 kg of P₂O₅ primarily from the TRUEX strip. The SESC estimates are based on a limiting P₂O₅ content of 10 percent in the Al calcine glass. If these estimates are correct, the Al calcine will produce an unacceptable amount of glass because the TRUEX process separates the aluminum and other chemicals from the TRUEX product. This problem could be corrected by blending the Al calcine with Zr calcine before the dissolution step or by recycling the TRUEX strip to the dissolvers after the HEDPA is destroyed and perhaps TRU is removed with acid side precipitation processes.

For the dissolved total calcine waste, two different composition envelopes were considered. The first group consists of the TRUEX and SREX strip solutions with 0.1 molar ammonium citrate in the SREX strip^5,6. The ESP model was used to establish that these solutions could be concentrated from 210 million liters to as little as three million liters, with HgMoO₄, MoO₃, (UO₂)₃(PO₄)₂-4H₂O and ZrO₂ as the main precipitates; and where (UO₂)₃(PO₄)₂-4H₂O is the only precipitate of criticality significance. Based on this model, it was found that the TRUEX and SREX solutions, with 0.1 molar ammonium citrate in the SREX strip, could be concentrated to about seven million liters before (UO₂)₃(PO₄)₂-4H₂O precipitation occurs. The second group is comprised of TRUEX and SREX strip solutions with 0.01 molar ammonium citrate in the SREX strip. The ESP model shows that these solutions could be concentrated from 210 million liters to as little as 1.48 million liters, with HgMoO₄ and (UO₂)₃(PO₄)₂-4H₂O as the main precipitates. The TRUEX and SREX strip solutions, with 0.01 molar ammonium citrate, apparently can be concentrated to about 1.7 million liters before (UO₂)₃(PO₄)₂-4H₂O precipitation occurs.

The solubility characteristics of the dissolved calcine TRUEX strip were also investigated. The results, as might be expected, were similar to those obtained for the combined TRUEX and SREX strip solutions. However, there appears to be some evidence, based on the relative scaling tendencies of various species, that oxalic acid may be a useful additive because it tends to induce the precipitation of uranium and plutonium species while forming a soluble complex with zirconium.

For SBW waste, four different composition envelopes were investigated. These envelopes include (1) the combined TRUEX and SREX strip solutions, (2) the TRUEX strip solution, (3) the TRUEX strip with various additives such as Fe(NO₃)₃, Al(NO₃)₃ and Ca(NO₃)₂, and (4) the combined TRUEX and SREX 1 & 2 strip solutions. Based on these results, the TRUEX and SREX strip solutions can be safely concentrated by a factor of 142 without forming any undesirable precipitates, such as (UO₂)₃(PO₄)₂-4H₂O. For the TRUEX strip solution, FePO₄-2H₂O and (UO₂)₃(PO₄)₂-4H₂O precipitate after approximately 60 cycles of concentration and Pu(HPO4)₂ after 300 cycles of concentration⁷. Fe(NO₃)₃ appears to be a useful additive for inducing phosphate precipitation in the concentrated TRUEX strip solution. Based on the ESP model, Fe(NO₃)₃ can be used to precipitate 92.7 percent of the phosphate from the SBW TRUEX strip solution after 46 cycles of concentration. Al(NO₃)₃ and Ca(NO₃)₂ were also investigated as possible additives for phosphate precipitation, but the ESP model indicates that these additives were relatively ineffective. The final composition consists of the combined TRUEX and SREX 1 & 2 strip solutions. The ESP model indicates that these solutions can be safely concentrated by a factor of 182 without forming any undesirable precipitates.

The solubility limit for dissolved calcine feed (TRUEX and SREX strip solutions with 0.01 molar ammonium citrate in the SREX strip) appears to about 85,000 liters (22,460 gallons) per year, based on the ESP model. For this simulation, HEDPA is assumed to have completely degraded in the boiling nitric acid solution The first identified precipitate of criticality significance is (UO₂)₃(PO₄)₂-4H₂O. Dissolved calcine concentrate is about two times more diluted than the solubility limit for (UO₂)₃(PO₄)₂-4H₂O.

Glass modeling studies estimate the composition and volume of HLW glasses produced from the Al calcine, Zr calcine and total calcine at INEEL. Based on these studies, the Al calcine glass is estimated to have 12.42 % waste oxide loading (WOL), with the composition being limited by the amount of P₂O₅ in the glass (10 % P₂O₅ and 10.58 % Al₂O₃). Zr calcine glass is projected to have 23.85 % WOL and is limited by P₂O₅ (10 % P₂O₅, 3.38 % Al₂O₃, 12.58 % ZrO₂ and 4 % Li₂O). For the total calcine case, the glass has 31.25 % WOL and is limited as in the earlier cases by P₂O₅ (10 % P₂O₅, 7.4 % Al₂O₃, 10.86 % ZrO₂ and 4 % Li₂O). Based on these projections , the optimum blend appears to be the total calcine case where Al calcine is uniformly blended with the Zr calcines from bin sets 2 through 6. Assuming that 25 percent of the waste consists of Al calcine and 75 percent Zr calcine, approximately 1,039 canisters of glass would be produced if the Al and Zr calcines are processed separately (405 canisters of Al calcine glass and 633 canisters of Zr calcine glass). However, in the total blend case, only 645 canisters of glass would be produced from the combined wastes, saving 394 canisters of glass and $551 million in associated disposal cost ($400,000 per canister in repository disposal cost and $1 million in shielded transportation cost).

Furthermore, because SO₄ is relatively insoluble in borosilicate glass (0.75 wt.%), the SBW solids (with 3,450 kg of SO₄) should be blended with the total calcine waste to minimize the volume of SBW glass, and the Li₂O concentration should be limited to a maximum of 4 percent in the dissolved calcine glass to prevent eucriptite (LiAlSi₂O₆) precipitation in the canister. The precipitation of eucriptite could drastically reduce durability of the glass by removing aluminum from the glass matrix.

The SBW and calcine glass compositions are generally limited by the amount of P₂O₅ in the glass (10 wt.%). At this concentration, a second phase could form (amorphous glass phase), which could be corrosive in the melter, or a molten salt phase (rich in SO₄ and P₂O₅) could form, creating a potential waste disposal problem or possibly affecting the melt-rate capacity of the melter. Based on the current glass models, it appears that the volume could be reduced by 20 percent if ferric nitrate is used to precipitate P₂O₅ from the vitrification feed. According to the literature, HEDPA readily decomposes in boiling nitric acid with hydrogen peroxide and vanadium catalyst, producing phosphoric acid and various organic residues.

A much larger reduction in waste volume might be obtained if the UO₂ and Pu are precipitated as (UO₂)₃(PO₄)₂-4H₂O and Pu(HPO₄)₂ from the TRUEX strip solution and immobilized in glass or grout. While Np may precipitate as Np(OH)₄, the behavior of Am (in the third valance) and other fission products in the TRUEX strip solution is uncertain. The supernate from this precipitation process could be recycled to the TRUEX separation process to recapture residual actinides in the waste. The glass volume might be reduced to five canisters of glass if Zr remains in solution (based on 10 percent PuO₂ and UO₃ in glass).

The ESP model, however, indicates that ZrO₂ is likely to precipitate from the concentrated TRUEX strip solution. If ZrO2 precipitates as expected, this would substantially increase the volume of glass produced from this sludge. By adding a small amount of oxalic acid is added to the TRUEX strip, the ESP model shows that actinides, such as UO₂ and Np, would be more likely to precipitate than Zr because the oxalate tends to form a soluble complex with Zr. Oxalic acid might also be considered as a holding complexant for Zr to reduce the amount of Zr that may otherwise be extracted in the TRUEX process. Other actinides, such as Pu, might be captured by recycling the supernate to the TRUEX process as a means to increase the concentration of Pu in the TRUEX strip solution

By using thermal dynamic equilibrium models like ESP and glass formulation, models based on statistical analysis of theoretical data will guide the waste treatment process development and testing work towards an optimal design for the given requirements and criteria for treatment of the INEEL high-level waste. Using these tools in evaluations and assessments will better define the optimal process conditions for the specified waste, defining the plant size which minimizes the overall life cycle costs (that of design, construction, operation, and decontamination and decommissioning) to achieve the objectives of the waste treatment effort.

1. Fluor Daniel, "Idaho Chemical Processing Plant Waste Treatment Facilities Feasibility Study," DOE Delivery Order DE-AD07-97ID60036, December 1997.

2. Meng, C. D., MacLean, G. T., Landeene, B. C., "Computer Simulation of Laboratory Leaching and Washing of Tank Waste Sludges," Westinghouse Hanford Company, WHC-SD-WM-ES-312, October 1994.

3. Lambert, S. L., G. E. Stegen, and J. D. Vienna, "Tank Waste Remediation System Phase I High-Level Waste Feed Processability Assessment Report," WHC-SD-TI-768, Rev. 0, Westinghouse Hanford Company, August 1996.

4. Hrma, P. R., et. al., "A Property/Composition Relationships for Hanford High-Level Waste Glasses Melting At 1150 ^oC," PNL-10359 Vol. 1 and 2, Pacific Northwest Laboratory, Richland Washington, December 1994.

5. Law, J. D., K. N. Brewer, R. S. Herbst, and T. A. Todd, "A Demonstration of the TRUEX Process for Partitioning of Actinides from Actual ICPP Tank Waste Using Centrifugal Contactors in a Shielded Cell Facility," INEL-96/0353, Lockheed Martin Idaho Technologies Company, September 1996.

6. Wood, D. J. et. al., "A Development of the SREX Process for the Treatment of ICPP Liquid Wastes," INEEL/EXT-00831, Lockheed Martin Idaho Technologies Company, October 1997.

7. Brewer, K. N., et. al., "A Elimination of Phosphate and Zirconium in the High-Activity Fraction Resulting from TRUEX Partitioning of Idaho Chemical Processing Plant Zirconium Calcines," INEEL/EXT-97-00836, Lockheed Martin Idaho Technologies Company, July 1997.