COMPLETION OF THE FIRST PHASE OF HLW VITRIFICATION AT THE WEST VALLEY DEMONSTRATION PROJECT

Transcription

1 COMPLETION OF THE FIRST PHASE OF HLW VITRIFICATION AT THE WEST VALLEY DEMONSTRATION PROJECT William F. Hamel, Jr., U. S. Department of Energy - West Valley Demonstration Project Paul J. Valenti and Daniel I. Elliott, West Valley Nuclear Services Company, Inc. ABSTRACT The first phase of high-level radioactive waste (HLW) processing at the West Valley Demonstration Project (WVDP) was completed on June 10, This accomplishment represents over 9.32 million curies of cesium-137/strontium-90 (approximately 85% of the total estimated curie inventory) removed from the underground waste storage tanks, leaving only tank heel wastes remaining to be removed. Through this first phase, starting with the initiation of HLW vitrification operations in July 1996, 211 canisters of borosilicate waste glass have been produced. Each canister holds approximately 2,000 kilograms of glass and has a contact dose rate of nearly 2,700 rem per hour. HLW vitrification at the WVDP encompasses waste mobilization and transfer, slurry feed preparation, joule-heated melter operation, canister handling and temporary storage, and melter off-gas treatment. Many challenges have been met throughout the first phase of processing, especially in the realm of remote operations within the vitrification process cell. This paper describes the successful first phase of HLW processing operations at this facility, the process technology applied, operating experience gained, and lessons learned from continuous operation. BACKGROUND AND HISTORY The WVDP was established in 1980 by the United States Congress as a U. S. Department of Energy (DOE) high-level radioactive waste management project. The mission of the project was to demonstrate solidification techniques that could be used to prepare HLW for transportation to, and disposal at, a federal repository. The DOE contracted with West Valley Nuclear Services, Co., Inc. (WVNS), a subsidiary of the Westinghouse Electric Company, to manage the project. The WVDP is located approximately 30 miles south of Buffalo at the site of the former nuclear fuel reprocessing plant at West Valley, NY. The plant was constructed in 1966 and ceased operations in 1972, after reprocessing approximately 640 metric tons of spent nuclear fuel. About 2,300 cubic meters (600,000 gal.) of high-level liquid plutonium extraction (PUREX) process waste was neutralized with sodium hydroxide and stored in an underground carbon steel tank. A sludge, consisting mostly of ferric hydroxide, precipitated to the bottom of the tank leaving a relatively clear liquid supernatant above the sludge. The primary radioactive isotope in the liquid was cesium-137. The other radioactive isotopes, primarily strontium-90, became part of the sludge. A reprocessing campaign for fuel containing thorium produced roughly 30 cubic meters (8,000 gal.) of acidic thorium extraction (THOREX) process waste which was stored in a separate underground stainless steel tank.

2 The supernatant portion of the HLW was pretreated by zeolite ion-exchange between 1988 and This process removed more than 99% of the cesium-137, allowing the remaining supernatant phase, that contained roughly 88% of the nonradioactive sodium, to be processed as low-level waste. Between 1991 and 1994, the remaining PUREX sludge underwent a series of washes to remove interstitial sulfate salts. The wash liquid was processed through the ionexchange columns. Both the decontaminated supernatant and the sludge wash solutions were processed, solidified in a cement matrix, and temporarily stored as low-level radioactive waste. The combination of supernatant and sludge washing pretreatment reduced the projected number of canisters required by nearly a factor of ten. Upon completion of the waste pretreatment, the cesium-loaded, ion-exchange media and the acidic THOREX wastes were combined with the washed PUREX waste to create a single homogeneous waste feed stream for the vitrification process. Between 1984 and 1989, the WVDP successfully performed a Functional and Checkout Testing of Systems (FACTS) program. The FACTS program demonstrated the ability of a full-scale vitrification process to produce high-quality glass on a production schedule. Approximately 150,000 kilograms (165 tons) of borosilicate glass was produced and 37 tests were performed using nonradioactive isotopes to produce a waste glass as close as practical to the projected HLW glass form. Lessons learned from FACTS led to improved designs on process components including: the submerged bed scrubber, the canister turntable, equipment to abate oxides of nitrogen in the melter off-gas, melter nozzle supports and nozzle liners, the melter off-gas nozzle film cooler, and the melter feed sampling equipment. Portions of the full-scale test facility (i.e., the concentrator feed makeup tank, the melter feed hold tank) were reassembled and reused in the final operating facility. Additionally, a Scaled Vitrification System (SVS) was designed, built, and operated in order to understand and evaluate the exact chemistry controls needed to process WVDP waste. The WVDP feed is a nitrate-based slurry. If the melter feed slurry is too reducing, metals may precipitate out of the mix and accumulate on the bottom of the melter. Significant metal accumulation within the melter could allow a short circuit between the melter electrodes, effectively ending the useful life of the melter. If the slurry is too oxidizing, gas generation could lead to foaming within the melter. Foam generation precludes adequate process control of the glass pool. At the WVDP, oxidation-reduction control is accomplished through the addition of sugar to the feed slurry. In the melter, sugar reacts with nitrated slurry and heat to form oxides of nitrogen that exit the melter in the off-gas stream. This reaction promotes reduction of oxygen in the molten glass, alleviating the concern for foaming. SVS proved to be a valuable asset in developing an acceptable range for oxidation-reduction indicators in the slurry feed. After construction of the Vitrification Facility (VF) was complete, an integrated test program was executed by WVDP test engineers. Testing progressed from component and subsystem demonstration first using water, then with nonradioactive slurry, and finally with fully integrated system test runs to produce filled canisters of nonradioactive glass. The performance testing incorporated extensive use of existing standard operating procedures and developmental operating procedures to allow for continual procedure refinement, procedure validation, and to provide operating experience and training opportunities for personnel prior to radioactive

3 operations. The integrated test runs culminated with approval for radioactive operations in June The first phase of the HLW vitrification campaign, representing the processing of approximately 85% of the total estimated curie inventory, was completed on June 10, See Table I for a production summary through this milestone. TABLE I WVDP VITRIFICATION PROCESS PHASE I OPERATING STATISTICS (As of 10 June 1998) 71% Plant Availability Y ~12,100 feeding hours Y ~ 4,950 nonfeeding hours 211 Canisters Filled Y 436,546 kilograms of glass produced Y 90.27% average fill height 68 HLW Transfers Y Over 9.46 million curies (Cs-137 and Sr-90) VITRIFICATION PROCESS OVERVIEW The goal of the vitrification process is to convert the HLW from its initial sludge/liquid form into a borosilicate glass waste form contained within stainless steel canisters. The canisters can then be stored temporarily in the High-Level Waste Interim Storage Facility (HLWISF) in the existing Main Plant (see figure 1). The following section provides a summary description of the vitrification process. The HLW is transferred from the Waste Tank Farm (WTF) HLW tank (Tank 8D-2) into the VF Concentrator Feed Makeup Tank (CFMT) where it is combined with an in-cell recycle stream from the off-gas scrubbing Submerged Bed Scrubber (SBS). CFMT samples are collected and the contents are concentrated by evaporation, removing excess water. The amount of glassforming chemicals to be added to the HLW is determined by analyzing the waste composition and predicting the volume of waste to be processed - one slurry batch in the CFMT usually produces just over three canisters of waste glass. Once ready, the chemicals are transferred into the CFMT and the CFMT is again sampled to ensure the feed composition will result in target glass characteristics. Following verification of the desired feed composition, the mixture is transferred into the Melter Feed Hold Tank (MFHT). The melter is pulse-fed from the MFHT using an air displacement slurry (ADS) pump.

4 Figure 1. Vitrification Process Flow Diagram In the slurry-fed ceramic melter (SFCM), the water will evaporate from the feed and the remaining solids will calcine. The calcined wastes and glass-formers accumulate on the surface of the glass pool to form a crust, referred to as the cold cap. The size of the cold cap is reflected in the temperature of the air space above it (melter plenum) and is managed by controlling the addition rate of the slurry feed. The feed rate is periodically manually adjusted to maintain the plenum temperature between 400 o C and 600 o C. The melter has two side electrodes and one bottom electrode, creating three circuits to pass electrical current through the glass pool and maintain its temperature within the range of 1,100 o C and 1,200 o C. The molten glass is periodically airlifted into a stainless steel canister. The canister is positioned under the melter with the canister turntable - a four-position, four-canister device - which allows two canisters to cool, while one canister is available for removal/replacement, and a fourth canister is being filled. During an airlift, the pour stream is visually monitored by closed circuit TV. The glass level within the canister is visually monitored using an infrared level detection system (ILDS), where a shielded infrared camera scans the canister surface to detect increased surface temperature and translates that information into a graphic image for the operator. Glass production rate typically exceeds 35 kilograms per hour, resulting in a normal canister fill time of roughly 58 hours. The cooled canister is removed from the turntable and moved to the weld station, where a stainless steel lid is welded to the canister. From the weld station, the canister is then moved to the canister decontamination station, where decontamination of the canister surface is accomplished by chemical etching. Welded and decontaminated canisters are moved from the Vitrification Process Cell to the HLWISF where they are placed into racks for interim storage.

5 During melting, steam, volatile elements evaporating from the glass pool, and feed particles entrained in the process off-gas will be vented to the off-gas treatment system. The first stage of the process is the SBS, where off-gases are quenched and particulate is scrubbed through a submerged bed of ceramic spheres. After the SBS, the off-gas is drawn through a high-efficiency mist eliminator (HEME) to remove mist and fine particulate. It is then heated and passes through a high-efficiency particulate air (HEPA) filter to remove particulate. At this point, the off-gas is essentially free of radiological pollutants. It then passes into another building via an underground trench, where oxides of nitrogen (NOx) are abated by catalytic reaction with ammonia. A final stage of HEPA filtering is also provided in this building prior to venting the off-gas to the environment through the plant stack. The primary process vessels are all maintained at relative vacuums by the vessel vent system. The vessel vent gas passes from a header through a condenser until it joins the off-gases prior to the HEME preheater. The vessel vent system also provides a means to bypass the SBS in the event the melter off-gas line becomes plugged. A canister load-in facility is provided for introducing canisters into the processing facility. Canisters are inserted horizontally through a cylindrical shield door into the Equipment Decontamination Room (EDR). The canisters are then upended and placed on the transfer cart using a crane. The radio-controlled transfer cart transports them into the primary process cell, where they are moved into a canister storage rack for eventual loading into the turntable. OPERATIONAL EXPERIENCES AND LESSONS LEARNED Melter Operation During startup testing with nonradioactive glass, the WVDP melter experienced chronic blockages in the glass pour stream. The blockages were the result of a failed barrier between the melt chamber and the discharge chamber, allowing glass migration and subsequent buildup at the bottom of the discharge chamber. The melter was repaired and placed back into service. Just prior to radioactive operations, excessive production and an accumulation of very thin glass fibers (angel hair) led to blockages in the glass pour stream. Prior to radioactive operations, the glass buildup was removed and a flow-reducing orifice was installed to reduce the airflow from the discharge chamber to the main melt chamber. Limiting the airflow through the discharge chamber seemed to successfully reduce angel hair production to an acceptable level. Angel hair formation, however, has not been, nor will it ever be, completely eliminated. Periodic accumulation of angel hair in the melter discharge port has been successfully overcome by rebalancing the discharge heater loads to maximize the temperature around the discharge port and melt the accumulated glass from the chamber. Shortly following the start of radioactive operation, a significant restriction in the melter off-gas piping (prior to the SBS) developed. The restriction was suspected to be caused by the collection of dried melter feed adhering to the inside surface of the piping. Remote radiation surveys showed the buildup to be occurring at an acute (45 degree) elbow just a few feet from the melter.

6 Design modifications incorporated to mitigate the blockage included the following: A water flush line was installed into the melter air injection line. Water is periodically introduced into the off-gas stream to steam clean the piping. A flow-restricting orifice was installed in the air line supplying motive force air for the melter ADS pump. Reduction in the airflow rate reduced the atomization of the melter feed as it was pulsed into the melter plenum. This made it less likely for the feed material to be swept into the off-gas piping. As melter operation progressed through the first phase of HLW processing, the electrical resistance of the molten glass pool decreased significantly. The electrical resistance values associated with each of the three electrode circuits are presently less than one third of their initial values observed at the start of radioactive processing. This phenomenon has been attributed to an accumulation of electrically conductive deposits, consisting mainly of ruthenium, rhodium, and palladium, within the WVDP melter. The shape of the melter cavity is essentially an inverted, truncated, pyramid. WVDP experience with the FACTS melter and supporting data from the testing of other joule-heated melters indicates that the conductive sludge layer may be accumulating primarily in the dihedral corners of the WVDP melter, forming preferential current paths between the melter electrodes. As relatively more current passes through these deposits, a lower percentage of the total electrode current is available to maintain molten glass temperature, leading to higher required electrode current demands. A mathematical model was created to predict the future trend of melter resistance. The model predicts, as shown in figure 2, that the rate of decrease in melter electrical resistance slows considerably over time. The actual observed trend in resistance correlates well with the predicted values. Furthermore, since the HLW tank sludge is the source of these noble metals, future metal deposition is limited to the remaining, limited, inventory in the HLW tank and should not impact continued operation of the WVDP melter.

7 Figure 2. Corner Model Validation The throughput capability of the WVDP melter requires a low average slurry feed flowrate. The pulsing design of the ADS pump has provided a low average feed rate while generating high slurry velocity and high back pressure. These characteristics have resulted in essentially no plugging problems normally encountered with slurry transport. The original ADS feed pump was finally replaced in July 1998 after more than two years and nearly a million operating cycles. In January of 1997, airborne contamination at levels slightly above ambient air was detected in an operating aisle outside the Vitrification Cell. Subsequent engineering assessment determined that the contamination had been transported from the melter discharge section, driven by a minor pressure transient in the melter, along electrical conduits that supply power to the melter discharge heaters. The barriers installed to seal the conduits from the ex-cell operating aisle had not been totally effective in preventing the contamination migration. To prevent future pressurization of the conduits, the conduits were vented within the Vitrification Cell. In addition, HEPA filter vents were installed on the ex-cell portion of the conduits to prevent any further spread of contamination. No further problems associated with these conduits have occurred to date. Feed Preparation and Sampling Initial waste transfers from the Waste Tank Farm (WTF) to the VF had lower curie contents than anticipated (i.e., less than 100,000 curies cesium-137/strontium-90 per batch). An additional

8 sludge mobilization pump was installed into an existing HLW tank riser in a low-flow area within the tank where solids were apparently mounding. Careful management of the HLW tank level also increased batch curie inventory. Leakage of seal water from one of the mobilization pumps added significant water to the HLW tank whenever it was operated. By periodically decanting the accumulated liquid, the solids concentration in the transfers to the VF was optimized while minimizing the transfer pump plugging problems associated with pumping slurries. Subsequent waste transfers typically exceeded 200,000 curies. Approximately 18 months after the initial waste transfer, the pump that transfers slurry from the HLW tank to the VF failed, effectively idling the vitrification process. Replacement of the nearly 50-foot long, highly contaminated, pump proved to be a complex task. Prior to removal of the failed pump, the pump and riser in which it was located were sprayed down in an effort to lower radiation levels near the work area. Radiation levels near the top of the riser were higher than initially predicted. This was presumably due to the process fluid, which is used to lubricate the close-tolerance shaft bearings, spraying the inner wall of the riser as it exits the bearing interface. The replacement pump was modified to direct this process fluid lubrication flow back to the tank without spraying the upper riser walls. The replacement pump was also designed with sideintake suction ports instead of a bottom-intake suction port to more efficiently transfer the suspended solids in the HLW slurry. In the feed preparation process, HLW is combined with SBS recycle liquid in the CFMT and then concentrated by evaporation prior to the addition of glass-forming chemicals. The batch cycle time associated with sample analysis and chemical mixing has been reduced since the start of radioactive operations on two fronts. First, due to the relative consistency of the waste composition, slurry acceptance engineers now generate chemical premix recipes well in advance of the final (post waste analysis) recipe. Independent verification of recipe calculations has virtually eliminated the need for corrective (and time consuming) chemical shims. Secondly, the turnaround times for laboratory chemical analyses have been drastically reduced. Early analytical cell mockup training and experience accumulated by the lab technicians has reduced the time required for various feed batch analyses by more than 50 percent, saving more than 66 hours on the total batch cycle time. However, as the curie content in the HLW tank diminished, the HLW stream became more dilute. This eventually required multiple HLW transfers and evapora-tion cycles to complete a single feed batch. This has increased the batch preparation time and led to periods where melter feed was idled while awaiting the completion of the next batch. Cell maintenance activities were then sched-uled in concert with these idle periods in order to reduce the impact of maintenance on process operating efficiency. A wider range of glass compositions has been studied to expand the range of acceptable glass recipes. These compositions include glasses with little or no uranium or thorium, which have become relatively more depleted from the waste as the tank empties. With a broader range of acceptable compositions, the necessity for small adjustments in the feed will virtually be eliminated. The Vitrification Cell in-line slurry sampling units were modified based upon input from plant operators to improve the ease of remote operation. Typical sampling time was reduced from

9 eight hours to one hour. Routine (daily) water flushing of the in-cell sampling lines was incorporated to preclude line blockages. One phenomenon observed during feed preparation involved the generation of NO x gases during the addition of glass-forming chemicals to the concentrated HLW in the CFMT. As the acidic chemicals were mixed with the basic HLW slurry, NO x gases were generated at rates that exceeded the system s capacity to abate the gases, resulting in visual emissions from the exhaust stack. Efforts to mitigate the emissions included: Establishing an air sweep on the CFMT to dilute the gases as they were generated. Ensuring that the catalytic converter bed was saturated with ammonia prior to the start of chemical mixing (i.e., when the melter is idle and not generating NO x gases). Making a separate addition of dilute nitric acid to the HLW slurry prior to the chemical mixing. Whereas the addition of glass-formers must be made at a flow rate of roughly 60 gallons per minute; to prevent solids settling in the transfer pipe, dilute nitric acid could be added at significantly lower flow rates, limiting the gas generation rate to a controllable range. Vessel level, density, and pressure indications within the Vitrification Cell are provided by bubbler assemblies. A control scheme that includes automatic periodic air blow-downs has maintained the bubblers in the CFMT and MFHT relatively free from plugging. Recent modifications to the automatic blow-down cycles have included a small (~100 milliliters) injection of demineralized water into the bubbler probes at the start of the cycle and are proving to be effective in maintaining clear bubbler lines. Canister Fill The WVDP HLW canisters are required to be filled at least 80% full. Originally, canisters were typically filled to the 85% level. Through the field experience gained with the video imaging capabilities of the ILDS system, canisters are now normally filled approximately 90% full. While the ILDS-based levels are backed up by mass balance calculations, the accuracy provided by the ILDS system for determining the canister fill level has been outstanding. The final (reported) canister fill verification is performed by direct measurement of the glass level (using a measuring stick device) at the vitrification weld station. The variation between the canister fill level provided by direct measurement and that provided by the ILDS is usually within 1%. A significant lesson learned is that a large canister opening is vital to the success of the WVDP vitrification process. The glass pour stream must fall a distance of over 1.5 meters from the end of the trough before entering the canister. Lateral movement in the pour stream can (and does) occur due to melter pressure fluctuations, variations in the airlift flowrate, or effects of air inleakage to the discharge chamber. The large canister opening (0.42 meters in diameter) has proven to be an effective design feature to accommodate pour stream deflections.

10 Canister Handling and Remote Operations Canister closure (lid welding) using a pulsed gas tungsten arc process has become routine. There have been only a few canister lid welds that were just outside the normal range for weld parameters. The welds were corrected by simply reperforming the automatic weld. The ease of reworking a canister lid by rewelding has proven to be one of the major advantages of this system. A significant challenge was offered when it was determined, after 177 canisters had been welded, that the current readings associated with the closure welds were inaccurate. Acceptance of production canister closure welds is based upon measured welding parameters, including weld electrode current values, being within ranges established by qualification testing. An unmonitored ground loop present in the welder control console resulted in unmeasured current being delivered to the weld electrode. Through subsequent testing, the magnitude of the previously unmonitored current was determined to be less than 10 percent of the total current. Additional test closure welds were made and evaluated (visual inspections, leak tests, burst tests, and metallographic examinations) to ensure that previous actual weld currents provided acceptable closure welds. Canister decontamination operations have also become routine. Early difficulties with tank temperature control and cooling coil pressurization were corrected with minor piping modifications and control loop tuning adjustments. The process generates significant volumes of neutralized decontamination and rinse solutions. Through careful scheduling of vitrification evolutions, decontamination and rinse solutions have been effectively managed by reintroducing them into the melter feed batch makeup process. Most of the down-time associated with the WVDP vitrification campaign can be attributed to maintenance and repair of remotely operated equipment. Maintenance of components within the Vitrification Cell is accomplished primarily by remote replacement. Most of the electrical and mechanical (piping) connections are made within the cell using replaceable jumpers with 3-jaw connectors called PUREX TM connectors and an electrical impact wrench suspended from a crane. Even with six, fixed, closed-circuit television (CCTV) cameras; one movable CCTV camera; and three CCTV cameras fixed on the process crane; good visual access to in-cell components is not always available. A significant lesson learned from the WVDP vitrification experience is the importance of proper visual access to in-cell components, especially those components with moving parts. A recent opportunity with replacement of the ADS feed pump bears this out. As the new feed pump and air-supply jumper were being reassembled in the cell, the PUREX connectors were not sealing properly. Because of the lack of adequate visual access, it was necessary to remove the pump from its tank, mount it on a fixture (specifically designed and fabricated to support this maintenance task) located directly in front of a cell viewing window, and attempt to mate the pump with its air-supply jumper. It was only then that a mechanical interference was identified on the ADS pump and corrective actions could be undertaken. The installation and fit of the rigid, in-cell jumpers requires close dimensional tolerances on jumper design and fabrication. WVDP experience with remote jumper placement has led to

11 design changes that provide some flexibility in the jumpers. By providing a spring-loaded support plate to connect the PUREX connector to the jumper s support structure, the PUREX connector is allowed to float as much as 19 mm (0.75 in.) in a direction perpendicular to the cell wall, while still maintaining the necessary rigidity of the jumper to allow proper remote handling. This has provided an additional margin of error in the design and fabrication of spare jumpers. This is especially important when considering that spare jumpers cannot be test fitted prior to placing them into the Vitrification Cell. SUMMARY On June 19, 1996 the Secretary of the Department of Energy approved the initiation of radioactive operations at the WVDP. On July 5, 1996 the first HLW canister was filled. As of June 10, 1998 the WVDP Vitrification Facility had operated at an overall availability (actual time feeding slurry to the SFCM) of approximately 71% and produced 211 HLW canisters of HLW glass. Over 9.46 million curies of cesium-137 and strontium-90 have been removed from underground storage tanks. This accomplishment represents just over 85% of the total activity to be processed and signifies the completion of the first phase of HLW processing at the WVDP. The vitrification process systems have been performing as designed since initiating radioactive operations. Exhaustive functional and integrated testing using waste simulant has limited the unscheduled downtime within the WVDP vitrification process. While the opportunities involving remote maintenance on vitrification system equipment continue to provide challenges, strong problem-solving skills and efficient use of unscheduled process downtime for scheduled maintenance has contributed greatly to the high system availability at the WVDP.