Evaluation of Demonstration Test Results of Alternative Technologies for Demilitarization of Assembled Chemical Weapons: A Supplemental Review (2000)

The General Atomics process uses a modified version of the baseline disassembly process and cryofracture of projectiles for munitions access. The agent and energetics are destroyed by hydrolysis. The hydrolysate is then treated by supercritical water oxidation (SCWO). Metal parts are subjected to caustic hydrolysis processing followed by 5X thermal treatment. Dunnage is shredded, mixed with caustic, and destroyed by SCWO.

Demonstration tests were conducted for the following operations:

• energetics rotary hydrolyzer (ERH)

• dunnage shredding and hydropulping

• SCWO

The objectives of the demonstration tests of the ERH are listed below (DOD, 1999b):

• Demonstrate the effective dissolution of aluminum and energetics in fuzes and bursters, as well as propellant in rocket motors, to allow downstream processing in the continuously stirred tank reactor, SCWO reactor, and heated discharge conveyor.

• Determine the deactivation of the energetics in fuzes and bursters and the propellant in rocket motors.

• Validate the retention times for aluminum and energetics in fuzes and bursters and propellant in rocket motors

• Characterize the gas, liquid, and solid process streams.

The General Atomics demonstration tests involved several different munition items and energetic materials. Complete destruction, (i.e., below the detection limit) was achieved for tetryl in M557 fuzes and M14 bursters and for tetrytol (tetryl/TNT) in M6 bursters. However, the following problems arose during the handling of other energetics (General Atomics, 1999a):

• Small quantities of fuze-train components remained unhydrolyzed; these were destroyed in the hot muffle furnace.

• Unhydrolyzed energetic material adhered to a flight drum during an M83 burster (RDX/TNT) validation test and burst into flame. ¹ (The technology provider claimfs that this was an artifact of the test; the flights in the ERH were designed to hold solids and liquids for sampling rather than to drop them into the hydrolyzing solution. An appropriate flight design will be used in the full-scale ERH).

• Excessive boiling and foaming was reported with the M83 burster, which could cause difficulties in processing.

• RDX and HMX were above the detection limit in the liquid analyte.

• Hydrolysis of M28 propellant in the motor casing was slower than anticipated; the NaOH solution concentration had to be raised to 12M. (The technology provider has suggested cutting the propellant into smaller pieces).

• During the processing of M28 rocket propellant, a yellow substance (identified as N-nitrosodiphenylamine) was generated and coated much of the interior of the explosive containment cubicle. The technology provider indicated that the coating was caused by the ventilation flow in that particular ERH test unit. The ventilation was sized to dilute hydrogen to below the lower explosive limit and was clearly inadequate to prevent

• fugitive emissions from the ERH. The technology provider reported that in the full-scale system, sufficient ventilation flow would be provided to prevent fugitive emissions (General Atomics, 1999b). The yellow material would be scrubbed from the ERH ventilation flow, and the scrubber solution would be combined with energetics hydrolysate and processed through the SCWO reactor. The committee was concerned that the proposed solution to the problem could result in the accumulation of similar energetic by-products in other parts of a full-scale system.

The committee’s earlier report contained the following finding concerning hydrolysis of energetics (NRC, 1999):

Finding GA-2. Hydrolysis of energetics at the scales proposed by the technology provider is a relatively new operation. Chemically, it is possible to hydrolyze all of the energetic materials; however, the rate of hydrolysis is limited by the surface area and, therefore, depends on particle size. (Smaller particles are more desirable because they have a higher surface-to-volume ratio.) The proposed method of removing and hydrolyzing the energetics appears to be reasonable, but further testing is required to determine the hydrolysis rates and to confirm that throughput rates can be achieved.

The demonstration tests substantively confirmed this finding. The test results demonstrated that the ERH could deactivate and dissolve the energetics and aluminum found in M557 fuzes and M83 bursters and could deactivate the energetics found in M6 and M14 bursters in two to four hours. Test data on the M28 rocket motor sections show that a residence time of 10 hours at 12M caustic concentration and 230°F were required for complete hydrolysis of the M28 propellant.

The demonstration program did not include the treatment of agent-contaminated solids. In the opinion of the committee, the ability of the ERH system to hydrolyze solid pieces of propellant supports the conclusion that similar treatment could successfully clean contaminated solids to a 3X condition. ² However, the demonstration results cannot be considered conclusive evidence that the required processing rates could be consistently achieved.

The committee’s earlier report included the following finding (NRC, 1999):

Finding GA-3. The rotary hydrolyzer appears to be a mature reactor configuration that is well suited for this application.

Although no test data on the reaction rate were provided, the tests did qualitatively demonstrate that the ERH could destroy energetic materials. However, some results indicate that the ERH did not completely wet the energetics with hydrolysis solution, which allowed some solid energetic material to exit the ERH before hydrolysis was complete. The explanation given by the technology provider (i.e., the shape of the flights) and the design modification proposed by the provider to address this problem (i.e., modification of the pitch and shape of the flights) should, in the committee’s opinion, decrease the amount of unexposed solid material that passes through the ERH.

No tests were conducted on the hydrolysis of energetics contaminated with agent; however, because of the long residence time in the ERH, the committee believes that chemical agent exposed to the caustic hydrolysis solution in the ERH would be hydrolyzed. Nevertheless, because the exact manner in which agent might penetrate energetic materials is not known, there is still some question as to whether chunks of unhydrolyzed energetic material, such as those that were found in the residue from the ERH, would be truly agent free. Agent embedded in the energetic solids might not have been exposed to the caustic solution and, hence, might not have reacted.

The purpose of the demonstration tests of the dunnage shredding/hydropulping system (DSHS) was to show that solid wastes (wooden dunnage, DPE suits, and butyl rubber) could be adequately reduced in size and pulped to a pumpable mixture. The objectives of the demonstration testing are listed below (DOD, 1999b):

• Validate that the shredders and hydropulper can adequately prepare the dunnage for downstream processing in the SCWO reactor.

• Qualitatively evaluate the operability (especially material handling) of the shredder/hydropulper unit operations.

• Validate that the shredders can process 1,000 lb/hr of pallets and, separately, 250 lb/hr of plastics.

Several commercial shredders identical in size to the units proposed for the full-scale system were used to achieve the size reduction of the solid materials of interest. In the initial report, the committee had stated the following (NRC, 1999):

Finding GA-4. Shredding of dunnage and injection of the slurry directly into a SCWO system is a new and unproven process. While General Atomics claims to have developed a proprietary pump capable of pumping the slurry at high pressures, it has not been tested under the

intense solids loading anticipated. Furthermore, the injection of large amounts of solid material, including wood shreds, cut-up nails, and complex organic materials, such as pentachlorophenol and other wood preservatives, into the SCWO system has not been demonstrated. Considering the difficulty SCWO reactors have encountered with deposition of solids when liquids are treated, the committee believes that this application of SCWO may encounter significant difficulties. (At the time of this writing, processing of solids with SCWO was being performed as part of the ACWA demonstrations.)

The individual components of the DSHS had been tested previously in their respective applications but had not been used collectively in the configuration used for the demonstration test program. Consequently, numerous, albeit surmountable, problems were encountered (e.g., wood “nesting” in the hammer mill and micronizer feed chutes and inadequate magnetic separation of metal from the shredded DPE suits prior to processing in the granulator). The technology provider was able to control both system and feed variables well enough to achieve the targeted feed processing rates and obtain the proposed objective for size reduction (< 1 mm for wood and < 3 mm for plastics). The 3-mm plastic material product was processed through a sieve to separate material that was less than 1 mm that could be fed to the SCWO reactor. The fullscale SCWO system will have larger feed nozzle diameters that should be capable of accepting the plastic dunnage material as shredded (i.e., without the need for sizing to less than 1 mm) (General Atomics, 1999b).

The demonstration tests did not validate that the hydropulper could consistently produce material that was smaller in diameter than the goal objective of 1-mm; however, the tests did determine that the hydropulper could blend energetics hydrolysates with size-reduced wood to yield a uniform, pumpable slurry for processing in the SCWO reactor.

The mass balance reported for the two validation test runs of the micronizer while processing wood pallets showed a 5.4 and 6.3 percent deficit (General Atomics, 1999a). The deficit was attributed to “Presumably . . . the loss of water due to heatup in the micronizer. ” This loss is not a problem for pallets that are not contaminated with agent. When contaminated wood is processed, however, the water vapor released could contain vaporized agent, and the gas stream will have to be managed accordingly.

The duration of the shredding tests was too short to allow for an evaluation of the long-term efficacy of this process. The demonstration was highly labor intensive and, because it was performed on uncontaminated material, did not require that the operators work in full protective clothing. Therefore, it cannot be concluded that a full-scale system would provide similar levels of materials segregation without further development of the process. For example, one of the technology provider ’s conclusions is that the metal parts in DPE suits would have to be manually cut out in glove boxes prior to processing and then decontaminated to a 5X condition in the metal parts furnace (General Atomics, 1999a). Because this step (which is necessary for successful processing) was not performed during the demonstration tests, the committee could not assess its efficacy.

The hydrolysates of energetic materials provided by the PMACWA were prepared using 12 percent sodium hydroxide (as specified in the technology provider’s proposal). The DREs from hydrolysis of energetic constituents of Comp B, tetrytol, and M28 propellant all exceeded 99.999 percent, except for the nitrocellulose component of the M28 propellant. The latter was set to measure a DRE of no greater than 99.988 because the analytical method has a high detection limit.

The SCWO system was demonstrated to validate its capability to destroy Schedule 2 and other organic compounds produced from agent hydrolysis. The objectives of the demonstration testing of the SCWO reactor concerning agent hydrolysate products are listed below (DOD, 1999b):

• Validate that the SCWO reactor can eliminate the Schedule 2 compounds present in the agent hydrolysate feed.

• Validate that the agent hydrolysis process and the SCWO reactor can achieve a DRE of 99.9999 percent for HD, GB, and VX.

• Demonstrate the long-term operability of the SCWO reactor with respect to salt plugging and corrosion.

• Characterize the gas, liquid, and solid process streams from the SCWO reactor.

The SCWO system was also demonstrated to validate its capability to destroy organic compounds from energetic hydrolysis products and to demonstrate the feasibility of destroying shredded dunnage. The demonstration tests included the following objectives (DOD, 1999b):

• Validate that the ERH, continuously stirred tank reactor (CSTR), and SCWO can achieve a DRE of 99.999 percent for tetrytol, Comp B, and M28 propellant.

• Determine the impact of the aluminum from the ERH process on SCWO operation.

• Determine how well organics in the shredded dunnage are oxidized in the SCWO reactor.

• Characterize the gas, liquid, and solid process streams from the SCWO system.

The committee’s initial report contained the following finding concerning General Atomic’s use of SCWO (NRC, 1999):

Finding GA-5. All of the findings in the [1998] NRC report, Using Supercritical Water Oxidation to Treat

Hydrolysate from VX Neutralization, apply to the General Atomics system.

The demonstration confirmed this finding (see ). Although the SCWO system successfully destroyed organic compounds in the liquids, the results did not demonstrate that the system is capable of operating without frequent shutdowns for repair or cleaning. This uncertainty could affect the system’s ability to treat the numbers of munitions located at a storage site within a reasonable length of time. For the destruction of agent and energetics hydrolysates and dunnage, the SCWO system performed reasonably well. However, corrosion and salt plugging both raised concerns about reliable long-term operation.

Operationally, the validation test runs for agent hydrolysate (all liquid feeds) proceeded smoothly, except for inconsequential leaks at some joints. Validation test runs for the energetic hydrolysates and dunnage feeds showed that these can be processed successfully, provided that aluminum hydroxide is removed from the feed (it caused severe plugging). Safety issues pertaining to the removal of aluminum hydroxide are noted later in the chapter.

Thus, the demonstration confirmed the concerns of the committee (and of another NRC committee that had previously evaluated the use of SCWO to treat VX hydrolysate) about the durability of components and the materials of construction in the highly corrosive SCWO system environment (NRC, 1998, 1999). Although the demonstration plan had called for the use of a platinum-lined reactor, because of problems encountered in fabricating the platinum liner, an unlined Inconel ^TM 718 SCWO reactor was used. This contributed to the corrosion and plugging of the downstream components with corrosion products (DOD, 1999b).

SCWO processing of the dunnage slurry was not demonstrated beyond a simple proof of concept. As described in the technology provider ’s report, a mixture of tetryl hydrolysate, aluminum hydrolysate, deionized water, phosphoric acid, micronized wood, granulated plastic (< 1 mm), ground activated carbon, and a stabilizing additive proprietary to the technology provider was fed to the SCWO reactor at an approximate rate of 6 kg/hr (General Atomics, 1999a). The committee concluded that this brief test constituted a proof of concept only and could not be considered a validation of the method.

The demonstrated treatment of shredded and slurried dunnage using SCWO resolved one of the committee’s concerns but raised new ones. The demonstration tests showed that the SCWO system’s pump can pressurize the slurry to the high pressure required for the SCWO reactor and that the SCWO reactor is capable of oxidizing the slurried dunnage. However, the testing did not demonstrate that tramp metal ³ would not prove to be a problem in extended operation. Furthermore, the demonstration tests of the SCWO system with dunnage feed was too short to demonstrate the long-term reliability of the system.

Finally, the demonstration tests used slurried solids of dunnage shredded to less than 1 mm (rather than less than 3 mm as proposed in the full-scale process), and the feed nozzles were smaller than those proposed for full-scale operation. Thus, the efficacy of the process with particles sized to full-scale specifications and larger nozzles was not demonstrated.

The demonstration tests revealed that additional processing steps to remove aluminum from energetics hydrolysate would be necessary to prevent plugging of the SCWO reactor. The technology provider has proposed using a neutralization and filtration process to remove aluminum hydroxide from the hydrolysate, with subsequent 5X treatment of the precipitated aluminum filter cake in an inductively heated metal parts furnace (General Atomics, 1999b). Aluminum hydroxide forms a very flocculent precipitate, however. Because this compound is also amphoteric, the pH will have to be carefully controlled and the precipitate carefully filtered. If other hazardous metal salts precipitate with the aluminum hydroxide, they may have to be treated under RCRA specifications.

The removal of aluminum hydroxide would require additional processing equipment, which would add to the maintenance and reliability burden of the plant and would increase worker maintenance time in DPE suits and opportunities for worker exposure to agent. This concern was raised in the committee’s initial report (NRC, 1999). It is repeated here to emphasize that modifications used in the demonstration tests would increase the potential of exposure.

The demonstration tests showed that condensable organics, such as nitroglycerine, will be evolved from the ERH and will be subsequently condensed and returned to the CSTR for hydrolysis. The committee notes that considerable care will be required to ensure that these condensable explosive materials are not initiated, thereby increasing the possibility of worker exposure to agent and damage to process equipment. The ERH demonstration tests using propellant feed also resulted in the release of volatile organic compounds (VOCs) into the explosive containment cubicle for the ERH. The walls of the cubicle were coated with this material as it condensed. As the technology provider noted, this experience reveals that the ERH design will have to control fugitive emissions (General Atomics, 1999b). The committee believes that the potential for worker exposure to agent would be increased during the maintenance of currently undefined control systems for fugitive emissions.

The technology provider also indicated that, to preclude dust explosions (which are extremely unlikely) in the micronizer component of the DSHS, additional safety

features will be required for the full-scale design of the system (General Atomics, 1999b). These features, too, would increase the opportunities for worker exposure to agent during maintenance.

In general, the demonstration tests revealed that more maintenance in DPE suits would be required, and, thus, the opportunities of exposure to agent by workers would be increased. At the baseline incineration disposal facilities operating on Johnston Island (in the Pacific Ocean) and at Tooele, Utah, workers in DPE suits are only allowed to remain in contaminated areas for two hours at a time and can only enter if another worker is present. In case of emergency, two more workers wearing protective clothing must be prepared to provide assistance (PMCD, 1998). Thus, an increase in maintenance in DPE suits can have a significant impact on productivity. Process design and the selection of reliable process equipment and materials, in conjunction with suitable training and procedures, should be used to minimize requirements for activities in DPE suits.

In the initial report, the committee concluded that the liquid effluent from the General Atomics process consists of pure water from the evaporator/crystallizer used to produce the solid filter cake (NRC, 1999). This effluent is essentially distilled water and should not pose a significant hazard to human health or the environment. The solid waste from the process, consisting of dried filter cake, was reasonably well characterized. The gaseous effluent from the SCWO process was not well characterized, however, and as a result, its hazardous characteristics could not be determined.

Tables 3.4-10 through 3.4-19 in the demonstration test report by General Atomics present some analytical results on the liquid and gaseous effluents from the SCWO reactors (General Atomics, 1999a). However, the reported characterizations are inadequate to determine if the solid filter cake could be stabilized adequately or to estimate the degree of risk to human health or the environment posed by the gaseous effluent from the SCWO process.

A further concern relates to the presence of sodium and other solid materials in the gaseous emissions from the SCWO reactor. The mechanism whereby solids are released into the gaseous effluents is not clear. One would expect that these inorganic materials would be found in the solid and liquid phases, but not in the gaseous phase. Small quantities of chromium in the gaseous emissions from the SCWO reactor are of potential concern for two reasons. First, it reinforces the importance of demonstrating the reliable operation of the platinum-lined reactor; second, it illustrates the need to test gaseous emissions from the SCWO system for particulates, as well as for gaseous contaminants. Chromium emissions reported in Table 3.5-11 of the technology provider ’s demonstration test report were at 3.1, 12.3. and 10.5 micrograms, respectively, during a five-hour test period for each of three test runs (General Atomics, 1999a). If the reported emissions pertain to chromium in the hexavalent form, the committee has serious concerns.

Table 3.4-8 of the demonstration test report by General Atomics shows that chemical analyses on VOCs and semi-VOCs were conducted on samples from the off-gas duct of the SCWO system during tests with HD hydrolysate tests (General Atomics, 1999a). The results of these measurements, however, do not appear to be adequate for evaluating the environmental impact of the process. Standard EPA methods for analysis of gaseous effluent samples generally produce full scans that can indicate the quantities of a large number of compounds of environmental concern. ⁴ These results, along with the results for emissions of metals (including chromium valency), can then be used to assess the environmental impact of a facility through accepted risk assessment methods (EPA, 1998a).

The committee’s earlier report included six required steps for implementation of the General Atomics overall technology package (NRC, 1999). These steps are reprinted below, followed by a description of the effects of the demonstration tests on them.

1. Conduct tests of the cryofracture process to ascertain if it provides better access to the agent cavity in projectiles and mortars then the baseline disassembly process.

Cryofracture was not part of the demonstration.

2. Sample and analyze air emissions from the demonstration system. The air emissions will have to be measured to a level of detail and accuracy that can be used for HRAs and environmental risk assessments required by EPA (1998a).

Some sampling and analyses of air emissions were conducted during the demonstration. However, additional data will be required to evaluate HRA and EPA emissions requirements.

3. Verify that energetic materials encased in metal (e.g., rocket or other munitions fragments) will be hydrolyzed.

The demonstration tests did verify that energetic materials encased in metal can be hydrolyzed. They also confirmed that the chemical reaction of the aluminum casings with the caustic solution is sufficient to gain access to and hydrolyze the contained energetic materials in the design residence time of the ERH.

4. Ascertain how well the SCWO process can handle high-solids materials (shredded dunnage).

The demonstration indicated that the SCWO process can handle materials with a high solids content (e.g., shredded dunnage). However, the SCWO system was not operated long enough to demonstrate reliable continuous operation.

5. Ascertain how well the SCWO system can treat hydrolysate containing large amounts of chlorides, sulfur, and phosphates on a continuing basis.

The ability of the SCWO system to treat hydrolysate containing large amounts of chlorides, sulfur, and phosphates on a continuous basis was not demonstrated.

6. Determine erosion and corrosion behavior of the components of the SCWO system.

General Atomics provided data on the types and quantities of metals found in the precipitates. Both the types and relative quantities matched those of Inconel^TM 718. These data provide a strong indication that Inconel^TM 718 was the source of the precipitates during the demonstration tests; they do not prove that other materials would not also form precipitates. In addition, the results do not confirm that a platinum-lined reactor could withstand the SCWO conditions and protect the underlying reactor wall during sustained operation.

Finding GA-1. Testing on the hydrolysis of energetic materials contaminated with agent will be necessary before a fullscale system is built and operated.

Finding GA-2. Testing will be required to verify that the larger diameter supercritical water oxidation (SCWO) reactor feed nozzles will be capable of accepting the dunnage material as shredded (i.e., without additional classification and segregation) and that the reactor will perform reliably under these conditions.

Recommendation GA-1. Operation of the size reduction and slurrying system, and long-term operation of the supercritical water oxidation (SCWO) reactor with slurry, should be conducted before proceeding with a full-scale system.

Recommendation GA-2. Before construction of a full-scale supercritical water oxidation (SCWO) system, additional evaluations of construction materials and fabrication techniques will be necessary because corrosion and plugging prevent continuous operation with the present design. If the new construction materials do not solve these problems, then alternative SCWO reactor designs should be investigated.

Recommendation GA-3. To determine the operability of the supercritical water oxidation (SCWO) reactor and the reliability of the materials of construction, long duration runs of a SCWO reactor should be conducted with slurry, with energetics hydrolysate, and with agent hydrolysate before full-scale implementation proceeds.

Recommendation GA-4. The efficacy and safety of the additional step to remove aluminum hydroxide from the hydrolysate produced from rocket propellants should be evaluated prior to construction of a full-scale supercritical water oxidation (SCWO) system.

Recommendation GA-5. Decontamination of solid munitions materials by flushing and immersion should be demonstrated prior to full-scale implementation.

Recommendation GA-6. The air emissions data from the demonstration tests should be used in a screening risk assessment. The results of the air effluent samples should be subject to (1) a human health risk assessment following the Human Health Risk Assessment Protocol (HHRAP) for Hazardous Waste Combustion Facilities from the Environmental Protection Agency (EPA) [EPA530-D-98-001(A,B,C)], and (2) an ecological risk assessment following a protocol that will be released by EPA in the very near future.