Assessment of Supercritical Water Oxidation System Testing for the Blue Grass Chemical Agent Destruction Pilot Plant (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

2 First-of-a-Kind Testing INTEGRATED OPERATIONS SCWO operating conditions; that is, variations from the planned normal operating conditions specific to each agent The committee’s first task in this study was to deter- being processed. The abnormal operating conditions tested mine whether the supercritical water oxidation (SCWO) included off-normal flow rates, temperatures, and hydro- system met its first-of-a-kind (FOAK) testing goals and lysate composition. These abnormal conditions and their whether the testing supports the systemization of the SCWO impact on the system are addressed in more detail in the fol- system at Blue Grass Chemical Agent Destruction Pilot lowing sections of the report. Also, as discussed below in the Plant (BGCAPP). The FOAK testing involved integrated section “Maintenance,” errors were induced to test system operation of one of the three SCWO reactor trains and was response to abnormal operating conditions. Despite numer- conducted at the system contractor’s facility in San Diego, ous minor maintenance issues, a system availability greater California.1 The functionality, operability, performance, than the target of 76 percent was achieved. The availabilities and maintainability testing were all completed. In summary, achieved during the 100-hr tests were as follows: GB, 88.5 four test campaigns were completed: integrated operations percent; VX, 96.5 percent; and H, 100 percent.2 (no simulant), blended simulants of sarin (GB) and energet- Both the chemical processes in the SCWO reactor and ics hydrolysates, blended simulants of VX nerve agent and the SCWO process equipment were reviewed to assess energetics hydrolysates, and blended simulants of H and en- whether the objectives described in the SCWO FOAK test- ergetics hydrolysates. Simulants of the agent and energetics ing presentation were met. Because various scheduled and hydrolysates were blended at the system contractor’s site to unscheduled maintenance events are inevitable in such a have chemical compositions as representative of the actual complex system, the contractor expressed test success as agent and energetics hydrolysates feeds as possible. Tables achieving an overall system availability goal of 76 percent. 2-1, 2-2, and 2-3 give the composition of the simulated The FOAK test objectives were to provide full-scale verifica- blended hydrolysates. The chemical reactions that occur on tion for H, GB, and VX blended hydrolysate simulant3 feeds the SCWO reactor are standard oxidation reactions. For the of the following items: interested reader, a detailed listing of the SCWO effluents resulting from the processing of H, GB, and VX hydrolysates •  eactor rinsing requirements, if any; R can be found in BPBGT, 2007. •  he replacement of sulfuric acid with elemental sul- T In the final FOAK tests, each of these simulants was fur in the GB campaign; tested in aggregate for between 96 and 100 hr of operation at •  ariations in a limited number of feed compositions V a feed rate of approximately 1,000 lb/hr. Stable and predict- and flow rates; able temperature and pressure control was achieved during •  eactor liner and thermowell corrosion; and R SCWO processing. Destruction of organics was confirmed to • Ability to meet 76 percent overall availability.4 less than 10 ppm by measuring total organic carbon (TOC) in the effluent. Any salt buildup was sufficiently minimized 2Personal communication between Steven Mantooth, Blue Grass Chemi- so as to not affect system availability goals. Variations in a cal Agent Destruction Pilot Plant program office, and James Myska, NRC limited number of compositional and operational parameters study director, on April 25, 2013. were explored. These variations were tests of abnormal 3Blended hydrolysate consists of a mix of agent and energetics hydrolysate. 4Kevin Downey, advanced process system manager, General Atomics, 1General Atomics is the system contractor for the SCWO systems that “SCWO Factory Acceptance Testing,” presentation to the committee on will be used at BGCAPP. January 7, 2013. 15 

16 ASSESSMENT OF SUPERCRITICAL WATER OXIDATION SYSTEM TESTING FOR THE BGCAPP TABLE 2-1  Simulated Blended GB and Energetics • Elemental sulfur additive; Hydrolysates • Limited variations in feed composition; Constituent Wt-% • High-pressure air compressor; • High-pressure quench water; Deionized water (H2O) 80.64 Sodium sulfate (Na2SO4) 9.14 • H  igh-pressure hydrolysate feed, high-pressure fuel Sodium chloride (NaCl) 3.81 feed; Sodium hydroxide (NaOH) 1.08 • Gas and liquid sampling and analyses; Dimethylmethyl phosphonate (C3H9O3P) 1.66 • Temperature and pressure letdown and effluent; 35 percent Hydrochloric acid (HCl) 1.38 • High-pressure water feed, preheater; Sodium formate (NaCHO2) 0.85 Sodium fluoride (NaF) 0.52 • M  aintainability (seven anticipated maintenance ac- 100 percent Isopropanol (C3H8O) 0.33 tivities were evaluated during FOAK testing); and Sodium nitrite (NaNO2) 0.28 • Integrated operations. Sulfur (S) 0.26 tri-n-Butylamine (C12H27N) 0.06 SOURCE: BPBGT, 2013. OVERVIEW OF TEST RESULTS This section discusses the test results that the committee believes to be the most significant. These are either areas of TABLE 2-2  Simulated Blended VX and Energetics particular criticality to the overall SCWO process, or areas Hydrolysates where significant challenges were encountered. This is not Constituent Wt-% intended to be a comprehensive review of the FOAK test report, so not every aspect of the FOAK testing and results Deionized water (H2O) 74.18 Sodium sulfate (Na2SO4) 7.40 is discussed here. If something worked well or a particular 56 percent Sodium isethionate (C2H5NaO4S) 4.34 challenge was judged by the committee to be minor based on Sodium hydroxide (NaOH) 1.77 its experience, it is not addressed here. The committee had Sodium chloride (NaCl) 4.00 access to a wide range of technical material while conduct- 85 percent Diethanolamine (C4H11NO2) 2.15 ing its work, including test plans and reports, process flow Dimethylmethyl phosphonate (C3H9O3P) 2.06 Denatured ethanol (C2H6O) 1.29 diagrams, briefings, and discussions with BGCAPP and 35 percent Hydrochloric acid (HCl) 1.25 system contractor staff. Please see the list of this material Sodium formate (NaCHO2) 0.68 in Chapter 1. 98 percent Sulfuric acid (H2SO4) 0.65 Sodium nitrite (NaNO2) 0.23 Corrosion SOURCE: BPBGT, 2013. The severe conditions of the SCWO process and the generation of molten salts have a corrosive effect on the in- TABLE 2-3  Simulated Blended H and Energetics terior of the reactor. This corrosion requires a relatively inert Hydrolysates metal for use in the reactor design. The committee assessed Constituent Wt-% the ability of the titanium liners and thermowells to withstand Deionized water (H2O) 88.34 the reaction conditions within the SCWO unit and evaluated Sodium hydroxide (NaOH) 3.04 whether the corrosion rates noted during testing support a Thiodiglycol (C4H10O2S) 2.52 SCWO availability of 76 percent or greater. Sodium chloride (NaCl) 1.38 The liner of the SCWO reactor is 10 ft long, has an Ferric chloride (FeCl3) 2.92 inner diameter of 7.625 in., and a wall thickness of 0.5 in. Sodium sulfate (Na2SO4) 1.11 Sulfur (S) 0.43 The material of construction is grade 2 titanium. The liner Sodium formate (NaCHO2) 0.18 is not a pressure vessel but is contained in one with an an- Sodium bicarbonate (NaHCO3) 0.09 nular space between the pressure vessel and the titanium SOURCE: BPBGT, 2013. liner. The liner mitigates the potential for penetration of the hydrolysate into the annular space and, thus, corrosion of This report evaluates the results of the FOAK from two the pressure vessel. Corrosion of the pressure vessel itself perspectives. First, the effectiveness of each step was as- is not an issue since failure of the liner results in shutdown sessed with respect to the individual operation of that step of the reactor. The liner and thermowells are subject to the (e.g., Did the air compressor work?). Then, each step was highest temperatures and pressures of the components of evaluated with respect to any overall impact on the process the SCWO system. (e.g., Did the reliability of the air compressor affect system Periodic reactor liner change-out is required because availability goals?). The processes/subsystems evaluated of corrosion. The highest liner corrosion occurs in the first include the following: several diameters from the top of the reactor and the high- 

FIRST-OF-A-KIND TESTING 17 TABLE 2-4  Summary of Reported Maximum Average Corrosion Data Reported from FOAK Testing Maximum Average Liner Maximum Average Thermowell Test Conditions Corrosion Rate (mil/hr) Corrosion Rate (mil/hr) GB 24-hour low sulfur 1.6 1.9 GB 24-hour high sulfur 1.8 1.6 GB 8-hour +2 percent sulfur 1.0 1.5 GB 8-hour –2 percent sulfur 1.0 1.7 GB 24-hour low velocity (800 lb/hr) 0.56 0.55 GB 24-hour high velocity (1,200 lb/hr) 1.2 1.7 GB, 1,200°F 1.8 1.9 GB, 1,150°F 1.2 1.7 VX 8-hour +2 percent HCl 1.2 0.67 VX 8-hour +2 percent NaCl 1.0 0.67 VX 24-hour low velocity (800 lb/hr) 0.97 0.94 VX 24-hour high velocity (1,200 lb/hr) 0.95 1.2 H 24-hour high velocity (1,200 lb/hr) 0.01 0.04 GB performance test, 100-hr (1,000 lb/hr) 1.1 1.5 VX performance test, 100-hr (1,000 lb/hr) 0.82 1.1 H performance test, 100-hr (1,000 lb/hr) 0.006 0.09 SOURCE: BPBGT, 2013. est thermowell corrosion measured occurs within the first reactor after the test. The failure of the thermowell had no 4.5 reactor diameters (4.5 × 7.625 in., or 34.3 in.) from the impact on operations.5 top of the reactor. The corrosion rates of the titanium liners Corrosion rates for the reactor liners were measured dur- were determined using ultrasonic measurements and com- ing and following the nominal 100-hr FOAK tests. They were paring them to a baseline measurement. Corrosion rates determined using ultrasonic measurements and comparing for thermowells were determined by direct micrometer the measurements to a baseline measurement taken before measurements of the disassembled components. Corrosion the liner was first used. Either 320 or 328 measurements were varied among the various FOAK tests as detailed in Tables made at each sampling, i.e., either 40 or 41 axial measure- 1-1 and 1-2 of the preliminary draft of the Test Report for ments and 8 circumferential locations, depending on the test. Supercritical Water Oxidation (SCWO) First-of-a-Kind The corrosion rate for each individual liner was determined (FOAK) Test (BPBGT, 2013). The reported corrosion data by dividing the change in liner thickness by the number of are compiled in Table 2-4. The highest corrosion rate noted hours of hydrolysate operation. The maximum average cor- was 1.9 mil/hr for thermowells in the high-temperature GB rosion rate for liners in a given test condition was the highest test. The lowest corrosion rate noted was 0.01 mil/hr for of the averaged 8 circumferential points for each liner at any the liner in the H tests. The maximum allowable corrosion distance along the length of the liner. is 67 percent of the initial wall thickness for the liners and 80 Thermowell corrosion was measured along the ther- percent for the thermowells, and replacement intervals for mowell’s thinnest orientation using a micrometer. Measure- liner and thermowell replacement were determined accord- ments were made at 1-in. intervals along the length of the ingly. Liner change-out requires a shutdown of the SCWO thermowell. The difference in thickness between 0.50 in. reactor for 10-12 hours while the liner is replaced. Ther- (the original diameter of the thermowell) and the microm- mowell change-out takes approximately 6 hours. While it eter measurement was divided by 2. The maximum average is desirable to prevent the failure of thermowells, such a corrosion rate was determined by using the highest of the failure would not necessitate an immediate shutdown of average of corrosion rates measured at the same point along the reactor because there are four redundant thermowells each of the thermowells and was determined by dividing and they are internally pressurized to the working reactor the maximum average corrosion by the number of hours of pressure with compressed air to prevent the contents of the simulated hydrolysate operation. SCWO reactor from migrating up a ruptured thermowell. The maximum average corrosion rates for the liners (A liner failure would, however, necessitate a shutdown.) across all the FOAK tests in simulated VX hydrolysate were During the performance test with simulated GB hydroly- sate, thermowells did fail after 85 hours of operation and 5Dan Jensen, advanced process system program manager, General Atom- broke apart. The pieces were found at the bottom of the ics, “SCWO System Testing and Lessons Learned,” presentation to the Committee on Chemical Demilitarization on April 10, 2013. 

18 ASSESSMENT OF SUPERCRITICAL WATER OXIDATION SYSTEM TESTING FOR THE BGCAPP 1.2 mil/hr, similar to those predicted from the Technical Risk Finding 2-2. The 300-hr liner change-out interval recom- Reduction Project testing of 1.2 mil/hr (BPBGT, 2005 and mended for the operational GB hydrolysate campaign is 2013). The maximum average corrosion rates for the simu- adequate if the operating temperature remains at the recom- lated GB hydrolysate varied as functions of flow velocity mended 1150°F and the flow rate is 1,000 lb/hr. The 300-hr and temperature. In addition to testing at the 1,000 lb/hr flow change-out interval is not adequate if a higher operating rate planned for use at BGCAPP, tests were conducted for temperature (1200°F vs. 1150°F) is reached in the reactor for all simulated hydrolysates at a flow rate of 1,200 lb/hr, and significant lengths of operation. It also may not be adequate also at 800 lb/hr for the simulated GB and VX hydrolysates. for flow rates above 1,000 lb/hr. For the simulated GB hydrolysate, the high velocity nozzles/ flow rates doubled the maximum average corrosion rate of Finding 2-3. The 400-hr liner change-out interval recom- the liner from 0.56 mil/hr to 1.2 mil/hr and tripled the maxi- mended for the VX nerve agent campaign is not supported by mum corrosion rate of the thermowell from 0.55 mil/hr to the worst-case corrosion data in the draft FOAK preliminary 1.7 mil/hr. Temperature also influenced corrosion rates in the test report. case of GB testing. The highest maximum average corrosion rates during GB testing occurred when running the SCWO Recommendation 2-1. The Blue Grass Chemical Agent De- at 1200°F and was 1.8 mil/hr for the liner and 1.9 mil/hr for struction Pilot Plant staff should shorten the liner change-out the thermowells. For the tests conducted at 1150°F, the maxi- period for VX processing from the 400 hr recommended in mum average corrosion rates were 1.2 mil/hr for the liner and the test report until the corrosion rates under actual operating 1.7 mil/hr for the thermowells (BPBGT, 2013). Visual exami- conditions are verified. Two hundred hours would be a better nation of the liner after exposure to the hydrolysate simulant initial change-out period. indicated that while the corrosion damage is localized, in the form of shallow rivulets extending longitudinally along the Corrosion of the thermowells is generally more severe wall of the liner, there were no exceptionally severe areas of than corrosion of the liners. The wall of one thermowell corrosion. The FOAK testing report recommended the fol- suffered full penetration (and accordingly failure) in ap- lowing liner and thermowell change-out periods to achieve proximately 76 hr during the GB simulant FOAK test.6 The the desired SCWO reactor availability goal: thermocouples contained in the thermowells provide critical process feedback and so it is obviously desirable that ther- • GB: liners, 300 hr; thermowells, 75-100 hr; mowell mechanical integrity be maintained. • VX: liners, 400 hr; thermowells, 130 hr; and Since the thermowell walls are about 180 mil thick, •  : liners, no change-out required; thermowells, H the measured maximum average corrosion rates will result 1,600 hr (BPBGT, 2013). in the near or actual penetration of the thermowell walls before 100 hours of operation at the higher corrosion rates According to the committee’s calculation, a 400-hr measured. Mechanical stresses from the turbulent flow in change-out period for the liners during VX processing would the reactor will only further shorten the thermowell lifetime result in the liners being almost entirely consumed. As noted by causing mechanical failure when only small amounts of above, the maximum average corrosion rate observed dur- titanium remain. While the thermowells met availability re- ing VX simulant testing was 1.2 mil/hr (BPBGT, 2013). A quirements with their current material and design, BGCAPP change-out period of 400 hr at this corrosion rate would has expressed a desire to increase the operating life of the result in the loss of 480 mil, or 0.48 in., of titanium. That is thermowells to bring their change-out periods more in line 96 percent of the liner thickness. A shorter change-out period with the change-out periods for the liners, thus increasing the would seem to be called for by these data. A change-out availability of the SCWO system and reducing maintenance period of 200 hr would result in only 240 mil, or 0.24 in., requirements.7 of the liner being consumed at the maximum average corro- BPBGT (2013) suggested several alternatives to extend sion rate. At 1.2 mil/hr, the maximum allowable 67 percent thermowell operating life, including evaluation of corrosion for the liners would be reached in approximately 279 hours. Thus a 200 hour change-out period would be •  eactor operation at <1150°F during GB blended R conservative until systemization and operational experience feed processing; can be used to adjust that number. •  ther titanium materials, including hardened tita- O nium alloys, Finding 2-1. The FOAK testing was adequate to establish the expected operating lifetime for the titanium liners. This 6BGCAPP project staff indicated that the thermowells had been exposed expected operating lifetime is sufficient to enable continued to various simulants for as many as 30 hours before the performance test safe operations and adequate operational throughput except was conducted. 7PEO ACWA, “Supercritical Water Oxidation Tech Review,” briefing to for VX hydrolysate. the Program Executive Officer, Assembled Chemical Weapons Alternatives supercritical water oxidation technical review, April 22, 2013. 

FIRST-OF-A-KIND TESTING 19 • Hastelloy or an alternative alloy; an alternative geometry for the titanium cap would likely •  ncreasing thermowell wall thickness by increasing I necessitate further testing to ascertain whether these changes the overall diameter to 0.75 in. from 0.5 in.; would affect SCWO processing operations. • Application of coatings to the thermowells; and • Alternative nozzle/insert geometries. Finding 2-7. Based on the test data, the committee believes that availability goals can be met with grade 2 titanium ther- Also, the BGCAPP project staff indicated in discussions with mowells of appropriate wall thickness. the committee that redesign of the thermowells, by increasing the diameter of the thermowells to 0.75 in. from the current Recommendation 2-3. The corrosion issue with thermo­ 0.5 in., would not be a major design change and even recom- wells should not be resolved by attempting to qualify another mended the change. This would lengthen the thermowell material. Grade 2 titanium should be used as planned. Other change-out periods and thus increase the availability of the potential strategies for extending thermowell life (i.e., coat- SCWO reactors. The project staff also raised the possibility ings, alternative geometries, reduced operating temperature, that a more corrosion-resistant alloy could be selected for the and hardened titanium alloys) need not be explored since thermowells. However, any new alloy that might be considered simply increasing the wall thickness of the thermowells will would have to be qualified for the corrosive environment of extend their operational life and thus reactor uptime. the SCWO reactor. There is no guarantee that changing al- loys would result in improved thermowell performance. The Elemental Sulfur Additive application of coatings to the thermowells was also presented as an option, but this would meet the same challenges as se- It was originally planned to add sulfuric acid to blended lecting a new alloy. Processing GB at less than 1150°F would GB/energetic hydrolysate before SCWO processing. The likely necessitate further testing to ensure that operations at sulfuric acid would perform two functions. First, along with the selected temperature would produce the required results, hydrochloric acid, it would neutralize the basic hydroxide because it would represent a change in the operating condi- used in the hydrolysis reaction. Second, it would provide tions of the reactor. The use of alternative nozzle and insert sufficient sulfate to form the sodium sulfate/phosphate geometries would also likely necessitate additional testing to eutectic system upon SCWO processing of the neutralized ascertain whether they would affect SCWO processing opera- hydrolysate.8 When sulfuric acid was used with GB hydro- tions since the reaction in the SCWO reactor is dependent to lysate, however, the resulting pH was low enough to cause some extent on geometry (BPBGT, 2013). concerns about the possible reformation of GB. This concern led to the testing of elemental sulfur as a replacement for Finding 2-4. Corrosion thinning of the thermowells (ac- the sulfuric acid in an attempt to raise the pH of the GB hy- companied by mechanical failure when minimal material drolysate. Elemental sulfur would eliminate the pH concern remains) is a critical failure path for the reactors and is ex- associated with sulfuric acid while still forming the desired pected to be the most frequent maintenance issue associated eutectic system. The oxygen in the SCWO environment acts with continuous operations. on the sulfur to form SO3–, which hydrates to form sulfuric acid. This sulfuric acid is neutralized by the caustic feeds. Recommendation 2-2. The diameter of the thermowells This is the source of the sulfate that forms and maintains should be increased to at least 0.75 in. from the current di- the eutectic. Four items were investigated when testing the ameter of 0.50 in. by increasing the wall thickness. The Blue elemental sulfur additive: Grass Chemical Agent Destruction Pilot Plant project staff should use existing corrosion data to estimate the lifetime of 1.  oes the elemental sulfur additive perform equiva- D these new thermowells. This operation is a simple modifica- lently to sulfuric acid during systemization by tion that would lengthen the thermowell life during operations similarly creating a eutectic salt concentration in the and help minimize maintenance-related reactor shutdowns. SCWO reactor that allows for simplified salt handling (i.e., less reactor rinsing?) Finding 2-5. As noted above, the corrosion of thermowells is 2. Can the replacement of the sulfuric acid with elemen- greater for higher feed velocities than for lower feed veloci- tal sulfur avoid unnecessarily low pH conditions? ties. Another way to address the corrosion of thermowells 3. Does the solid elemental sulfur remain suspended and could be by reducing the feed velocity. dispersed so that the feed has uniform composition, as would the solute sulfuric acid? Finding 2-6. Any new alloy used for the thermowells, or any coatings applied to the thermowells, would have to be quali- 8A eutectic system is a chemical mixture that has a freezing/melting point fied before use in the SCWO system. There is no guarantee that corrosion performance would improve. Using a lower lower than any of the chemicals in the mixture would normally have. The effect in this application is to keep the salts molten and flowing through processing temperature for sarin (GB) processing and using the SCWO reactor. 

20 ASSESSMENT OF SUPERCRITICAL WATER OXIDATION SYSTEM TESTING FOR THE BGCAPP 4.  ould addition of elemental sulfur change any cor- W tial for the sulfur to change the corrosion parameters of rosion parameters? the blended GB hydrolysate feed, several sets of tests were conducted. In one set of tests the sulfuric acid was replaced To avoid unnecessarily low pH, testing was carried out by enough elemental sulfur to reach the target pH of 6. This with the sulfuric acid replaced by elemental sulfur, avoid- resulted in a maximum average corrosion rate for the liner ing formation of strongly acidic hydrolysate. The minimum of 1.6 mil/hr and 1.9 mil/hr for the thermowells. When the acceptable pH to prevent GB reformation was determined sulfuric acid was completely replaced with elemental sulfur to be 5. An initial target pH of 6 was chosen to provide a in a second set of tests, the maximum average corrosion safety margin (BPBGT, 2012). The elemental sulfur was rate for the liner was 1.8 mil/hr and the maximum average added in the form of sublimated flowers of sulfur with a 200 corrosion rate for the thermowells dropped to 1.6 mil/hr. In mesh particle size (BPBGT, 2013). When 83 percent of the a subsequent set of tests, the amount of sulfur in the feed sulfuric acid was replaced with 0.2 wt-% elemental sulfur, stream was varied by ±2 percent. These variations produced the neutralized hydrolysate simulant had a pH of 6. When all no change in the maximum average corrosion rate for the the sulfuric acid was replaced with elemental sulfur, the pH liners (1.0 mil/hr). During the −2 percent sulfur test, the of the neutralized hydrolysate simulant was approximately maximum average thermowell corrosion rate was 1.7 mil/hr, 7.5. The simulants containing elemental sulfur showed pro- and it was 1.5 mil/hr during the +2 percent sulfur test. In the cess equivalency to simulants with sulfuric acid. The simu- final 100-hr test with simulated blended GB and energetics lants with elemental sulfur were processed in an equivalent hydrolysate, the maximum average corrosion rate for the manner with similarly complete oxidation of the organics liner was 1.1 mil/hr and for the thermowells was 1.5 mil/hr in the FOAK testing. This addresses the first two concerns (BPBGT, 2013). These corrosion data are summarized above (BPBGT, 2013). in Table 2-4. The third concern above pertains to uniform distribu- It is also useful to look at this in the context of the tion of the sulfur in the blended hydrolysate. The solubility simulated H blended hydrolysate recipe from Table 2-11 in of nonpolar sulfur in water is low (2 × 10–8 mol S8/L H2O, BPBGT, 2013. The 0.43 wt-% solid sulfur addition amounts or about 5 µg/L) and likely to be slightly lower in the hy- to less net sulfur than is present in the simulated blended drolysate salt solution (Boulegue, 1978). As sulfur does not H hydrolysate. The corrosion rates in the SCWO testing of dissolve in aqueous solutions, agitation is needed to keep it blended H simulant (which had the highest sulfur concen- in suspension so that compositionally it behaves equivalently trations of all the simulants) were the lowest of the three to a solution during SCWO processing. Consistent values for feedstocks (simulated blended VX, GB, or H hydrolysate SCWO effluent pH and the lack of salt buildup during GB simulants). Therefore, sulfur does not significantly add to surrogate processing confirm sufficient uniformity of the the SCWO corrosion rates (BPBGT, 2013). blended surrogate feed.9 The draft FOAK report provided a description of the Finding 2-8. The elemental sulfur additive did not have any heating and mixing systems to keep the elemental sulfur adverse impact on FOAK testing and successfully resolved dispersed. The initial FOAK test heaters caused sulfur ag- the issues it was intended to. glomeration due to melting of sulfur (sulfur begins to melt at about 240°F). Agglomeration of the sulfur could lead to Recommendation 2-4. Elemental sulfur additive should be inhomogeneous composition within the batch, clogging of used instead of sulfuric acid additive. the hydrolysate feed apparatus, and, potentially, a noneutec- tic salt mixture (off-spec sulfate/phosphate ratio) within the Variations in Feed Composition SCWO. To keep this from happening, the original FOAK heaters were replaced with lower wattage elements, which There is a direct relationship between any additions to resulted in lower temperatures and eliminated the problem the SCWO feed streams and the chemistry within the SCWO. of agglomeration. The potential effects of any variability in SCWO feed com- Any variation in the testing caused by the use of el- position could be a concern and should be evaluated as the emental sulfur did not disrupt the SCWO processing. As no process parameters were evaluated for a specific simulant salt buildup was present in the SCWO reactor, any minor composition. For example, some variations in feed compo- compositional changes which may have occurred during the sition did result in changes in corrosion, as summarized in SCWO processing did not affect the ability of the reactor to Table 2-4. Variations in any of the compositions or concentra- efficiently move salts through the reactor. tions within the hydrolysate feed beyond those tested during Regarding the fourth concern listed above, the poten- FOAK testing could yield SCWO conditions different from 9Kevin Downey, advanced process system manager, General Atomics, “SCWO FOAK Test Status,” presentation to the committee on February 4, 2013. 

FIRST-OF-A-KIND TESTING 21 those evaluated in the FOAK tests. As part of FOAK testing, compressor, manifold, and feed systems worked as designed variations in the SCWO feed composition and flow rate for a and were able to feed the liner purge, reactor feed, and limited set of parameters were tested as follows:10 thermowell/blowdown line purges as long as the compres- sor remained operational. That said, the compressor had •  lemental sulfur (for GB hydrolysate simulant): E numerous failures throughout the testing period. Events 0.20 wt-% and 0.26 wt-%, and ±2 percent from the such as a high-pressure cooling loop failure, expansion amount planned to be used at BGCAPP. tank bladder failures, and a low oil level/demister failure •  Cl: ±2 percent from the amount planned to be used H repeatedly interrupted the test cycles. Makeshift measures at BGCAPP. such as wiring dummy transducers, warrantied repairs, and •  ydrolysate flow into the SCWO: 1,000±200 lb/hr. H other workarounds enabled target availability to be met, but operations were not as smooth as desired. Additionally, all The sulfur concentration varies from low (hydrolysate the failures occurred in new equipment. feed at pH = 6) to high (hydrolysate feed at pH = 7.5). As an Another concern related to the high-pressure air com- example of how significant the impact of feed variations can pressors is the oil-water separator (OWS). While four air be, the apparently minor difference of 0.06 wt-% between compressors are planned for use at BGCAPP (three opera- the low and high sulfur additive concentrations represents a tional, one spare), only one OWS is planned for use with all change of 1.5 pH units, or a factor of approximately 32 in three of the operational air compressors. There were several acid concentration. The final FOAK tests were completed instances where workarounds in the OWS system or in the at nominal 1,000 lb/hr simulant feed rates. The variations lubrication system were necessary for the system contrac- in SCWO feed that were tested did not have any negative tor to be successful in keeping the system operational and impacts on the SCWO process itself. achieving the operational goals of the test (BPBGT, 2013). Upon systemization, the hydrolysate storage area will The committee’s concern is that if there are problems with hold the contents from multiple 750-gal energetics neutrali­ the OWS during systemization and BGCAPP operations, the zation batches. The combined energetics hydrolysate batches entire SCWO process could be shut down because all three will be diluted, neutralized, filtered to remove aluminum air compressors would have to be shut down. There could salts, and then blended with agent hydrolysate to create the also be a risk of sympathy shutdowns, where perturbations SCWO feed. The agent hydrolysate will itself be a mixture in one system could ripple across the other systems, caus- of multiple hydrolysis batches and the additives to those ing shutdowns. This seems to present a risk for a significant batches. All of this will serve to homogenize any minor single-point failure. variations in the feedstock that might come from initial varia- tions in the munition agent fills or energetics. Additionally, Finding 2-10. There was a much higher failure rate for the neutralization to a specific pH also dampens any fluctuations high-pressure air compressor than would be expected from a caused by variations in the original feedstocks. mature system. These failures repeatedly interrupted FOAK testing. The multiple issues with the air compressor, which Finding 2-9. The limited feed variations tested during FOAK is a single, critical component of the SCWO system, are of testing did not result in any negative impacts to the SCWO concern to the committee. process. Variations other than those tested during FOAK testing could be a concern and are addressed in Chapter 3. Recommendation 2-5. The Blue Grass Chemical Agent Destruction Pilot Plant project staff should review the main- tenance and start-up plan for the air compressors so that they Functionality During FOAK Testing are maintained to industry standards. The project staff should This section of the report discusses specific pieces of use the time between compressor delivery and systemization SCWO process equipment. Not every piece of equipment to run the compressors, to discover any problems similar to or process is discussed here. Only those pieces and circum- those which occurred during FOAK testing, with the goal of stances that stood out to the committee are discussed here. obtaining reliability comparable to that of the supercritical water oxidation reactors. High-Pressure Air Compressor Finding 2-11. The oil-water separator appears to present the The SCWO system needs high-pressure air to provide risk of a critical single-point failure that could shut the entire the oxygen for the SCWO process. The air is provided by SCWO process down. a modified commercial compressor. The high-pressure air Recommendation 2-6. In addition to the action recom- mended in Recommendation 2-5, the Blue Grass Chemical 10Kevin Downey, advanced process system manager, General Atomics, Agent Destruction Pilot Plant project staff should consider “Status of SCWO FOAK Testing,” briefing to the Committee on Chemical Demilitarization on December 4, 2012. mitigation strategies such as having an additional oil-water 

22 ASSESSMENT OF SUPERCRITICAL WATER OXIDATION SYSTEM TESTING FOR THE BGCAPP separator (OWS) in parallel that could run should the main some equipment modifications were performed. All of the OWS require maintenance. Additionally, a spare OWS could actions and troubleshooting activities were expected and be kept on hand. normal for FOAK testing. No issues are anticipated with this equipment. High-Pressure Quench Water Finding 2-13. Particulates in the blended hydrolysate simu- In SCWO systems in general, salts are always a poten- lant, which was not filtered, caused the check valves to stick tial issue in the supercritical portions of the reactor, yet they during testing. mostly dissolve in the subcritical portions of the reactor. If salts are not handled properly, they can lead to excessive Gas and Liquid Sampling and Analyses downtime due to frequent rinsing or clogging of the SCWO reactor. In the SCWO system to be used at BGCAPP, salts The FOAK test process was monitored by gas analyzers move through the reactor and dissolve in the warm sub- measuring O2, CO, CO2, and total hydrocarbons as well as critical solution as the effluent leaves the SCWO reactor. by liquid analyzers measuring TOC and pH. These monitors High-pressure quench water will be injected into the end of are planned for use at BGCAPP as part of systemization and the SCWO reactor to facilitate salt removal. The test results operations. The gas sampling is used to confirm that sufficient indicated that no rinse was needed to remove residual salts O2 is present for the oxidation process (the O2 analyzer) and from the SCWO reactor. The consistent conductivity of the that complete oxidation is occurring (the CO analyzer) and effluent reveals a well-controlled system. The fact that a reac- to monitor the effluent concentration of oxidized organics tor rinse is not necessary directly addresses the original test (the CO2 analyzer) and the concentration of residual hydro- objective, to determine the rinsing requirements. carbons (the total hydrocarbon analyzer). The liquid effluent Despite the fact that reactor rinses were not necessary is analyzed to confirm destruction of the target organics in the concentration of sodium hydroxide in the H blended the hydrolysate and destruction of the isopropanol (the TOC simulant was reduced by 8 percent to “reduce build up of analyzer) to a target of less than 10 ppm organic carbon in the salts in the reactor” (BPBGT, 2013, p. 87). Additionally, the effluent. The pH of the effluent is measured as a simple check temperature of the H processing was raised to 1275°F from on processing consistency (the pH analyzer). All these mea- the original target of 1200°F. surements delivered relatively consistent results. Some scatter and occasional outlier points were observed, but the general Finding 2-12. The processing of the modified blended H and specific trends of the gas concentrations were clear. hydrolysate simulant recipe at 1275°F proceeded smoothly The TOC measurement takes several minutes and is an with little corrosion and complete oxidation. important indicator of SCWO reaction completion. Regard- less of minor, normally expected process variations, if the Recommendation 2-7. The Blue Grass Chemical Agent TOC value is low then the organics have been destroyed. Destruction Pilot Plant supercritical water oxidation system During different FOAK tests, the TOC analyzer inlets be- test plan should ensure that the recipe for the blended H came plugged. A backup TOC analyzer was then used to hydrolysate is the same as the recipe used in the FOAK tests. obtain the TOC measurement. High-Pressure Hydrolysate Feed Temperature and Pressure Letdown and Effluent Heat Exchanger The hydrolysate flows through check valves as it is pressurized for delivery to the SCWO unit. Particulates in The gas-rich effluent salt solution in the SCWO reac- the blended hydrolysate simulant, which was not filtered, tor must be cooled from the SCWO operating temperature caused the check valves to stick, reducing the efficiency to a warm temperature and the pressure reduced to levels of the pressurization operation. In preliminary lessons appropriate for post-SCWO operations in a controlled man- learned, system contractor personnel recommend that ner. The quench water flow at the bottom of the SCWO ­Hastelloy check valves be replaced by a more wear-resistant reactor reduces the temperature of the SCWO effluent from alloy such as Stellite, because the pump must handle solids supercritical to subcritical values and dissolves the entrained inherent to the SCWO process such as iron oxide and sulfur salts. Subsequent heat exchangers, pressure letdown valves, (BPBGT, 2013). gas-liquid separators, and two-stage gas pressure letdown The final function of the high-pressure hydrolysate feed valves take the liquid to its final effluent state. Preliminary pump is to flush the piping/reactor. The flush conditions are indications are that the high-temperature, high-pressure reac- mild compared to the hydrolysate feed conditions during tor effluent is cooled, reduced in pressure, and stripped of its operation of the SCWO, 109°F for more than 10 min. No excess gas in an efficient manner. The target temperature is issues are anticipated with the flushing operations. 140°F at the outlet of the heat exchanger, which was achieved Minor adjustments to control set points were made and during testing (BPBGT, 2012). 

FIRST-OF-A-KIND TESTING 23 High-Pressure Water Feed, Preheater 5. High-pressure gas-liquid separator solids cleanout; 6. Preheater thermocouple replacement; and The high-pressure water feed and the preheater bring the 7. Liquid pressure letdown valve inspection. reactor and its contents to supercritical condition. When the isopropanol is added as the reactor warms (at 700-801°F), it All of the maintenance activities were performed in the reacts with the feed air and is oxidized. The preheater is then expected manner. The titanium liners were replaced and the allowed to cool. The air and isopropanol flows are increased reactor returned to fully operational status within 12 hr. The until the reactor reaches 1099-1150°F, when temperature is thermowells were replaced and the reactor returned to fully then controlled by monitoring O2 in the effluent and adjust- operational status within 7 hours. Numerous adjustments ing the effluent O2 to 5 percent by varying the compressed were made to equipment and to control set points. These air feed.11 The reactor comes to rough thermal equilibrium adjustments are a normal result from learning to efficiently as the heat of reaction becomes sufficient to maintain the operate FOAK equipment. Other maintenance activities temperature and pressure conditions necessary for efficient were routine industrial tasks and were accomplished using SCWO. Then, hydrolysate is added to start the destruction standard industrial procedures/practices. of the organics it contains. During the testing, the preheater/ high-pressure water pump was able to reliably initiate oxi- Finding 2-14. The overall first-of-a-kind testing objectives dation in the SCWO reactor. The various high-pressure air for the supercritical water oxidation system to be used at the compressor failures gave the operators additional practice Blue Grass Chemical Agent Destruction Pilot Plant were in shutting the reactor down and bringing the reactor up to met. The processes and subsystems, inclusive of mainte- operational (steady-state) conditions. nance activities, performed individually and collectively to meet test objectives and exceed the target of 76 percent Isopropyl Alcohol Fuel Issues availability. There were problems during FOAK testing with accu- rate SCWO fuel composition. The contractor did not provide REFERENCES adequate specifications for the fuel, and the incorrect fuel Boulegue, J. 1978. Solubility of elemental sulfur in water at 298 K. Phos- was delivered and used during FOAK testing (70 percent iso- phorous and Sulfur and the Related Elements 5(1): 127-128. propyl alcohol by weight instead of 70 percent by volume). BPBGT (Bechtel Parsons Blue Grass Team). 2005. Technical Risk Reduc- This error and attempts to correct for the error resulted in tion Project (TRRP) 07 and 09 Report on Supercritical Water Oxidation Blended Feed Performance Tests, Rev. 0, April. Richmond, Ky.: Blue some problems in achieving SCWO conditions. The commit- Grass Chemical Agent Destruction Pilot Plant Program Office. tee includes this as an example of the simple things that could BPBGT. 2007. SCWO Building Water Recovery - R.O. Unit Process Flow create problems during SCWO systemization and operation Diagram, October 2. Richmond, Ky.: Blue Grass Chemical Agent if they are not watched. D ­ estruction Pilot Plant Program Office. BPBGT. 2012. Test Plan for Supercritical Water Oxidation (SCWO) First- of-a-Kind (FOAK) Test, Rev. 1, July. Richmond, Ky.: Blue Grass Chemi- Maintainability cal Agent Destruction Pilot Plant Program Office. BPBGT. 2013. Test Report for Supercritical Water Oxidation (SCWO) First- Thirty-two system shutdowns and start-ups were includ- of-a-Kind (FOAK) Test, April, Preliminary Draft. Richmond, Ky.: Blue ed in part of the FOAK test plan resulting from the simulation Grass Chemical Agent Destruction Pilot Plant Program Office. of a wide variety of system errors. Simulated errors included low and high levels in various pieces of process equipment, high temperatures beyond what is safe, and low and high flow levels for various feeds and effluents (BPBGT, 2012). There were also unplanned shutdowns due to real system errors, such as the failure of a thermowell, discussed above. During these shutdowns seven maintenance activities were demonstrated during FOAK testing: 1.  eactor liner change-out (reactor liners were re- R moved, inspected, and/or replaced 17 times); 2. Reactor thermowell change-out; 3. Reactor feed nozzle removal and replacement; 4. Effluent heat exchanger inspection; 11Kevin Downey, advanced process system manager, General Atomics, “SCWO FOAK Test Status,” presentation to the committee on February 4, 2013.