Review and Assessment of Developmental Issues Concerning the Metal Parts Treater Design for the Blue Grass Chemical Agent Destruction Pilot Plant (2008)

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.

4 Thermal Testing, Modeling, and Predicted Throughput of the Metal Parts Treater The objective for thermal testing and modeling of the profiles in the prototype TRRP MPT. The numeric desig- metal parts treater (MPT) was to demonstrate that the metal nations of the projectile locations and the location of the parts at all locations in the tray could be heated to 1000°F for thermocouples are shown in Figure 4-1. An example of the at least 15 minutes within a total time duration that supported temperature-time profile for thermocouples numbers 4 and 6 the design production rate. Heat-up times for projectiles located on projectile 14 and comparison with the CFD pre- located in the trays passing through the Technical Risk Re- dictions are shown in Figure 4-2. Figure 4-2 is representative duction Program (TRRP) MPT were measured and predicted. of the agreement between the model and experiment for final The key elements of this effort were as follows: versions of the CFD code, although some earlier versions of the code had greater differences. The thermocouples were • Thermocouple measurements of the surface tempera- usually placed on projectiles 6 and 7, because the modeling ture were made on a limited number of projectiles at indicated that these were the slowest to heat. Thermocouples specific locations in the tray as they passed through were also placed on projectile 14 because this location was the MPT and were heated. used in early testing at PCAPP. • Predictions of the temperature distribution in the Thermocouple measurements are subject to errors projectiles were made by using a computational created by (1) heat transfer in the thermocouple wires and fluid dynamics (CFD) thermal modeling program shielding, (2) the creation of junctions not in the intended that was compared with a previous model used in location, and (3) calibration. In the experimental TRRP tri- initial Pueblo Chemical Agent Destruction Pilot als, thermally sensitive paint and thermal dosimeters were Plant (PCAPP) MPT testing. However, all model also used. Both the paint and the dosimeters confirmed that results must be validated with experimental data as the peak temperature had indeed exceeded 1000°F, and the discussed elsewhere in this report. dosimeters confirmed that the criterion requiring 1000°F for 15 minutes was met (BPBGT, 2007d). The experimental data and the modeling focused on the The shielded thermocouples were inserted into ¼-in.- temperature-time profiles for projectiles at specified loca- diameter holes in the surface, and the leads were strapped tions in the tray. Figure 4-1 shows the projectile numbering on the surface by hose clamps (see Figure 4-3). The ther- system and the locations of the test thermocouples. In this chapter, the reliability of both temperature mea-  The thermally sensitive paints selected for use are Tempilaq °G surements and model predictions is evaluated, as well as temperature-indicating liquids manufactured by Tempil, Inc. They have the ability of the model to estimate the heating rates for the individual colors that indicate their temperature specification. When the projectiles in the full-scale unit. Conclusions on the ability specified temperature is reached, they require about 10 minutes to change of the full-scale MPT to meet throughput requirements are color, usually to black. Four temperature-level paints were used in the tests, then presented. paints that changed at 950°F, 1000°F, 1022°F, and 1050°F (according to the manufacturer, these are accurate to +/– 1 percent). One-inch-wide stripes were painted along the length of the projectiles tested. The thermal dosimeters use a crystalline solid inside a metal shell. The EXPERIMENTAL TEMPERATURE MEASUREMENTS solid begins to melt at 1000°F. When melted completely (which takes a Type K thermocouples encased in stainless steel sheaths minimum of 15 minutes), the result is a glassy solid, indicating performance were mounted to the top and bottom areas of three projectile of the MPT within specifications. Tests were run on May 8, 2007, with five dosimeters; four indicated that the 1000°F/15 minute criterion was reached; casings in order to allow the monitoring of temperature-time the fifth failed because it could not be removed from the projectile. 27 

28 Review and Assessment of Developmental Issues Concerning the Metal Parts Treater Design 17 9 1 M110 M110 M110 PCAPP - Equivalent 21 13 Munition Position 5 M 110 M110 M110 18 10 2 M110 14 6 22 Redundant M 110 X 45 Top of Round 1 2 X 19 6 Bottom of Round 11 X 3 INLET 3 In tray 7 M110 M110 DOOR 23 15 END M 110 X Moved 4 20 12 4 To Top of Round #14 M110 M110 M110 for 15 June Test 24 16 8 M 110 M110 M110 1 = Projectile Position Number per CFD Model M110 = M110 H Projectile used as placeholder = M125 VX Projectile used as test article 14 = Instrumented Projectile X = Thermocouple location relative to projectile and tray = “Cold Spots” Predicted by CFD Model FIGURE 4-1 Location of thermocouples (X) and computational fluid dynamics model “cold spots” on test rounds and in tray for June 14 testing. Source: BPBGT, 2007a. 4-1 Color mocouples are in direct contact with only the sheath, which tEMPERATURE PREDICTION By COMPUTATIONAL is in contact with the projectile surface. The measurements FLUID DYNAMICS THERMAL Modeling were taken to be surface temperatures when compared with The Bechtel Parsons Blue Grass Team (BPBGT) used a the CFD predictions. However, the sheath is also in contact mathematical model for comparison with the TRRP experi- with the outside environment and is subject to both direct mental measurements and to predict the performance of the convective and radiative heating. It is a matter of specula- full-scale MPT for the Blue Grass Chemical Agent Destruc- tion whether or not the extra heat transported through the tion Pilot Plant. The BPBGT’s model gives spatial and tem- outside sheath to the thermocouple lead is significant, but poral temperature behavior of the parts being processed in it could lead to erroneous temperature measurements. The the MPT. The purpose of the modeling was to show that the preferred method of measuring surface temperature is to feed MPT design was adequate for treating munitions at 1000°F the thermocouple wires through holes from the inside of the for 15 minutes while meeting operational and schedule re- projectile to the surface. Then, the sheath and leads are not quirements and that the design could guide the scale-up and exposed to the outside environment. the testing of the full-size unit. Comparison with the experimental measurements was Finding. Errors may exist in the experimental tempera- used to validate and modify the model. The improved model ture measurements with thermocouples from heat transfer appears to be fairly rigorous; it uses a commercial CFD through the leads. package, AcuSolve Version 1.7b (Acusim Inc.), which is a 

Thermal Testing, Modeling, and Predicted Throughput of the Metal Parts Treater 29 #14 Temperature (°F) Time (min) FIGURE 4-2  Comparison of computational fluid dynamics (CFD) model predictions with experimental results for thermocouples 4 and 6 on projectile 14 with CFD predictions. Source: Jonathon Berkoe, Manager, Bechtel Advanced Simulation and Analysis Group, “Blue Grass 4-2 MPT CFD Modeling Comparison to Pasco TRRP Experiment Measurements,” presentation to the committee, September 5, 2007. this is a fixed image we can replace type but not make changes to original general-purpose finite-element-based incompressible flow •  Reflected and emitted radiation from the adjacent solver. The code is based on a nonlinear solver that (1) is projectiles and other structures inside the MPT accurate to second order in space and time; (2) globally and chamber. locally conserves mass, momentum, and energy; and (3) al- lows a choice of finite-element shape function. The model These effects must be summed over all surface elements handles multimode heat transfer, including graybody radia- at each time interval. A typical set of parameters used in the tion with view factor computation. model for the TRRP MPT is given in Table 4-1, and typical The model includes the coupled effects of (1) convec- boundary conditions are shown in Table 4-2. The model is tive heat transfer rates between the injected steam and the highly nonlinear in temperature, and it includes first- and projectiles, (2) conduction within the projectiles, and (3) fourth-power temperature dependences and their effect on radiative transfer rates between the inductively heated cham- the heat-up rates. ber surfaces and the projectiles, and between the projectiles The present CFD model uses constant values for the pro- themselves. These rates are used to predict the local projec- jectile emissivity and for the specific heat of the projectiles. tile temperature-time profiles while the projectiles are in the Because of the variety of coatings used on the projectiles chamber. The model is complex because the instantaneous and the unknown change that occurs during coating py- heat transfer rate depends on the following: rolysis, it is probably not possible to model the emissivity more accurately. However, measured values of temperature- • The temperature difference between the local steam dependent specific heat cp were obtained by the BPBGT (Fig- and the projectile surface temperatures at any time; ure 4-4), but these were not used in the CFD model except • The radiative transfer rate, which depends on the to extract a representative constant value. The specific heat difference between the fourth power of the absolute nearly doubles between room temperature and 700oC. The temperature of each element on the inductively increasing values of cp with temperature will tend to cause heated surface and a projectile element; the heat-up rate to be slower at higher temperatures. This • The emissivities of all the surfaces; and may contribute to the difference in behavior between the 

30 Review and Assessment of Developmental Issues Concerning the Metal Parts Treater Design TABLE 4-1  Metal Parts Treater Unit’s Material Properties Component Materials Tray, fins ASTMa Projectiles Carbon steel Insulation on MPT doors CS 85 MPT conveyor, conveyor support, rollers ASTM MPT door frames ASTM Emissivities Projectiles 0.65 MPT coils and walls 0.9 Conveyor and conveyor support 0.65 Rollers 0.3 MPT doors (insulation) 0.4 Material Properties Steam Density 0.2673 kg/m³ Conductivity 0.0261 J/m-s-˚K Specific heat 2014 J/kg Expansivity 0.003472 1/˚K Viscosity 0.00005 kg/m-s ASTM Density 8000 kg/m³ Conductivity 16.3 J/m-s-˚K Specific heat 500 J/kg CS85 Density 85 kg/m³ Conductivity 0.29 J/m-s-˚K Specific heat 1260 J/kg Carbon Steel Density 7850 kg/m³ Conductivity 51.9 J/m-s-˚K Specific heat 485 J/kg FIGURE 4-3 Thermocouple installation on the projectile. SOURCE: BPBGT, 2007a. aAmerican Society for Testing and Materials RA 330 Steel. SOURCE: BPBG, 2006b. experiments and the CFD model that is observed. (Figure Finding. The CFD model developed by the BPBGT predicts 4-2 is a typical example.) the trends in temperature behavior over time and space and The committee originally had concerns that the modeling the location of cold spots within the system. predictions were inaccurate because the model did not take into consideration the effects of the door openings and clos- Finding. The specific heat of the projectile material varies by ings nor of the purge flows during these transients because a factor of two over the temperature range of interest, which the effect was found to be small. The TRRP report indicates may affect the predicted temperature-time profiles. that door-opening transients resulted in a “flat spot” of from 1 to 7 minutes in the measured temperature profiles. These Recommendation 4-1. For more accurate prediction of were not significant over the 90-105 minute test period. The temperature heat-up, the dependence of specific heat on inlet and exit chamber purge flows of nitrogen are maintained temperature should be included in further modeling. during door openings and enter the MPT and exit through the MPT exhaust headers during door openings (BPBGT, COMPARISON OF TEMPERATURE MEASUREMENTS 2007b). The CFD model was modified for prediction of AND MODELING full-scale performance by changing the physical properties and boundary conditions to conform with the full-scale de- Because the experimental data could also be inac- sign. The boundary conditions and properties for full-scale curate, comparisons between the CFD predictions and the analysis are shown in Tables 4-3 and 4-4. data cannot differentiate between errors in the model and 

Thermal Testing, Modeling, and Predicted Throughput of the Metal Parts Treater 31 TABLE 4-2  Computational Fluid Dynamics Model Boundary the measured data. The BPBGT did not present any error Conditions for the Technical Risk Reduction Program analysis in either case. However, the temperature trends in time and distributions in space are in general agreement with Model Region Condition the experimental measurements, although there are some Steam Inlet differences in magnitude. The differences in the model and   Full model 150 lb/hr 1000oF   Four-munitions sub-modela 11.1 lb/hr 1000oF the measurements appear to be within the usual differences expected in such modeling. Steam outlet 0 pressure Induction-heated MPT wall temperature Finding. The error bounds have not been reported on the   Constant 1350oF experimental data for the TRRP test; therefore, it is not pos-   Temperature gradientb Distance (ft) Temperature (oF)   1.03 1233.5 sible to assess the significance of differences between the ex-   3.00 1329.5 perimental measurements and the CFD model predictions.   4.67 1310.5   6.93 1326.5 Recommendation 4-2. In the testing of the full-scale   8.90 1362.0 MPT unit, error analysis should be performed and reported 10.87 1174.5 on the temperature measurements from multiple runs. Air leakc 2.72 ft2/min 65oF aFour-munitions sub-model only. Finding. The CFD model has been used to aid in the bBaseline with variable MPT wall temperature case only. design of the full-scale MPT and should be adequate for cAir leak model only. predicting cold spots and thus guiding thermocouple place- SOURCE: BPBGT, 2007a. ment for testing of the full-scale MPT. 0.9 data input points 0.8 0.7 measured Cp extrapolated data Heat capacity (J/kg-°C) 0.6 0.5 0.4 constant value used in baseline 0.3 0.2 0.1 0 0 100 200 300 400 500 600 700 800 Temperature (°C) Specific Heat 1.000 painted V 0.900 painted II constant value 0.800 used in baseline case 0.700 painted IV averaged data Emissivity 0.600 input points 0.500 extrapolated data 0.400 0.300 0.200 0.100 0.000 0.0 100.0 200.0 300.0 400.0 500.0 600.0 700.0 800.0 Temperature (°C) Emissivity FIGURE 4-4 Variation of specific heat and emissivity with temperature. SOURCE: Jonathon Berkoe, Manager, Bechtel Advanced Simulation and Analysis Group, “Blue Grass MPT CFD Modeling Comparison to Pasco TRRP Experiment Measurements,” presentation to the committee, September 5, 2007. 4-4 

32 Review and Assessment of Developmental Issues Concerning the Metal Parts Treater Design TABLE 4-3  Computational Fluid Dynamics Model 070806 sitioning of the steam header) has improved the agreement Boundary Conditions between observed and predicted ramp-up rates in the three- quarter-scale TRRP MPT.,, Boundary Condition Quantity 2. Improvements in the CFD model have resulted in Steam temperature 1000˚F better agreement between observed and predicted rates in the Wall temperature 1250˚F in Zone 1 and three-quarter-scale MPT. Predicted values of 108 minutes  ���������������� 1350˚F in Zone 2 to reach 1000oF now compare with experimental values of 105 to 114 minutes. Steam flow rate 0.0189 kg/s 3. Ramp-up rates are chiefly affected by the temperature attained by the radiating internal surface of the MPT. The Rail temperature Varies in axial direction full-scale MPT is designed to have a controllable interior Inlet door temperature 140˚F wall temperature of 1450oF (1910oR) or higher in compari- son with the 1350oF (1810oR) available in the TRRP proof- Initial temperature of-concept unit. Because of the dependence of radiative   Cold munition 65˚F transfer on the fourth power of the absolute temperature,   Hot munition 1000˚F   Inside MPT 1000˚F this increase in wall temperature translates to an appreciable increase in heat transfer rate to the projectiles (or secondary SOURCE: BPBGT, 2007a. waste containers) of (1910/1810)4 = 1.24, or a potential 24 percent increase in heat transfer rate to the projectiles. TABLE 4-4  Computational Fluid Dynamics Model 070806 This should allow proportionally faster ramp-up rates in the Component Masses full-scale MPT unit. Because the most recent TRRP MPT ramp-up rates were close to design values, the full-scale MPT Component Mass (kg) should be capable of meeting design requirements. Munition 75.52 4. The projected required full-scale MPT munitions Tray 448.86 processing rates are quite low. Peak operating rates are pro- jected to be 15 8-in. GB projectiles per hour, 39 155-mm VX Rail and support 1080.20 shells per hour, or 40 155-mm H projectiles per hour. These SOURCE: BPBGT, 2007a. rates are based on charging one munitions tray at a time. The BPBGT has stated: “For the VX case, the tray can hold 40 rounds, but one out of 40 positions is left open to hold nose closures. The duration of processing each tray in the MPT Recommendation 4-3. The Army should depend on ac- main chamber is 2 hours” (BPBG, 2006c, p. 22). tual testing of the full-scale MPT, rather than solely on modeling. The TRRP MPT testing was terminated before all planned testing was completed on waste streams because ABILITY TO SCALE UP and Meet throughput the BPBGT concluded that the available test results were requirements adequate for designing the full-scale unit. The BPBGT throughput analysis states: Early tests of the three-quarter-scale TRRP MPT indi- cated that measured ramp-up heating rates to 1000oF were As stated earlier, there are two MPTs. The spare MPT, al- slower than anticipated by modeling. This raised concerns though processing secondary waste, is assumed to be read- that the full-scale MPT might be unable to meet required MPT throughput rates. The pertinent design goal is that the John Ursillo, Pasco Resident Engineer, Bechtel Parsons Blue Grass MPT be “designed to support decontamination of one muni- Team, “MPT Technical Risk Reduction Program (TRRP) Testing,” presenta- tion body every 1.5 minutes maximum” (BPBG, 2006c). tion to the committee, September 5, 2007. Assurance of meeting throughput rates has been ad- Jonathan Berkoe, Manager, Bechtel Advanced Simulation and Analysis dressed by the BPBGT. This conclusion is reached on the Group, “Blue Grass MPT CFD Modeling Comparison to Pasco TRRP Experiment Measurements,” presentation to the committee, September 5, basis of the following: 2007. Jonathan Berkoe, Manager, Bechtel Advanced Simulation and Analysis 1. Incorporation of the proposed modifications in the Group, “FOAK CFD Models Planning,” presentation to the committee, mechanical configuration of the three-quarter-scale MPT September 5, 2007. Jonathan Berkoe, Manager, Bechtel Advanced Simulation and Analysis unit (e.g., better seals on the inlet and outlet doors and repo- Group, “Blue Grass MPT CFD Modeling Comparison to Pasco TRRP Exper- iment Measurements,” presentation to the committee, September 5, 2007. Bechtel Parsons Blue Grass Team, “Blue Grass Chemical Agent De- Samuel Hariri, Process Design Lead, Bechtel Parsons Blue Grass Team, struction Pilot Plant (BGCAPP),” presentation to the committee, September “Thermal Modeling to Support OTM Design,” presentation to the commit- 5, 2007. tee, September 5, 2007. 

Thermal Testing, Modeling, and Predicted Throughput of the Metal Parts Treater 33 ily switchable (within 2 hours) to process projectile bodies TABLE 4-5  Metal Parts Treater/Metal Parts Treater should the need arise. Testing of the MPTs has shown startup Cooling System Projectile Throughput Rates from a cold condition to take 4-6 hours. There is an 8 tray buffer between the MWS and MPT. One MPT can process Installed Average 8 trays in 8 hours. Negating the GB case because of the Operating Throughput fact that there are very few munitions to process compared Rate Rate to rockets, the MPT can process rounds at a faster rate than System Munition Type (rounds/hr) (rounds/hr) that scheduled for the NCR-MWS or the LPMD (39 vs 21 MPT/MCSa 8-in. M426 (GB) 15 7.5 projos/hr for VX, and 40 vs 26 projos/hr for H). For VX, it MPT/MCS 155-mm M110 (H) 40 9.9 takes the MPT 8 hours to process a full buffer that the NCR- MPT/MCS 155-mm M121A1 (VX) 39 8.8 MWS took 15 hours to produce; and for H it takes 8 hours aMCS, metal parts treater cooling system. to process what took the LPMD 12 hours. Therefore, even SOURCE: BPBG, 2007. if the second MPT were cold when the first failed, because the buffer size is larger than the heat up time and because the MPT can catch up on a backlog, the NCR-MWS or production rates are not a critical factor in plant production. LPMD could continue operating normally and there would However, it is possible that lower-than-anticipated ramp-up be no net effect change in the processing schedule (BPBG, rates could affect plant closure operations. 2006c, p. B-5). Finding. Based on the four factors specified above in this Average throughput rates are even lower than stated section (Ability to Scale Up and Meet Throughput Require- above (see Table 4-5). ments), the full-scale MPT is expected to meet design The required processing rates are so low that extended throughput rates, and, even if it falls somewhat short of processing times due to lower ramp-up rates will have no design values, overall plant throughput is not expected to effect on overall plant munitions throughput. Thus, MPT be affected.