TRANSMITTED TO DOE/M. WHITAKER 07/05/95 DEFENSE NUCLEAR FACILITIES SAFETY BOARD March 27, 1995 MEMORANDUM FOR: G. W. Cunningham, Technical Director COPIES: Board Members FROM: David T. Moyle SUBJECT: Radiolytic Hydrogen Generation in Rocky Flats Plutonium- Nitric Acid Solution Tanks 1. Purpose: This report documents an independent Defense Nuclear Facilities Safety Board (Board) staff review of hydrogen generation in actinide solution tanks at the Rocky Flats Environmental Technology Site (RFETS). This is a follow-up action to concerns raised in a recent Board staff visit to Rocky Flats (November 28-December 1, 1994). The staff intends to evaluate whether passive venting of tanks is sufficient to prevent hydrogen accumulation and potential risk of hydrogen explosions in the tanks. 2. Summary: Assuming diffusion limited transport as a bounding calculation, analysis shows that even though vent lines to the actinide solution tanks are open, hydrogen and oxygen will accumulate in tank headspaces. Without headspace sampling, all tanks may be assumed to contain explosive mixtures of hydrogen and oxygen. Detonations and/or deflagrations may occur in vent lines and tanks if ignition sources arise. Detonation effects are similar for a wide range of hydrogen concentrations where the maximum reflected pressures at tank walls could exceed the failure pressure by a factor of two. Even if the tank wall remains intact, fittings and sight glasses may likely fail, breaching containment. The analysis indicates that within relatively short times hydrogen gas can build up to explosive concentrations. Due to the relatively long stagnant storage of these solutions, a hydrogen detonation in a tank is believed to be a credible hazard. 3. Background: Alpha decay of plutonium isotopes causes radiolysis reactions in solutions which produce hydrogen and oxygen gases as major products. Build up of this flammable gas mixture in tank void spaces poses several hazards to the facility, resulting from possible explosions, including: a. Loss of containment which could cause spills and airborne releases from vaporization. b. Missiles / Shrapnel. c. Criticality from settling of broken Raschig rings, geometry changes. The Board trip report dated December 8, 1994, discussed Rocky Flats' past efforts to resolve this issue.[1] In 1993, Los Alamos Technology Office at Rocky Flats (LATO) performed a safety study of plutonium and uranium solutions at RFETS, which concluded that in unvented high plutonium concentration tanks, sufficient hydrogen could be generated to reach the lower flammability limit (LFL) in about 12 hours. Further, LATO recommended that it was "extremely important that ventilation be maintain on all solutions in tanks." At Department of Energy / Rocky Flats Office (DOE/RFO) request, vent line outlet hydrogen concentrations were measured for tanks in building 771. All readings were zero except one at 17% of LFL. LATO assumed that outlet concentrations through several meters of vent line reflected the status of the tank void spaces, and concluded, "these measurements indicate that the potential for an explosion in a tank is extremely low." Furthermore, they concluded that the consequences of a tank explosion would be minimal, possibly blowing out gaskets and causing a leak, but not breaching the tank itself.[2] In response to Board staff questions about the hydrogen explosion scenario, EG&G representatives stated that there would be no off site consequences, and since only workers would be affected, EG&G considered it was not necessary to pursue.[1] Occurrence report numbers RFO-EGGR-371OPS-1995-0037 and RFO-EGGR-771OPS- 1995-0064 referenced a potential unreviewed safety question regarding hydrogen gas generation and buildup in stagnant actinide solution tanks in buildings 371 and 771.[3][4] More details were offered in Operating Experience Weekly Summary Report 95-09, reporting the completion of a draft Unreviewed Safety Question Determination (USQD) on February 28, 1995. The study concluded that a detonation could rupture the tanks if they were not vented and an ignition source was present. If vented, however, hydrogen accumulation was determined to be a manageable hazard. Further, an ignition source could not be identified.[5] The Board staff is concerned that this draft USQD does not adequately address the hydrogen accumulation and explosion issue, and this paper summarizes results of an independent staff analysis. Diffusion calculations, presented in Appendix A, determine theoretical worst case concentrations of hydrogen in tanks and vent lines, while the explosion analysis in Appendix B estimates maximum pressures that could be experienced by containment in event of an explosion. 4. Discussion: a. Industry Standard Design and Operating Requirements: National Fire Protection Association (NFPA) 69, Standard on Explosion Prevention Systems, states that, "The combustible concentration shall be maintained below 25 percent of the lower flammability limit," when no automatic safety interlocks are provided.[6] Implicit in this requirement is the general assumption that an ignition source will be present. It is generally not acceptable practice to rely on the lack of an obvious ignition source unless a safety system is in place to assure suppression of ignition sources. b. Identification of Tanks: This analysis assesses hydrogen generation in a total of 14 actinide solution tanks in buildings 371 and 771. Building 371 contains four of these tanks (D49B, D49C, D55A, D134C), while building 771 houses the remaining ten tanks (D452, D472, D550, D931, D933, D971, D972, D974, D1007, D1810). c. Hydrogen Generation Rates: Hydrogen generation rates (G-values) for the alpha radiolysis of nitric acid solutions are taken from N.E. Bibler's experimental work at Savannah River.[7],[8] Oxygen G-values are estimated by relative trends observed at Rocky Flats in Kazanjian's research with actual plutonium nitric acid solutions.[9] See Appendix A for more details. d. Diffusion Analysis: A Rocky Flats report has established that all solution tanks are vented, i.e. the vent lines are not blocked.[10] However, the vents are "passive", and in the absence of pressure variations, the escape of hydrogen from tank vapor spaces is limited by diffusion down the vent line. Concentration measurements at the vent line outlet are not representative of the tank vapor concentration, because diffusion limitations will cause a concentration gradient to develop down the length of the vent line. Appendix A develops a three component, one dimensional model for radiolytic hydrogen and oxygen diffusing through non-diffusing air in a horizontal vent line. This model predicts that vent lines longer than approximately 1 meter will result in a flammable tank atmosphere (4 volume % hydrogen). For realistic vent line lengths, equilibrium concentrations in several tanks may approach the stoichiometric ratio of hydrogen and oxygen. Hydrogen is quite easy to ignite over a wide range of concentrations, requiring less energy than most other flammable gases. In oxygen, hydrogen concentrations between 4 and 94 volume % are flammable, with a detonable range between 15 and 90 volume%.[11a] All tanks are expected to have equilibrium concentrations in the detonable range for hydrogen-oxygen mixtures. Appendix A also gives diffusion results and hydrogen buildup times for Kazanjian's hydrogen generation rate data (less conservative than Bibler's), as estimated in the Rocky Flats draft USQD calculation.[12] With these generation rates, explosive mixtures are still expected in all solution tanks. e. Flammable Gas Buildup Time: Using the hydrogen generation rates from Bibler's research, worst case hydrogen gas build-up times were approximated, neglecting diffusion. Results in Appendix A show that LFL (4 volume %) can be reached in 1 to 10 days in some tanks, and equilibrium concentrations are reached in 1 month to 8 years. Based on the number of years that the tanks have remained idle, it could be assumed that all tanks are currently at explosive concentrations. f. Explosion Analysis: The method followed for the explosion analysis in Appendix B primarily came from references used in a seminar on the calculation and evaluation of fire and explosion hazards sponsored by the American Institute of Chemical Engineers. For confined gas explosions, the deflagration pressure wave is generally assumed to be 10 times the initial pressure. Calculations for a stoichiometric hydrogen-oxygen mixture show the deflagration pressure is approximately 143psia. When a pressure wave strikes a surface, a reflected pressure wave is developed. This reflected pressure is greater than the incident pressure and results from a momentum change, due to a change in direction when the moving air strikes a dense surface. The reflected pressure of a deflagration wave striking a surface normal to the incident pressure wave is approximately twice the deflagration pressure or, 285psia.[11],[13] The detonation pressure for a confined gas can be estimated as twice the deflagration pressure or, 285psia in this case. The maximum reflected pressure from the detonation shockwave striking a surface, such as the inner tank wall, normal to the direction of propagation, will be approximately 1800psia.[11],[13a] This worst case reflected pressure results in a tensile stress nearly twice the ultimate stress of a 42 inch diameter, 1/4 inch thick wall, 304L stainless steel cylindrical tank. Due to the ductility of 304L stainless steel,[14] it is difficult to determine if the impulse of a reflected detonation pressure will rupture the tank, but it is likely to cause deformation and blow out fittings. The calculated detonation pressure for a stoichiometric hydrogen- oxygen ratio corresponds to that reported in Bureau of Mines Bulletin 627. Further, a graph of detonation pressures shows that the effects of a hydrogen detonation will be essentially the same for much of the explosive range (20 to 80 volume %).[15] g. Deflagration Versus Detonation: By definition, deflagrations propagate at subsonic velocities, and detonations propagate at supersonic velocities. Deflagrations transition to detonations when the reaction front accelerates to the speed of sound. If a deflagration were to occur in a storage tank, it is possible that pressures could be vented enough to avoid extensive damage to system components. However, 90 degree elbows in vent lines and large length to diameter ratios will limit the effectiveness of venting.[11],[13] Therefore, deflagrations may cause some structural damage. Based on current analyses, the possibility of a detonation can not be ruled out. Acceleration to detonation in horizontal pipes generally occurs in distances proportional to the square root of the pipe diameter. However, depending on the strength of the ignition source, detonation can be almost instantaneous.[11],[13] Therefore, a hydrogen detonation in an actinide solution tank or vent line can not be ruled out. Furthermore, deflagrations and/or detonations are likely to affect other tanks connected through common vent lines. See appendix B for further discussion. 5. References [1] Bamdad, F., memorandum for G. W. Cunningham, "Nuclear and Criticality Safety at Rocky Flats", Trip Report (November 28-December 1, 1994), December 8, 1994. [2] Los Alamos Technology Office at Rocky Flats, "Plutonium and Uranium Solutions Safety Study", October 14, 1993. DNFSB Control # RF: 94-3523. [3] Occurrence Report RFO-EGGR-371OPS-1995-0037. [4] Occurrence Report RFO-EGGR-771OPS-1995-0064. [5] "Operating Experience Weekly Summary 95-09", February 24 through March 2, 1995. [6] "NFPA 69, Standard on Explosion Prevention Systems", 1992 Edition. [7] Bibler, N.E., "Curium-244 Alpha Radiolysis of Nitric Acid. Oxygen Production from Direct Radiolysis of Nitrate Ions", Savannah River Laboratory, E.I. duPont Nemours & Co., DP-MS- 72-68. DNFSB Control # SR:94-257. [8] Bibler, N.E., memo to R. Maher (plus attachments), "Radiolytic Hydrogen Production from Process Vessels in HB Line - Production Rates Compared to Evolution Rates and Discussion of LASL Reviews", November 12, 1992, WSRC-RP-92- 1312. DNFSB Control # SR: 94-258. [9] Kazanjian, A.R., & D.R. Horrell, "Radiologically Generated Gases in Plutonium-Nitric Acid Solutions", Radiation Effects 1972, Vol 13, pp. 277-280, Gordon and Breach Science and Publishers Ltd., Glasgow, Scotland. [10] Adams, A.B., memo to C. Sprain, "Potential for Line Blockage Resulting in Hydrogen Pressurization of Actinide Solution Holding Tanks in Building 771 - ABA-001-95". [11] Tunkel, Steven J., "Methods for the Calculation of Fire and Explosion Hazards", AIChE Today Series. Course notes and excerpts, including: [11a] Handbook of Compressed Gases, Third Edition, Compressed Gas Assn., Inc., Van Nostrand Reinhold, NY, 1990. [11b] Kuchta, Joseph M., Investigation of Fire and Explosion Accidents in the Chemical, Mining, and Fuel-Related Industries - A Manual, U.S. Department of the Interior, Bureau of Mines, Bulletin 680, 1985. [12] Colwell, R.G., "Analysis of Hydrogen Generation, Explosivity and Pressure Rise in Unvented Pu-HNO3 Solution Tanks Due to Radiolysis", (DRAFT) Calc. No. CALC-RFP-95.0386-RGC-USQD. [13] Grelecki, Dr. Chester, "Fundamentals of Fire and Explosion Analysis", AIChE Today Series. Course notes and excerpts, including: [13a] Glasstone, Samuel (Editor), The Effects of Nuclear Weapons, Chapter III: "Air Blast Phenomena", United States Atomic Energy Commission, April, 1962. [13b] Cook, Melvin A., The Science of High Explosives, Appendix II, American Chemical Society Monograph Series. Reinhold Publishing Corp., New York, 1958. [14] Perry, Robert H., & Don Green (Editors), Perry's Chemical Engineers' Handbook, Sixth Edition, McGraw-Hill Book Company, New York, 1984. [15] Zabetakis, Michael G., Flammability Characteristics of Combustible Gases and Vapors, U.S. Department of the Interior, Bureau of Mines, Bulletin 627, 1965. [16] Bird, Stewart, & Lightfoot, Transport Phenomena, John Wiley & Sons, Inc., 1960. NOTE: Please contact the Defense Nuclear Facilities Safety Board directly for Appendix A and B which contains charts and equations.