1. 1 Chemical Process Safety Runaway Reactions

2. 2 Two CSB Videos: Review 1.Reactive Hazards (31 July 2007)31 July 2007 2.Runaway: Explosion at T2 Laboratories (19 Dec 2007; video: 22 Sep 2009)22 Sep 2009 “167 serious uncontrolled reactions with 108 deaths from 1980 – 2001”

3. 3 Two CSB Videos: Review 1.Reactive Hazards: a)What do you remember about the video? b)Lessons “learned”

4. 4 Two CSB Videos: Review 1.Reactive Hazards: a)1984 Bhopal CSB formed & established chemical process safety b)Synthron: butyl acrylate (solvents: toluene, cyclohexane) 1500 gal reactor HE was used to condense solvent vapors & cool exothermic reaction Batch size increased HE couldn’t remove enough heat c)BP Amoco: HP nylon Polymerization reactor bypass to 750 gal waste tank Overfilled waste tank; no PI or vent Secondary decomposition reaction d)MFG Chemical: allyl alcohol vapor release 30 gal test reactor (3 rd test significant heat generation) Production in 4000 gal reactor (SA/vol ratio: HE inadequate) e)1 st Chemical Corporation: mono-nitro toluene (MNT) 145’ distillation tower; MNT left in reboiler Leaking steam valve Heated to 450 oF – decomposition reaction

5. 5 Two CSB Videos: Review 1.Reactive Hazards: a)What do you remember about the video? b)Lessons “learned”

6. 6 Two CSB Videos: Review 2.Runaway: Explosion at T2 Laboratories: a)What do you remember about the video? b)Lessons “learned” Producing a gasoline additive: methylcyclopentadienyl manganese tricarbonyl (MCMT) Reactor

7. 7 Two CSB Videos: T2 Laboratories Brief overview of process steps Added to reactor –sodium metal in mineral oil –methylcyclopentadiene dimer –diethylene glycol dimethyl ether (diglyme) close the vessel set pressure to 3.45 bar and heating oil temp to 182.2 C heating melted sodium that reacted with methylcyclopentadiene forming sodium methylcyclopentadiene, hydrogen, and heat Hydrogen gas was generated when mix reached 100°C, agitation was shut off at 150°C hot oil flow stopped at 180°C cooling was initiated with water admitted to the reactor jacket. maintain temperature from the exothermic reaction via water evaporation.

8. 8 Two CSB Videos: T2 Laboratories 175 th batch exploded Former Reactor Site

9. 9 Figure 2. Control room.* * From CSB final report; Sep 2009.

10. 10 Figure 4. Injury and business locations.* * From CSB final report; Sep 2009.

11. 11 Figure 5. Portion of the 3-inch-thick reactor.* * From CSB final report; Sep 2009.

12. 12 Figure 4. Injury and business locations.* * From CSB final report; Sep 2009.

13. 13 Two CSB Videos: T2 Laboratories CSB Investigation Runaway exothermic reaction Occurred during the first metalation step of the process An uncontrollable rise in temperature and resultant pressure lead to the burst of the reactor Upon bursting, contents ignited in air Creating an explosion equivalent of 635 kg (1420 lb) of TNT exploding from a single point

14. 14 Two CSB Videos: T2 Laboratories CSB Investigation Possible causes for the explosion Investigation considered: –cross-contamination of the reactor –contamination of raw materials –wrong concentration of raw materials –local concentration of chemical within the reactor –application of excessive heat –insufficient cooling

15. 15 “The CSB determined insufficient cooling to be the only credible cause for this incident, which is consistent with witness statements that the process operator reported a cooling problem shortly before the explosion. The T2 cooling water system lacked design redundancy, making it susceptible to single-point failures including water supply valve failing closed or partially closed. water drain valve failing open or partially open. failure of the pneumatic system used to open and close the water valves. blockage or partial blockage in the water supply piping. faulty temperature indication. mineral scale buildup in the cooling system. Interviews with employees indicated that T2 ran cooling system components to failure and did not perform preventive maintenance. * From CSB final report; Sep 2009.

16. 16 Two CSB Videos: Review 2.Runaway: Explosion at T2 Laboratories: a)What do you remember about the video? b)Lessons “learned” “T2 did not recognize the runaway reaction hazard associated with the MCMT it was producing.” Contributing causes: 1.“The cooling system employed by T2 was susceptible to single point failures due to a lack of design redundancy. 2.The MCMT reactor relief system was incapable of relieving the pressure from a runaway reaction.”

17. 17 Two CSB Videos: T2 Observations Scaled up from 1 liter to 9300 liter directly Batch 42 the recipe was increased by 1/3 (testing?) Periodically experienced problems with cooling No “backup” cooling system Used city water supply (minerals?) Did not recognize and control reactive hazards No evidence found by CSB that T2 performed a recommended HAZOP. There was a need for reactive chemistry testing.

18. 18 CSB Testing on T2 Recipe CSB testing completed with a VSP2 (Vent Sizing Package 2) Adiabatic Calorimeter (116 ml test cell) * From CSB final report; Sep 2009. reaction 1 exotherm diglyme decomposition

19. 19 * From CSB final report; Sep 2009.

20. 20 Follow-up Topics Key Findings of CSB investigation:investigation: Cooling discussion Overpressure Runaway reactors Hazard analysis

21. 21 * From CSB final report; Sep 2009. A second exothermic reaction occurred This reaction became uncontrollable around 200°C The reaction was the uncontrolled decomposition of diglyme (the solvent used) Probably catalyzed by the presence of sodium. By the time the rupture disk opened (28.6 bar) It was too late If the rupture disk had opened at 6.2 bar, then no explosion would have occurred

22. 22 Over pressure Wave Profile, 1 Psi=0.07 bar psi 0.017 Bar 0.14 Bar 0.017 Bar 1.7 Bar * From CSB & SACHE module by R. Willey, 2012.

23. 23 Combustion Behavior – Most Hydrocarbons Slide courtesy of Reed Welker. Smoke and fire are very visible!

24. 24 Combustion Behavior – Carbon Disulfide Slide courtesy of Reed Welker. No smoke and fire, but heat release rate just as high.

25. 25 Combustion Behavior – Methane Methane burns mostly within vessel, flame shoots out of vessel.

26. 26 Combustion Behavior – Dusts Much of the dust burns outside of the chamber.

27. 27 Definitions - 1 LFL:Lower Flammability Limit Below LFL, mixture will not burn, it is too lean. UFL:Upper Flammability Limit Above UFL, mixture will not burn, it is too rich. Defined only for gas mixtures in air. UNITS:

28. 28 Definitions - 2 Flash Point:Temperature above which a liquid produces enough vapor to form an ignitable mixture with air. Defined only for liquids at 1 atm. pressure. Auto-Ignition Temperature (AIT): Temperature above which adequate energy is available in the environment to provide an ignition source.

29. 29 Definitions - 3 Limiting Oxygen Concentration (LOC): Oxygen concentration below which combustion is not possible, with any fuel mixture. Expressed as volume % oxygen. Also called: Minimum Oxygen Concentration Max. Safe Oxygen Conc. Others

30. 30 Explosion: A very sudden release of energy resulting in a shock or pressure wave. Shock, Blast or pressure wave: Pressure wave that causes damage. Deflagration: Reaction wave speed < speed of sound. Detonation: Reaction wave speed > speed of sound. Speed of sound in air: 344 m/s, 1129 ft/s at ambient T, P. Deflagrations are the case with explosions involving flammable materials. Definitions - 4

31. 31 Minimum Ignition Energy (MIE): Smallest energy to initiate combustion. Higher for dusts & aerosols than for gases Many HC gases have MIE ~ 0.25 mJ Auto-oxidation: slow oxidation and evolution of heat can raise T and lead to combustion. i.e. liquids with low volatility. Adiabatic compression: of a gas generates heat, increases temperature, and can lead to autoignition. Ignition sources: usually numerous and difficult to eliminate. Objective is to identify and eliminate, but not to solely rely on this step to eliminate combustion risk. (Table 6-5; Crowl) Definitions - 5

32. 32 Typical Values - 1 LFLUFL Methane:5.3%15% Propane:2.2%9.5% Butane:1.9%8.5% Hydrogen:4.0%75% See Appendix B Flash Point Temp. (deg C) Methanol:12.2 Benzene:-11.1 Gasoline:-43

33. 33 Typical Values - 2 AIT (deg. C) Methane:632 Methanol:574 Toluene:810 LOC (Vol. % Oxygen) Methane:12% Ethane:11% Hydrogen:5% Great variability in reported AIT values! Use lowest value. Appendix B Table 6-2

34. 34 Flammability Relationships Figure 6-2

35. 35 Aerosol Flammability Too rich Too lean M. Sam Mannan, Texas A&M, Mary Kay O’Conner Process Safety Center

36. 36 Minimum Ignition Energies What: Energy required to ignite a flammable mixture. Typical Values: (wide variation expected) Vapors: Dusts: Dependent on test device --> not a reliable design parameter. Static spark that you can feel: about mJ Lightning: about 500 megajoules Table 6-4 Or ~ 500,000,000,000 mJ

37. 37 Minimum Ignition Energies

38. 38 Ignition Sources of Major Fires

39. 39 Experimental Determination - Flashpoint Cleveland Open Cup Method. Closed cup produces a better result - reduces drafts across cup. Figure 6-3

40. 40 Experimental Determination - Flashpoint

41. 41 Setaflash Flashpoint Device

42. 42 Setaflash Flashpoint Device – Close-up

43. 43 Setaflash Flashpoint Device – Close-up Window

44. 44 Setaflash Flashpoint Device – Close-up

45. 45 Auto-Ignition Temperature (AIT) Device

46. 46 Auto-Ignition Temperature (AIT) Device

47. 47 0 2 4 6 8 10 02468 Fuel Concentration in air (vol%) Maximum Explosion Pressure (barg) LFLUFL Run experiment at different fuel compositions with air: Experimental Determination - LFL, UFL Need a criteria to define limit - use 1 psia pressure increase. Other criteria are used - with different results! Flammability limits are an empirical artifact of experiment! See Figure 6-5

48. 48 Experimental Determination: P versus t 0 2 4 6 8 10 050100150200250 Time (ms) Pressure (bar-abs) P max (dP/dt) max PI TI Ignitor Final experimental result:

49. 49 Experimental Apparatus

50. 50 Experimental Determination - LFL, UFL

51. 51 Flammability Limit Behavior -1 As temperature increases: UFL increases, LFL decreases --> Flammability range increases Approx. for many hydrocarbons Equations 6-4, 6-5

52. 52 Flammability Limit Behavior -2 As pressure increases: UFL increases LFL mostly unaffected P is pressure in mega-Pascals, absolute Pressure and temperature effects on flammability limits is poorly understood – estimation methods are poor. No theoretical basis for this yet!

53. 53 Flammability Limits of Mixtures Le Chatelier Rule (1891) y i on a combustible basis only n is the number of combustible species Assumptions: 1) Product heat capacities constant 2) No. of moles of gas constant 3) Combustion kinetics of pure species unchanged 4) Adiabatic temperature rise the same for all species Details provided in Process Safety Progress, Summer 2000.

54. 54 Flammability Limits - Le Chatelier LeChatelier’s rule shows that the LFL can be approximated by: Where C p is the product heat capacity, is the adiabatic temperature rise, and is the heat of combustion. 1200 K is frequently used as the adiabatic temperature rise at the flammability limit. A similar expressions is written for the UFL.

55. 55 Flammability Limits of Mixtures From this equation, a plot of the flammability limit vs. 1/(Heat of Combustion) should yield a straight line if Le Chatelier’s rule is valid. If this is done, one finds that: Le Chatelier’s rule works better at the lower flammability limit than the upper flammability limit. Assumptions are more valid at LFL.

56. 56

57. 57 Upper Flammability Limit and Heat of Combustion 0 20 40 60 80 100 0.0000.001 0.002 1/(-  h c ) [kJ/mol] UFL [vol. % fuel in air] Hydrocarbons Oxygen Compounds Nitrogen Compounds Sulfur Compounds

58. 58 Estimating Flammability Jones equation where the stoichiometric concentration, C st, is vol% fuel in fuel plus air. From the general combustion equation, C m H x O y + zO 2 = mCO 2 + x/2 H 2 O It follows that z = m + x/4 – y/2, where z has the units of moles O 2 /mole fuel Therefore, The Jones equation can now be converted to LFL = 0.55C st UFL = 3.50C st

59. 59 Estimating Flammability Suzuki and Koide correlation where: LFL and UFL are the lower and upper flammability limits ( vol% fuel in air ), respectively, and is the heat of combustion for the fuel (in 10 3 kJ/mol) NOTE that the accuracy of this and Jones methods are modest.

60. 60 LOC limiting oxygen conc. [vol% O 2 ] Typically 8 - 10% Estimating LOC (1)Fuel + (z) Oxygen --> Products Concentration required to generate enough energy to propagate flame Reduce O 2 concentration below LOC to prevent the fire/explosion If data for LOC is not available, estimate using the stoichiometry of the combustion process and the LFL For example, the stoichiometry for butane: The LFL for butane is 1.9% by volume, therefore from stoichiometry By substitution, we obtain,

61. 61 LOC’s for Various Substances

62. 62 Flammability Diagram

63. 63 Flammability Diagram Useful for: Determining if a mixture is flammable. Required for control and prevention of flammable mixtures Problems: Only limited experimental data available. Depends on chemical species. Function of temperature and pressure. Flammability diagram can be approximated.

64. 64 Flammability Diagram (1) Fuel + (z) Oxygen ---> Products Fuel Oxygen Nitrogen 0 1000 0 Flammable UFL LFL Stoichiometric CH 4 + 2 O 2 --> Products z = 2

65. 65 Drawing an Approximate Diagram 1. Draw LFL and UFL on air line (%Fuel in air). 2. Draw stoichiometric line from combustion equation. 3. Plot intersection of LOC with stoichiometric line. 4. Draw LFL and UFL in pure oxygen, if known (% fuel in pure oxygen). 5. Connect the dots to get approximate diagram.

66. 66 Example Methane: LFL: 5.3% fuel in air UFL: 15% fuel in air LOC: 12% oxygen CH 4 + 2 O 2 --> CO 2 + 2 H 2 O --> z = 2 Pure Oxygen: LFL: 5.1% fuel in oxygen UFL: 61% fuel in oxygen % oxygen

67. 67 Flammability Diagram - Example Fuel Oxygen Nitrogen 0 1000 0 66.7% O 2 Stoichiometric UFL = 15% fuel LFL = 5.3% fuel LOC = 12% oxygen 61% Methane 5.1% Methane

68. 68 Flammability Zone Non-Flammable Flammable Transition Boundary

69. 69 Flammability Zone Ethylene Stoichiometric Line Transition Boundary Flammable Non-Flammable

70. 70 Removal of Vessel from Service

71. 71 Explosions - Definitions Explosion: A very sudden release of energy resulting in a shock or pressure wave. Shock, Blast or pressure wave: Pressure wave that causes damage. Deflagration: Reaction wave speed < speed of sound. Detonation: Reaction wave speed > speed of sound. Speed of sound in air: 344 m/s, 1129 ft/s at ambient T, P. Deflagrations are the case with explosions involving flammable materials.

72. 72 Explosions Rapid release of energy Damage due to dissipation of energy in the form of pressure wave, projectiles, sound, radiation, etc Reaction front moves out from ignition source preceded by shock wave or pressure front. Once combustible material consumed, reaction front terminates, but pressure wave continues. Shock wave (results from abrupt pressure change) and is associated with highly explosive materials Most damage due to blast wave (shock / pressure wave followed by wind)

73. 73 Detonations Energy releases short, < 1 ms, associated with abrupt rise in P Shock and reaction front > speed of sound Reaction front provides energy to shock wave and drives it at sonic or greater speeds P of shock wave: ~ 10 - 100 atm.

74. 74 Deflagrations Energy release longer than detonation ~ 0.3 s, Pressure front = speed of sound; reaction front behind at < speed of sound Mechanism: turbulent diffusion, mass transfer limited P of wave: ~ a few atmospheres Can evolve, especially in pipes but not open spaces, to a detonation due to adiabatic compression and heating leading to pressure rise

75. 75 Comparison of Behavior Reacted gases Unreacted gases Deflagration: Detonation: Pressure Wave Reaction / Flame Front Ignition Reaction front moves at less than speed of sound. Pressure wave moves away from reaction front at speed of sound. Reaction front moves greater than speed of sound. Pressure wave is slightly ahead of reaction front moving at same speed. X X

76. 76 Comparison of Behavior Reacted gases Unreacted gases Deflagration: Detonation: Pressure Wave Reaction / Flame Front Ignition P Distance P Shock Front

77. 77 Comparison of Behavior Detonation Deflagration Localized Damage No wall thinning Lots of pieces Damage all over Wall thinning A few pieces

78. 78 Confined Explosions Occurs in process or building. Almost all of the thermodynamic energy ends up in the pressure wave. Cubic Law: K i Deflagration index (bar-m/s) G gas St dust (Staub) Deflagration index: Measure of explosion robustness, higher value means more robust. Depends on experimental conditions. Not a fundamental property.

79. 79 Deflagration Indexes

80. 80 Deflagration Indexes

81. 81 Data: Max. P and K G

82. 82 Damage Estimates from Overpressure Table 6-9; Crowl

83. 83 Dust Explosions Finely divided combustible solids dispersed in air encounter an ignition source Examples: flour milling, grain storage, coal mining, etc Initial dust explosion produces secondary explosions Conditions for explosion : a) particles < certain size for ignition & propagation b) particle loading between certain limits c) dispersion in air fairly uniform for propagation

84. 84 Occur in the open. Only 2 to 10% of thermodynamic energy ends up in pressure wave. Use for this class: Unconfined Explosions Prevention VCE: Vapor Cloud Explosion - sudden release flammable vapor - dispersion and mixing with air - ignition vapor cloud Flixborough - smaller inventories - milder process conditions - incipient leak detection - automated block valves

85. 85 BLEVE BLEVE: Boiling Liquid Expanding Vapor Explosion - Release large amount of superheated liquid after vessel rupture (e.g. fire) BLEVE: Explosive vaporization of a liquid at a temperature above its normal boiling point caused by container rupture. Ex: from external fire If liquid is flammable, a VCE can result Boiling liquid can behave as rocket fuel, propelling vessel fragments Fraction of liquid vaporized from Chapter 4, T o > T b

86. 86 BLEVE Liquid Vapor Vessel with liquid stored below its normal boiling point Below liquid level – Above liquid level – Effects: Blast + thermal

87. 87 BLEVE Consequences

88. 88 Mechanical Explosions Rupture of vessel containing an inert gas at high pressure. Where: W e is the energy of explosion, P is abs. gas pressure in vessel, P E is abs. ambient pressure, T is abs. temperature. Max. Mechanical Energy Eqn. 6-31

89. 89 Batch Reactor Explosion Consequences

90. 90 Overpressures Blast Origin Blast wave PI Side-on Overpressure Direct-on Overpressure

91. 91 Peak Side-on Overpressures

92. 92 Peak Side-on Overpressures

93. 93 Consequences of Explosions: Table 6-9 Peak Side-on Overpressure (psig)Consequence 0.03Large glass panes shatter 0.15Typical glass failure 0.7Minor house damage 1.0Partial house demolition 3Steel frame building distorted > 15100% fatalities 3 psig: Hazard zone for fatalities due to structure collapse. P Distance

94. 94 P Distance

95. 95 TNT Equivalency Method Scaled distance

96. 96 P Distance

97. 97 TNT Equivalency for VCEs Where: m TNT is the equivalent mass of TNT is the explosion efficiency m is the total mass of fuel E c is the heat of combustion E TNT is the heat of combustion for TNT (1120 cal/gm = 4686 kJ/kg = 2016 BTU/lb)

98. 98 TNT Equiv. - Explosion Efficiency Use a default value of unless other information is available.

99. 99 Other Methods Other methods are based on degree of congestion or confinement. Basis is that confinement leads to turbulence which increases the burning velocity. TNO Multi-Energy Model (see pages 271-274) Baker - Strehlow Model Both produce essentially the same answer. Need much more information, i.e. confinement info.

100. 100 TNT Equivalency Procedure 1. Determine total mass of fuel involved. 2. Estimate explosion efficiency. 3. Look up energy of explosion (See Appendix B in text). 4. Apply Equation 6-24 to determine m TNT. 5. Determine scaled distance. 6. Use Figure 6-23 or Equation 6-23 to determine overpressure. 7. Use Table 6-9 to estimate damage. Problem: Determine consequences at a specified location from an explosion.

101. 101 TNT Equivalency Procedure The problem with the application of this approach to exploding vapor is that: Overpressure curve developed from detonation data, i.e. TNT, and flammable vapor explodes as a deflagration. The TNT method applied to vapor explosions tends to underpredict overpressures at some distance from the explosion, and over-predicts the overpressures near the explosion. P Distance P Shock Front Detonation   Deflagration

102. 102 Example Determine the energy of explosion for 1 lb of n-butane? What is the TNT equivalent? Use an explosion efficiency of 2%.

103. 103 Example

104. 104 Example

105. 105 TWA - 800: July 17, 1996

106. 106 TWA - 800: July 17, 1996

107. 107 Example Determine equivalent TNT mass for TWA 800 explosion. Assume: 18,000 gallon fuel tank, P = 12.9 psia, T = 120 F, Concentration of fuel = 1%, Energy of explosion for jet fuel = 18,850 BTU/lb, M = 160. Mass of fuel in vapor:

108. 108 Example Moles of fuel = (0.01)(4.99 lb-moles) = 0.0499 lb-moles = 7.98 lb of fuel Assume 100% efficiency (confined explosion).

109. 109 Questions?

110. 110 Flammability Diagram - 3 Air line always extends FROM: Fuel: 0%, Oxygen: 21% Nitrogen: 79% TO: Fuel: 100%, Oxygen: 0%, Nitrogen: 0% Equation for this line: Fuel = -(100/79) Nitrogen + 100

111. 111 Fuel/Air Explosive CBU-72 / BLU-73/B Fuel/Air Explosive (FAE) The the 550-pound CBU-72 cluster bomb contains three submunitions known as fuel/air explosive (FAE). The submunitions weigh approximately 100 pounds and contain 75 pounds of ethylene oxide with air-burst fuzing set for 30 feet. An aerosol cloud approximately 60 feet in diameter and 8 feet thick is created and ignited by an embedded detonator to produce an explosion. This cluster munition is effective against minefields, armored vehicles, aircraft parked in the open, and bunkers. During Desert Storm the Marine Corps dropped all 254 CBU-72s, primarily from A-6Es, against mine fields and personnel in trenches. Some secondary explosions were noted when it was used as a mine clearer; however, FAE was primarily useful as a psychological weapon. Second-generation FAE weapons were developed from the FAE I type devices (CBU-55/72) used in Vietnam.