Evaluating Vapor Cloud Explosions Using a Flame Acceleration Method 7 42

FM Global Property Loss Prevention Data Sheets

7-42 October 2012 Page 1 of 24

EVALUATING VAPOR CLOUD EXPLOSIONS USING A FLAME ACCELERATION ACCELERATION METHOD

Table of Contents Page 1.0 SCOPE SCOPE ..................................................................................................................................................... 3 1.1 Change Changes s ......... ................... .................... .................... ..................... ..................... .................... .................... .................... .................... .................... .................... ..................... ................... ........ 3 2.0 LOSS PREVENTION RECOMMENDAT RECOMMENDATIONS IONS ........................................................................................ 3 2.1 Human Eleme Element nt .......... .................... ..................... ..................... .................... .................... .................... .................... .................... .................... ..................... ..................... ............... ..... 3 2.2 Constr Construction uction and Locati Location on ......... ................... .................... ..................... ..................... .................... .................... .................... .................... .................... ..................... ........... 4 3.0 SUPPORT FOR RECOMMENDATIONS ................................................................................................. 4 3.1 Va Vapor por Cloud Explos Explosion ion Princi Principles ples .......... .................... .................... .................... .................... .................... ..................... ..................... .................... ................... ......... 5 3.1.1. Lami Laminar nar Burni Burning ng Ve Velocity locity (LBV) .......... .................... .................... ..................... ..................... .................... .................... .................... .................... ............ 5 3.1.2 Flame acceleration .................................................................................................................. 5 3.1.3 Blast Effect ......... ................... ..................... ..................... .................... .................... .................... .................... .................... .................... ..................... ..................... ............... ..... 7 3.2 Estima Estimating ting the Vapor Cloud Explos Explosion ion Hazard .......... .................... .................... .................... .................... .................... ..................... .................... ......... 7 3.2.1 Source Materials ...................................................................................................................... 7 3.2.2 Relea Release se .......... .................... .................... .................... .................... ..................... ..................... .................... .................... .................... .................... .................... ................... ......... 9 3.2.3 Dispersion and Drifting of Released Vapor Clouds ............................................................... 11 3.2.4 Accumulation Area ................................................................................................................. 11 3.2.5 Igniti Ignition on Source Sources s .......... .................... .................... ..................... ..................... .................... .................... .................... .................... .................... ..................... .............. ... 15 3.2.6 PES Volume Threshold ......................................................................................................... 16 3.2.7 Energy Relea Release se ......... .................... ..................... .................... .................... .................... .................... .................... ..................... ..................... .................... ............... ..... 17 3.2.8 Blast Effect Effects s and Ensuin Ensuing g Fire Damag Damage e ......... ................... .................... .................... ..................... ..................... .................... ................... ......... 17 3.3 Loss Histor History y ........... ..................... .................... .................... .................... .................... .................... ..................... ..................... .................... .................... .................... .................... .......... 17 3.3.1 Overview and Statistics ......................................................................................................... 17 3.3.2 Proces Process s incid incidents ents .......... .................... .................... .................... .................... .................... ..................... ..................... .................... .................... .................... ............ .. 18 3.3.3 High-Pressure Transmission Pipeline Incidents .................................................................... 19 3.3.4 Fixed Storage Tank Incidents ................................................................................................ 20 3.3.5 Mobile Transportation Incidents .................. ... .............................. ............................. ............................. .............................. ............................. ................ 21 3.4 Preven Prevention tion and Contro Controll of Conseq Consequences uences ......... .................... ..................... .................... .................... .................... .................... .................... ................ ...... 22 4.0 REFERENCES ....................................................................................................................................... 22 4.1 FM Globa Globall .......... .................... .................... .................... .................... .................... .................... ..................... ..................... .................... .................... .................... .................... ............... ..... 22 4.2 Other ................................................................................................................................................ 22 APPENDIX APPEND IX A GLOSS GLOSSARY ARY OF TERMS ....................................................................................................... 23 APPENDIX B DOCUMENT REVISION HISTORY ....................................................................................... 24

List of Figures Fig. 1. Simplified VCE mechanism ................................................................................................................. 5 Fig. 1a. Sketch of flame propagation causing expansion and flow ahead of the flame .......... .................... .................... ............ .. 6 Fig. 2. Flame acceleration in obstructed channel caused by flame area increase and turbulence turbu lence generation generation ahead of the flame ......................................................................................... 6 Fig. 3. Examples of experimental pressure and impulse profiles from VCEs .............................................. 7 Fig. 4. Decision tree to determine need for a vapor cloud explosion study ................................................. 8 Fig. 5. No confinemen confinementt ................................................................................................................................. 13 Fig. 6a and 6b. Partia Partiall singl single-surfa e-surface ce confin confinement ement ..................................................................................... 13 Fig. 7a and 7b. Single-surface confinement ................................................................................................ 14 Fig. 8a and 8b. Multi-surface confinement .................................................................................................. 14 Fig. 9a and 9b. Low congestion .................................................................................................................. 15 Fig. 10a and 10b. Medium congestion ........................................................................................................ 15

©2012 Factory Mutual Insurance C ompany ompany.. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, or transmitted, in whole or in part, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without written permission of Factory Mutual Insurance Company.

7-42 Page 2

Evaluating Vapor Cloud Explosions FM Globa Globall Prope Property rty Loss Preve Prevention ntion Data Sheet Sheets s

Fig. 11a and 11b. High congestion .............................................................................................................. Fig. 12. Process area damage at Flixborough .......................................................................................... Fig. 13. Damag Damage e to Phillips plant following following the VCE ................................................................................... Fig. 14. Blow-down drum at BP, Texas City (photo from the U.S. CSB final report) ................................... Fig. 15. Buncefi Buncefield eld terminal; fire follo following wing VCE ...........................................................................................

16 18 19 20 21

List of Tables Table 1. Threshold Matrix (PES Volume and Mass Released) * ................................................................... 17

©2012 Factory Mutual Insurance Company Company.. Al l rights reserved.

Evaluating Vapor Cloud Explosions FM Global Property Loss Prevention Data Sheets

7-42 Page 3

1.0 SCOPE

This loss prevention data sheet describes the theory and practice of a flame acceleration methodology (FAM) to evaluate the effects of a vapor cloud explosion (VCE) as well as basic approaches that can be taken to minimize the chance of such an event. This data sheet includes rules and thresholds that produce VCE in occupancies that present the greatest likelihood of occurrence, such as confined, congested outdoor process areas. One exception is pyrophoric gases, where open area releases may create significant blast effects. Events in forests and rural areas are excluded. Flammable vapor releases with accumulation and delayed ignition inside a building will likely have widespread damaging effects in the building and possibly impact the nearby area. Refer to Data Sheet 1-44, Damage Limiting Construction , for information on protection against this exposure, and Data Sheet 7-32, Ignitable Liquid Operations , for information on identifying the potential for building explosion hazards. 1.1 Changes

October 2012. This document has been completely rewritten to reflect the use of a flame acceleration method (FAM) rather than a TNT equivalency method to predict a VCE. FAM analysis uses the flammable cloud outside the release structure and pressure-impulse analysis, rather than overpressure only, to determine damage. The only major change in excluded materials is natural gas, which is now recognized as a possible VCE source; however, large amounts are needed and this is reflected in the release amount thresholds given in Section 3.2.6. Threshold quantities for completing a study have been revised. 2.0 LOSS PREVENTION RECOMMENDATIONS

The following recommendations are directed at methods to maintain the integrity of the chemical process with the goal of preventing uncontrolled releases of flammable vapors that could result in a VCE. 2.1 Human Element

2.1.1 Implement inherently safer design strategies. Refer to Data Sheet 7-43, Loss Prevention in Chemical Plants , for guidance on inherently safer design. 2.1.2 Operate chemical process industry (CPI) facilities under principles of process safety management (PSM). Refer to Data Sheet 7-43, Loss Prevention in Chemical Plants , for guidance on PSM. In particular, address the following: A. Conduct process hazard analysis (PHA) on any operation involving processing or handling of flammable or reactive chemicals. The PHA needs to consider the effects of the chemicals used and both normal and abnormal operating conditions as well as the impact of organizational and staffing changes. B. Implement a comprehensive management of change (MOC) program. Specifically address temporary modifications and clearly define “replacement-in-kind.” C. Conduct pre-startup safety reviews (PSSR) after implementing process changes, the addition of any new equipment, and after extended operational outage to ensure all equipment is ready for operation as expected. D. Train operators to handle normal, upset, and startup procedures. E. Investigate and analyze all process upsets and near-misses to determine their root causes and to develop and implement procedures to prevent reoccurrence. 2.1.3 Provide additional operators and supervision during startup and other nonstandard operations. 2.1.4 Limit and strictly control the ability to disable or bypass process safety control systems and interlocks as follows: 2.1.4.1 Use a written tag-out procedure that is authorized at management level and tracked by a responsible person with follow-up to ensure the disabled or bypassed device is cleared.

©2012 Factory Mutual Insurance Company. Al l rights reserved.

7-42

Evaluating Vapor Cloud Explosions

Page 4


2.1.4.2 Attach warning tags to the disabled or bypassed device where possible. 2.1.4.3 Post a follow-up card or other reminder where readily visible to remind operators and staff of the impaired device and the importance of return to service; for example, in the control room of the affected system. 2.1.5 Implement a positive material identification (PMI) program as a key component of a mechanical integrity (MI) program to ensure materials suitable for the service are installed in vessels, equipment. and piping. 2.1.6 Implement preventive and/or predictive maintenance programs in accordance with Data Sheet 9-0, Maintenance and Inspection . In particular, for piping and equipment handling ignitable liquid and flammable gas, address the following: A. Maintain an inspection, testing, and repair database. B. Implement proper bolting procedures to ensure all required bolts are in place, properly tightened, and with gaskets suitable for the service. C. Eliminate open-ended piping wherever possible by installing blind flanges or other reliable sealing mechanisms. A single closed valve is not a suitable mechanism. D. Implement a program to ensure the integrity of the piping and hangers, to evaluate the security of all connections, and to assess physical conditions with respect to erosion, corrosion, and leakage. E. Label or color-code all piping to ensure recognition of the hazardous contents and reduce the chance of introducing incompatible materials (see ANSI/ASME A13.1, Scheme for Identification of Piping Systems). F. Analyze repair records for trends with a view to inhibiting long-term deterioration, or repairing or replacing suspect equipment. G. Require repair of malfunctioning equipment before permitting a return to operation. 2.1.7 Prepare written emergency response plans for unexpected releases of ignitable liquid or flammable vapor or gas; train operating personal in their expected actions, and hold emergency drills to see that the actions can be performed effectively and accurately. 2.2 Construction and Location

The rules for developing the VCE event should not be used to establish space requirements for chemical process units. Such spacing could only reasonably be applied in new construction and should be established by conducting a risk analysis. Additional guidance on space separation can be found in Data Sheet 7-44, Spacing of Facilities in Outdoor Chemical Plants . The Center for Chemical Process Safety (CCPS) also publishes guidelines for spacing and design of important process and utility buildings in chemical plants (see Section 4.2). 3.0 SUPPORT FOR RECOMMENDATIONS

The TNT methodology was based on the results of testing on high explosives conducted by military agencies to predict blast effects with different masses of explosive. It was then applied to vapor cloud explosions, a different type of event, as the simple way to predict consequences. In the TNT equivalency method, a mass of flammable gas is related to an equivalent mass of TNT by comparing heats of combustion and assigning an efficiency factor. However, there are several major drawbacks to the TNT methodology. The efficiency factor can only be estimated. TNT produces a point source detonation, while a vapor cloud usually covers a large area and begins as a deflagration. TNT produces a sharp rise in overpressure and has a very short duration, while a VCE produces a more gradual rise in pressure over a longer period of time. As a result, the TNT methodology over-predicts VCE pressure in the near field (within a few hundred feet [meters] of the epicenter), and under-predicts it in the far field.



7-42 Page 5


3.1 Vapor Cloud Explosion Principles

Flame speed in a vapor cloud enhanced by confinement and/or congestion and the expansion of the burned gases determine the pressure-impulse (P-I) effects that can be produced during a VCE. In a vapor cloud deflagration (most common), the flame propagates through the unburned fuel-air mixture at subsonic speed. A flame accelerated to a speed greater than about 0.2 times the speed of sound (approximately 70 m/sec) is generally needed to produce damaging P-I effects. At flame speeds lower than this, the vapor cloud will usually burn as a flash fire. Flash fire effects are direct flame and thermal radiation heat transfer, without significant P-I effects. In a vapor cloud detonation (rare), the flame propagates through the unburned fuel-air mixture at supersonic speed. Figure 1 provides an overview of the steps leading up to a VCE.

Release Dispersion Delayed ignition

Very strong ignition

Flame acceleration Strong

Weak Low speed deflagration

High speed deflagration

DDT

Detonation

Flash Fire

Vapor Cloud Explosion

Direct flame and radiation thermal effect

Pressure-Impulse (P-I) effects Direct flame and radiation thermal effect

Fig. 1. Simplified VCE mechanism

3.1.1. Laminar Burning Velocity (LBV)

Flammable gases and vapors have a physical property known as the laminar burning velocity (LBV). It is the speed at which a flame moves through an unburned fuel-air mixture and is measured in a quiescent, non-turbulent state. The maximum LBV occurs at an optimum fuel concentration, which is close to the stoichiometric concentration for most fuels. While LBV is a fundamental property of a fuel of specific composition, flame propagation in an unintended release is influenced by many factors. 3.1.2 Flame acceleration

Propagating flames are inherently unstable and accelerate due to expansion of combustion products and flame instabilities. Common fuel-air combustion reactions expand the mixture volume by about a factor of eight. Flame instabilities (e.g., hydrodynamic, thermo-diffusive, acoustic, Richtmyer-Meshkov, and Rayleigh-Taylor) can be seen as flame surface area wrinkle effects. See Figure 1a. Confinement and congestion (see Sections 3.2.4.3 and 3.2.4.4) are the most important factors when considering flame acceleration and VCE potential. This is due to the following mechanism (see Figure 2): •

Flame speed increase depends on the expansion ratio, burning velocity, flame area, and flow cross-section.

•

Interactions of the flame with obstacles result in strong increase of the flame area and acceleration of gas flow ahead of the flame.


7-42


Page 6


Burning velocity, ST

Flow speed, U

Expansion

Flame speed, Vflame Fig. 1a. Sketch of flame propagation causing expansion and flow ahead of the flame

•

Interaction of the gas flow with obstacles also creates turbulence in the flow due to shear instabilities.

•

Turbulence increases burning velocity and gas flow velocity ahead of the flame.

•

The gas flow ahead of the flame and obstacle-induced turbulence are due to the expansion of the combustion products.

•

This creates the positive feedback loop and results in continuous flame acceleration and pressure rise.

•

Any confinement that prevents the combustion products’ expansion in directions other than that of the flame propagation promotes further flame acceleration.

Flame

Flow

Turbulence Fig. 2. Flame acceleration in obstructed channel caused by flame area increase and turbulence generation ahead of the flame

Flame accelerating turbulence can also be directly initiated from a very high-pressure fuel release, which is known as pre-existing turbulence. In rare cases, flame acceleration can even occur without obstacles or preexisting turbulence due to hydrodynamic flame instabilities that are relatively weak. This process requires long run-up distances for a flame to accelerate to speeds that can produce measurable overpressures. Such run-up distances are impractically large for most source materials.



7-42 Page 7


3.1.3 Blast Effect

A blast effect is created when the chemical energy in the fuel-air mixture is released as the flame moves through the cloud from the ignition source. This energy is released in the form of thermal energy of the combustion products, mechanical energy of expanding gas movement and thermal radiation. The gas movement, if sufficiently intense, generates a blast effect that can be damaging to process structures, buildings, and equipment. The mechanical energy generated during combustion is highly dependent on flame acceleration within the cloud as described previously. A pressure wave of finite amplitude is generated in air by a rapid release of energy. Usually, the initial pressure wave from a vapor cloud explosion has a smooth pressure rise with time, which is followed by pressure decrease (see Figure 3, example on left). Under some conditions of confinement, congestion, and turbulence, the pressure wave builds sharply until a shock front is formed (see Figure 3, example on right). The two parameters of the blast wave that are of primary importance for evaluation of its damage potential are the maximum overpressure (Pmax) and the impulse (I) of the positive phase of the pressure-time curve.

Pmax Pmax

l p

l

p

t

t

Fig. 3. Examples of experimental pressure and impulse profiles from VCEs

3.2 Estimating the Vapor Cloud Explosion Hazard

For a VCE with damaging overpressures to occur, the following conditions need to be present: •

Inventory of a sufficient amount of source material (fuel) in a pipeline, process, storage, or transportation system (see Section 3.2.1)

•

Release of the source material from its containment, vaporization if the source material is liquefied, and mixing of the source material with air forming a vapor cloud with some part of the cloud in the flammable range (see Section 3.2.2)

•

Dispersion or drifting of the vapor cloud into a potential explosion site (PES) (see Section 3.2.3)

•

Delayed ignition of the cloud followed by flame acceleration through the PES and the cloud; important conditions within an ignited vapor cloud that must be present for flames to accelerate are spatial confinement and congestion (see Section 3.2.4)

Use Figure 4 to decide whether the fundamental criteria that create a potential for a VCE are present at a specific site. Where the preliminary analysis indicates the potential for a vapor cloud is present, further evaluation should be conducted to determine if there is the potential to release a sufficient amount of that source material. 3.2.1 Source Materials

A source material is an ignitable liquid or flammable gas and could be a chemical process feedstock, intermediate, product or by-product, or used in a support system. The following materials should be considered to present a VCE exposure unless specifically excluded in Section 3.2.1.1:


7-42


Page 8


Process, storage, or piping system contains flammable gas, liquefied flammable gas, or ignitable liquid

No

No VCE study needed

Yes

Flammable gas or liquefied flammable gas

Yes

Material excluded per Section 3.2.1.1?

Yes

No No

Ignitable liquid with flash point less than 200°F (93°C)

Yes

Heated above atm boiling point and kept liquid by pressure

No

Yes

No

No VCE study needed

Fig. 4. Decision tree to determine need for a vapor cloud explosion study

•

Flammable gas or liquefied flammable gas at any temperature and pressure


Go to threshold matrix Table 1

Evaluating Vapor Cloud Explosions FM Global Property Loss Prevention Data Sheets

•

7-42 Page 9

Ignitable liquids with flash points below 200°F (93°C) processed above their atmospheric boiling point and maintained as a liquid by pressure

Mixtures of two or more liquid or gaseous materials may be present. There is no easy solution for dealing with mixtures, and judgment must be used. Generally, the more hazardous material should be selected as if it were the entire volume. Material reactivity is determined by fundamental properties of mixtures of flammable gases and vapors with air. The most important of these properties include the LBV, heat of combustion, ratio of densities between reactants and products, and flammability limits. For the purposes of qualitative comparison of various materials, one can approximately attribute materials to low, medium, high, and very high ranges of reactivity. •

Low reactivity materials (20 cm/sec ≤ LBV < 40 cm/sec) include flammable gases such as methane and ethylamine.

•

Medium reactivity materials (40 cm/sec ≤ LBV < 60 cm/sec) include flammable gases such as propane, propylene, butane, and vapors of boiling ignitable liquids such as cyclohexane, naphtha, and gasoline.

•

High reactivity materials (60 cm/sec ≤ LBV < 100 cm/sec) include ethylene, ethylene oxide, propylene oxide, and vinyl acetylene.

•

Very high reactivity materials (LBV

≥

100 cm/sec) include hydrogen, silane, and acetylene.

3.2.1.1 Excluded Materials FM Global has interpreted that the following materials do not present a significant or credible outdoor VCE exposure. These exclusions are based on many factors including heat of combustion, LBV, and loss history. •

Materials with an LBV less than 20 cm/sec, which includes ammonia, carbon monoxide, coke/coal gas and blast furnace gas

•

Synthesis gas used as feed for ammonia, methanol, and other processes

•

Gaseous hydrogen at any temperature and pressure

•

Materials processed above their auto-ignition temperature

•

Ignitable liquids with a viscosity greater than 1 x 10 5 centipoise (100 Pa-s) (e.g., bitumen, pitch)

3.2.1.2 Hydrogen: Liquid Hydrogen is highly reactive and should be considered a VCE source material when stored or used as a liquid. 3.2.1.3 Silane and Other Pyrophoric Gases Pyrophoric gases commonly ignite immediately upon release, but small-scale incidents and research have demonstrated that delayed ignition can occur with an uncontrolled release of high purity gas. 3.2.1.4 Hybrid Mixtures VCEs have occurred in hybrid systems (mixture of solids/powders in a flammable gas), notably at polyolefin manufacturing plants (Phillips, Texas City, TX, 1989). While it is possible that the suspended dust enhances the explosion effect, there is no model to account for this. The flammable gas is the major component of the release and the VCE should be evaluated assuming a pure gas is released. 3.2.2 Release

The release phase of the scenario is dependent on the duration of the release, the size of the release opening, the physical properties of the source material, operating pressure and temperature and, to a small degree, the outside environment. In determining the release source, primary emphasis should be on looking for release points within a PES. Only when these obvious sources in the PES don’t exist should one look for piping or large tanks outside the PES where drift and dispersion are needed to get to a PES. While there are many scenarios that can result in a release of a VCE source material, there is no attempt to consider or quantify them all. Historically, most VCEs are from a release through an existing pipe connection


7-42


Page 10


to a vessel or transport system (vehicles, pipelines, etc.). For the purpose of considering the worst credible case, only releases from piping will be analyzed as defined below. 3.2.2.1 Process System (Vessel and Piping) or Fixed Storage Tank Assume rupture of the largest interconnecting pipe or pipe connection that will allow the most discharge from the largest vessel/tank or train of interconnected vessels. Where there is a possibility of release from a liquid or a vapor space, in almost all cases an opening in the liquid space will create the largest mass of released material. (See Section 3.2.2.5 to determine if credit may be taken for flow-limiting devices and/or block valves.) 3.2.2.2 Supply Pipelines Supply pipelines are those that carry flammable gas or ignitable liquids from remote sources to supply one or multiple customers. Aboveground pipelines containing flammable gas or ignitable liquids can be considered as a release source. Assume the pipeline is completely severed (full-bore break with flow from both ends). Do not credit any automatic or manual block valve with limiting this flow. Buried flammable gas pipelines should be considered as a release source only when the pipeline pressure is greater than 250 psig (17 barg). While there is significant history of failures of buried pipelines allowing for a massive release of material, ignition of the release is usually prompt. 3.2.2.3 Mobile Transportation Vehicles When considering the release from mobile vessels such as railcars, trucks, ships, and barges assume breakage of the largest connected line during transfer operations with gravity and storage pressure as the driving force. (See Section 3.2.2.5 to determine if credit may be taken for flow-limiting devices and/or block valves.) 3.2.2.4 Release Duration The release duration is modified depending on the PES congestion. This modification is based on studies conducted to determine the likelihood of ignition for ordinary spills and releases of gas and ignitable liquid in chemical and petroleum plants. For the purposes of this document, the release duration should be considered 10 minutes for low congestion PES, and 5 minutes for medium and high congestion PES. Cryogenic (liquid) hydrogen should be considered as a normal release with duration based on the congestion of the PES as noted above. For liquid or gaseous silane and pyrophoric gases the duration should be 30 seconds. 3.2.2.5 Flow-Limiting Devices and Block Valves Flow-limiting devices, such as orifice plates, pipe reducers, and excess flow valves (EFVs), may be credited to limit the amount of material released if they are reliably installed and protected against mechanical damage. Excess flow valves rely on moving parts to achieve shutoff and there are a number of failure modes that would result in discharge at a reduced rate without shutoff. This may occur due to blockage by foreign objects, mechanical failure, or simply piping that limits the flow to less than the closing capacity. Use the maximum opening inside the EFV body (i.e., opening acts as an orifice) when developing the discharge rate. Credit can be given for either internal or external devices. To credit an external EFV it should be welded to the vessel or bolted to the first flange of any pipe connection on the vessel In order to credit flow-limiting devices, all connections to the vessel should be provided with flow-limiting devices EXCEPT those with pressure-relief devices. Generally, the presence of emergency shutoff valves in a chemical process system, even if automatic, cannot be credited for limiting the flow from a multi-vessel system. Considerable judgment is needed on very large systems with redundant automatic block valves.



7-42


Page 11

3.2.2.6 Partially Full Systems and Vessels Process systems and tanks may not be used or filled to capacity due to process or safety considerations. When it can be reliably concluded that this is a normal condition, as verified by operating records, standard operating procedures, etc., and where management systems are in place to ensure changes are not made without awareness of plant safety professionals, then the actual amount in the vessel or system may be used for determining maximum release, rather than the nominal (full capacity) amount. When solid material (e.g., catalyst) represents a significant portion of the volume of a reaction system, then the volume occupied by the solid should be excluded in the calculation of the liquid release volume. In many polymerization processes, the polymerization begins very quickly after catalyst introduction. The liquid or gaseous monomers and co-monomers are usually unreacted in the process system for only a short time prior to reaction initiation. At these times, pressures and temperatures may be substantially lower and conditions less severe than under normal operating conditions. Judgment must be used when developing a worst-case scenario. Usually the partially full operating conditions are a more reasonable choice for worst case. 3.2.3 Dispersion and Drifting of Released Vapor Clouds

For the purposes of this standard, a hazardous cloud is one that reaches the LEL and enters the flammable range, as opposed to a cloud at ppm levels that presents a hazard to people or the environment. Dispersion of the released cloud is highly dependent on the physical properties of the source material, including the force and direction of the release, weather conditions, surrounding terrain, and physical plant layout. High momentum releases, such as from an opening in a pressurized pipe or vessel, that do not directly impact the ground or other solid surface, will create an extended cloud in the flammable range. The flammable envelope develops in a matter of seconds after the release starts and then becomes reasonably stable in size for the duration of the release. Low momentum clouds, such as those coming from a pool of released liquid or from a pressurized jet release, impacting directly on the ground may form large clouds over hundreds of ft (meters) from the release. These clouds take longer to develop due to the action of wind and mixing with air. The portion of a hot liquid not flashed on release can be included in the cloud volume by estimating the pool vaporization, but this portion is usually small compared to the original release and flash. Instead of the prior technique of a table providing fixed drift distances depending on the mass released, numerical computing methods are recommended to estimate the size of flammable clouds from releases of gas/vapor, atomized liquid, and vaporizing pools, but are not part of this standard. Assume the release of a source material within a PES is capable of filling the PES with a uniform stoichiometric mixture. The effect of piping, small platforms, and vessels is not considered to have an impact on the distribution of this cloud within the PES. Drift calculations are unnecessary when a release can occur within a PES. (This also assumes the pipe is capable of releasing the necessary quantity of material in the defined duration.) Where there is no feed pipe or vessel in the process area (PES) that can be a source of a release material for a vapor cloud, drift from large fixed or mobile storage vessels should be considered. For the purpose of these studies, atmospheric conditions (in particular, normal or prevailing wind direction) should not be considered as a deciding factor on the cloud drift direction. Where cloud drift must be considered to reach a PES, assume whatever direction is necessary to move across open areas from the release point to the PES. Do not artificially force the cloud around or through buildings or through other areas containing process pads that could break up or impede the drift. 3.2.4 Accumulation Area

3.2.4.1 Potential Explosion Site (PES) The PES is a defined volume with sufficient confinement and/or congestion to create flame acceleration. A vapor cloud generally needs confinement and/or congestion to transit from a slow flame (flash fire) to a fast deflagration or detonation. Confined, congested, elongated outdoor process structures provide the best


7-42


Page 12


conditions to produce these results. A PES could also be an open or partially open process structure, an open equipment pad, or heavily-loaded, multi-level pipe racks where the multiple layers of pipes can create the effect of a solid roof/confinement. Another way for the vapor cloud to transit from slow flame to a fast deflagration is with long run-up distances that can be achieved with a very large cloud and limited congestion or confinement or a small cloud of highly reactive fuel. With low-reactivity fuels, a very large cloud is needed to achieve the same blast effect as a cloud in a relatively small PES. Where the PES has a volume of less than approximately 35,000 ft3 (1,000 m3), the VCE potential need not be evaluated unless a large cloud of source material can be released (see Section 3.2.6). This is not an indicator that a VCE cannot occur; only that severe damage is unlikely to extend beyond the unit initially involved with damage, similar to an uncontrolled fire event. Because material releases are typically located near ground level, for the purposes of this document, structure and equipment over 40 ft (12 m) will not be included in the PES volume. Often very tall process structures have very low congestion at the higher levels, contributing little to flame acceleration. In evaluating the effects of a VCE, the epicenter of the explosion will be considered as the approximate center of the congested PES. Damage rings will be determined based on this epicenter. Epicenter locations in non-congested areas will not be considered. 3.2.4.2 Selecting a PES Where there is more than one PES on a plant site, it may be necessary to analyze more than one scenario to determine which presents the worst case. Because confinement and congestion can cause effective flame acceleration, plant layout can directly affect the potential for a VCE. A plant with mostly open space and minimal or small process structures generally does not have the right conditions for effective flame speed acceleration. An example is a tank farm with no pipe racks. Even though suitable amounts of source materials may be present and a vapor cloud release could occur, the layout would likely not support flame acceleration that contributes to overpressure. If the spaces between tanks or roadways were covered with canopies or dense pipe racks, the layout might have the conditions needed for a VCE. 3.2.4.3 Confinement Spatial confinement defines how a flame accelerates within the process structure. The greater the restriction on the ability of the combustion products to expand out of the PES during flame acceleration, the greater the potential pressures generated by the PES. Most outdoor chemical process structures are limited to confining horizontal surfaces such as platforms, solid or grated floors, roofs, and substantial pipe racks. Where there are substantial vertical surfaces, these should also be considered as affecting the degree of confinement. Confinement can be measured as the percentage of the PES external surface (excluding the ground) that is confined with rigid surfaces that can restrict the ability of combustion products to expand during flame acceleration. FM Global will consider three levels of confinement in evaluating the impact on flame acceleration: 1. None or partial single-surface confinement 2. Single-surface confinement 3. Multi-surface confinement 3.2.4.3.1 None or Partial Single-Surface Confinement An open process pad with only equipment and structural framing has very little or no confinement. An example is a pad with no solid roof or horizontal obstacles and no intermediate grated floors or a tank farm with well spaced vertical tanks (see Figure 5). Partial single-surface confinement is characterized by partial, but not complete horizontal blockage that prevents the flame front from expanding through the partially closed surface. Examples include pipe racks, ©2012 Factory Mutual Insurance Company. Al l rights reserved.


7-42


Page 13

Fig. 5. No confinement

grated floors, and partial intermediate solid platforms or floors. This is a typical arrangement found in many chemical process areas. (See Figures 6a and 6b.) Confinement is measured to be in the range of 0% to approximately 20%.

Fig. 6a and 6b. Partial single-surface confinement

3.2.4.3.2 Single-Surface Confinement Single-surface confinement is characterized by blockage caused by a solid horizontal or vertical surface. The flame front can expand out of the structure in two directions. Examples include extensive solid platforms above process equipment, multi-level piping arrays or cable trays that cannot be seen through, and less than full-height spaces below compressor platforms. (See Figures 7a and 7b.) Confinement is measured to be in the range of 20% to 40%.


7-42


Page 14


Fig. 7a and 7b. Single-surface confinement

3.2.4.3.3 Multi-Surface Confinement Multi-surface confinement is characterized by confinement with horizontal and vertical planes. There are many and extensive solid platforms and walls on the sides of process equipment. (See Figures 8a and 8b.) Tunnels are included in this category. This is a worst-case situation and flame can accelerate to detonation speeds. The confinement is measured to be in the range of 40% to 80%.

Fig. 8a and 8b. Multi-surface confinement

3.2.4.4 Congestion Congestion is defined by obstacle density through which the flame must move. Repeated and closely spaced obstacles, such as piping, structural columns or beams, vessels, and pumps, in the center of a cloud can provide very efficient geometry for flame acceleration. Generally, a high obstacle density will produce increased flame acceleration. There is some data to indicate that small obstacles such as pipes, columns, beams, and pumps are more effective at accelerating flame than are large process vessels (Cleaver and Shale 1999). Low congestion/obstacle density is defined as less than 2% of volume blockage and more than 6.5 ft (2 m) between obstacles. If you can walk through an area relatively unimpeded, it is likely one with low congestion. (See Figures 9a and 9b.)



7-42


Page 15

Fig. 9a and 9b. Low congestion

Medium congestion/obstacle density is defined as between 2%and 6% of volume blockage, and from 1.5 to 6.5 ft (0.5 to 2 m) between obstacles. If you can walk though an area by taking an indirect path but it is awkward to do so, it is likely one with medium congestion. (See Figures 10a and 10b.)

Fig. 10a and 10b. Medium congestion

High congestion/obstacle density is defined as more than 6% of volume blockage and less than 1.5 ft (0.5 m) between obstacles. If you cannot walk through an area, or if little light penetrates into it, then it is likely one with high congestion. (See Figures 11a and 11b.) 3.2.5 Ignition Sources

The presence or lack of ignition sources should not be considered when calculating material release. For a credible worst-case scenario, the total amount that might be spilled in the specified duration should be used in estimating the cloud size. Loss experience has shown that wind patterns may allow the formation of large clouds without their being ignited by nearby ignition sources.


7-42


Page 16


Fig. 11a and 11b. High congestion

3.2.6 PES Volume Threshold

Incident history has demonstrated that most damaging VCEs have been the result of large congested areas (PES) and release of a large fuel cloud. Minimum PES volumes and/or threshold amounts of source material subject to release are assigned by most practitioners of VCE analysis to ensure that only credible events are considered. However, just because threshold limits are met does not imply that a VCE cannot occur from a smaller PES and/or smaller source material release. At a minimum, there should be enough mass of source material released to achieve at least an average of the stoichiometric concentration throughout the PES. After determining whether a VCE prone material is available as detailed in the criteria in 3.2.1, threshold PES volumes have been set which identify credible cases for study of potential consequence from a VCE. After determining the release amount as directed in Section 3.2.2, use the following criteria (or Table 1) to determine the need for a VCE consequences study: 1. Where the PES is over 250,000 ft3 (7,000 m3), a VCE consequences study is needed for any congestion level or fuel reactivity. 2. Where the PES is between 35,000 ft3 and 250,000 ft3 (1,000 and 7,000 m3), a VCE consequences study is needed for any fuel reactivity and high or medium congestion level. 3. Where the PES is between 35,000 ft3 and 250,000 ft3 (1,000 and 7,000 m3), a VCE consequences study is needed where the congestion level is low and the fuel reactivity is high or very high. 4. Where the PES is between 35,000 ft3 and 250,000 ft3 (1,000 and 7,000 m3), a VCE consequences study is needed where the congestion level is low and the total release amount meets the following criteria: a. 60 tons (54 tonnes) or more with low reactivity (20 cm/sec ≤ LBV < 40 cm/sec) Examples: methane/natural gas, butane b. 7 tons (6.4 tonnes) or more for materials with medium reactivity (40 Examples: propane, butadiene and cyclohexane

≤

LBV < 60 cm/sec)

5. Where the PES volume is less than 35,000 ft3 (1000 m3) with any congestion level, a VCE consequences study is needed where the total release meets the following criteria: a. 60 tons (54 tonnes) or more with low reactivity (20 cm/sec ≤ LBV < 40 cm/sec) Examples: methane/natural gas, butane b. 7 tons (6.4 tonnes) or more for materials with medium reactivity (40 Examples: propane, butadiene and cyclohexane

≤

LBV < 60 cm/sec)

c. 1200 lb (540 Kg) or more for materials with high reactivity (60 cm/sec ≤ LBV < 100 cm/sec) Examples: ethylene, ethylene oxide and propylene oxide



7-42


Page 17

d. 100 lbs (45 kg) or more for materials with very high reactivity (LBV ≥ 100 cm/sec) Examples: hydrogen, acetylene and silane Table 1. Threshold Matrix (PES Volume and Mass Released) * ≥

PES Fuel reactivity Congestion Low

Congestion Medium or High

3

Low > 60 tons (54 tonnes) > 60 tons (54 tonnes)

3

< 35,000 ft (1,000 m ) Medium High Very high

>7 tons (6.4 tonnes)* > 7 tons (6.4 tonnes)

>1200 lbs (540 kg) > 1200 lbs (540 kg)

Low

> 100 lbs >60 tons (45 kg) (54 tonnes) > 100 lbs Yes (45 kg)

35,000 ft 3 (1,000 m 3 ) and < 250,000 ft 3 (7,000 m 3 ) Medium High Very high

>7 tons (6.4 tonnes) Yes

250,000 ft 3 (7,000 m 3 ) Any

≥

Yes

Yes

Yes

Yes

Yes

Yes

* Table shows criteria indicating a need to conduct VCE Study (Yes) or when a study is needed only when release amount exceeds specified values.

3.2.7 Energy Release

The chemical energy in the fuel-air mixture is released during combustion as flame expands through the cloud from an ignition source. The strength of blast waves generated during combustion depends on the rate of energy release. In this document, energy release rate is estimated using flame acceleration method (FAM) as generally represented by Baker-Strehlow-Tang or TNO multi-energy models. Flame acceleration models that include the effects of acceleration in the PES and energy release in the unconfined cloud surrounding the PES are known to provide better modeling of both near and far field effects than the TNT methods that have been used in the past. Some practitioners model only the effects produced by the PES or define multiple PES within the cloud and combine the individual components of the event. Several commercial software models are available to do the complex calculations. FM Global has developed proprietary software, BlastCalc, which is used by FM Global engineers for VCE evaluations. These models use correlations or science-based calculation models to estimate the effects of confinement, congestion, and the size and properties of the cloud on flame acceleration. FM Global will consider both the PES and surrounding cloud for VCE modeling. The flame acceleration and energy release are then used by these models to create pressure and impulsedistance profiles that can then be used to evaluate the effect on structures and equipment based on knowledge of the building and equipment response to these phenomena. Chapter 6 of the CCPS Guideline book on VCE (See Section 4.2) provides some guidance on calculation methods as well as worked examples for evaluating blast effects produced by these events. 3.2.8 Blast Effects and Ensuing Fire Damage

No specific guidance is provided in this document on blast damage effects but it can be obtained based on generally available data or specific commercial software. The BlastCalc tool provides this information for FM Global engineers using a concept of damage zones defined by the combination of overpressure and impulse. The potential for ensuing fires and secondary explosions must be considered to fully evaluate the damage potentials of a vapor cloud explosion. 3.3 Loss History 3.3.1 Overview and Statistics

Lenoir and Davenport listed 103 suspected VCE incidents that produced observable overpressures during the period 1921 to 1991 (see Section 4.2). Of these events, 96 were well documented; an FM Global review of them determined the following: A. Percentage by Release Source •

62% were process related (pumps, compressors, piping, etc)

•

17% were railcar or tank truck incidents


7-42


Page 18


•

11% were high pressure pipeline failures

•

10% were storage tank incidents

B. Percentage by Material Released •

28% involved liquefied petroleum gases other than ethylene

•

14% involved ethylene (liquid or gas)

•

11% involved ignitable liquids

•

5% involved methane/natural gas

•

3% involved hydrogen or synthesis gas

•

39% involved various other gases

3.3.2 Process incidents

3.3.2.1 Bypass Failure at Caprolactam Plant— Nypro UK Ltd, Flixborough, England This incident was the watershed event that brought the chemical process industries to the realization that flammable gas releases and subsequent explosions were capable of doing catastrophic damage to a large plant, and caused industry to investigate methods to prevent future events. On June 1, 1974, cyclohexane vapor was released after the rupture of a 20 in. (500 mm) process pipe bypassing a reactor. One of the six reactors in series began to leak and the decision was made to bypass that reactor. Bellows were installed in the piping between the reactors to address alignment issues and operated for several months. The plant was shut down to repair a leaking sight glass. The bellows in the bypass failed on startup. The VCE and subsequent fire destroyed the entire facility. Property damage was estimated at US$200 million (indexed to 2011 values). The plant was not rebuilt. (Parker 1975.)

Fig. 12. Process area damage at Flixborough

3.3.2.2 Release from Polyethylene Unit— Phillips Petroleum, Pasadena, TX, USA This event in the United States was the equivalent of the Flixborough event in the UK, in that it brought renewed effort to prevent releases that could result in a large VCE, and instigated the OSHA PSM regulation. On October 23, 1989, ethylene and isobutane were released from a ball valve on the settling leg of a vertical loop reactor. The vapor cloud drifted toward the center of the high-density polyethylene unit (HDPE) before ignition, which is believed to have occurred 1 to 2 minutes after the release. Pressure effects were evident at 6 miles (9.6 km) from the epicenter. Two HDPE units were destroyed. Investigations revealed that a maintenance lock-out device had been removed, the ball valve was open, and the settling leg was open to



7-42


Page 19

atmosphere where a spool leading to the product take-off valve should have been connected. Property damage was about US$1 billion (indexed to 2011 values). It took almost two years to restore full HDPE production capacity. (U.S. OSHA 1990.)

Fig. 13. Damage to Phillips plant following the VCE

3.3.2.3. Discharge from Atmospheric Vent—BP, Texas City, TX, USA This incident was a leading factor in the revision of API 521 (ISO 23251) in 2007, which put new restrictions and limits on the use of blow-down drums and added guidance on recognizing the way systems can present a vapor cloud hazard. (See Section 4.2.) On the morning of March 23, 2005, the raffinate splitter tower (mix of C5-C8 hydrocarbons) in the refinery’s isomerization unit was restarted after a maintenance outage. During the startup, operations personnel pumped liquid hydrocarbons into the tower for over three hours without any liquid being removed, which was contrary to startup procedure instructions. Critical alarms and control instrumentation provided false indications that failed to alert operators of the high level in the tower. The 170 ft (52 m) tall tower was overfilled and liquid overflowed into the overhead pipe at the top of the tower. The overhead pipe ran down the side of the tower to pressure-relief valves located 148 ft (45 m) below. As the pipe filled with liquid, the pressure at the bottom rose rapidly from about 21 psi to 64 psi. The three pressure-relief valves opened for six minutes, discharging a large quantity of hydrocarbon liquid to a blow-down drum (Figure 14) with a vent stack open to the atmosphere. The blow-down drum and stack overfilled, which led to a geyser-like release out the 113 ft (34 m) tall stack. According to the CSB investigation, this blow-down system was an antiquated and unsafe design originally installed in the 1950s that had never been connected to a flare system to safely contain liquids and burn flammable vapors released from the process. Severe damage occurred to the isomerization unit, resulting in more than US$1.5 billion in gross loss costs. 3.3.3 High-Pressure Transmission Pipeline Incidents

Rupture of high-pressure transmission pipelines for petroleum gases and liquids is a potential VCE exposure that should be considered in chemical plants (11% of the incidents reported by Lenoir & Davenport; see Section 3.3.1). They happen on a regular basis; there have been 10 major incidents with detailed investigation reports posted on the US National Transportation Safety Board website since 2004. In most cases these


7-42


Page 20


Fig. 14. Blow-down drum at BP, Texas City (photo from the U.S. CSB final report)

incidents occur in rural, forest, or farmland areas and do not involve refinery, chemical, or petrochemical plants and are thus outside the scope of this document. Three prominent events that caused widespread damage were a natural gas liquids release in Brenham, Texas, USA (1992); a propane pipeline in Port Hudson, Missouri, USA (1970); and propane and mixed gas liquids pipeline in Siberia, Russia (1989). One event involving a refinery is summarized below. 3.3.3.1 Ethane Propane Mix—Baton Rouge, Louisiana, USA On December 24, 1989, an 8 in. (200 mm) above-ground pipeline at approx 700 psig (48 bar) carrying a mix of ethane and propane from a refinery to nearby off-site storage facilities catastrophically failed during record low temperatures. The resulting vapor cloud that was released in a remote tank farm travelled an estimated 1500 ft (460 m) across open land into the refinery, where it was ignited at a fired heater. The resulting burning cloud flashed back to the tank farm where obstacles and confinement in the form of pipe racks and a rail underpass were sufficient to cause flame speed transition to deflagration. The explosion produced blast pressures felt 15 miles (24 km) away, ruptured 17 of 70 pipelines in the pipe rack, ignited two large storage tanks containing 3.6 million gal (14,000 m 3) of diesel fuel, and 12 smaller tanks with approximately 900,000 gal (3,400 m3) of lube oil. The loss cost was estimated at US$69 million (Marsh & McLennan 2012). 3.3.4 Fixed Storage Tank Incidents

VCE events initiated as a result of release from a storage tank have been relatively few compared to other sources (10% of the incidents reported by Lenoir & Davenport; see Section 3.3.1), but a number have occurred in the recent past that have cause widespread damage. 3.3.4.1 Light Hydrocarbon Liquids—Pernis, Netherlands From 60 to 120 tons (55 to 109 tonnes) of light hydrocarbon liquids were released January 20, 1968 when a slop oil tank frothed over due to breaking of an oil water emulsion in the tank. The low-pressure tank (a



7-42


Page 21

few inches [cm] above atmospheric) failed at the seam due to the frothing action. The resulting cloud diameter was 450 ft (140 m). The explosive yield was estimated to be 22 tons (20 tonnes) TNT equivalent. The loss exceeded US$100 million in property damage. 3.3.4.2 Winter Grade Gasoline—Buncefield, Hertfordshire, UK The Hertfordshire Oil Storage Terminal supplied 8% of the oil and fuel supplies of the UK, including about 40% of the aviation fuel to London’s Heathrow and Gatwick airports. It covered 50 acres (20 hectares) and was supplied by three pipelines from a number of refineries in the UK. Starting at approximately 7 p.m. on December 10, 2005, Tank 912 began receiving gasoline by pipeline at a rate of about 2400 gpm (550 m3 /hr). From approximately 3 a.m. on December 11, the tank level gauge was unchanged but fuel continued to be received. Calculations indicate the tank began to overflow at about 5:20 a.m., and CCTV showed a vapor cloud that had spread to an adjoining commercial property and reached almost 6.6 ft (2 m) deep shortly before the 6:01 a.m. explosion. The release was estimated to be approximately 300 tons (272 tonnes) of gasoline. Official accident reports estimate that approximately 10% of this fuel vaporized. The explosion caused widespread damage in the surrounding office and residential areas (windows broken as much as 1 mile [1.6 km] away), and the subsequent fires destroyed much of the tank farm. (See Figure 15.)

Fig. 15. Buncefield terminal; fire following VCE

3.3.5 Mobile Transportation Incidents

VCE incidents causing damage to industrial facilities that have involved transportation vehicles, such as railcars, tank trucks, or waterway vessels, are not unusual (17% of the incidents reported by Lenoir & Davenport; see Section 3.3.1). 3.3.5.1 Railcar, Liquefied Petroleum Gas—East St. Louis, Illinois, USA On January 22, 1972, a railcar full of liquefied petroleum gas (LPG) (primarily propylene) rolled into another railcar during rail yard switching operations. The collision occurred at approximately 15 mph (24 kph), roughly 10 mph (16 kph) faster than normal practice. The coupling on a stationary car punctured the moving LPG


7-42


Page 22


car and released 24,300 lbs (53,500 kg) of vapor at 220 psig (15 barg) into the rail yard. The vapor cloud covered an area of more than 215,000 ft2 (20,000 m2) and reportedly ignited in two different locations. The explosion was enhanced by the tightly congested arrangement of railcars throughout the large rail yard. An estimated 1,000 buildings were damaged to various degrees. The property damage was estimated at US$45 million (U.S. NTSB 1973). 3.4 Prevention and Control of Consequences

Many features of chemical process plants control the occurrence of vapor cloud releases and explosions, but complete elimination of the hazard potential is not possible. The recommendations in this document focus on the key human element features and a mechanical integrity program to prevent the release, limit or control the size of the release, prevent ignition, and respond promptly to any unexpected events. Some additional factors that could be considered are in a number of other FM Global data sheets. Limiting quantities of hazardous materials in the operation, and locating hazardous operations remotely with respect to lower hazard ones, is usually not possible with an existing facility; new construction, however, should consider these features. A few documents and significant factors are highlighted below. •

Data Sheet 7-32 describes the proper location of ignitable liquid operations and safeguards for proper operation.

•

Data Sheet 7-43 is directed at chemical process plants and includes information on PSM and inherent safety.

•

Data Sheet 7-49 provides guidance on the proper sizing of overpressure protection systems, including proper control of the releases. Overpressure protection and process pressure control systems present an opportunity to limit the release of hazardous liquid or vapor; however, these systems can be the unplanned source of the vapor cloud release if improperly designed or operated. The incident described in Section 3.3.2.3 is an example of such an occurrence. Additional guidance on these systems can be found in CCPS guideline books on process vents and emission control systems or pressure-relief effluent handling systems (see Section 4.2).

4.0 REFERENCES 4.1 FM Global

Data Sheet 7-0, Causes and Effects of Fires and Explosions Data Sheet 7-14, Fire Protection for Chemical Plants Data Sheet 7-32, Ignitable Liquid Operations Data Sheet 7-43, Loss Prevention in Chemical Plants Data Sheet 7-44, Spacing of Facilities in Outdoor Chemical Plants Data Sheet 7-49, Emergency Venting of Vessels 4.2 Other

American Petroleum Institute (API). Guide for Pressure-Relieving and Depressuring Systems . API 521. Fifth Edition, 2007. (Identical to ISO 23251, Petroleum Petrochemical and Natural Gas Industries Pressure Relieving and Depressuring Systems . First Edition, 2006.) American Society of Mechanical Engineers (ASME). Scheme for Identification of Piping Systems . ANSI/ASME A13.1. Buncefield Major Incident Investigation Board. The Buncefield Incident, 11 December 2005: The Final Report of the Major Incident Investigation Board. Volume 1. December 2008. Center for Chemical Process Safety (CCPS). Guidelines for Evaluating Process Plant Buildings for External Explosions and Fires . 2010. Center for Chemical Process Safety (CCPS). Guidelines for Facility Siting and Layout . 2003. Center for Chemical Process Safety (CCPS). Guidelines for Pressure Relief and Effluent Handling Systems . 1998. Center for Chemical Process Safety (CCPS). Guidelines for Vapor Cloud Explosion, Pressure Vessel Burst, BLEVE and Flash Fire Hazards . Second Edition, 2010.



7-42


Page 23

Center for Chemical Process Safety (CCPS). Safe Design and Operation of Process Vents and Emission Control Systems . 2006. Cleaver, R. P., and G. A. Shale. “Mathematical Models for Gas Explosions on Industrial Sites.” Proceedings of the International Conference and Workshop on Modeling the Consequences of Accidental Releases of Hazardous Materials . American Institute of Chemical Engineers (AIChE). 1999. Daycock, J.H., and P.J Rew. Development of a Method for the Determination of On-Site Ignition Probabilities. Health and Safety Executive . RR226. 2004. Dorofeev, S. B. “A Flame Speed Correlation for Unconfined Gaseous Explosions.” Process Safety Progress . 26:2 (June 2007): p. 140-149. Lenoir, E. M., and John A. Davenport. “A Survey of Vapor Cloud Explosions: Second Update.” Process Safety Progress . 12:1 (January 1993): p. 12-33. Marsh & McLennan. 100 Largest Losses, 1972-2011 (in the Hydrocarbon Industry) . 22nd Edition. 2012. Parker, R. J. The Flixborough Disaster: Report of the Court of Inquiry. H.M. Stationary Office,1975. U.S. Chemical Safety and Hazard Investigation Board. “Statements of CSB Chairman John Bresland and CSB Investigations Supervisor Don Holmstrom Updating the Public on the Investigation of the Nov. 4 Explosion at the Silver Eagle Refinery in Woods Cross, Utah, November 17, 2009.”www.csb.gov/newsroom/ detail.aspx?nid=296 (accessed August 13, 2012). U.S. Chemical Safety and Hazard Investigation Board. Refinery Explosion and Fire , BP, Texas City, Texas, March 23, 2005. Report No. 2005-04-I-TX. March 2007. U.S. National Transportation Safety Board (NTSB). Highly Volatile Release from Underground Storage Cavern and Explosion , Mapco Natural Gas Liquids, Inc., Nov. 4, 1993. PAR-93-01. U.S. National Transportation Safety Board (NTSB). Railroad Accident Report: Hazardous Materials Railroad Accident in the Alton and Southern Gateway Yard in East St. Louis, Ill ., January 22, 1972. NTSB-RAR-1. 1973. U.S. Occupational Safety and Health Administration (OSHA). The Phillips 66 Company Houston Chemical Complex Explosion and Fire: A Report to the President. April 1990. APPENDIX A GLOSSARY OF TERMS Blast: A transient change in the gas density, pressure and velocity of the air surrounding an explosion. A discontinuous change is known as a shock wave. A gradual change is known as a pressure wave. Deflagration: A propagating chemical reaction of a substance in which the reaction front advances into the unreacted substance rapidly, but at less than sonic velocity in the unreacted material. Detonation: A propagating chemical reaction of a substance in which the reaction front advances into the unreacted substance at or greater than sonic velocity in the unreacted material. Explosion: A release of energy that causes a blast. Flash Fire: The combustion of a flammable gas or vapor and air mixture in which the flame propagates through that mixture in a manner such that negligible or no damaging overpressure is generated. Ignitable Liquid: Any liquid or liquid mixture that is capable of fueling a fire, including flammable liquids, combustible liquids, inflammable liquids, or any other reference to a liquid that will burn. An ignitable liquid must have a fire point. Impulse: A measure that can be used to define the ability of a blast wave to do damage. It is calculated by the integration of the pressure-time curve. Overpressure: Any pressure above atmospheric caused by a blast. Potential Explosion Site (PES): An open process structure with congestion and confinement that is the location of a vapor cloud release and the epicenter of the blast effects once the cloud is ignited Turbulence: A random-flow motion of a fluid superimposed on its mean flow.


7-42


Page 24


Vapor cloud explosion: The explosion resulting from the ignition of a cloud of flammable vapor, gas or mist in which flame speeds accelerate to sufficiently high velocities to produce significant overpressures. APPENDIX B DOCUMENT REVISION HISTORY

October 2012. This document has been completely rewritten to reflect the use of a flame acceleration method (FAM) rather than a TNT equivalency method to predict a VCE. FAM analysis uses the flammable cloud outside the release structure and pressure-impulse analysis, rather than overpressure only, to determine damage. The only major change in excluded materials is natural gas, which is now recognized as a possible VCE source; however, large amounts are needed and this is reflected in the release amount thresholds given in Section 3.2.6. Threshold quantities for completing a study have been revised. January 2012. Terminology related to ignitable liquids has been revised to provide increased clarity and consistency with regard to FM Global’s loss prevention recommendations for ignitable liquid hazards. May 2008. Minor editorial changes May 2005. Editorial corrections to Table 7B January 2001. Editorial corrections to Figure 2, Table 1 added to compounds. September 1998. Minor reformatting and re-issue as Data Sheet 7-42. January 1998. Editorial corrections September 1997. Editorial corrections to Table 1 April 1994. Initial issue as Data Sheet 7-0S and 7-0SC


Evaluating Vapor Cloud Explosions Using a Flame Acceleration Method 7 42

Recommend Documents