Vent sizing for fire considerations: external fire duration, jacketed vessels and heat flux variations owing to fuel composition

Journal of Loss Prevention in the Process Industries 14 (2001) 403–412 www.elsevier.com/locate/jlp

Vent sizing for fire considerations: external fire duration, jacketed vessels and heat flux variations owing to fuel composition J. Hauser *, W. Ciolek, H. Fisher, H. Forrest, M. Grolmes, S. Grossel, A. Keiter, A. Muller, F. Nazario, F.P. Nichols, J. Stipanovich, J. Wilday, J. Windhorst PROSAF Inc., 103 Yorktown Road, McMurray, PA 15317, USA Received 6 March 2001; accepted 13 March 2001

Abstract The methods for calculating heat flux to a vessel exposed to fire are well documented. API RP 520, 521, Standard 2000 and NFPA 30 are the basic industry guidelines providing correlations to calculate this heat flux. Other guidelines are available for less general cases. Given the heat flux, a relieving rate can be calculated to prevent vessel damage caused by over pressure. Considering the height and geometry variables, heat flux is calculated based on the exposed vessel wall area that is wetted by internal liquids. Some add a contribution owing to heat flux through unwetted surfaces. The above references provide guidance for the basic cause of calculating external fire heat flux to a typical vessel for the purpose of vent sizing. The following paper provides guidance for the three special external fire cases. The authors believe that these three cases are not addressed in the Codes or elsewhere, thus justifying their presentation here. These papers were presented in summary at the 2000 DIERS Users Group meetings.  2001 Elsevier Science Ltd. All rights reserved. Keywords: Vent sizing; Fire; Fire duration; Jacketed vessels; Fuel composition

1. External fire duration

1.2. Issue

1.1. Background

Both the NFPA and the API provide guidance on calculating the steady state heat flux caused by an external fire. However, they are silent on estimating how long a fire may burn. The various factors to be considered are: heat up times; fuel inventories; drainage; curbing; dikes; firefighting capabilities; and an overall risk to the public, operating personnel, the environment and to the equipment and supplies. Each of these factors and others must be considered when making design decisions on the predicted fire duration. This paper will try to elaborate on the information already published to provide designers or HAZOP teams with guidance on this difficult issue. Even in an obvious fire-risk area, it may be possible, as explained in guideline (3) below, to support designing the relief system for a non-fire contingency.

The methods for calculating heat flux to a vessel exposed to fire are well documented. API RP 520, 521Standard 2000 and NFPA 30 are the basic industry guidelines providing correlations to calculate this heat flux. Other guidelines are available for less general cases. Given the heat flux, a relieving rate can be calculated to prevent vessel damage caused by over pressure. Considering the height and geometry variables, heat flux is calculated based on the exposed vessel wall area that is wetted by internal liquids. Some add a contribution owing to heat flux through unwetted surfaces.

1.3. Guidelines * Corresponding author. Tel.: +1-724-942-3717; fax: +1-724-9423717. E-mail address: [email protected] (J. Hauser).

1. All these guidelines apply to vent sizing for the external fire case. As always, there may be additional

0950-4230/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved. PII: S 0 9 5 0 - 4 2 3 0 ( 0 1 ) 0 0 0 0 9 - 2

404

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

over pressure scenarios that may apply. If one of the alternate scenarios requires a larger vent than in the fire case, then the larger vent should be installed. If this is carried out, then the fire case and its relieving rate should still be documented. 2. Refer to the documents, “Vent Sizing for Fire Considerations for Special Equipment and Piping” (Hauser & Becker, 1995), “Vent Sizing for Fire Considerations for Jacketed Vessels” and “Vent Sizing for Fire Considerations for Heat Flux Variations Due to Fuel Composition” for additional guidelines that may apply to the case of small or unusually shaped equipment, exchangers and vessel jackets exposed to fire or fuels with unusual burning characteristics. 3. Fire hazard. The classic fire hazard is an accumulation of flammable liquid that could be ignited resulting in an extended duration pool fire that exposes the process equipment. While areas containing flammable vapors or gases can have releases resulting in fires, the fire will normally be either a flash fire of very short duration or a directed flame at the leak point that can impinge on other process equipment. The flash fire does not last long enough to heat the contents of a vessel. The impingement fire can overheat a small portion of adjacent equipment and cause failure at the hot spot — pressure relief devices will not protect the equipment from this event. Consequently, industry practice is to not size relief devices for fire exposure for facilities that handle flammable vapors or gases only (no flammable or combustible liquids). However, note that pool fires are possible and have been documented for gases liquified by pressure. LPG, e.g. is known to cause huge pool fires causing vessels to rupture because of excessive pressure and temperature. While fires can occur wherever combustible materials are handled, the likelihood of fire occurring from a high flash point, Class IIIB, combustible liquids (flash point at or above 200°F), when the liquid is not heated to within 30°F of the flash point, is very small. (The flash point is the temperature of the liquid at which a flame will flash through the vapor in air when ignited. It corresponds roughly to the temperature at which the vapor pressure of the liquid equals the lower flammable limit in air). Flash points of chemicals can typically be obtained from MSDSs. For definitions of the Flammability Classes of liquids, see NFPA 30, paragraph 1.2, Definitions. Consequently, this document does not recommend sizing relief devices for fire exposure where the only fire hazard is Class IIIB combustible liquids that are not heated to within 30°F of their flash points except for lube oil systems for large compressors or other equipment where an oil leak could be ignited by a hot bearing or another hot surface. This recommendation is supported by OSHA 29 CFR 1910.106, “Flammable and Combustible Liquids” (OSHA 1910.106 is based on NFPA 30, 1969) which excludes Class IIIB combustible

liquids. However, if a storage tank containing a Class IIIB combustible liquid or a non-combustible liquid is exposed by another fire hazard source, then fire exposure relief is to be provided in accordance with the OSHA requirements. Combustible solids are not normally considered to be a fire exposure hazard. 4. Heat up time. Heat up time (HUT) is defined as the time between the fire start with the vessel at operating conditions and the time where the vessel contents reach the bubble point temperature at relieving pressure or the ‘no return’ (or onset) temperature for reactive systems. HUT is a function of vessel geometry, vessel inventory, vessel fluid composition, the effective heat flux to vessel contents and the occurrence of any reactions which might accelerate or impede fluid temperature rise. HUT can range from just a few minutes to hours or even days. In general, HUT is compared to probable fire duration. If HUT is substantially longer, it may be a reasonable decision to design the relieving system for a non-fire contingency. In general, HUT can be calculated by modeling the system to be protected. In the simplest case, HUT may be (conservatively ignoring the heat needed to raise the vessel metal by ⌬T): HUT⫽

M CP ⌬T Q

where HUT is the heat up time (h); M the liquid inventory in vessel (lb); CP the liquid heat capacity at the midpoint of the temperature rise (BTU/lb-°R); ⌬T the temperature rise from operating temperature to bubble point or ‘no-return’ temperature at relieving pressure, °R; and Q is the steady state heat flux to the vessel (BTU/h). 5. Vessel inventory. The assumed vessel inventory can have a significant impact on the calculated heatup time. It may be necessary to consider a range of vessel inventories. Of course, when using API RP 520/521 methods, fire heat flux is a function of the vessel fill level. If the NFPA methods are used, then the heat flux values are determined independent of the fill level. However, it is probably unreasonable to use NFPA heat fluxes with low vessel fill levels when calculating heat up time. Therefore, HUTs at several inventory levels need to be checked. If the NFPA or API-2000 heat fluxes are used, then HUT is calculated at the maximum likely inventory. If this results in an HUT low enough to justify vent sizing calculations, then we can proceed and on the contrary if it results in an excessively long HUT, then HUTs at lower inventories, say at 50 and 25% of the maximum inventory are calculated. At lower inventories lower HUTs may result. Definitely at lower inventories complete vapor–liquid disengagement is more likely and the vent may be sized for all-vapor flow. As has been stated in many other places, this is probably a reasonable

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

minimum vent size for most vessels exposed to external fire. 6. Viscous systems. Be very careful when taking credit for long heat up times for highly viscous systems. Long heat up times are based on uniformly heating the entire vessel inventory to boiling or reaction onset temperature. Achieving uniform vessel fluid temperatures requires the establishment of convective currents to distribute the external fire heat from the vessel wall to its interior. As viscosity increases, the strength of these convective currents decreases. When this happens, vessel inventory temperatures become less and less uniform. At high viscosities, there can be significant differences between the fluid temperature at the vessel wall and the fluid temperature at the vessel interior. For non-reactive, boiling systems, this means boiling and pressure rise will be significantly less than in the calculated HUT. For reactive systems, it can mean that if fluid temperatures at the vessel wall reach runaway reaction onset temperatures, then the runaway can start even if vessel average temperatures are less than the onset temperature. Then, the reaction could continue even if the fire stops. In this case, it may be necessary to size the vent for an external-fire-driven runaway reaction. 7. Fire duration. A calculated fire duration is used only to determine the size of a relief device for a decomposition or runaway reaction in a vessel or for determining the possibility of the two-phase flow venting of liquid-filled vessels. For reactivity cases, if the fire is of sufficient duration, then a decomposition or runaway reaction may be initiated. On the other hand if the fire is not of sufficient duration, then the relief device is sized on the basis of an indefinite duration fire without considering the decomposition and runaway reaction as the sizing basis. The same type of reasoning is followed for determining the possibility of a two-phase flow venting. If the fire does not last long enough to cause the relief device to open, then the vent is not sized for two-phase flow, but is sized for all-vapor flow. Table 1 shows the guidelines for determining if a fire area is a high-, moderate-, or low-fire hazard. Fire durations shown later in these guidelines will be shown to depend on this degree of fire hazard in a given area. Once the fire hazard level has been established, the next step is to calculate the fire duration for each vessel of concern based on fire exposure from leakage of their own contents, leakage from vessels in adjacent areas, or pipeline leaks where the liquid flash point is less than 200°F or where liquids are being handled at or above their flash points minus 30°F. The burn area for consideration from adjacent vessels is the sum of the pad areas of the vessel of concern and the adjacent vessel. The highest calculated fire duration for each vessel is used unless it falls outside the limits for the degree of

405

fire hazard as shown in Table 2, in which case, the minimum or maximum fire duration value is used instead. Minimum and maximum durations are considered for storage areas and for process area fires. Areas that only handle non-combustibles or high flash point combustibles (Class IIIB liquids) at temperatures below 30°F below their flash points are not considered to be a fire hazard area. Liquids located close enough for a leak to spray onto hot surfaces (above 500°F) should be considered as having fire hazard potential. Experience has shown that high pressure combustible hydraulic fluids and lube oil for machine lube oil consoles can be ignited by hot bearings or other hot surfaces. The area where this type of equipment is located is considered to be a low-fire-hazard area. 8. Fire duration calculations. A great deal of experimental data are available on the burning rate of flammable liquids in a pool. The experimental burning rate is not affected by the pool diameter for fires larger than 2 m. The size of the pool is often constrained by the dike containing the flammable material. In the case of continuous spills in undiked areas, the liquid will spread and increase the burning area until the total burning rate equals the spill rate (valid for liquid hydrocarbons released at temperatures below their ambient boiling point). A fire duration can be calculated by determining the depth of the liquid in a dike or drainage (collection) area and assuming that the liquid burning rate caused by a pool fire is 1 in./7 min (Fisher & Forrest, 1994). Liquid runoff due to drains and/or trenches is usually ignored to obtain a conservative value. However, if the drainage or grading is designed to quickly remove the liquid spills to minimize exposure to a vessel of concern, the fire duration can be significantly reduced. Five situations have been identified which can result in a fire of sufficient duration to raise the temperature of the contents of storage tanks or process vessels enough to initiate a runaway reaction. A. Catastrophic failure of one tank in a dike/area which contains other tanks that are exposed to a subsequent fire. The contents of the largest tank in the dike/area and the ground surface area are used to determine the liquid depth (see Section 1.4.1). B. Self-leakage from a tank in a dike/area with the rate of leakage equal to the burning rate. The liquid burning rate and ground surface area are used to determine the leakage rate. The leakage rate and the volume of the contents of the tank are used to determine the fire duration. C. Leakage from a pipe (transfer line) with the resultant spill providing the fuel for fire exposure of the tanks in a dike/area. The leakage rate, duration of the leak, and the ground surface area are used to determine the liquid depth. An estimate is made of the time

406

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

Table 1 Fire hazard ratings for liquids Flammability classa

Handling temperature ⬍(Flash point minus 30°F)

I (FP⬍100°F)

II (100°FⱕFP⬍140°F

IIIA (140°FⱕFP⬍200°F

ⱖ(Flash point minus 30°F)

Quantityb (lb/gpm)

Fire hazard

Quantityb (lb/gpm)

Fire hazard c

ⱖ10,000/50 1000–10,000/5–50 ⬍1000/5 ⱖ10,000/50 ⬍10,000/50

High Moderate Low Moderate Low

ⱖ10,000/50 ⬍10,000/50

Low Low No Fire Hazard

ⱖ10,000/50 1000–10,000/5–50 ⬍1000/5 ⱖ10,000/50 1000–10,000/5–50 ⬍1,000/5 ⱖ10,000/50 ⬍10,000/50 ⱖ10,000/50 ⬍10,000/50

High Moderate Low High Moderate Low Moderate Low Moderate Low

IIIB (FPⱖ200°F)

a

FP means flash point. Use closed cup flash point where available. Quantity in a single vessel or in vessels manifolded together/the potential flow into the area via pipeline. c Table 1 shows the guidelines for determining if a fire area is a high, moderate, or low fire hazard. Fire durations shown later in these guidelines will be shown to depend on this degree of fire hazard in a given area. b

Table 2 Fire duration limits Duration

Minimum Maximum

Storage areas

30 min 4h

Process area fire hazard Low

Moderate

High

30 min 2h

1h 3h

2h 4h

it takes to detect the fire and stop the flow (minimum of 15 min). The calculated fire duration is the time it takes to stop the leak plus the time it takes to burn the accumulated liquid (see Section 1.4.2). D. Overfilling a tank with the resultant spill providing the fuel for fire exposure of the tanks in a dike/area. The liquid pumping transfer rate, duration of the spill (15 min minimum), and the ground surface area are used to determine the liquid depth. E. Leakage from a process vessel in a processing area that exposes another process vessel in an adjacent drainage area (see Section 1.4.3). Based on the availability of adequate fire fighting equipment, water and foam supplies, and trained personnel, minimum and maximum fire durations other than those calculated by the method described above can often be justified. The calculation for determining fire duration is: Fire Duration in Minutes ⫽

(Spill Volume in Gallons)(12 in./ft)(7 min/in.) (7.48 gal/ft3)(Burn Area in ft2)

Note: a first pass can be run using the total containment area and the largest vessel contents. If the fire duration calculated exceeds the maximum, then use the maximum. The following considerations are to calculate the burn area of the spill: A. The calculation considers the largest vessel in the containment area that contains a material with a flash point under 200°F, unless there is a material with a higher flash point handled at or above its flash point minus 30°F. B. The vessel is considered to leak at a rate equal to the burn rate. This provides the longest fire duration. C. The burn area is the area contained by curbing, diking, sloping pad, or artificial barrier, or an estimated circular area that would be expected to contain the vessel contents in the event of a failure. The area available is reduced by any area taken up by the equipment, i.e. tanks supported on pads or skirts, located within the flooded volume. D. All drains, trenches, sewers, etc. are normally assumed to be plugged. E. Vessel of concern is assumed to be fully engulfed by the fire. F. Duration of pipe line leaks are estimated for each case considering the time to detect the fire and stop the leak with a minimum duration of 15 min. G. For a process area, the burn area for the calculation is the total area of the sloped pad that directs the spills to the nearest sewer drain. H. For self-leakage cases where no natural or installed containment exists and a spill could engulf the piece of equipment in question, the spill should be modeled

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

as a circle with the radius extending from the spill point and reaching 5 ft beyond the vessel. I. To calculate the burn area size, the following information and documentation should be available: 1. Drawings that define the containment area for any spill. 2. Sketches with the number, type and size of all tanks or equipment located in the burn area. Note the size of the largest vessel containing a flammable or combustible liquid. 3. Materials being handled for process areas, or materials stored in tank farm areas. Where the spill would subject other vessels to potential fire involvement, use a radius of 30 ft for the burn area with the center of the burn area located in such a way that the vessel is engulfed by the fire. If the radius still does not engulf the vessel, then the radius should be extended to reach 5 ft beyond the vessel being reviewed with the center of the burn area located between the spill and the vessel. 9. Reduction in fire duration. Fire duration can be often reduced by isolating the equipment, by providing intermediate dikes or curbs, by sloping process areas to prevent spills from flowing towards critical vessels, and by enlarging the burn area. 10. Tank cars and tank trucks. Tank cars and tank trucks present a unique problem in determining the fire exposure potential. Because of their mobile nature, it is hard to determine the area that should be considered in calculating the fire exposure and the fuel source that should be considered. Drainage and diking cannot be considered as always being available to control fuel. For the purpose of fire duration, tank cars should be considered as being subjected to full fire involvement for a duration of 4 h and tank trucks for 2 h. 11. Fire duration in small pilot plants. In small pilot plant operations where the total fuel potential in an area is less than 50 gal (based on half an hour of feed rate plus all the interconnected vessel contents), the 30 min minimum fire duration can be replaced by the actually calculated value or 10 min, whichever is greater. 12. Maximum contents temperature with water spray protection (reactive case). Under ideal conditions, water films covering the metal surfaces of a vessel can absorb substantially all incident radiation from fire exposure to the vessel. Water spray protection requires creation of a film of water on the top, sides and bottom of protected equipment. Individual nozzles mounted on a piping grid are used to provide complete coverage of the equipment. Single water spray protection requires sufficient nozzles to provide 0.24 gpm/ft2 of coverage over the entire exposed surface of the vessel. A total of 0.04 gpm/ft2 of water is assumed to be lost owing to splashing. The remaining water is adequate to prevent excessive heating of the equipment. Double water spray protection

407

(0.48 gpm/ft2) requires twice the number of nozzles as those provided for single water spray. In both cases, nozzles must be provided every 12 ft of the vertical height and underneath horizontal vessels to ensure that the falling water film does not boil to dryness. When designing emergency relief for an uncontrolled reaction, as opposed to a boiling liquid case, the maximum wall temperature for a water-sprayed vessel can be used to eliminate the runaway reaction as a credible scenario by maintaining the temperature of the process material below the exothermic onset temperature. Typical maximum temperature estimates are given below: Single water spray: 194°F Double water spray: 158°F Note: these values have been intentionally made conservative; it has been calculated that the maximum temperature of the vessel contents because of fire exposure with single water spray coverage is 140°F. Caution: for reactivity cases, water spray protection is assumed to be effective 5 min after initiation of the fire. Small vessels can be heated substantially above the aforementioned temperatures during the the 5 min delay in the water spray application. 13. Literature. The above methods may be supplemented by the information provided by Lees (1980, sections 12.10.8 and 16.7.2) and Fisher and Forrest (1994). 14. Firefighting capabilities. Also consider the capabilities of the responding fire brigade. Is it a dedicated, on-site unit familiar with the layout, equipment and processes used? Or, does this site depend on response from the local public fire company? In either case, do they regularly practice fighting the kinds of fires likely in this unit? If so, review the practice response time data and refer to it in the design considerations documentation. Review the fire water system: are there adequate hydrants, monitors, sprinklers or water sprays? 15. In any case, keep the fire brigade personnel informed. This is especially important if a maximum fire duration is used as the basis for designing emergency relief systems. 16. The quality of fire protection of the plant structure will determine how long the structure can survive an extended fire. If the fire duration is long enough that the structure would weaken, it may not be necessary to provide fire relief for times after the structure is damaged beyond repair. At this point, the plant has been lost anyway. However, pressure relief may be appropriate to try to prevent injury to people (if they have not been evacuated) from exploding vessels. Further, depressuring the vessel by automated controls may be more appropriate than pressure relief by pressure relief valves. 17. Note that providing a properly sized relief device

408

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

does not necessarily assure that the vessel will not ultimately fail in a fire. Reclosing relief devices (pressure relief valves, etc.) maintain pressure in the vessel. If the fire continues long enough for vessels in liquid service, the metal of the vessel will become dry and develop hot spots if there is no reliable provision for water spray. Ultimately the vessel will fail near the set pressure of the device (API 520-I, parts 3.3.2, 3.3.3, D.3.2 and D.5.2.4; API 521, Appendix A). For vessels in gas service, the rate of heating of the metal must be considered (API 521, Appendix A). See part 3.19 of API 521 for systems to lower the venting pressure level and thus reduce the severity of the consequences of vessel failure. Non-reclosing devices (rupture disks, etc.) do not hold pressure on the vessels if there is no superimposed back pressure. If a fire has the potential to last long enough to cause vessel failure, other means of protection should be considered. Some possible other protection measures include water sprays, use of rupture disks instead of pressure relief valves and remotely actuated depressuring facilities. 18. When considering design options as a result of this document, it may appear desirable to avoid designing for the fire case by specifying higher vessel design pressures. In general, this option is not recommended as higher relieving pressures also involve higher relieving temperatures. Further, at elevated pressures and temperatures, fluid latent heats decline. Fire-sized relief devices protect vessels because of the cooling effect of boiling a liquid inside the vessel. In an extreme case, once the vessel fluid becomes supercritical, the boiling effect is lost entirely and vessel wall cooling only occurs through natural convection. Consequently, designing a vessel for supercritical fire relief should be avoided where possible. 19. Consider any other risk issues that may be present such as: A. Quantity and toxicity of fluids present. B. Number and locations of workers and on-site personnel. C. Proximity to public receptors such as houses, schools, hospitals, and roads. D. Consider the environmental risk if the process fluid escapes. 20. The relief system design documentation package should evaluate HUT vs fuel inventories, an estimated fire duration and the other considerations mentioned above. Try to be consistent within a plant site or corporation. Where appropriate, it may be reasonable to judge that an external fire may not cause a two-phase flow or runaway reaction over pressure event. Make sure that such a judgement has consensus in the affected operating organization and that it is well documented. Then, design

the relieving system for fire-sized, all-vapor flow as well as all other applicable contingencies. 1.4. Examples 1.4.1. Fire duration calculation example 1 Determine the fire duration for the following case: Tank A contains 20,000 gal of a reactive flammable liquid (flash point of 70°F). Tank OD=12 ft. Tank B contains 15,000 gal of a non-reactive flammable liquid (flash point of 90°F). Tank OD=10 ft. Tank C contains 30,000 gal of ethylene glycol (flash point of 240°F). Tank OD=15 ft. All three tanks are in a diked area (30 ft×80 ft with a 2 ft high dike). All the three tanks are flat bottom and have a pumpin rate of 200 gpm.

1.4.1.1. Solution The largest spill for a fire is 20,000 gal. (Ethylene glycol has a flash point of 240°F which is above 200°F and is not a hazard unless heated above 210°F.) The burn area is the area of the dike minus the area taken up by the three tanks (2032 ft2) The calculated fire duration is: 20,000 12 ⫻ ⫻7⫽111 min 7.48 2032 From Table 2, the allowable fire duration range for storage areas is 30–240 min. So, use 111 min as the fire duration time. Notes. 1. The same fire duration applies for all the three tanks. 2. The tanks for which we need to know the fire durations for reactivity considerations are Tank A and possibly Tank C. 3. The pump-in rate did not enter into the calculation because it was assumed the tanks were full and not being filled. 4. The 2 ft dike is high enough to contain the entire spill (1.3 ft needed). If the dike was only 1 ft high, it would be assumed that the spill would occur slowly enough that it would not overflow the dike and the fire duration would be the same.

1.4.2. Fire duration calculation example 2 Determine the fire duration for the following case: A 20,000 gal, 12 ft OD, flat bottom storage tank containing a reactive Class IIIB combustible liquid stored at ambient temperature is in a curbed area by itself.

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

The curbed area is 20 ft×20 ft with a 6-in. high curb and a normally open drain. A flammable liquid line passes over the curbed area; the normal flow rate in the line is 200 gpm. Several operators work in this area so it is unlikely a fire will go undetected for more than 20 min. It will take about 5 min more to shut off the flow.

1.4.2.1. Solution The Class IIIB combustible liquid is not a fire hazard as it is handled at temperatures below the flash point. However, the flammable liquid line passing through the diked area does constitute a fire hazard. The burn area is the area of the curb minus the area taken up by the tank (287 ft2). The liquid will burn at a rate of 1/7 in./min or about 25 gpm (287 ft2×1/7 in./min⬇25 gpm). As it is assumed that the drain is plugged for fire duration calculations, the liquid will overflow the curb until the flow is shut off. The fire duration will be the time from when the fire starts until the flow is shut off plus the 42 min it takes to burn off 6 in. of liquid depth. Fire duration=42+25=67 min. From Table 2, the allowable fire duration range for storage areas is 30–240 min. So, use 67 min as the fire duration time. 1.4.3. Fire duration calculation example 3 Determine the fire duration for the following case: A process vessel containing 1000 gal of flammable reactive liquid (flash point of 80°F) is located in a 20 ft×20 ft bay sloped to a central drain. A vessel in an adjacent area contains 2500 gal of combustible liquid (flash point of 120°F). It is located in a 20 ft×25 ft bay sloped to a central drain. This vessel has a feed rate of 25 gpm that could continue during fire exposure; flow out of the vessel is level controlled so it will stop on leakage from the vessel.

1.4.3.1. Solution There are two cases to examine, selfleakage and exposure from the vessel containing the reactive chemical and exposure because of the leakage from the adjacent vessel. For the self-leakage case, it is assumed that the drain is plugged and the leak fills the 20 ft×20 ft sloped area and leaks out at the burn rate. This produces the largest fire duration. The calculated fire duration is: 1000 12 ⫻ ⫻7⫽28.1 min 7.48 400 For exposure from the adjacent vessel it is assumed

409

the spill flows into both the drainage areas (total floor area=900 ft2), the drains are plugged and the leakage rate equals the burn rate. Assuming that it takes 30 min for someone to stop flow to the vessel, the total quantity that can leak out is 2500 gal plus 25 gpm×30 min=3250 gal. The calculated fire duration is: 3250 12 ⫻ ⫻7⫽40.6 min 7.48 900 From Table 1, it can be determined that (for the fluid from the 2500 gal tank) this is a moderate fire hazard area and so from Table 2, the fire duration limits are from 1 to 3 h. So, use a minimum of 1 h as the fire duration time.

2. Jacketed vessels 2.1. Background The methods for calculating heat flux to a vessel exposed to fire are well documented. API RP 520, 521, Standard 2000 and NFPA 30 are the basic industry guidelines providing correlations to calculate this heat flux. Other guidelines are available for less general cases. Given the heat flux, a relieving rate can be calculated to prevent vessel damage caused by over pressure. Considering height and geometry variables, the heat flux is calculated based on exposed vessel wall area which is wetted by internal liquids. Some add a contribution owing to heat flux through unwetted surfaces. Finally, a credit (environmental factor) is sometimes taken when the vessel is protected by an adequate insulation or other heat transfer mitigation system. 2.2. Issue How to determine the wetted surface area, heat flux and therefore relieving rates for the jacketed vessels exposed to fire? In most cases the jacket is also considered as a pressure vessel. Over pressure protection for the jacket itself must therefore be provided for the fire case. Relieving rate for the jacket vent is based on the jacket area. Heat flux into the jacket may be reduced by an environmental factor assuming an acceptable insulation system exists. This document assumes that the appropriate jacket relieving system has been designed and installed. Given that, the question remains on how to calculate the vessel relieving rate for the external fire case. This document provides guidance on whether or not the jacket itself performs as a form of thermal insulation for the vessel.

410

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

2.3. Guidelines 1. All of these guidelines apply to vent sizing for the external fire case. As always, there may be additional over pressure scenarios that may apply. If one of the alternate scenarios requires a larger vent than the fire case, then the larger vent should be installed. If this is carried out, then the fire case and its relieving rate should still be documented. 2. Refer to the document, “Vent Sizing for Fire Considerations for Special Equipment and Piping” Hauser and Becker (1995) for additional guidelines that may apply to the case of small or unusually shaped equipment exposed to fire. 3. The jacket vent should be sized for fire with heat flux based on jacket outside area. Jacket vent sizing is normally done assuming homogeneous flow in the jacket. For additional information on sizing jacket vents for the fire case, refer to Schiappa and Winegardner (1994a,b) and Forrest (1995). 4. Since the jacket volume is small, assume that any liquid contents are quickly vented. That is, when sizing the vessel vent, assume the jacket has no liquid remaining. 5. The position presented by this paper is the result of theoretical heat transfer calculations, Nichols (1998), engineering experience, engineering judgement, several drafts and re-drafts and technical discussion. There are no (to our knowledge) test data. Consequently, it has become necessary to take a conservative approach. Three styles of jacket construction were examined: half-pipe, dimple and annular. In the first two cases, there was inadequate support for taking a heat flux reduction credit owing to the presence of the empty jacket. On the other hand, for certain annular jackets, there is theoretical support for taking some heat flux reduction credit owing to the presence of the jacket. More details are given below. A. Half pipe and dimpled jackets. These jacket types have a lot of jacket metal to vessel metal contact. With these heat conduction paths plus heat transfer through the jacket space by radiation, the committee has concluded that there is little theoretical justification for expecting the jacket to provide significant heat transfer resistance. B. Annular jackets. It is expected that, under fire conditions, the primary means of heat transfer will be radiation and the establishment of convection currents in the annular space. Consequently, there may be certain geometries (inner diameter, outer diameter and vessel height or length) where these convection currents will be negligible. While theoretical calculations may support taking heat transfer resistance credit for some geometry range, there will still be the question of test data support. As noted above, this committee is unaware of any test data on this subject.

However, for vessels with annular jackets constructed according to the German DIN 28136, the following environmental credits are suggested: 1. Use a jacket environmental factor of 0.7. 2. Calculate the vessel effective area as wetted area not under jacket+0.7×wetted area under jacket. 3. Then apply an insulation environmental factor of 0.3 to the effective area (if properly insulated). 4. These credits are suggested only for vessels with unbaffled jackets such as those built according to the German DIN 28136. 6. In conclusion, when sizing vessel vents for the external fire case for almost all vessels, this committee recommends taking no heat transfer resistance credit for the presence of a vessel jacket. Therefore, when determining the heat flux for most jacketed vessels, calculate the exposed area as though the jacket is not present. 7. As usual, heat flux to the vessel may still be reduced by an appropriate environmental factor if there is an acceptable insulation system covering the jacketed vessel. The environmental factor chosen may come from API RP 520, NFPA 30 or elsewhere. The value of the environmental factor used is also outside the scope of this document.

3. Heat flux variations owing to fuel composition 3.1. Background The methods for calculating heat flux to a vessel exposed to fire are well documented. API RP 520, 521, Standard 2000 and NFPA 30 are basic industry guidelines providing correlations to calculate this heat flux. Other guidelines are available for less general cases. Given the heat flux, a relieving rate can be calculated to prevent vessel damage caused by over pressure. Considering height and geometry variables, the heat flux is calculated based on exposed vessel wall area which is wetted by internal liquids. Some add a contribution owing to heat flux through unwetted surfaces. These API and NFPA heat flux correlations are based on tests conducted using petroleum refinery stocks as the fuel. Is it possible that the heat flux to a vessel exposed to a significantly different fuel fire is higher or lower than these correlations predict? 3.2. Issue These NFPA and API heat flux correlations were based on test data where the fluid burning outside the exposed vessel was a petroleum-based hydrocarbon mix such as gasoline or kerosene. It is well known that other fluids have substantially different heats of combustion.

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

411

In fact, there is already some code guidance on this subject. NFPA 30, 2-3.6.7 permits an additional environmental factor of 0.5 for ethanol systems. This document will expand on that point for other fluids and provide a more detailed, quantitative technical basis for calculating a burning rate factor. This document will focus on heat flux reductions based on fuel compositions. If a particular fuel can cause significantly higher heat rates than typical hydrocarbon fuels, then this aspect should be investigated separately.

flux reduction factor be lower than 0.15 under NFPA methods or 0.026 under API methods. 9. The burning rate factor is calculated using the following formula

3.3. Guidelines

BRFⱕ1

1. All of these guidelines apply to vent sizing for the external fire case. As always, there may be additional over pressure scenarios that may apply. If one of the alternate scenarios requires a larger vent than the fire case, then the larger vent should be installed. If this is carried out, the fire case and its relieving rate should still be documented. 2. Refer to the documents, “Vent Sizing for Fire Considerations for Special Equipment and Piping” Hauser and Becker (1995), and “Vent Sizing for Fire Considerations for Jacketed Vessels” for additional guidelines that may apply to the case of small or unusually shaped equipment, exchangers and vessel jackets exposed to fire. Refer also to Center for Chemical Process Safety of the American Institute of Chemical Engineers (1998) and to Fisher et al. (1992). 3. As mentioned above, NFPA 30, 2-3.6.7 permits a reduction of the environmental factor (F) by 0.5 when the burning fluid is ethanol. The minimum acceptable resulting value for F is 0.15. 4. NFPA 30 permits this heat flux reduction for other “liquids whose heats of combustion and rates of burning are equal to or less than those of ethyl alcohol (ethanol)”. 5. NFPA does not provide any guidance on how to determine a fluid’s heat of combustion or rate of burning. The guidelines that follow provide one means to quantify a fluid’s burning rate and heat of combustion. 6. The burning fluid that provides a basis for the API and NFPA heat flux equations is not clear. Many of the tests carried out to determine heat flux data used gasoline or kerosene as the burning fuel. 7. Calculations were carried out comparing ethanol with hexane and nonane. Using the proposed formula (below), the comparison to nonane was more conservative in that it yielded a higher reduction coefficient. Consequently, this paper will recommend a method that compares a burning fluid’s characteristic with nonane. 8. Therefore, this paper proposes calculating and using a burning rate factor based on a predicted spilled liquid’s burning characteristics. This factor is to be multiplied by the NFPA or API environmental factor before calculating an overall heat input rate to a vessel exposed to an external fire. In no case shall the resulting overall heat

where BRF is the burning rate factor, dimensionless; 2 the conservatism factor; HC the liquid heat of combustion, BTU/lb (lower heating value); HV the liquid heat of vaporization at TBP, BTU/lb; CP the liquid heat capacity at TBP, BTU/lb-°R; TBP the atmospheric pressure boiling point, °R; Ta the ambient or ground temperature, °R; 19,054 the nonane heat of combustion, BTU/lb; 124.5 the nonane heat of vaporization at TBP, BTU/lb; 0.821 the nonane liquid heat capacity at TBP, BTU/lb°R; and 763 is the nonane normal boiling point, °R. 10. If there is a potential for more than one fluid to spill or leak and begin burning near the subject vessel, for design, use the fluid that results in the higher burning rate factor. 11. Example: ethanol gets the following BRF.

2HC2 HV+CP(TBP−Ta) BRF⫽ 19,0542 124.5+0.821(763−Ta) subject to the limitation of

HC: liquid heat of combustion, 11528 BTU/lb (lower heating value). HV: liquid heat of vaporization at TBP, 367.7 BTU/lb. CP: liquid heat capacity at TBP, 0.717 BTU/lb-°R. TBP: One atmosphere boiling point, 632.5°R. Ta: Assumed ambient or ground temperature, 530°R. Insert these values into the BRF formula: 2(115282) 367.7+0.717(632.5−530) BRF⫽ 190542 124.5+0.821(763−530) where BRF=0.52 (NFPA 30 recommends 0.5 for ethanol). 12. Overall environmental factor examples (with the above BRF of 0.52). A. Uninsulated tank; F=1. Overall F=0.52×1=0.52. B. NFPA valid insulation; F=0.3. Overall F=0.52×0.3=0.156. C. NFPA valid insulation, water spray and drainage; F=0.15. Overall F=0.52×0.15=0.078 → use a minimum value of 0.15.

412

J. Hauser et al. / Journal of Loss Prevention in the Process Industries 14 (2001) 403–412

D. API valid insulation with thickness and conductivity good for F=0.15. Overall F=0.52×0.15=0.078. E. API valid insulation with thickness and conductivity good for F=0.03. Overall F=0.52×0.03=0.0156 → use minimum value of 0.026.

4. Disclamer Information contained in these considerations was developed by the authors and contributors, based on their experience. Your company or organization is responsible for determining whether these considerations would be suitable for your use. The authors and contributors make no warranties, either express or implied, as to the accuracy, reliability or appropriateness for your use of this information or that it is complete and without omissions. The authors and contributors disclaim any liability arising out of your use of these considerations.

References American Petroleum Institute (1990a). Guide for pressure-relieving and depressuring systems. API Recommended Practice 521 (3rd ed.). American Petroleum Institute (1997). Guide for pressure-relieving and depressuring systems. API Recommended Practice 521 (4th ed.). American Petroleum Institute (1990b). Sizing, selection and instal-

lation of pressure-relieving devices in refineries. API Recommended Practice 520 (5th ed.). American Petroleum Institute (1993). Sizing, selection and installation of pressure-relieving devices in refineries. API Recommended Practice 520 (Part 1, 6th ed.). American Petroleum Institute (1992). Venting atmospheric and lowpressure storage tanks. API Standard 2000 (4th ed.). Center for Chemical Process Safety of the American Institute of Chemical Engineers (1998). Guidelines for pressure relief and effluent handling systems. Fisher, H. G., Forrest, H. S., Grossel, S. S., Huff, J. E., Muller, A. R., Noronha, J. A., Shaw, D. A., & Tilley, B. J. (1992). Emergency relief system design using DIERS technology (The Design Institute for Emergency Relief Systems (DIERS) Project Manual). The Design Institute for Emergency Relief Systems of the American Institute of Chemical Engineers. Fisher, H. G., & Forrest, H. S. (1994). Protection of storage tanks form two-phase flow due to fire exposure. AIChE Summer National Meeting. Hauser, J. J., & Becker, M. L. (1995). Vent sizing for fire considerations for special equipment and piping. Process Safety Progress, January, 1995. Lees, F. P. (1980). Loss prevention in the process industries (Vol. 1, 1st ed.). London: Butterworths. National Fire Protection Association (1996). ANSI/NFPA 30, flammable and combustible liquids code (1996 ed.). Forrest, H. S. (1995). Emergency Relief System Design for Fire Exposure with Consideration of Multiphase Flow. Proceedings of the International Symposium on Runaway Reactions and Pressure Relief Design, August, Boston, MA. Nichols, F. P. (1998). Fire relief of jacketed vessels. Report to the DIERS Users Group. Schiappa, C. A., & Winegardner, D. K. (1994a). The Dow Chemical Company, Design of pressure relief systems for vessel jackets exposed to fire. Presented at the AIChE Summer National Meeting, Denver, CO, 15 August. Schiappa, C. A., & Winegardner, D. K. (1994b). The Dow Chemical Company, Pressure relief system alternatives for vessel jackets. Journal of Loss Prevention in the Process Industries, 7 (1).