You are on page 1of 3
Operations & Maintenance Improve the Wing Y. Wong UOP LLC Fire Protection of Pressure Vessels During a plant fire, are! f valve may not open in time. Designing for this contingency can save a vessel from destruction ressure relief valves (PRVs) are a favored protection against explosions of pressure vessels due to external fires. Because two Recommended Practices (RPs) issued by the American Petro- Teum Institute (API; Washington, D.C.) endorse PRVs, many engineers do not realize that PRVs are only one ‘component of a total defense strategy. This article details alternative ap- proaches to protect vessels in addition to using a PRV. ‘The first step towards understand- ing starts with a look at the process fluid inside a closed vessel. It can be gas, liquid plus gas, or liquid. Protec- tion techniques depend on the fluid volume and its thermal properties at the first instant of fire. This article will discuss gas-filled and quasi-gas- filled vessels. We will outline the prob- Jem and then solutions. PRVs for gas-filled vessels A PRY on a gas-filled vessel, exposed to an external fire very likely will not protect the vessel — it wil just provide a false sense of security. A flame has hot spots and cool spots. The heat- transfer coefficient between the wall and the gas in the vessel is very small. ‘The stee! wall cannot cool down. Tests and actual fire incidents have con- firmed that when a gas-filled vessel is exposed to a fire, the vessélall devel: ‘ops overheat spots where the flame is hottest. The vessel will fail due to heat stress on the wall long before its inter- nal gas pressure reaches the set pres- sure of the PRV. Section 3.15.12 of API RP 521 [1] warns design engineers about the po- tential consequences of such a thermal failure. API RP 521 does indicate that a PRV alone may not be adequate to protect gas-filled vessels. At the same time, API RP 520 [2] provides detailed fire-load calculation procedures for gas-filled vessels. Most petroleum re: finers use these equations to size PRVs. However, a gas-filled vessel cannot be protected by a PRY alone. Equations 1 and 2 are used to size PRVs. They are based on the following assumptions: + The vessel is not insulated + The vessel wall temperature (rec- ommended maximum: 1,100°F) will not weaken the wall below its rated rupture stress. 4 n= RT, a (aon | ay Te ‘These assumptions are the source of the problem. The biggest shortcoming of using a PRY is a product of real-life materials. Most market-available PRVs will fail below 1,100°F. The highest temperature capability of an available steam valve from valve man- ufacturers is about 1,050°F, Most W =0.1406./MP;| Ae CHEMICAL ENGINEERING / OCTOBER 1999 NOMENCLATURE La, Exposed surface area of the vessel, 2 LAtoret Tota surface area ofthe vessel, fi Honea Wate foe ore of he ae heat of vapor or gas, omen MMolecitlar weight of the gos or vapor, lb/lb-mole Total moles of material in the vessel, Ib-moles ng Moles of liquid in the vessel, lb-moles| 1, Moles of vapor in the vessel, lb-moles| Roving pres, ic. Ts the et pressure she ol ‘overpressure (21% eT flo amespheric prose, pala Penal pss of «vse wh st drop Sheu, psa ,, Normal operating vapor pressure, Se Tafel heat absorption across weted surface area, Btu, R Gas constant, 10.73 {lb/in. 2\(43)/(°RIIb moles) ‘Average boiling point of liquid ventory ofthe releving pressure, Tq Normal vapor operating tempera tres Vener Nats, Ty Vessel wal omperature, °R ‘V Initial liquid volume in the vessel, f Vy lnitol vapor volume in the vessel, Vrotet Total vessel volume, 2 'W Recpired relieving capacity, o/h We Initial quid inventory, fb 2 Latent heat of fization of the liquid, Btu/lb o iid density, f/f tTime, min ep Meatoas bon it tp Boilff ine, nin + The recommended maximum vessel wall temperature, Ty, for the usual carbon-steel plate matarsal .'1,100°F. veel are Fabricated from alloy materials, the value for TT should be changed to.» more-appropriate longercire at the reliving ey other valves with the best available elastomér or plastic seals (such as stainless-steel-filled Teflon and Kalrez) do not work above 550°F. ‘Some carbon steel vessels fail below 1,100°F. In refinery fires, the flame temperature can exceed 1,400°F. API RP 520 indicates that an unwetted steel vessel with ASTM A515 Grade 70 steel will rupture in about 2.5 min- tutes at 1,300°F [3]. Figure 1 shows an example of the interaction of time and temperature at 1 atm Effective fire protection ‘There are other measures that can be taken to effectively protect gas-filled vessels against external fire: 193 1, Proper location of the vessel. This is key in planning for’ protection against external fire. Refineries and chemical plants should seriously con- sider the measures below, especially for vessels containing Hp, or other ex plosive or flammable gases: (a) Isolating the high-risk, gas-filled vessels by surrounding them with a firewall and moving them away from the main process area. (b) Burying the gas-filled vessels underground and covering them with earth (©) Raising the gas-filled vessel to an elevation 25 ft above grade. Fire re- ports confirm that flame height usu- ally stays below 25 ft. 2. Install external fireproof insula- tion on a gas-filled vessel. Insulation can reduce the heat absorption from a fire to 30% or less of that of a non-in- sulated vessel. However, design engi- neers must be aware that: (a) The insulation materials must function effectively while withstand- ing flame temperatures up to 1,660°F during a fire. (b) The insulation should withstand fire exposure for two h. (©) Insulation should resist dislodg- ment by high-pressure water streams used for fire fighting. Therefore, stain- less steel jacketing and bonding to re- tain the insulation shape is required. 3. Design an automatic vapor-de- pressurizing system [1] Vapor depres- surizing is one of the mosteffective emergency approaches for gas-filled vessels exposed to an external fire. Such a system can greatly reduce the severity of the fire, with subsequent consequences of vessel failure, includ~ ing high-pressure gas erupting from the weakest overheated spot of a ves- sel. Consideration should be given to installing an automatic vapor depres- surizing system on all major gas-filled vessels with operating pressure above 50 psig. 4. Equip a reliable fire-monitoring system and a rapid-response fire fighting team. 5, Install a well-designed water del- tuge system (WDS). This should effee- tively allow very little or no heat input to reach a vessel. A carefully designed WDS can effectively lengthen the time of exposure to the external fire while 194 = 140° finim FIGURE 1. Steel (ASTM A515, Grade 70) can ail from the wrong combination of time and temperature [2] limiting serious damage to the insula- tion and vessel wall, However, API Recommended Prac- tives do not incorporate a WDS in the PRV calculation, because most of these systems would either not func- tion properly or be destroyed during a fire. A WDS is a good measure to pro- tect a vessel against a small fire, but if itis not well-maintained, it will be un- reliable. It will not resist a large fire. Design engineers should read the NFPA (National Fire Protection Assn.; Norwood, Mass.) Handbook to learn the basic requirements for an ef- fective WDS. High-boiling point liquid Engineers should also take extra pre- cautions when dealing with a high- boiling-point liquid. The design tem- perature must be above the fluid’s boiling point at its relieving pressure. Ifit is difficult to build a vessel of suit- able metal, then apply the measures for protecting gas-filled vessels, The calculation of the fire-relieving capacity for the high-boiling-point lig uid is different from the rest of this ar- ticle. There are two scenarios for a ‘vessel with high-boiling point liquid: * Ifthe vessel is not full, the situation is vory similar to the gas-filled vessel because there is no or very little vapor generation at the relieving pressure. ‘The same sort of hot spots show up on the walls, while the contents stay cool. * If the vessel is full, the required fire-relieving capacity is equal to the thermal relieving capacity, since there is very little or no vapor generation after its initial thermal expansion. Engineers can refer to API 520 and Reference 4 for sizing the required thermal relief valves. In both cases, avoid connecting the PRV discharge line to the common CHEMICAL ENGINEERING /OCTOBER 1999 flare header. It is better to use a sepa- rate discharge line. The problem is that high-boiling-point liquid might crack during a fire, form coke and plug the line. If the vessel has no vent, it will probably break apart. As with gases, the most conservative approach is to isolate or bury such vessels. Be aware that some viscous high- boiling-point liquids may need hours to boil. The equations say that no PRV is required; nevertheless, hot spots on the vessel wall can still develop, and eventually rupture, before fire-fight- ing teams arrive. Good fire protection approaches are still needed. Low-boiling-point liquids inside a vessel will boil off into the vapor phase when the vessel is heated. There is a good heat-transfer coefficient associ- ated with this phase change. It buys time before the vessel becomes a gas- filled vessel. It will be the designer's task to figure out if the firefighting team can arrive before the vessel wall fails. The amount and properties of this liquid determine the time. Aboiling liquid buys time ‘When a fire is catastrophic or uncon- trolled, even’a WDS can be melted down. Basically, everything in the fire zone could be wiped out, But, for ves- sels with large liquid inventory (“wet- ted-surface”), there is usually more time to hold the wall temperature below dangerous levels. For an external fire, the only fune- tion of a pressure relief valve is to limit the damage by having the PRV open prior to vessel wall rupture. Don't count on a PRV being good pro- tection in a large fire, even if the ves- sel has a large inventory of liquid. When a vessel with a low liquid in- ventory is exposed to a fire, the enor- mous amount of heat absorption from the fire could quickly deplete the liq- uid inventory by evaporation. After that short period of time, the vessel becomes a gas-filled vessel. The in- stalled PRVs can not protect low-liq- uid inventory vessels from external fire any better than they can protect gas-filled vessels, ‘The first design task is to classify the category of low inventory vessels as ei- ther belonging to wetted-surface ves- sels or gas-filled vessels. In other words, what is the boundary separating small and large inventories? A wetted-surface vessel will have ‘enough liquid inventory that it will un- dergo the vaporization process until the firefighters arrive. Ifthe liquid in- ventory will be depleted in less than 15 =20 min, treat the low-liquid-inven- tory vessel as a gas-filled vessel. ‘The actual unprotected time should be defined by design engi- neers based on the average response time from the fire fighting team, the location of available fire fighting equipment, and the fire protection conditions of the plant. The low limit of the total response time should not be less than 15 min, Calculation method Let us set up calculations for the time required for vaporizing all the liquid in- ventory ina vessel. This makes it possi- ble totest the time and the setting ofthe PRY. ‘The following calculations. as- sume that the liquid level is below the 25 feet of effective fire height specified by Reference 3. Step 1— Calculate the actual liq: uid volume in a vessel. For simplic- ity, assume the partial volume of liq- uid in a vessel is proportional to the ratio of wetted surface area to total area. Ifa vessel has a boot, its volume should be added. wet) Step 2 — Calculate the time re- quired for vaporizing all the liquid ina vessel. First, find the time required to heat the liquid to the boiling point: 1, = Veo Top ~T : @ Calculate the time required to va Porize the liquid inventory 196 3) ) 60 (a) 6) Step 3 — Obtain the total time re- quired to deplete the liquid inventory: tent © If the calculated time is less than 16 min, then treat it as a gas-filled ves- sel. The PRVs on the vessel can not Prevent a vessel-wall meltdown. If the calculated time is more than 15 to 20 minutes, we can feel confident using the wetted-surface-area proce dures. Size the PRV with Equations 1 and 2. Fine-tuning the calculation, using Equation 6, may add credit for ‘the metal wall. Figure 2 can be used to estimate that. Step 4 — Determine if the vessel pressure could reach its set pressure. In certain cases, after the last drop of the liquid is vaporized, the vessel Pressure may reach its set pressure and the PRV on the vessel will lift. It is especially true if the set pressure of the vessel is low. Therefore, in addi- tion to the above time criterion, design engineers should also perform a rough calculation to determine the final References 1 meri Paloum Inaate, Gide. for ressure-Releving an suring Sy tems * API RP 521, Fourth eaition, Wachis- ton, hitaret 1997 2 Amorican Petroleum Institute, “Sizing, Se- lection and installation of Pressure Relies fgg Devieos in Refineries. Part |, Sising and Selestion," APL RP 590, Sixth edition, Wash ington, March 1998, 3. op. cit, Part 11, Installation.” 4 Wong, W. ¥., Safer Relief Valve Sizing, Chem: Eng..98,5, pp. 137-140, May, 1989. CHEMICAL ENGINEERING /OCTOBER 1999 late 1/2 in. thick ate 1 in thick FIGURE 2. This shows the average rate of heating of steel plates exposed to an open gasoline fire on one side [2] Itcan be used toestimate the timefora vessel wall toheat up pressure of the vessel with the last drop of liquid. Assume: (1) The temperature inside vessel has reached the boiling point. (2) The liquid and vapor inside the vessel isa single-component hydrocar- bon, or other pure material (3) The vapor is an ideal gas For the initial vapor phase, the vapor moles can be calculated as: BY, nae The liquid moles can also be caleu- lated by @ ape Vee. (8) Peal = 30. ‘The total moles in the vessel is: nem +n @ The final pressure of the vessel, when all the liquid is vaporized, is: nay Vewct A final confirmation: Check this against the PRV set point. . Edited by Peter Silverberg (10) Wing ¥; Wong is » msi specialist m-pressure rellev- ae svstems for UOP LLC East Algonquin Road. Des Plaines, IL. 60017, Phone" Seis Pan ot 8K. 2266; E-mail: ‘wywongauop som) Mr Wong aw wtien wing’ systems He has an NS. tn chemical engineering ‘rom the University of Alberts and @ BS. frm East China Inetutite of Chemical ‘Technology. He ea ropistered P-E. so iin,

You might also like