Hazards Of Vacuum

Nomenclature

F1 Friction loss in equivalent length, ft g Acceleration of gravity, ft/s2

gc Conversion factor, ft- lbm/ft- lbf s2

H Total tank vertical height, ft ho Initial fluid height, ft hi Height between tank base and centerline of pump suction, ft hf Final height of liquid above tank base, ft

NPSH Net positive suction head, ft

P Final tank vacuum, in Hg

Po Atmospheric pressure, lbf /in2

p Liquid density, lbm/ft3

R Tank radius, ft

Vo Initial airspace volume, ft3

Vp Vapor pressure of tank fluid equivalent, ft

Y Ratio of molar specific heats, Cp/Cv (about 1.4 for diatomic gases)

Introduction Ask any chemical engineer who has had some plant experience what he or she knows about vacuum and the engineer will probably smile and tell some tale about some piece of equipment that tried to turn itself inside out. Usually no one was hurt, and often there was no massive leakage—but not always!

Causes of Vacuum Hazards The design for the internal pressure condition of vessels is usually straightforward and well understood. The design for external pressures is more difficult. The devious ways in which external pressure can be applied can often be overlooked.

Following are some obvious causes of vacuum collapse:

• Liquid withdrawal by pump or gravity draining

• Removal of gas or vapor by withdrawing with a blower, fan, or jet

• Siphoning of liquids Less obvious causes include:

• Condensation of vapor

• Combination of cooling and condensation of a mixture of gas and condensable vapor

Sometimes obscure causes of vacuum collapse include:

• Absorption of a gas in a liquid, for example, ammonia in water or carbon dioxide in water.

• Reaction of two or more gases to make a liquid or solid, for example, ammonia plus hydrogen bromide to form ammonium bromide.

• Reaction of a gas and a solid to form a solid, for example, corrosion in a tank. Air plus Fe or FeO may give Fe2O3 in the presence of water.

• Reaction of a gas and a liquid to give a liquid, for example, chlo-rination, hydrogenation, or ethylation.

• Sudden dropping of finely divided solids in a silo, creating a momentary vacuum that can suck in the sides of the silo.

• Plugging of flame arresters, for example:

In styrene service, vapor may condense in flame arresters, and the liquid formed is low in inhibitor. Liquid may polymerize and plug off the arrester. Possible solutions include cleaning the arrester frequently or using a PVRV (pressure-vacuum relief valve).

In liquid service in cold weather, vapor may condense in a flame arrester and the liquid formed may freeze and plug the arrester. A

ossible solution is to heat and insulate the arrester to prevent con-ensation.

• Maintenance and testing. It is not a good idea to apply vacuum on a vessel during maintenance or testing without full knowledge of the external pressure rating, unless a suitable vacuum relief device is in place and operable.

Location of Vacuum Relief Device (Carl Schiappa, Michigan Engineering, The Dow Chemical Company, Midland, Mich., personal communication, March 20, 1992.) If a vacuum relief device is used, locate the device at the highest point on the top of the tank. If the vacuum relief device is not installed in this location and the tank is overfilled with liquid, the relief device will be sealed in liquid and will be ineffective in protecting the tank. This is especially true for the part of the tank above the vacuum relief device if it is sealed in liquid, the liquid level is lowered, and the tank goes into a partial vacuum.

Protective Measures for Equipment There have been many incidents where vessels were designed for internal pressures of 25 psig or higher and the tank collapsed under vacuum. The internal pressure rating is not a good indication of the vacuum rating. If equipment may be subject to vacuum, consideration should be given to designing the equipment for full vacuum. This may eliminate the need for complicated devices such as relief valves and instruments; if they are used but fail or plug, designing the equipment for full vacuum will prevent collapse of the vessel. For vessels where steam is used in the tank, such as steam-sterilized sanitary-service tanks, a full vacuum rating is advisable under any circumstances.

A disadvantage of this approach is that it may appear at first to be more expensive to design equipment for full vacuum. The cost differential of adding vacuum rating is usually modest compared to the tank's value. It can be less than 10 percent for 15 psig tanks of up to 3000-gal nominal capacity (Wintner, "Check the Vacuum Rating of Your Tanks," Chem. Eng., February 1991, pp. 157-159). When the total cost of a suitably instrumented vessel not designed for vacuum is compared with the cost of a vessel designed for vacuum but without the extra equipment, the difference may be small or negligible, and the vessel designed for vacuum will be inherently safer. If a vessel is designed for vacuum, precautions should be taken to ensure that internal or external corrosion will not destroy the integrity of the vessel. Dimpled jackets mayprovide an economical way of providing vacuum protection when jacketed vessels are involved.

Personnel Hazards The following case history illustrates how vacuum can be harmful and dangerous to personnel. A plant superintendent was checking an open nozzle on a glass-lined reactor on which there was a vacuum pump pulling vacuum, when suddenly his arm was sucked into the nozzle, up to his shoulder. He could not remove his arm until help arrived to release the vacuum on the vessel. He was injured painfully, though not seriously. The injury could have been very serious if help had not been nearby. Personnel hazards can also result from vacuum conveyor systems for solids handling.

Examples of Vacuum-Related Accidents Figure 26-47 shows a jacketed tank, where the jacket was designed for low-pressure steam. When the steam was turned off and the drain valve and trap were closed, the steam condensed, causing the jacket to collapse. The jacket should have been designed for full vacuum, or a suitable vacuum relief device should have been installed on the jacket.

Figure 26-48 shows the collapse of a large storage tank containing acetone. The overflow and vent line had recently been changed, so it would vent through a vapor seal of water to remove acetone emissions from the vent when the tank was being filled. When the tank was being emptied, water was sucked into the vent pipe, creating a vacuum in the tank, which collapsed the top of the tank. A suitable vacuum relief device on the tank should have been installed to prevent this incident. Venting the tank through a liquid seal of this type is probably not very effective and a better method of controlling emissions should have been selected.

Low-Pressure Storage Tanks Low-pressure storage tanks are fragile. Even an eggshell can withstand more pressure and vacuum (Sanders, "Don't Be Another Victim of Vacuum," Chemical Eng. Prog., September 1993, pp. 54-57). Low-pressure storage tanks do not require much pressure difference between the inside of the tank and the atmosphere to buckle the relatively thin tank walls. Pressure

FIG. 26-47 Collapse of vessel jacket due to condensation of steam. (W T. Allen, Michigan Engineering, The Dow Chemical Company, Midland, Mich., personal communication, May 1988.)
FIG. 26-48 Collapse of storage tank due to adding a liquid vapor seal to reduce vapor emissions. (Allen, 1988.)

differences as low as 10 mbar (0.01 atm, or 0.15 psig, or 0.7 in H2O) between the inside and outside of the tank can buckle some tanks. The rate of handling of product and the breathing volume flow rate due to weather effects must be taken into account in designing the necessary pressure compensation devices.

A critical situation arises in summer when the tank is heated by strong radiation, then cooled by sudden rainfall. Heavy rainfall results in a rapid drop in ambient temperature and the formation of a rainwater film that flows on the top of the tank and down the tank wall. The wall and, with a certain lag, the gas in the tank are cooled, and air must flow into the tank to prevent a significant pressure difference from arising between the inside and outside of the tank. If vapors in the tank are condensed, more air must flow into the tank.

The initial gas temperature in a tank can reach a value of 55°C (131°F) as a result of strong solar radiation. Ambient rainfall is assumed to be 15°C (59°F). The maximum flow rate of air into the tank is reached some minutes later. At the start of the rainfall, after a certain lag, the flow rate at first increases, then reaches a maximum, then decreases.

A study has been made to allow the prediction of the rate at which air must enter a tank with and without internal condensation to prevent a pressure difference from arising (Fullarton, Evripidis, and Schlünder, Institut für Thermische Verfahrenstechnik, Universität Karlsruhe (TH), "Influence of Product Vapour Condensation on Venting of Storage Tanks," Chem. Eng. Process., 22(3), 1987, published by Elsevier-Sequoia, New York). The results are too involved to be presented in detail here. The reader is referred to this paper for details of the calculations.

The results of a specific case study are shown in Fig. 26-49. This depicts the change in inbreathing volume flow rate as a function of time. The middle curve describes the case when the tank is filled with dry air: that is, no condensation occurs. When the air is saturated with water vapor at 55°C (131°F) and condensation occurs, the top curve is obtained. The bottom line represents the volume flow rate brought about by thermal contraction alone, not including the amount condensed. Because of the heat of condensation released, this fraction is less than the volume flow rate without condensation, but this effect is more than compensated for by the additional volume flow rate due to condensation.

Experimental data in small equipment has shown that condensation of water vapor causes a twofold increase in the maximum flow rate compared to dry air, and a fourfold increase in condensation of methanol vapor.

API 2000 lists the venting capacity for inbreathing (vacuum relief) and outbreathing (pressure relief) for oil tanks up to 180,000 barrels (7,560,000 gal or 2.86 x 104 m3) capacity at 14.7 psia and 60°F. Tanks larger than 180,000 barrels require individual study (API 2000, "Venting Atmospheric and Low-Pressure Storage Tanks, Non-Refrigerated and Refrigerated," API Standard 2000, 3d ed., American Petroleum Institute, Washington D.C., January 1983).

Vacuum Requirements for Draining Tanks (Wintner, 1991, by permission.) A shortcut method of calculating the vacuum that can occur when a tank is being drained while the vent line is closed can be performed by measuring the head in the tank, assuming it is completely full. This is the maximum vacuum that would exist in a gravity-drain tank before air would begin to enter it. If the tank's overall height is designated H, then this vacuum is 2.036Hp(g/gcPo) in Hg.

If the tank has some headspace, as is usually the case, it is desirable to get a better estimate of the actual level, since tanks usually have some gas headspace even when filled with liquid. Two tank configurations are considered: the gravity discharge tank (discharge is open to the atmosphere) and the pumped discharge tank. These calculations assume that the process is so rapid that an adiabatic model for the gas in the headspace is the correct choice. This is true when the drainage

FIG. 26-49 Effect of water vapor condensation on volume flow rate of air into tank. (Fullarton, Evripidis, and Schlünder, 1987, by permission of Elsevier Science S.A., Lausanne, Switzerland.)

time is short (on the order of a few minutes for a tank of several thousand gallons capacity). An isothermal model is the best choice when the drain time is long.

For the gravity discharge case, the height of the fluid at maximum vacuum, which is the point at which air would begin to backflow into the tank, is determined by Eq. (26-54). Equation (26-55) calculates the corresponding vacuum in the tank's headspace at this liquid height. Since the drain nozzle is open to the atmosphere, this solution is a static force balance that is satisfied when the sum of the internal pressure and the remaining fluid head is equal to the atmospheric pressure.

144P0(1 - {V„/[V„ + nR2(h0- hf )]}Y) Pg /gc p = 29.92 hf pg

Pg/gc

For the pumped-discharge case, internal pressure and final fluid height are calculated by Eqs. (26-56) and (26-57). The final fluid level is the point at which the net positive suction head (NPSH) equation is satisfied.

The solutions of Eqs. (26-54) and (26-56) involve a trial-and-error technique or a numerical method. This can be solved using a computer program for multivariable equations, or it can be calculated by hand. In either equation, assume a reasonable value for hf and insert it on the right-hand side of the equation. The left-hand value obtained is then substituted until the values guessed at and those calculated are in close agreement. The number of trials is strongly dependent on the initial guess; for realistic tank dimensions, between four and ten iterations should produce good agreement.

The pumped-discharge case is generally more difficult to solve because of the uncertainty in dealing with negative numerical results. As a final answer, a negative value could indicate that the pump has completely emptied the tank; however, as an intermediate value, it could mean that it is not a true solution. A simple check is to try a different initial estimate and see if the intermediate negative results disappear.

Example Assume the tank in Fig. 26-50 has a diameter of 4 ft and a capacity of 1000 gal, is filled with water, and discharges to the atmosphere. The shortcut calculation (tank is initially completely full) indicates that the internal pressure would be 10.65 in Hg. An initial fillage of 70 percent of the tank's volume would produce a vacuum of 6.93 in Hg, which is 65 percent of the shortcut result.

In the case of pumped discharge, assume that a centrifugal pump is used. Its NPSH will determine the height at which vacuum is released by the backflow of air through the pump. Detailed information about the pump characteristics is needed to evaluate the potential vacuum. For these design calculations, assume that the pump will stop delivering liquid and air backflow will begin when the pump's NPSH requirements are no longer met (Sommerfield, "Tank Draining Revisited," Chem. Eng., May 1990, p. 171).

The precise flow-decay pattern will depend on the type, size, and dimensions of the pump. Flow for a typical centrifugal pump will begin to decay at the NPSH point, but some additional fluid transfer will usually occur before a steady backflow of air through the pump begins. At that point, the pump's priming is completely lost.

The mathematical solution for maximum vacuum is based on Eq. (26-56), which solves the NPSH equation for this value of the fluid height. The nomenclature used contains only positive numbers for elevation, with the base point being set at the tank's discharge nozzle (analogous to the gravity-discharge case).

For this example, assume the following parameters:

Pump capacity = 50 gal/min

FIG. 26-50 Tank configuration used in example calculation.

For these parameters, the equations predict a much higher vacuum (24.5 in Hg or 230 percent of the shortcut method) than the gravity-discharge case. Of course, different tank dimensions and pump characteristics could give different comparisons between cases. If conditions are such that the pump can completely empty the tank before backflow occurs, the vacuum is best calculated from Eq. (26-57).

If proper instruments are provided, the tank and pump can be interlocked, so the pump will stop when abnormal conditions are detected. This may help keep the tank from collapsing, but the gravity-discharge case should also be checked to ensure that failure will not occur after the interlock stops the pump. In all cases where instrumentation is used, the consequences of potential failure should be considered.

+1 0

Post a comment