How Trays Work: Flooding
3. How Trays Work: Flooding 3.1. History of Distillation The ancient Egyptians produced beer from barley that had a few percent of alcohol. Next, Next, wine wine was produced produ ced by fermenting g rape juice, juice, which had a greater sugar content than barley. This brought the content of alcohol up to about 14 percent. percent . Next, Next, fortified wines (like (like sherry sh erry or port) were made b y adding extra sugar and yeast to the fermenting grape juice. This increased the alcohol up to abou t 17 percent. Much above abo ve this point, alcohol kills kills off the yeas t. Next, distillation was used. I visited a primitive distillation plant in Peru. It was a single-stage evaporation proces s. The Th e fermented grap e juice juice is partially partially vaporized vaporized and the th e alcohol—water alcohol—water vapors are totall tot ally y condensed. condens ed. The resulting condensat cond ensate e is 40 to 44 percent alcohol. The higher h igher alcohol conten contentt is obtained b y vaporizi vaporizing ng less of the stil st ill' l's s conten t. To go much beyond the 44 percent alcohol, one needs ne eds to introdu introd uce modern p rocess engineering technology: Partial condensation conden sation Reflux Reboiler I've 've shown sh own a sketch sket ch of such a facility facility in Fig. Fig. 1.3. The Th e idea is to gen erate reflux to improve the separation between water and ethanol. To generate reflux, a partial (rather (rather than th an a total) condens er is required. Also, a way of adding
more heat to the t he s till, till, to match up with th e capacity of the partial conden ser, ser, is need ed. I've 've now introd uced a complex complex control loop into my plant. Alcohol levels levels of 60-plus 60-plus p ercent can be obtained ob tained with this t his two-stage evaporator. evaporator. Finally, Finally, we have h ave th e Patent Still, intro introdu duced ced b y the Scots in the th e 1830s. 1830s. Now ow,, distill distillatio ation n trays equipped with bubb le caps and feed preheat are used. I visited visite d an apple ap ple orchard orch ard in Englan En glan d, which wh ich use u sed d the t he original orig inal des d esign ign of the th e Patent Still Still to produce ap ple brand y. A sketch s ketch of this app aratus, whos e design has not no t been altered in 180 180 years, years, is shown in Fig. 3.1. The sketch has been copied from the original patent app lication lication filed in London in the 1830s. 1830s.
Figure Figure 3.1. The very first distillation distillation tower was the Patent Still. till. Th is drawing was filed with the original patent ap plication plication submitted su bmitted in 1835 1835.. Column on the t he right is a bubble-c bub ble-cap ap trayed tower. tower. Column on the left is a feed preheater preheate r vs. an overhead vapor cond enser ens er.. With With the multitrayed multitrayed distill d istillation ation column, ethanol ethan ol concentrations (as limited limited by the alcohol—water azeotrope) of 9090-plus plus p ercent can be obtained, ob tained, if enough reflux and enough trays are used.
3.2. Tray Types Distillation Distillation towers are the heart hea rt of a process plant, and the working component of a distillation distillation column is the tray. tray. A tray consists of the following following components, as sh own in Fig. Fig. 3.2: Overflow, Overflow, or outlet ou tlet weir w eir Downcomer
Tray deck
Figure 3.2. Perforated trays. There are two types of tray decks: perforated trays an d bubble-cap trays. In this chapter, we describe on ly perforated trays, examples of which are Valves or flutter caps V grid, or extruded-valve caps Sieve decks Jet trays Possibly 90 percent of the trays seen in th e plant are of these types. Perforated tray decks all have one feature in common; they dep end on the flow of vapor through the tray deck perforations, to prevent liquid from leaking th rough the tray deck. As we will see later, if liquid bypasses the outlet weir and leaks th rough the tray deck onto the tray below, tray separation efficiency will suffer.
3.3. Tray Efficiency Distillation trays in a fractionator operate between 10 and 90 percent efficiency. It is the process person 's job to make trays operate as close to 90-
percent efficiency as poss ible. Calculating tray efficiency is s ometimes simple. Compare th e vapor temperatu re leaving a tray to the liquid temperature leaving the trays. For example, the efficiency of the tray shown in Fig. 3.3 is 100 percent. The efficiency of the tray in Fig. 3.4 is 0 percent.
Figure 3.3. Hundred-percent tray efficiency.
Figure 3.4. Zero-percent tray efficiency. How abou t the 10 trays sh own in Fig. 3.5? Calculate their average efficiency (the answer is 10 percent). As the vapor temperature rising from the top tray equals the liquid temperatu re draining from the bottom tray, the 10 trays are behavin
as a sin le
erfect tra with 100- ercent efficienc . But as there are
10 trays, each tray, on average, acts like one-tenth of a perfect tray.
Figure 3.5. Average tray efficiency = 10 percent. Poor tray efficiency is caused by one of two factors: Flooding Dumping In this chap ter, we discuss p roblems that contribute to tray deck flooding.
3.4. Downcomer Backup Liquid flows across a tray deck toward the ou tlet weir. The liquid over-flows the weir, and drains throu gh the downcomer to the tray below. Vapor b ubbles up through the sieve holes, or valve caps, on the tray deck, where the vapor comes into intimate contact with th e liquid. More precisely, the flu id on the tray is a froth or foam—that is, a mixture of vapor an d liquid. In this s ense, the function of a tray is to mix the vapor and liquid together to form a foam. This foam shou ld separate back into a vapor and a liquid in the downcomer. If the foam cannot drain quickly from a downcomer onto the tray below, then the foamy liquid or froth will back up on to the tray above. This is called flood ing.
3.5. Downcomer Clearance Referring to Fig. 3.6, note that the downcomer B is flooding. The cause is loss of the downcomer seal. The h eight of the outlet weir is below the bottom edge of the downcomer from the tray above. This p ermits vapor to flow up downcomer B. The up-flowing vapor displaces the downflowing liquid. That is, the vapor pushe s the liquid up onto the tray above—which is a cause of flooding. On the other hand, Fig. 3.7 shows what happens if the bottom edge of the downcomer is too close to the tray below. The h igh press ure drop needed for the liquid to es cape from downcomer B onto tray deck 1 causes the liquid level in downcomer B to back up onto tray deck 2. Tray 2 then floods. Once tray 2 floods, down comer C (shown in Fig. 3.7) will also b ack up and flood. This process will continue until all the tray decks and down comers above downcomer B are flooded.
Figure 3.6. Flooding d ue to lack of a downcomer seal.
Figure 3.7. Flooding caused by inadeq uate down comer clearance. On the oth er hand, all trays in a tower below down comer B will lose liquid levels and d ry out when flooding s tarts in downcomer B. Thus , the following
rules apply: When flooding starts on a tray, all the trays above that point will also flood, but trays below that point will go dry. An early ind ication of flooding in a distillation column is los s of liquid level in the bottom of the column. If the downcomer clearance—which means the distan ce between the bottom edge of the d owncomer and the tray below—is too great, the downcomer becomes unsealed. Vapor flows up the downcomer, and the trays above flood. If the downcomer clearance is too small, then liquid backs up in the downcomer, and the trays above flood. To calculate th e height of liquid in the downcomer, due to liquid flowing throu gh th e downcomer clearance:
where ΔH = inches of clear liquid backup in t he downcomer, due to head loss under the downcomer V = horizontal component of liquid velocity, in ft/s, as the liqu id es capes from the downcomer To guarantee a proper downcomer seal, the bottom edge of a downcomer should be about 0.5 inch below the top ed ge of the outlet weir. This dimension should be carefully checked by process pers onnel when a tower is opened for inspection. It is quite eas y for sloppy tray installation to distort this critical factor.
3.5.1. Height of Liquid on Tray Deck As the liquid level on a tray increases, th e height of liquid in the downcomer feeding this tray will increase b y the same amount. Again, excessive downcomer liquid or froth levels res ult in flooding and loss of tray efficiency. The liquid level on a tray is a function of two factors: Weir height Crest height
The weir height on many trays is adjustab le. We usually adjust the weir height to between 2 and 3 inches. This produces a reasonable depth of liquid on the tray to promote good vapor-liquid contact. The crest h eight is similar to the h eight of water overflowing a dam. It is calculated from
where crest height = inches of clear liquid overflowing the weir GPM = gallons (U.S.) per minute of liquid leaving the tray The sum of the crest height plus the weir height equals the dep th of liquid on the tray deck. One might now ask, "Is not the liquid level on the inlet side of the tray higher th an the liquid level near the ou tlet weir?" While the answer is "Yes, water does flow downhill," we des ign the tray to make this factor small enough to neglect.
3.6. Vapor-Flow Pressure Drop We have yet to discus s the most important factor in determining the height of liquid in the downcomer. This is the p ressu re drop of the vapor flowing through the tray deck. Typically, 50 percent of the level in the downcomer is due to th e flow of vapor through the trays. When vapor flows through a tray deck, the vapor velocity increases as the vapor flows through the small openings provided by the valve caps, or sieve holes. Th e energy to increase th e vapor velocity comes from the pressure of the flowing vapor. A common example of this is the p ressure drop we measure across an orifice plate. If we have a p ipeline velocity of 2 ft/s and an orifice plate hole velocity of 40 ft/s, then the energy needed to accelerate the vapor as it flows through the orifice plate comes from the pressure drop of the vapor itself. Let us as sume that vapor flowing through a tray deck undergoes a press ure drop of 1 psi (lb/in 2). Figure 3.8 shows that the pressure be low tray deck 2 is 10 psig and the press ure above tray deck 2 is 9 psig. How can the liquid in downcomer B flow from an area of low press ure (9 psig) to an area of high press ure (10 psig)? The answer is gravity, or liquid head pres sure.
Figure 3.8. Vapor ΔP causes d owncomer backup. The height of water needed to exert a liquid head pres sure of 1 psi is equal to 28 inches. of water. If we were working with g asoline, which has a specific gravity of 0.70, then the height of gas oline needed to exert a liquid head press ure of 1 psi would be 28 inches/0.70 = 40 inches of clear liquid.
3.6.1. Total Height of Liquid in the Downcomer To summarize, the total height of clear liquid in the d owncomer is the s um of four factors: Liquid escape velocity from the downcomer onto the tray below. Weir height. Crest height of liquid overflowing th e outlet weir. The pressu re drop of the vapor flowing through the tray above the downcomer. (Calculating this press ure drop is discus sed in Chap. 4.) Unfortunately, we do not have clear liquid, either in the downcomer, on the tray itself, or overflowing the weir. We actually have a froth or foam called aerated liquid. While the effect of this aeration on th e specific gravity of the liquid is largely unknown and is a function of many complex factors (surface tension, dirt, tray design, etc.), an aeration factor of 50 percent is often used for many hydrocarbon services.
This means that if we calculated a clear liquid level of 12 inches in our downcomer, then we would actually have a foam level in the downcomer of 12 inches /0.50 = 24 inches of foam. If the height of the downcomer plus the h eight of the weir were 24 inches, then a downcomer foam height of 24 inches would correspond to downcomer flooding. This is sometimes called liquid flood. This discussion assumes that the cross-sectional area of the downcomer is adequ ate for reasonable vapor-liquid separation. If the downcomer loading (GPM/ft2 of downcomer top area) is less th an 150, this assumption is okay, at least for most clean services. For dirty, foamy services a downcomer loading of 100 GPM/ft 2 would be safer.
3.7. Jet Flood Figure 3.9 is a realistic picture of what we would see if our towers were made of glass. In addition to the d owncomers and tray decks containing froth or foam, there is a quantity of sp ray, or entrained liquid, lifted above th e froth level on the tray deck. The force that generates this entrainment is the flow of vapor through the tower. The s pray height of this entrained liquid is a function of two factors: The foam height on the tray The vapor velocity through the tray
Figure 3.9. Entrainment causes a jet flood. High vapor velocities, combined with high foam levels, will cause the spray height to hit the underside of the tray above. This caus es mixing of the liquid from a lower tray with the liquid on the upper tray. This backmixing of liquid reduces the separation, or tray efficiency, of a distillation tower. When the vapor flow through a tray increases, the heigh t of froth in the downcomer draining the tray will also increase. This does not affect the foam height on the tray deck until the downcomer fills with foam. Then a further increase in vapor flow causes a noticeable increase in the foam height of the tray deck, which then increases the spray height. When the s pray height from the tray below hits th e tray above, this is called the incipient flood point, or the initiation of jet flooding. Note, though, that jet flood may be caused by excessive downcomer backup. It is simple to see in a glass column separating colored water from clear methanol how tray separation efficiency is reduced as s oon as the s pray height equals the tray spacing. And while this observation of the ons et of incipient flood is straightforward in a transp arent tower, how do we observe the incipient flooding point in a commercial distillation tower? The reason I can write with confidence on this su bject is that I worked with a 4-inch demonstration transp arent column at the Chevron Refinery in Port Arthur, Texas , in 1989. I used th e little dis tillation tower to explain to plant
technicians how distillation towers worked. The tower's feed was windshield wiper fluid with blue dye. The alcohol went overhead an d the blue water was the bottoms product.
3.8. Incipient Flood
3.8.1. A Fundamental Concep t Figure 3.10 illustrate s the operation of a simple propane -butane s plitter. The tower controls are such that both the pressu re and bottoms temperature are held constant. This means that the percent of propane in the butane bottoms product is held constan t. If the operator increases th e top reflu x flow, here is what will happen: 1. The tower-top temperature drops. 2. The amount of butane in the overhead propane product drops. 3. The tower-bottom temperature starts to fall. 4. The reboiler duty increases, to restore the tower-bottom temperature to its set point. 5. The weight flow of vapor and the vapor velocity through the tray increase. 6. The spray height, or entrainment, between the trays increases. 7. When the spray height from the lower trays impacts the upper trays, the heavier, butane-rich liquid contaminates the lighter liquid on the upp er trays with heavier butane. 8. Further increases in the reflux rate then act to increase, rather than decrease, the butane content of the overhead propane product.
Figure 3.10. A simple depropanizer. Figure 3.11 illustrate s this point. Point A is called the incipient flood point, that point in the tower's operation at which either an increase or a decrease in the reflux rate results in a loss of separation efficiency. You might call this the optimum reflux rate; that would be an alternate description of the incipient flood point.
Figure 3.11. Definition of the incipient flood con cept.
3.8.2. Bypassing Steam Trap Stops Flooding I wake up early to answer email questions be fore breakfast. Here's today's ques tion from South Africa: Hi Norman. We have a distillation tower that floods. Delta P on trays below feed point is stable; delta P above feed (trays 16—22), increase from 9 to 19 KPA. Condenser and reflux drum is internal in tower, and we cannot measure the reflux rate. Yesterday, bypassed s team trap on reboiler outlet, and flooding s topped. Conclude that flooding tower due to defective steam trap. What's your opinion? Note tower fractionation also improved after trap bypassed. Regards, Jon Sacha Dear Jon: You're quite wrong. When you bypass ed the steam trap, you blew the condensate seal on the reboiler outlet. This permitted uncondensed steam to blow through th e reboiler, thus reducing the reboiler duty. The reduction in the reboiler duty reduced the vapor flow up the tower and hence the internal reflux rate. This un loaded the trays and stoppe d the flooding. Your obse rvation that the tower fractionation improved as a consequence of bypassing the steam trap was a positive indication that you h ad d egraded tray efficiency due to entrainment. That is, you were operating above the tower's incipient flood point. Certainly, there is nothing amiss with your reboiler steam trap. You should try to water wash the trays above the feed point ons tream, as the trays in this service are typically subject to NH4Cl salt sublimation. Hope this helps.
Best Regards, Norman P. Lieberman
3.9. Tower Pressure Drop and Flooding It is a characteristic of process equ ipment that the best operation is reached at neither a very high nor a very low loading. The intermediate equipment load that results in the most efficient operation is called the best efficiency point. For distillation trays, the incipient flood point correspond s to the b est efficiency point. We have correlated this bes t efficiency point for valve and sieve trays as compared to th e measured p ressu re drops in many distillation towers. We have derived the following formula:
where ΔP = pressure drop across a tray section, psi NT = the number of trays TS = tray spacing, inches s.g . = specific gravity of clear liquid, at flowing temperatu res On the basis of hundreds of field measurements, we have observed K = 0.18 to 0.25: Tray operation is close to its best efficiency point. K = 0.35 to 0.40: Tray is suffering from entrainment—increase in reflux rate noticeably reduces tray efficiency. K = ≥0.5: Tray is in fully developed flood—ope ning a vent on the overhead vapor line will blow out liquid with the vapor. K = 0.10 to 0.12: Tray deck is suffering from low tray efficiency, due to tray deck leaking. K = 0.00: The liquid level on the tray is zero, and quite likely the trays are lying on the bottom of the column. K = 1.00: Tower is completely full of liquid.
3.9.1. Carbon Steel Trays
One of the most frequent causes of flooding is the us e of carbon steel trays. Especially when th e valve caps are also carbo n steel, the valves have a tendency to stick in a partially closed p osition. This raises the pres sure d rop of the vapor flowing through the valves, which, in turn, pus hes u p the liquid level in the down comer draining th e tray. The liquid can th en back up onto the tray deck and promote jet flood due to ent rainment. Of course, any factor (dirt, polymers, gums, salts) that causes a reduction in the op en area of the tray deck will also promote jet flooding. Indeed, most trays flood below their calculated flood point, because of these sorts of problems. Trays, like peop le, rarely perform quite up to expectations. The us e of movable valve caps in any service where d eposits can accumulate on the tray decks will cause th e caps to stick to the tray deck. It's best to avoid this potential problem. Use of grid trays with fixed cap ass emblies is preferred for most services.
3.10. Optimizing Feed Tray Location From the d esign persp ective, the optimum feed tray, for a feed with only two components, is that tray where the ratio of the two components matches the ratio in the feed. If the feed is at its bub ble point temperature, then the feed temperature and the tray temperature will be the same, at the same pressure. But that's only for a binary feed composition. In multicomponent distillation, the ratio of the key components in the feed will typically not coincide with the ratio of the key componen ts in the liquid on the tray, even though the tray temperature is the same as the feed at its bu bble point temperature. So the question is, which of the following two criteria should be used to determine the feed tray location: Where the ratio of the key components in the liquid on a tray matches th e ratio of the key components in the feed? Or where the feed temperature matches the temperature of the tray? My practice is to identify both p ossible locations and t hen locate the feed nozzle halfway between the two options. I would provide two alternate feed
nozzles at each of the other above locations, except the operators would likely never use them. Inciden tally, when I refer to "Key Components," I mean, for example: Debutan izer—Normal Butane and iso-Pentan e Depropanizer—Propane and iso-Butane De-Ethanizer—Ethane and Propylene Gasoline Splitter—iso-Hexane and cyclo-Hexane All of the above s ervices have feeds with dozens of other non-key components.
3.11. Catacarb CO 2 Absorber Flooding I've never told this story to anyone. Not even to Liz or my mom. It occurred in Lithuania in 2006. I had been hired to expand the capacity of the hydrogen plant that was limiting refinery capacity. The bottleneck was the absorber that removed CO2 with catacarb solution from the hydrogen product. This absorber was subject to flooding as th e catacarb circulation rate increased in proportion to H2 production. That is, the solution was carried overhead with the hydrogen product. I stud ied the design of the tower, but could not see an explanation for the flooding. Nevertheless, I decided to modify all the 40 trays in the abs orber. The materials were ordered, and the labor force organized. However, the morning the ab sorber was opened, I received a call from my assistant, Joe. "Hey, Norm, there's kind of a plate in front of the solution inlet nozzle (see Fig. 3.12). It don't show on the drawings. What you want to do with th at plate?"
Figure 3.12. Restriction of the inlet distributor causes entrainment of the catacarb solution. "Joe," I answered, "I'll be there in 10 minutes." I looked at the plate. Dimension "x" was on ly about ¾ inch. Evidently, the plate was intended as an inlet solution distributor. I calculated th at the d elta P, as the s olution flowed underneath the plate, was abou t 15 inches: Delta H (inches) = 0.6 (V)2 Where V was the solution velocity through the ¾-inch gap in feet per second. The plate was 12 inches h igh. As th e solution rate increased , the liquid would back up over the top edge of the plate and be blown out of the top of the ab sorber. So, I told Joe to have the bottom 2 inches of the plate cut off to increase "x" to 2¾ inches. "And, Joe," I continued, "Also, close up the tower afterward. "
"Norm, but wh at about all the tower tray changes ?" Joe protes ted. "Don't argue. I know what I'm doing." When the tower started up a week later, the Hydrogen Plant bottleneck was gone. The plant manager never found out what I did, or that I had wasted $20,000 for unused tray materials. Perhaps , since I had achieved the objective, he wouldn't have cared. Anyway, the alternate propos al to expand H2 plant capacity, submitted by a major engineering contractor, would have cost $3,000,000. "All's well that ends well." Citation EXPORT
Norman P. Lieberman; Elizabeth T. Lieberman: Working Guide to Process Equipment, Fourth Edition. How Trays Work: Flooding, Chapter (McGraw-Hill Professional, 2014), AccessEngineering
Copyright © McGraw-Hill Global Education Holdings, LLC. All rights reserved. Any use is subject to the Terms of Use. Privacy Notice and copyright information . For further information about this site, contact us. Designed and built using Scolaris by Semantico. This product incorporates part of the open source Protégé system. Protégé is available at http://protege.stanford.edu//
