A further hike in vapor rate may trigger flooding by expanding the liquid holdup to the point at which the phases begin to invert, with the liquid phase becoming the continuous phase. After flooding has begun, the liquid can continue to build up above the fioodpoint m the column.
Vapor, being the continuous phase in a packed column, does not bubble through the liquid, as in a trayed column. Because of this and the absence of vapor-flow orifices (as in a tray), packed columns operate at much lower pressure drops than trayed columns. Because liquid and vapor contact in a packed column is less agitated than in a trayed column, the contents of packed columns are less likely to foam.
Pressure drops through random packing typically range from 0.3 to 0.9 in. water/ft of packing. At the higher end of these ranges are smaller packings and packings having less void space (such as ceramic packing): at the lower end. larger and more-open packings (Figure 2).
Low pressure drops. 0.3 to 0.5 in./ft. are normal for absorption, scrubbing and vacuum columns. Structured packing usually operates at lower pressure drops, typically 0.1 to 0.7 in. water/ft of packing.
Between 0.8 and 1.4 in. of pressure drop/ ft of packing, liquid entrainment may begin to reduce the apparent efficiency of the packing. Flooding often occurs when the pressure drop exceeds 1.3 to 2.5 in./ft of packing.
Predicting the efficiency of packing, or HETP'. in a particular column with accurately is difficult. HETP is a function of packing type and size, and the physical properties, relative volatilities, distributions and flowrates of the liquid and vapor. Efficiency can best be assessed be means of plant columns in the same service, models scaled from such columns, software models based on data from such columns, or measurements from laboratory models.
Hri'.'lu emuvaieiu intforelit'^i niale i> ;i im-asuiv ni ):;;i»-lrans[i.T <.'lhvv:H".: m a ¡jaCKed liis'.iilation ominin. it * h'-:Lrii; ->: :iarki:'.i' 'nat makes a -inanition equivalent liia*. "I a ui.'Hivtk'a: mute, rnr nuu-r ivivtu ar'.iae* m Chemical EnL'iiteenii'.' on :laikeii .iistulauon cumins. >ee "Practical Ti|p> on Tower r'atkin-.'. ! 'v. 7. l!tv7. ... 101: "Boostinu Tower lYrior* .«¡law i»v Mure Than a Trickle." May 27. liKi. |>. Si: -inic;invu .> uic bv-.vorti in Tower-Packinc Worii!." Mar. . ::».'.-¿: and "Parked Column Internal." Mar.."». i'.—:.
For troubleshooting purposes, how-j ever, a particular packing type and size | can be categorized by a typical HETP. As a general rule, one can assume for random packings an HETP of 18 in. for 1-in. packings; 26 in. for ll/2-in. packings: and 36 in. for 2-in. packings. For structured packings, typical HETPs are 9 in. for 0.25-in. crimp: 18-in. for 0.5-in. crimp: and 33 in. for 1-in. crimp.
Random packings, of which there are many types, sizes and materials, are so called because each packing in a bed need not be oriented in any specific way — i.e.. they can be dumped into a column. Common structured packings are manufactured into mats of thin, corrugated metal sheets, layered together and custom installed. Both ge-, neric and proprietary packings are ! available.
Once chieflv solid-walled (such as
: Raschig rings and Berl saddles) and ; constructed of ceramic, metal or plas-i tic. random packings are now predomi-| nantly the slotted-ring type fabricated mostly from metal or plastic (Figure 2). I Newer proprietary packings having as-! pect ratios that differ from those of standard ring packings have improved I column performance. However, older packings, such as ceramic rings and i saddles, are still used in corrosive services.
' The capacity of a particular packing depends on its geometry, and liquid and vapor flowrates and physical proper-I ties. In general, capacity increases with packing void-space. I As might be expected, corrosion is | common with metal packings. For the j typical metal-ring wall thicknesses of | 15 to 35 mils, a corrosion rate of 10 mils/yr is severe because packing is at-. tacked from two sides. Even if corro-I sion rates are moderate, it is not unusu-j al for packing to be replaced ! frequently. Corrosion is an even great-| er problem with structured packing, i The corrugated-sheet and knitted-wire | packings are manufactured mostly | from 5-to-8-mil-thick material. ! Plastic and ceramic packings usually I resist corrosion better, but present other : problems. For example, heat and solva-; tion can soften plastic packing, and me-
FIGURE 1. Liquid enters the column via a feedpipe (which is sometimes a perforated pipe), flows through, in sequence, a distributor (in this case, the V-notch type), a packing holddown plate (which keeps the topmost packing from fluidizing), an upper bed of packing, a bed support plate, and is spread out by another distributor (sometimes call a flashing-feed distributor) before it passes through a lower bed of packing, which is also sandwiched between a holddown plate and a bed-support plate
! chanical shock and thermal cycling can j disintegrate ceramic packing. Distorted ( packing and packing chips can reduce bed voidage, limiting column capacity.
; Potential troublemakers
The liquid distributor is the single most i important internal structure of a i packed column. Although a packed bed I could not exist without a support plate. ! the distributor is critical for initiating I uniform contact between liquid ana va-, por. It. therefore, strongly influences packing efficiency. Packings with fair I to poor ability to redistribute liquid will i not correct poor initial distribution. ■ Local fluctuations in the ratio of liquid i to vapor flowrates can vary component I separation, as if the column were divided into many smaller ones, each operating i at a different reflux ratio. In the same i way, a malfunctioning distributor can | undermine the performance of the entire ; bed. This cannot happen in a trayed col-i umn. because each tray redistributes the j liquid from the tray above, minimizing ; the effect of a single bad tray, j The liquid distributor must spread the ! liquid uniformly, resist plugging and ! fouling, provide free space for gas flow, j and allow operating flexibility. Distribu-j tors fall into two categories: gravitation-: al — such as orifice and weir types, ana ' pressure-drop types — such as spray ; nozzles and perforated pipes.
Orifice-tvpe distributors, perhaps the j most prevalent in smaller columns, con-i sist of flat trays having risers for vapor flow and orifices for liquid flow (p. 124). Common in columns larger than four feet in diameter are troughs having bottom orifices. The feed liquid and liquid downflow from upper beds should ! build a uniform head over all of the ori-| flees, so as to balance the flow from j each orifice (Figure 4). The key to orifice-distributor operation is the liquid head and flow relationship that limits a distributor's liquid flowrate flexibility. The basic orifice-distributor hydraulic relationship is given below:
Here, Q = distributor flow, gal/min: n = number of orifices in distributor: d = orifice dia., in.; and H = liquid head loss. in. of the liquid, j The actual liquid level on the distrib-! utor is the sum of liquid head loss through the orifices plus the pressure drop of the opposing gas flow through the distributor (usually less than 0.25 in.) For uniform liquid distribution, a minimum liquid head of 0.75 to 1 in. is recommended, which allows for the distributor being somewhat out of level.
By matching the minimum acceptable liquid head with the minimum expected liquid fiowrate. the engineer designs the distributor for some combination of orifice count and size. Four to ten orifices per ft2 of distributor cross-section is common.
As the foregoing correlation indicates, orifice distributor hydraulics follow the usual head-to-fiow-sauared relationship. Therefore, for a maximum liquid flow-rate of four times the minimum liquid fiowrate. the distributor height must accommodate a liquid level of 16 times the minimum head. Similarly, for liquid turnup ratio of six. the distributor height must be sized for a maximum liquid level of 36 times the minimum liquid head. For this reason, a turnup or tumdown ratio of three to four represents a practical limit for the orifice distributor, and a ratio of 2:1 is common.
At a liquid fiowrate less than the design minimum, the liquid head will begin to drop below the lower design level. Below this level — for example, at 0.25 inch head — a tilted distributor could cause liquid to flow preferentially on one side of the column. Furthermore. a tilt combined with a low liquid head could allow liquid to g cling to the bottom of the distributor. abetting the maldistribution. At higher than design liquid rates,* the liquid level could ^ rise above the design maximum level and overflow into the vapor risers. This could result in liquid entrainment and flooding.
A drawback of orifice redistributors is the difficulty of ensuring mixing of liquid downflow and feed liquid. Liquid feedpipes that uniformly distribute liquid onto the distributor enhance liquid mixing. To ensure good mixing, the feed liquid and liquid downflow are sometimes collected in a common header before being supplied to the distributor.
The V-notch weir distributor (above)
FIGURE 3. Liquid replaças vapor as the continuous phase as a column changes from normal to flooded operation
FIGURE 2. Presenting less void, ceramic and smaller packing create a higher pressure drop
FIGURE 2. Presenting less void, ceramic and smaller packing create a higher pressure drop
resists fouling and plugging bet ter than does the orifice distributor. An inferior variation is the notched-riser distributor. whose circular risers for vapor flow are notched to allow counter-current liquid flow. Less common are gravity-flow perforated-pipe distributors.
The V-notched distributor would appear to offer a greater turnup or turndown ratio than the orifice distributor because liquid height (H) is proportional to flow-to-the-exponent-2.5 i^-5) in its hydraulic relationship (compared to Q2 for orifice distributors). However, the increased horizontal thrust of liquid from a notch ^0at excessive liquid T' rates can cause poor distribution. At low flowrates and liquid levels less than 0.75-in. of head in the notch, liquid may flow erratically down the side of the trough. Distribution is also hindered by liquid-level gradients in the troughs. Thus, the V-notch distributor, like the orifice distributor, is limited to operating within the design liquid fiowrate.
Because of the head and flow limitations of the gravity-flow distributor, the troubleshooter should make sure that such a distributor is operating within design limits. The troubleshooter should refer to design specifications.
vendor drawings, or field calculations.
The perforated-pipe distributor (p. 125) avoids the level problems of gravity-flow distributors. However, a high pressure drop across discharge orifices can cause spray, entrainment and maldistribution. Discharge velocities should be about 3 to 6 ft/s. Because of the higher discharge pressure differential, the discharge holes are smaller than for gravity distributors, increasing the likelihood of plugging.
Spray nozzles, the most common distributor in scrubbing columns, are inexpensive but produce a mist that is easily
entrained. In addition, they are apt to plug, and feed rates to them usually must be high to compensate for poor distribution.
With flashing feeds, the entrance en-erg}- of the entering vapor and liquid must be absorbed to prevent internal damage and entrapment or flooding. This requires a special distributor.
Distributors are custom designed for uniformly distributing two liquid phases. Also, special narrow-trough distributors having a high concentration of drip points per square foot are frequently installed in columns of structured packing.
Redistribution is necessary whenever liquid from a bed above must be collected and distributed to the next bed — whether or not feed is introduced to the bed. Bed heights are typically limited to between 20 ft and 30 ft. mainly so that redistribution can correct poor liquid distribution and radial concentration gradients.
A redistributor collects all the liquid downflow, including wall flow, allowing no bypassing. Riser covers prevent bypass through the gas risers. A distributor only receives liq^ uid directed to it. It does not collect downflowing liquid. A distributor generally is installed above the top bed in a column, or below a special support plate or collector tray.
So that the support plate will not bottleneck the flow of liquid and vapor, its open area should equal, or exceed, that of the packing void fraction. Most plates have either slotted vapor risers or slotted ridges that provide separate paths for liquid ana vapor passage. With these types of plates, the effective free area for gas flow can exceed 957< of a tower's cross-sectional area. However. this area of many plates falls in the 80 to 907< range, which is usually acceptable if the ridges or risers promote separate liquid and vapor flowpaths.
High gas velocities can fluidize packing on top of a bed. Unless restrained, the packing could be carried into the distributor, become unlevel or. if frag
ile. be damaged. If the packing is metallic or plastic, a bed limiter should be installed at the surface of the bed. The limiter should be secured to the wall or otherwise anchored, so that column upsets will not dislocate it. A slight settling of the bed below the bed limiter need not be a concern.
With ceramic packing, which can be broken by movement, even slight fluidization must be suppressed. A weighted holddown plate (left) resting on the packing surface will prevent fluidization. The plate is not attached to the wall so that it can settle with the bed. Collection and chimney trays pass vapor while collecting liquid for sidedraws or other purposes. Vapor distributors, sometimes an integral part of the support plate, are occasionally installed to reduce anticipated vapor maldistribution from vapor inlets. They are often recommended if the inlet kinetic energy of the vapor exceeds 1.5 in. of liquid. Mist eliminators may be installed at the top of the column to collect and return liquid entrainment.
Liquid and vapor inlets and outlets can be the source of trouble. A nonflashing feed is usually delivered to a distributor through a perforated pipe. The perforations should distribute feed onto the distributor to minimize hydraulic gradi ents and to maximize nujang^iih-.uid Jowngjpv tfce&aflsbovi'--The ■ feedpjp« must be elevated and oriented so as to prevent feeding liquid into vapor risers or onto riser covers. Elbows and slotted pipes tend to not completely direct liquid downward. and should not be elevated more than a couple inches above nearby vapor risers. Pipe nipples around feedpipe discharge orifices are often preferred.
A flashing feed should not be delivered through a perforated pipe sized for liquid flow. The pipe entrance may be open or deflected, depending on the type of distributor fed by the pipe.
Liquid outlets, particularly column sidedraws. should allow the liquid to degas before it is withdrawn. This is often done by means of a recessed sump. Exit nozzles and pipes must be sized to prevent flashing in the pipes. Overflow protection are often included in total-collection trays, co prevent liquid from backing up in the column, if the side-draw flow is stopped. Nozzles for feeding vapor, or for returning vapor or va-por-and-liquid mixtures from the reboiler, should be located far enough below the bed-support plates to mini
mize the liquid and vapor expansion into the support plates or distributors that could disrupt the bed.
Among the causes of flooding or excessive pressure drop in packed columns are: packing hydraulic limit, restrictive support plate, fouled packing (caused cnGinecfliriG FCOTURC
by precipitation, lodgment of loose material and debris), damaged packing, foaming, improper feed introduction.
(such as a collapsed bed. or a fallen distributor) and restricted liquid outlet.
Packing hydraulic limit — On the basis of demonstrated or design rates, the troubleshooter should have some idea of a column's maximum capacity. If the column is flooding at rates equal to. or exceeding, the column's expected capacity, the troubleshooter should suspect that the packing capacity is limiting the column. This can be checked by comparing predicted capacity and pressure drop with those observed.
Internal vapor and liquid rate and physical property data, particularly density, are necessary for capacity and pressure-drop correlations. For a quick check, the top and bottom liquid and vapor rates can be determined by means' of an overall heat and material balance. Of course, complete stage-by-stage calculations will more accurately identify the most loaded point in the column.
The operating pressure drop for random packing is readily estimated by means of the generalized pressure-drop correlation. Compare predicted and actual pressure drops carefully, because pressure drop is a function of the pressure drop causes a deviation froir. this relationship, espeetite at moderate to high liqu^fiowrates.
.dive port plater that lack sufficient vapor ,|ow area. or that have become fouled "or pit ged. will create a capacity pinch-point. Such a restriction will often cause premature flooding before other confirmed in laboratory distillation apparatus. Fouled packing can sometimes be washed online with a solvent, or dumped and washed una high-pressure water, for instance), rather than replaced.
Damaged packing — Packing car, be damaged during operation or by improper installation (Figure 5i. Column upsets, such as flooded operation or va-
FIGURE 6. Fouling reduces packing void space, increasing tendency to flooding
square of vapor velocity — thus, a 15% error in the calculated vapor rate can result in a 32% error in predicted pressure drop. The liquid contribution to
FIGURE 6. Fouling reduces packing void space, increasing tendency to flooding
FIGURE 7. Crushed packing drastically boosts column pressure drop indications of a loaded column appear (such as high bed pressure-drop). The actual open area for gas flow can be as low as the support-plate open area multiplied by the packing void fraction, because some packing comes to rest partially in the support-plate openings.
Fouled packing — Solids in feeds or. more often. from precipitation and crystallization in the column can foul and plug packing (Figure 5). Solids I dissolved in the feed stream can precipitate below the feedpoint after the solvent has been stripped from the downflowing liquid. By reducing the void fraction of the packing, fouling boosts the pressure drop and can lead to flooding. Precipitation problems can be por surges that fluidize the bed, can break ceramic packing. Ceramic packing ground into smaller pieces reduces the void fraction of the bed. hiking the column pressure drop and increasing the likelihood of flooding. Small pieces of packing have often been found in pump suctions.
Plastic packings can be overheated or solvated. Both can soften the packing, compacting the bed. reducing the void fraction, and thus column capacity. Packing softening and excessive bed depth can cause the packing to be extruded through the support-plate openings. further restricting the flow area.
Foaming—Foaming systems will flood early if the foam does not readily break down into its liquid and vapor phases. The rising vapor flow will push this higher-density pseudo-phase up the column, creating high resistance to the liquid downflow. Laboratory tests with feed and product samples at the operating temperature and pressure will often confirm a foaming problem.
as will the temporary addition of an anti-foaming agent.
Improper feed introduction — A flashing feed or a vapor feed that can expand upward into a support plate can interfere with normal liquid and vapor traffic. Flashing-feed energy not absorbed by distributors can generate massive entrainment that can initiate flooding. Reducing the feed temperature may alleviate this problem.
Flooded gas riser — If a distributor or collection tray is flooded with liquid, such as by a blocked sidedraw or high-er-than-design liquid flowrates. the injection of vapor from the risers into the liquid may generate entrainment. This can rapidly lead to flooding because there is often little space between the top of gas risers and the bottom of bed-support plates for disengaging the liquid from the vapor. A distributor hav-
A packed column was installed to separate methanol and water from ethylene glycol. The separation is easy, but a control problem was expected because of the sharp temperature profile and the difficulty of controlling such a profile in a packed column having a low liquid holdup. (In comparison with traved-column operation. the temperature and composition profile move rapidly and is sensitive to small changes in the heat balance.)
Problem: Soon after the column was comissioned, a process engineer noticed that, in stripping all the methanol and water from the ethylene glycol underflow, an excessive amount of ethylene glycol was being taken overhead.
Troubleshooting: Increasing the reflux did not seem to improve the separation. Because column control remained steady, it appeared that the column's efficiency was much lower than had been expected. A check revealed that the internal drawings did not indicate how the top orifice-pan liquid distributor was to be attached to its support ring. One explanation for the lower efficiency was that an upset had dislocated the distributor. Another was that the distributor's orifices were plugged, causing poor distribution.
With a contact pyrometer having a probe sharp enough to penetrate through the insulation to the vessel wall, the crafty process engineer measured the column's radial and vertical temperature profiles. One side of the column being colder than the other side for several feet below the reflux distributor confirmed the hypothesis of liquid maldistribution.
Corrective action: An inspection of the shutdown column showed that the reflux distributor was tilted. This caused it to dump all the reflux down one side (the cold side) of the column. The distributor was securely and evenly clamped to its support ring.
Outcome: Started up, the column achieved the desired separation. A check showed the radial temperatures just below the distributor to be uniform in profile.
ing -less than 20c/< of its cross-sectional area for vapor flow will tend to act like a sieve tray in that it will not allow liquid to pass down the gas risers.
Internal damage — Disassembled distributors, perhaps from column upsets. can fall on a packed bed and cause flooding. A damaged support plate can cause an entire bed to collapse onto the bed below or into the base of the column (Figure 6).
Restricted liquid outlet — A restricted bottoms or sidedraw nozzle can -cause liquid to backup into a packed bed. A wide-open bottoms or sidedraw control valve may indicate a flow restriction.
Most efficiency problems can be traced to an improperly selected, designed or operated, poorly installed or degraded distributor, excessive entrainment. or vapor maldistribution. The trouble-shooter should be aware of the types of distributors in a problem column and whether they meet design objectives.
Distributor features — A weir-trough or orifice-pan (or trough) distributor installed between beds without provisions for collecting and redistributing liquid from the bed above constitutes a common error in new or revamped columns. Such an installation allows liquid from the bed above to flow into the next bed without being uniformly distributed or mixed with feed liquid. In orifice-pan distributors, vapor risers lacking covers represent a typical liquid-bypass route.
Process designers too often provide liquid feedrates to distributor manufacturers without mentioning that the feed will flash upon entry into the column. Flashing feeds require special feed distributors and compatible feedpipes. to prevent unstable distributor performance and excessive feed velocities. The number and layout of orifices, weirs or spray nozzles will influence the uniformity and coverage of the distributed liquid.
Distributor operation — Solids or other fouling ingredients in the feed can plug distributor openings, causing maldistribution or overflowing (Figure 7). Even a brief introduction of solids to a column, such as from a tank heel or through an upstream process upset.
€nGin€€RinG fcaturc can plug distributor openings. The in- j stallation of screens or filters to cap- !
ture construction debris should be con- j sidered at startup. Some fouling i materials, such as scale dislodged by ; thermal expansion from the column wall, are more difficult to control.
The liquid rate distributed to each !
bed. including downflow and feed, 1
must be within the maximum and mini- j mum design rates for the distributor. !
As noted previously, operating outside j of these limits can result in liquid mal- j distribution. This is a common problem i for columns that are operated at differ- ;
ent internal liquid flowrates or in sever- i al services. Distributors can often be j modified by adding or plugging orifices, ]
to better accommodate liquid flowrates. i
Distributor installation — Gravity- j flow distributors must be installed level I
to ensure uniform distribution at low j liquid rates and low levels on distribu- !
tors. A distributor more than 0.25 in. out j of level is cause for concern if the col- i umn will be operated at minimum rates.
Distributors and other internal structures should be secured to the column, to prevent their being disturbed during normal operation or column upsets. Gross defects, such as a poorly assembled distributor (e.g., one installed tilted), can cause serious liquid maldistributions. This problem can sometimes be detected by radial temperature measurements indicating hot or cold sides of a column or showing inverted temperature profiles (see box on p. tk).
Distributors oriented poorly with respect to feedpipes and bed limiters can interfere with liquid distribution or cause a mismatch between feedpipe discharge points and vapor-riser locations. Structured-packing distributors are specifically oriented with the packing. Exchanging distributors in multiple-bed columns mismatch of distributor type and liquid flowrates.
Corrosion — This can enlarge orifices and notches, dissolve gaskets or even cause the total collapse of a distributor. Even if well secured to the column. a distributor (as well and other in-| ternal structures) can be bent and even torn apart by flooding and vapor > surges (Figure 6V
Excessive entrainment — This iiq-I uid backmixing in the bed. which is i common as the floodpoint is ap-! proached. will reduce the apparerr effi-j ciency of the column. I Vapor maldistribution — This oc-J curs when an internal structure de-! fleets vapor flow, or when the packing i pressure drop is not sufficient to cor-: rect an inlet-vapor velocity profile. Typical obstructions include deep support-plate I-beams, unusual inlet-nozzle ! deflectors and deep drawoff sumps. Columns operated at very low pressure drop, are susceptible to vapor maldistribution if the inlet vapor velocities are l high, such as when vapor lines are un-i dersized as a result of capacity being ! boosted via a revamping. ■
Reonnted from CHEMICAL ENGINEERING. Aoril 1989. convrioht iaso hu .u ——------
Tracking the causes of flooding and loss of efficiency
Distribution wt if
Clearance under downcomar
Haat-anct-masa transfer area,
Bottom Mai pan
Was this article helpful?