Maldistribution fraction f (b)

FIG. 14-66 Loss of efficiency due to maldistribution as a function of maldistribution fraction f and the number of stages per bed for a given case study. (a) fmax and reduction in number of stages. (b) Showing larger efficiency loss at higher number of stages per bed and higher f. The steep drops indicate pinching associated withfmax. (From Lockett and Billingham, Trans. IChemE 80, Part A, p. 373, May 2002; reprinted courtesy of IChemE.)

Maldistribution fraction f (b)

FIG. 14-66 Loss of efficiency due to maldistribution as a function of maldistribution fraction f and the number of stages per bed for a given case study. (a) fmax and reduction in number of stages. (b) Showing larger efficiency loss at higher number of stages per bed and higher f. The steep drops indicate pinching associated withfmax. (From Lockett and Billingham, Trans. IChemE 80, Part A, p. 373, May 2002; reprinted courtesy of IChemE.)

1. Three factors appear to set the effect of maldistribution on efficiency:

a. Pinching. Regional changes in L/V ratio cause regional composition pinches.

b. Lateral mixing. Packing particles deflect both liquid and vapor laterally. This promotes mixing of vapor and liquid and counteracts the pinching effect.

c. Liquid nonuniformity. Liquid flows unevenly through the packing and tends to concentrate at the wall.

2. At small tower-to-packing diameter ratios (DT/Dp < 10), the lateral mixing cancels out the pinching effect, and a greater degree of maldistribution can be tolerated without a serious efficiency loss. At high ratios (DT/Dp > 40), the lateral mixing becomes too small to offset the pinching effect. The effects of maldistribution on efficiency are therefore most severe in large-diameter columns and small-diameter packings.

A good design practice is to seek a packing size that gives a DT/Dp between 10 and 40. This is often impractical, and higher ratios are common. When DT/D„ exceeds 100, avoiding efficiency loss due to maldistribution is difficult. Either ratios exceeding 100 should be avoided, or a special allowance should be made for loss of efficiency due to maldistribution.

3. Wall flow effects become large when DT/Dp falls below about 10. Packing diameter should be selected such that DT/Dp exceeds 10.

4. Columns containing less than five theoretical stages per bed are relatively insensitive to liquid maldistribution. With 10 or more stages per bed, efficiency can be extremely sensitive to maldistribution (Strigle, Packed Tower Design and Applications, 2d ed., Gulf Publishing, Houston, Tex., 1994) (Fig. 14-66). Beds consisting of small packings or structured packings, which develop more theoretical stages per bed, are therefore more sensitive to maldistribution than equal-depth beds of larger packings. This is clearly demonstrated by FRI's experiments [Shariat and Kunesh, Ind. Eng. Chem. Res. 34(4), 1273 (1995)]. Lockett and Billingham (Trans. IChemE, vol. 81, Part A, p. 131, January 2003) concur with these comments when their procedure (above) indicates high sensitivity to maldistribution, but allow a higher number of stages per bed when the sensitivity is low.

5. Maldistribution tends to be a greater problem at low liquid flow rates than at high liquid flow rates [Zuiderweg, Hoek, and Lahm, I. ChemE Symp. Ser. 104, A217 (1987)]. The tendency to pinch and to spread unevenly is generally higher at the lower liquid flow rates.

6. A packed column has reasonable tolerance for a uniform or smooth variation in liquid distribution and for a variation that is totally random (small-scalemaldistribution). The impact of discontinuities or zonal flow (large-scale maldistribution) is much more severe [Zuiderweg et al., loc. cit.; Kunesh, Chem. Eng., p. 101, Dec. 7, 1987; Kunesh, Lahm, and Yanagi, Ind. Eng. Chem. Res. 26(9), 1845 (1987)]. This is so because the local pinching of small-scale maldistribution is evened out by the lateral mixing, and therefore causes few ill effects. In contrast, the lateral mixing either is powerless to rectify a large-scale maldistribution or takes considerable bed length to do so (meanwhile, efficiency is lost).

Figure 14-67 shows HETPs measured in tests that simulate various types of maldistribution in FRI's 1.2-m column containing a 3.6-m bed of1-in Pall® rings. The y axis is the ratio of measured HETP in the maldistribution tests to the HETP obtained with an excellent distributor. Analogous measurements with structured packing were reported by Fitz, King, and Kunesh [Trans. IChemE 77, Part A, p. 482 (1999)]. Generally, the response of the structured packings resembled that of the Pall® rings, except as noted below.

Figure 14-67a shows virtually no loss of efficiency when a distributor uniformly tilts, such that the ratio of highest to lowest flow is 1.25 (i.e., a "1.25 tilt"). In contrast, an 11 percent chordal blank of a level distributor causes packing HETP to rise by 50 percent.

Figure 14-67b compares continuous tilts with ratios of highest to lowest flow of 1.25 and 1.5 to a situation where one-half of the distributor passes 25 percent more liquid than the other half. The latter ("zonal") situation causes a much greater rise in HETP than a "uniform" maldistribution with twice as much variation from maximum to minimum.

Figure 14-67c shows results of tests in which flows from individual distributor drip points were varied in a gaussian pattern (maximum/mean = 2). When the pattern was randomly assigned, there was no efficiency loss. When the variations above the mean were assigned to a "high zone," and those below the mean to a "low zone," HETP rose by about 20 percent. With structured packing, both random and zonal maldistribution caused about the same loss of efficiency at the same degree of maldistribution.

7. A packed bed appears to have a "natural distribution," which is an inherent and stable property of the packings. An initial distribution which is better than natural will rapidly degrade to it, and one that is worse will finally achieve it, but sometimes at a slow rate. If the rate is extremely slow, recovery from a maldistributed pattern may not be observed in practice (Zuiderweg et al., loc. cit.). Even though the

Was this article helpful?

## Post a comment