Different operating regimes. The operating regime on the tray governs tray efficiency (17,99,106-110,116,117,147). Scaieup of efficiency data from one regime to another is therefore dangerous and should be avoided. Further, the test data must be at similar column loading to the prototype (18,135).
Scaieup from an existing commercial column. As long as data are for the same system under similar process conditions, loadings, and operating regime, this can give excellent efficiency predictions. If both columns are about the same size, data obtained in one column directly extend to the other. Fractional hole area and the number of passes have a small but significant effect (Sec. 7.3.4), and any changes in these parameters need to be allowed for in the scaieup. The empirical information in Sec. 7.3.4 is useful for estimating the magnitude of these effects. Weir height and hole diameters also have a small effect, and these need to be considered as well.
There have been suggestions in the literature (145) to allow for tray geometry using a theoretical model. This procedure is satisfactory only if the model correctly predicts the trends of variation of tray efficiency with geometry. Considering the unreliability of current theoretical efficiency models (Sec. 7.2.1), it is doubtful whether they can be trusted for this purpose. Further, it was shown (15) that many flood correlations givegoodpredictions but poor trends of variation of flood points with tray geometry; the same may apply to tray efficiency. Few studies compare measured to theoretically predicted trends of variation of tray efficiency with geometry.
Scaieup from pilot column. The most common scaieup is from a small column (e.g., pilot plant) to a commercial column. No reduction in efficiency on scaieup is expected (24,99) as long as
1. Both pilot and commercial columns are operated at the same approach to the flood point (18,134,135). Also, the flood point in the pilot column must not be premature (e.g., due to foaming). This can be checked by flood testing the pilot column using a standard nonfoaming system, and comparing to the test system flood point.
2. Both pilot and commercial columns are operated in the same hydraulic regime. Larger columns, in general, tend to operate in the froth regime (Sec. 6.4.2). A parameter that can (and should) be manipulated in the pilot column to produce the desired operating regime is the tray spacing. With the pilot column running at the same percent flood as the prototype (1 above), a higher tray spacing implies higher vapor load and a tendency to generate spray, while a smaller tray spacing will promote froth regime operation.
Both the bubble (deep pool) and the cellular foam regimes (Figs. 6.25a,b; 6.26a) are common in pilot and lab columns, but rare in commercial practice. Cellular foams need the small-column wall stabilization, while a deep pool forms only behind tall outlet weirs in small columns. In both these regimes, the dispersions are tall and the residence times high, leading to excellent efficiencies. Surface tension gradients further boost efficiency in the cellular foam regime (116,206). Higher efficiencies were reported in both the bubble (116) and cellular foam (116,206) regimes than in the froth or spray regimes. Pilot column data collected in the bubble or cellular foam regimes will therefore give optimistic efficiency predictions.
4. Any relevant process factors (Sec. 7.3.6) are allowed for. Also, it is a good policy to pilot-test over several composition ranges engulfing all those expected in the commercial column. A pilot test is never an exact replication of a commercial column, and differences may lead to poor scaleup.
5. Pinching is avoided. It has been recommended to pilot-test at total reflux (18). At finite reflux, pinching can convert small measurement errors into major errors in efficiency estimates (130). However, finite reflux testing is useful in supplementing a total reflux test and providing information on pinch-point location.
6. Watch out for stagnant regions on the larger tray (Sees. 7.3.2, 7.3.3). A scaling-up process that failed to meet its objectives, presumably for this reason, has been described (24,170). Techniques for dealing with stagnant sections are in Sec. 7.3.3.
7. If more than two passes are used in the larger trays, watch out for possible maldistribution (Sec. 7.3.4).
8. In general, as column diameter increases, tray efficiency first increases because of the increased liquid plug flow character on the tray, and then slowly decreases because of stagnant regions and vapor plug flow. In general, it was stated that scaling up from pilot columns would be a conservative means of predicting column efficiency if the above precautions are observed (24,99).
Scaleup from Oldershaw columns. One laboratory-scale device that found wide application in supplying efficiency data is the Oldershaw column (Fig. 8.13; Ref. 207). This column is available from a number of laboratory supply houses and can be constructed from glass for atmospheric operation or from metal for superatmospheric separations. Small hole diameters and small tray spacings are used. Typical column diameters are 1 to 3 in.
Fair et al. (208) investigated scaleup of Oldershaw column data to commercial columns. Over the region of practical interest (50 to 85 percent of flood), the commercial point efficiency was either equal to or slightly higher than the Oldershaw column efficiency. Fair et al. concluded that the Oldershaw column efficiency is essentially a point efficiency measurement, and recommend this point efficiency for the design of commercial columns. A mixing model can be used to convert the point efficiency to overall column efficiency. This will enhance the commercial column efficiency. In a later paper (145), Chan and Fair include an additional correction for weir height; this correction will also enhance the commercial column efficiency. A conservative approach proposed by Fair et al. is to apply the Oldershaw column effi-
ciency as the overall column efficiency of the commercial column. This approach takes no credit for the greater plug-flow character and taller weir upon scaleup. The author prefers this conservative approach, considering the poor reliability of mixing models (Sees. 7.2.1 and 7.3.2),
Previous work with Oldershaw columns (209-211), however, spells a note of caution to Fair et al.'s conclusion. For a fixed system, higher Oldershaw column efficiencies were measured under cellular foam conditions than under froth conditions. For this reason, Gerster (212) warned that when cellular foam can form, scaleup from an Oldershaw column may be dangerous. The conclusions presented by Fair et al. (208) do not extend to Oldershaw columns operating in the cellular foam regime. Other considerations for scaleup from pilot columns (above) may also be important. The scaleup procedure recommended by Fair et al. (208) is
1. Run the real system in the Oldershaw column,
2. Determine the Oldershaw column flood point.
3. Establish the Oldershaw column operation at 60 percent of flood.
4. If good VLE data are available, run the Oldershaw column at total reflux and calculate efficiency. Obtain overall Oldershaw column efficiency and assume it is the same as the commercial column point efficiency. Use a mixing model to calculate the overall column efficiency for the commercial column. A conservative alternative is to assume that overall commercial column efficiency is the same as overall Oldershaw column efficiency,
5. If good VLE data are not available, vary reflux ratio and find by test the combination of reflux and stages that will give the desired separation. Assume that a commercial column with the same number of trays and operating at the same reflux ratio will give the same separation as the Oldershaw column. The number of plates thus calculated can sometimes be reduced by estimating the efficiency enhancement from point to column efficiency.
Some limitations to this procedure are mentioned above and in Ref. 208. In addition, it has been reported that reflux ratio may have a marked effect on Oldershaw column efficiency (213) and this variation must be studied carefully, especially if measurements are not conducted at total reflux. Also note that the formation of the cellular flow regime is composition-dependent (116,206), and it may occur under some conditions, but not under others.
7.4 Nomenclature (Chapters 6 and 7)
7.4.1 English letters
Aa Active area (the same as bubbling ft2.
ADB Area at the bottom of the downcomer, ft2. (Note: In multipass trays, ADB is the sum of the bottom areas of all downcomers transporting liquid from the tray.)
Apr Same as AD.
Ada Area under the downcomer apron, ft2.
Ado Area of holes on the deck of a valve tray, ft2.
Ar Fractional hole area ( = AJAa).
Ah Hole area, ft2.
a, Interfacial area per unit volume of liquid and gas holdup, ft2/ft3.
a'i Interfacial area, ft2.
AN Net area (column cross-section area less downcomer top area), ft2.
As Slot area, i.e., the total vertical curtain area through which va por passes in a horizontal direction as it leaves the valves or bubble caps, ft2.
As,, Open slot area, i.e., slot area when the valve units are fully open, ft2.
AT Total tower cross-section area, ft2.
b Intercept on linearized equilibrium relationship, Eq. (7.11).
C Constant in Bennett et al.'s pressure drop correlation, defined by
Cd Coefficient in the clear liquid height correlation, defined by Eq.
Cs C-factor, a parameter describing vapor load, defined by Eq. (6.4),
Was this article helpful?