the debit side, the constants required for using the correlation are unavailable for many common packings.

Mersmann's correlation and Malkowlak's correlation. Mersmann (73) postulated that a thin liquid film forms in the flow channel of the packing. The ratio of film thickness to equivalent packing diameter is a function of the liquid load. Mersmann combined this function with a trickle flow model to yield an expression for dry packing pressure drop at flood as a function of liquid rate. Ma6kowiak (73a) surveyed sources that followed up and improved on Mersmann's initial model.

Madkowiak (736) compared predictions from Mersmann's correlation to a massive data bank. He found that Mersmann's correlation gave good prediction for the low-capacity first-generation packings (for which it was derived), but grossly optimistic flood predictions for modern high-capacity random and structured packings. As packing capacity increases, the flood pressure drop did not stay constant, as predicted by the correlation, but became smaller as described earlier. These problems are identical to those reported with the GPDC chart.

Mackowiak (73a, 736) derived a new flood correlation. Like the Billet and Schultes correlation, it is based on the drop entrainment model and takes liquid holdup into account. Unlike Billet and Schultes, Mackowiak uses a different set of premises and expressions. Ma6ko-wiak's correlation applies for both random and structured packings, has a good theoretical basis and was shown (73a,736) to predict a large number of flood data to within ±8 percent. On the debit side, the correlation is complex and requires the availability of four constants for each packing. Mackowiak also states (73a) that for high liquid rates, Mersmann's film model is more suitable than his drop model.

Flood prediction by interpolation. GPDC interpolation plots are used to interpolate actual flood data. Data interpolation gives accurate flood-point prediction, but can only be used where sufficient flood point data are available.

Until recently, interpolation using GPDC plots was the only published method for predicting structured packing flood points. Figure 8.18 shows GPDC flood plots by Fair and Bravo (50), Billet (3,50), Spiegel and Meier (21), and Dolan, Hausch, and Petschauer (2,24, 31e) for several structured packings. Billet (3,56), Fair (41), and Maikowiak (736) present GPDC plots also for many random packings. MacDougall (58) checked Billet's plots for Raschig and Pall® ringB against data and found them to give good flood predictions.

Chapter 10 presents a compendium of GPDC data interpolation charts for flood, MOC, and pressure drop prediction, both for random and structured packings. When flood data are absent, pressure drop data can be used for approximating the flood point using Eq. (8.1).

flaurn (Sulzer BX) Sfleetmelal (Mallapak 2S0 Y)

V Sakaia (1972), xylsnes A Meier, 1979 (aii-water)

Fractionation Research. Inc. + Meier, 1979 (methanol-water) T Billet, (J969)(E0/siyrene) O Meier. 1979 (cMorobenieneeihylbemeni)

Rgur* 8.18 GPDC interpolation plots for structured packings flood points, ia) The Fair and Bravo plot for Sulzer BX® and Mellapak® 250Y. (Part a, from J. R. Fair and J. L. Bravo, I. Chem. E. Symp. Ser. No. 104, p. A183, 1987; reprinted courtesy of the Institution of Chemical Engineers (UK).)

For random and structured packings, all the flood data plotted on the GPDC interpolation charts of Chapter 10 are based only on definitions that describe incipient flooding (Sec. 8.2.3). Unfortunately, for grids, all literature flood data are based on the definition "the gas velocity at which the packing pressure drop reaches 2 in of water per foot of packing." This definition describes "fully developed" rather than "incipient" flood (Sec. 8.2.3). For lack of alternative, those data were plotted on charts 10.8005 to 10.8108. Therefore, for grids only, interpolation of flood data points (charts 10.8005 to 10.8218, four charts only) is likely to yield capacities roughly 10 to 15 percent higher than incipient flooding.

Others. Flood correlations are often available in the manufacturer literature (e.g., 8,31,82) or on manufacturers' computer disks. A flood correlation by Beck (83) was shown by MacDougall (58) to be accurate, but applies only for Pall® and Raschig rings and Berl and Intalox® saddles.

Which method to use. Data interpolation is generally the most accurate and should be preferred whenever flood data are available. Otherwise either if pressure drop data are available or when pressure drop can be reliably predicted, Eq. (8.1) is recommended. When the packing factor Fp exceeds 60, the Eckert correlation, Fig. 8.17 is recommended. At

Was this article helpful?

## Post a comment