Figur« 7.S The O'Connell correlation for overall column efficiency, (a) Plot for distillation; (i>) plot for absorbers. (From H. E. O'Connell, Trans. AIChE, 42, p. 741, 1946. Reprinted courtesy of the American Institute of Chemical Engineers. )

The volatility and viscosity are evaluated at the average arithmetic temperature between the column top and bottom temperatures. The relative volatility is between the key components. For absorbers, O'Connell generated a second correlation (Fig. 7.56), using a solubility function instead of the relative volatility.

The O'Connell correlation was based on data for bubble-cap trays, and it was stated (131) to predict 90 percent of the efficiency data within ± 10 percent, both for distillation and absorption. For sieve and valve trays, its predictions are likely to be slightly conservative (151). Ludwig (4) warns that O'Connell's absorber correlation (Fig. 7.5b} sometimes predicts efficiencies that are too high. He believes that it can be used for stripping of gases from rich oils and for absorbers provided care is exercised not to accept too high values.

The combination of reasonable accuracy, good reliability, and simplicity, together with the weakness of theoretical tray efficiency correlations, rendered the O'Connell distillation correlation (Fig. 7.5a' the standard of the industry. It has been recommended by most literature sources (4,10,18,33,126,131,151,152) as one of the best empirical methods available for tray efficiency prediction. The author has had extensive favorable experience with the distillation correlation (Fig. 7.5a), and heard the same from many others in the industry. Frank (10) and the author believe that the O'Connell plot is the best computational method for estimating distillation tray efficiency; others (4,12,33), however, prefer theoretical methods.

The MacFarland, Sigmund, and Van Winkle correlation (153). Another empirical efficiency correlation often mentioned in the literature (12,131,152), it expresses efficiency in terms of dimensionless groups. It is based on binary data for sieve and bubble-cap trays.

Was this article helpful?

0 0

Post a comment