IMPROVED LIQUID HOLD-UP
CORRELATION
FOR
PACKED TOWERS
Simon Piché, Faïçal Larachi,*Bernard P.A. Grandjean
Department
of Chemical Engineering & CERPIC
Laval
University, Québec, Canada G1K 7P4
Chem. Eng. Res. Des. (Trans IChemE part A), 79, 71-80 (2001)
Abstract:
The state-of-the-art tools for the evaluation of the total liquid holdup in
gas-liquid counter-current randomly dumped packed beds are critically evaluated
by thoroughly interrogating a wide hydrodynamic database. This database
consisting of ca. 1,500 experiments on liquid hold-up below the flooding point
represents an important portion of the non-proprietary information released in
the literature since the 1930s. Providing access to diversified
information, it is dedicated to embrace wide-ranging temperature and gas
density levels, and packing shapes extending from the classical ones to the
modern third generation packings. Furthermore, a total of eleven
correlations on the total liquid hold-up extracted from the literature are
cross-examined with the database. Many limitations, regarding the level
of accuracy and generalization, come to light with this investigation.
Artificial neural network modeling and dimensional analysis are then proposed
to improve the accuracy in predicting the total liquid hold-up in the
pre-loading and the loading regions of packed beds. A combination of five
dimensionless groups, comprising the liquid Reynolds (ReL), Froude (FrL) and
Ohnesorge (OhL) numbers as well as the gas Froude (FrG) and Stokes (StG)
numbers are used as the basis of the correlation. The correlation yields
an absolute average relative error of ca. 14 % for the whole database and
remains in accordance with the trends reported in the literature.
Keywords:
randomly packed bed, counter-current flow, liquid hold-up, neural network,
database
You can get the packedbedsimulator.zip
file that contains an Excel worksheet simulator to compute pressure drop,
liquid holdup along with loading and flooding capacities.
You may also download our Excel worksheets simulators for Trickle-bed or Flooded
Bed reactors.
The neural
correlation was developped with the software NNFit