Page 309 - Applied Process Design For Chemical And Petrochemical Plants Volume II

298 Applied Process Design for Chemical and Petrochemical Plants

1. Establish liquid and vapor rates. such as in Figure 9-21C may not be consistently accurate
2. Determine fluids physical properties. due to the variations in data used for correlation, i.e., data
3. Select design pressure drop for operations. Suggested just as flooding begins, and then at full flooding. The data
values of below 1 .O in. water/ft. Low-pressure, atmos- presented is for gas-liquid systems and not liquid-liquid
pheric, and pressure columns usually require 0.5 to extraction as Strigle [82] recognizes in his Chapter 11.
0.7 in. water/ft, with absorbers and strippers around Strigle [82] identifies a regime 20% above point F on
0.2-0.6 in. water/ft. For vacuum distillation low val- Figure 9-22 as the maximum hydraulic capacity and is
ues of 0.05-0.6 in. water/ft are often necessary, usu- termed the flooding point for atmospheric operations.
ally depending on the required boiling point of the Kister [93] has correlated large quantities of available
bottoms. data for flooding and offers a new correlation based on
4. Calculate Lf/Gf, values of Fpd cancel out. the same GPDC correlation, which is, as Kister refers to it,
5. Using Equation 9-31B trial and error, calculate using the Sherwood-Leva-Eckert (SLE) correlation chart as
Gf, for given Lf/Gf until the desired AP is obtained. Lf developed by Strigle [82], Figure 9-21H, for semi-log plot.
must be below 20,000. For higher Lf use chart in the Kister [go, 931 has presented plots showing the plotted
original article not included here. data points on the Strigle or SLE charts for a wide selec-
6. Select packing for column, and establish packing fac- tion of packing. In effect this illustrates how the data fit
tor, Fpd. the generalized charts.
7. Calculate column cross-section area using the opera- For example, Figures 9-26 and 9-27 from Kister [93]
tional gas rate, G, and the calculated value of Gf (gas illustrate the collected data superimposed on the SLE
loading factor) : chart for the specified pressure drops, specific packing
G = Gf/[(0.07?i/pg)0.5 (Fpd/20)o.5], lb/hr/ft2 (9 - 31F) and column size, and packing heights at designated pack-
ing factors. Kister [93] recommends using the specific SLE
8. Establish tower diameter; Robbins [96] recommends chart (also see Kister [go] for a wide selection of charts)
that the tower diameter should be at least 8 times the and interpolating and extrapolating the curves when the
packing size; if not, repeat the calculations with dif- design/operating requirements fall close to the data
ferent packing. points on the selected chart. Extrapolating too far can
ruin the validity of the pressure drop results.
where A = tower cross-sectional area, ft2
CO = constant specific to a particular packing When the design/operating requirements are far from
C1 = constant specific to a particular packing the chart’s data points, contact the manufacturer of the
c3 = 7.4 x 10-8 packing for data and also consider selecting another type
c4 = 2.7 10-5 and/or size of packing to provide a better fit. In selecting
D = tower diameter, ft a chart to use, do not overlook the nature of the physical
F, = packing factor, dimensionless process system, i.e., whether predominantly aqueous or
Fpd = dry-bed packing factor, dimensionless non-aqueous, and whether there are foaming characteris-
F, = V, (pg)O.j, (ft/sec) (lb/ft3)o.5 tics as well as high or low viscosity fluids. Ester’s assess-
G = gas loading, lb/hr-ft2 ment is that both the study in Reference 93 and Strigle
GA = design vapor flow rate, lb/hr [82] show that the large volume of published data does fit
Gf = gas loading factor the Strigle [82] charts (Figure 9-21E, F) quite well and
L = liquid loading, lb/hr-ft2 gives “good pressure drop predictions.” A significant vari-
Lf = liquid loading factor ation is that the curves predict an optimistic (too low)
AP = specific pressure drop, in. water/ft of packing value for non-aqueous systems at high flow parameters of
hppb = specific pressure drop through dry bed, in.
water/ft of packing the chart. Also, similar optimistic values are noted for non-
ITs = superficial gas velocity, ft/sec aqueous systems at low flow parameters, such as for vacu-
pg = gas density, Ib/ft3 um distillation, for example. Unfortunately, most data
pL = liquid density, lb/ft3 from the data banks were obtained on small scale as com-
p = liquid viscosity, centipoise pared to industrial size equipment, and so the statistical fit
is still not adequate for industrial design with confidence
Strigle [82] and Kister [931 point out the importance of and reliability [93].
evaluating data where available to reduce the need for
interpolating the GPDC charts. The question of reason- Pmfmance Comparisons
ably accurate (+lo to 15%) flooding pressure drop data
has been studied by Kister [93], and the results suggest Strigle [94] presents some helpful comparisons refer-
that the establishment of flooding pressure drop curves ring to Figure 9-22, and Tables 9-28-31 from [94] present