Page 173 - Gas Purification 5E

160 Gas Purification

(tar continiredfrompage 157)
An approach for generating approximate LLE data, which could be used with convention-
al graphical or analytical methods for estimating the required number of stages or packing
height, was suggested by Honerkamp (1975). The method is based on the principle that when
two phases are in equilibrium with the same third phase, they must also be in equilibrium
with each other. In Honerkamp’s method, a hypothetical gas phase is assumed to be in equi-
librium with both the aqueous amine and LPG phases. This allows the large amount of pub-
lished gas-liquid equilibrium data to be used for liquid-liquid applications. For example, the
vapor pressure of COz, Pm2, above the amine solution is estimated by extrapolating avail-
able amine-carbon dioxide gas-liquid equilibrium data to the appropriate range. Then,
assuming that the LPG is in equilibrium with the same hypothetical vapor phase, the concen-
tration of C02 in the LPG is estimated using Raoult’s law:

(2-47)

where VP,, is the vapor pressure of pure COz liquid at the LPG treating temperature (up to
the critical temperature of COz). A similar procedure was proposed for H2S. This approach
was found to underestimate the carbon dioxide concentration in the LPG by a factor of about
6. To correct for this error and add a factor of safety, Honerkamp suggested that equilibrium
acid gas concentrations in LFG estimated by this method be increased by a factor of 12 to
obtain values to be used in design.
Holmes et al. (1984) modified the Honerkamp approach to improve its accuracy and adapt it
to computer simulation. They used the Kent-Eisenberg correlation to calculate the vapor pres-
sures of carbon dioxide and hydrogen sulfide over the amine solution, and the SoaveRedlich-
Kwong equation of state to correlate the vapor pressures of the acid gases over the LPG phase.
Equating the two correlations for each acid gas produces an expression that relates the concen-
tration of the acid gas in the amine solution to its concentration in the LPG phase at equilibri-
um. This expression can then be used to generate liquid-liquid equilibrium curves and, with
conventional column design techniques, to estimate the required number of theoretical stages.
If additional precision is desired, the basic approach of Honerkamp and Holmes could be
Mer improved by using a more rigorous model for predicting the acid gas-amine solution
equilibrium (e.g., the Deskmukh-Mather (1981) or Austgen et al. (1991) correlation coupled
with the Peng-Robinson equation of state to defiie the acid gas-LPG equilibrium.
Holmes et al. (1984) describe the use of their LPG/amine equilibrium model in the
TSWEET process simulation program to design LPG treaters. The TSWEET program,
which was originally written for gas treatment plants, is based on rigorous tray-by-tray cal-
culations. A comparison of the calculated results with actual plant operating data is given in
Table 2-30. The operating data for plants 1 through 7 were originally reported by Hon-
erkamp (1975). Although the measured product LPG acid gas concentrations are limited in
number and precision, the results indicate that the calculation method provides a reasonable
prediction of plant performance. Additional operating data for LPG treaters using random
packing are reported by Tse and Santos (1993).
The use of the Holmes et al. calculation method to evaluate contactor design alternatives
is described by Fleming et al. (1988). They considered MEA, DEA, and MDEA solutions for
reducing the carbon dioxide concentration in 50 gpm of LPG from 7.7 mole 8 to 0.16 mole
o/c. Both packed columns and static mixer/coalescer systems were evaluated. As a result of
the study, a singlestage static mixer/coalescer was selected. Operating data showed the unit
to be capable of reducing the C02 concentration to 0.10 mole % using 70 gpm of 25%. DEA.