MODELING OF A LIQUEFIED NATURAL GAS (LNG)DEPROPANISER COLUMN

The removal of the condensable higher molecular weight constifuents from natural gas
prior to its liquefaction is very important both from the technical and economic aspects-of
the process. The higher molecular weight constituents like propanes, butanes, and
pentanes have to be removed to very low levels from the methane content, so that the
subsequent cryogenic
_units can operate properly. Further, the higher molecular weight
components have higher economic value and need to be sepaiated to realize betJer
economic potential from the natural gas (NG) facility. The major process in the NG
pretreatment to achieve the above objective is the distillation chain consisting of a chain
of 4 distillation columns, namely, the demethaniser, deethaniser, depropaniser and the
debutaniser' The high value of propane as a refrigerant for internal use in the cryogenic
s-y,stemsm, akes its operation very critical to the overall economics of the plant. In viJw of
this, Advanced Process Control techniques are often used for the control of this column.
To develop proper APC strategies for the column it is essential to have a distillation
column model. This paper presents certain preliminary work in developing a suitable
column model for the depropaniser column. The column modeled'is
"a
+7-stage
depro-panisecro lumn processing6 26.4 torslday of feed.T he column pressureis 1g.56b ar.
The feed contains 6 components propane, i- and n- Butane, I and n- pentane and n-
!9xqe The inputs to the simulation were the feed concentrations, flow rate and the
distillate concentration. The model was solved from the bottom plate up, going from stage
to stage. The material balance and temperature convergen.. rv.i. obtainJ usi"ng the theia
convergence method. After the temperature has converged, the liquid and vapor flow rates
were corected using the energy balances. The model developed in the study was able to
predict the plate compositions of an existing industrial column to good accuracy. The
model was used for generating answerst o many WHAT-IF tlpe of q:uestionsw; hilh can
be used for plant optimization.