A Mixed Integer Optimization Strategy for Process Retrofit of Ethane Plants- The Debottlenecking Problem

MARIA SOLEDAD DIAZ; ADRIANA SERRANI; RUBEN DE BEISTEGUI

LATIN AMERICAN APPLIED RESEARCH

PLAPIQUI(UNS-CONICET)

Lugar: Bahia Blanca; Año: 1996 vol. 26 p. 177 - 183

0327-0793

Natural gas processing plants are subject to frequent changes in feed flowrate, CompoSlTIOn, temperature and pressure. In the present study, a typical plant fed with gas from several pipelines is considered. The naturalgas from each pipeline comes from different gas fields and, consequently, there are considerable variations in gas composition and flowrate. T'ne plant prodnces ethane, LPG and C02. The process considered fur ethane and LPG recovery is a low temperatme turboexpander plant. The process economics requires the optimization of ethane and LPG recovery for varying feed compositions noder equipment capacity constrsints. Besides, an important constrsint in this type of process is C02 precipitation due to the low operating temperature of the demethanizing colunm. The main process optimization variables are demethanizer pressure and bottom flowrate and cold tank temperature. However, sigoificant structural changes can be introduced in the process scheme by minor plant modifications. This problem constitutes a restricted class of retrofit design called 'debottIenecking' problem; the Objective is to flod the optimal retrofit design where the topology of the flowsheet remains fixed and the replacing or modification of existing equipment is represented by discrete decisions. In this work, a flexible rigorous simulator of the cryogenic sector of the ethane plant has been developed so as to interface the optimization algorithm. The presence of discrete decisions to acconot for different process configorations leads to a Mixed Integer Nonlinear Programming (MINLP) optimization model. The optimization program is an implementation of the Outer Approximation algorithm that can be coupled to a simulation routine.with gas from several pipelines is considered. The naturalgas from each pipeline comes from different gas fields and, consequently, there are considerable variations in gas composition and flowrate. T'ne plant prodnces ethane, LPG and C02. The process considered fur ethane and LPG recovery is a low temperatme turboexpander plant. The process economics requires the optimization of ethane and LPG recovery for varying feed compositions noder equipment capacity constrsints. Besides, an important constrsint in this type of process is C02 precipitation due to the low operating temperature of the demethanizing colunm. The main process optimization variables are demethanizer pressure and bottom flowrate and cold tank temperature. However, sigoificant structural changes can be introduced in the process scheme by minor plant modifications. This problem constitutes a restricted class of retrofit design called 'debottIenecking' problem; the Objective is to flod the optimal retrofit design where the topology of the flowsheet remains fixed and the replacing or modification of existing equipment is represented by discrete decisions. In this work, a flexible rigorous simulator of the cryogenic sector of the ethane plant has been developed so as to interface the optimization algorithm. The presence of discrete decisions to acconot for different process configorations leads to a Mixed Integer Nonlinear Programming (MINLP) optimization model. The optimization program is an implementation of the Outer Approximation algorithm that can be coupled to a simulation routine.