Article ID: | iaor20113403 |
Volume: | 12 |
Issue: | 1 |
Start Page Number: | 153 |
End Page Number: | 174 |
Publication Date: | Mar 2011 |
Journal: | Optimization and Engineering |
Authors: | Camponogara Eduardo, Castro Pereira, Plucenio Agustinho, Pagano Juan |
Keywords: | programming: integer, programming: nonlinear, location, scheduling |
In gas‐lifted oil fields, high pressure gas is injected at the bottom of the production tubing of the wells to artificially lift oil to the surface. Lift‐gas should enter each well at a certain mass flow and pressure, giving rise to the problem of deciding which compressors (facilities) should be installed and how they supply the demands of the wells (clients). This compressor scheduling is a mixed‐integer, nonconvex, nonlinear programming problem that generalizes the standard facility location problem. By piecewise‐linearizing the performance curve of each compressor–a function relating output mass flow and discharge pressure, the problem is recast as a mixed‐integer linear program. This paper presents this linear reformulation, proposes families of valid inequalities, and reports on results from the application of these inequalities to solve representative instances of the compressor scheduling problem.