ABSTRACT

Engineering design and planning studies of pipeline networks are often involved with the task of optimizing the performance of the overall system, in some sense. In recent years, optimization packages based on pipeline network simulators have become available for the engineer to use as tools in these studies. Although there is a growing literature on the technology used in these optimizers, little has been said about their use in practice. In particular, their use is rarely fully automatic, and all of the algorithms currently available benefit from the guidance of a trained engineer. We aim to offer some guidelines for the use of optimization software in practical system design studies. Some of the questions we address are: How do we define and set up the optimization problem? How can the problem be simplified to clarify and speed up the results? How can the optimizer be helped past plateaus and barriers to convergence? What kind of new technologies are emerging that will genuinely be useful in the future? Examples are given covering some general categories of optimization studies to illustrate each of these questions, and the discussion is kept independent of the actual software used in the studies.

1. Introduction

Any pipeline system with control facilities involves settings which influence the efficiency of the network. The correct allocation of compression through the system, and the correct setting of pressure or flow regulators can dramatically influence the total throughput of a network. Similarly, the correct utilization of compression facilities, coupled with regulator settings, can have a large impact on the overall cost of fluid delivery. Process simulators, used in facilities engineering for refineries and other industrial applications, have long had optimization packages attached to them. These have relatively recently been adapted for use in the pipeline area. The new application has been a challenge mainly because of the large number of extra nonlinear constraints imposed by the detailed compressor models used by pipeline simulators. Nevertheless, there is a growing literature on this subject, especially with regard to the mathematical techniques used. In this paper, we concentrate on the practical application of optimization software, but also give a literature review at the end. Typically, automated optimizing simulators aim to improve one of three functions of the state of the pipeline:

  1. The total fuel used in compression.

  2. The total throughput of the system.

  3. A function, usually related to pollutant emissions, of the compressors' power, efficiency and fuel consumption.

Other objectives have also been considered, in particular relating to transient studies [l, 141, but are rare in day-today work. These functions need very careful definition, as we discuss in the next section below, to make any sense.

This content is only available via PDF.
You can access this article if you purchase or spend a download.