ABSTRACT

This paper reviews a decade of progress and presents new results. Short-term transient loads significantly degrade average (steady) pipeline capacity. Several papers show that transient optimization (T.O.) can recover much of this lost potential. The goal of T.O. technology is to support gas pipeline operators in optimally managing line-pack under time-varying loads, even when the location and timing of large peaking loads is uncertain. While in a transient environment the approach can minimize fuel and operational cost, its greatest potential lies in maximizing average attainable pipeline capacity in pipelines forced by users into transient operation. We review prior work showing a broad variety of circumstances in gun-barrel, Y-structures and simple loops where T.O. can be useful for both analysis and real time support. We then summarize progress of current work to turn "back room experimentation" into technology useable by operators on arbitrary pipelines. Finally, this paper presents new work using optimization to determine an accurate pipeline state quickly from only the most recent ½-hour SCADA history. Examples show this functionality can be effective even if some data are bad or missing. Robust state-finding is essential for initializing transient optimization in real-time support. The paper concludes by showing how new technology works in general, complex, previously T.O.-untreatable pipeline structures.1.

AUTHORS' PREVIOUS WORK

Steady state optimization (SSO) of gas pipelines is now a mature technology. This approach can determine the optimum pressure set-points and station bypass configurations of a general pipeline system to deliver the most gas possible using the least fuel, given supply and delivery pressures and a model of the system (including constraints). The resulting steady state of the pipeline (the flow and pressure by milepost) is useful as an ideal target. But SSO cannot address the issue of how to control the pipeline to maximize profit while delivering transient loads and achieving an optimal state for making tomorrow's deliveries. A decade ago a client asked how to control a pipeline to transit from a current to an optimum SSO-defined target state at day's end while maintaining all safety margins and successfully meeting rapidly changing client deliveries. Dialing in tomorrow's optimum target set-points and waiting does not guarantee attaining the desired state. It can also lead to pressure violations or delivery failure during the transition period. Moreover, the wait may be unacceptably long. We first addressed operational optimization in work reported at PSIG in 2000 using a hypothetical 25-station 1000-mile transmission gun-barrel pipeline as an example. To make deliveries and achieve the target goal efficiently and safely, we learned that the linepack must be managed in a coordinated way to move it continuously to the time-dependent optimum locations and release it in the right way. This pack management was achieved by manipulating station setpoints over time to move pack to the desired flowing state. Indeed we further learned how to manage linepack to deliver any number of transient loads in the meantime, all the while satisfying constraints1. This new technology was termed Transient Optimization (T.O.), and appeared to be a useful innovation. In the past there were no tools to aid operators in the sometimes difficult job of dynamically manipulating line-pack, or in rigorously determining the feasibility of time-dependent load patterns imposed by shippers.

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