This paper addresses the use of a consecutive steady-state analysis technique to solve the problems of season analysis of a natural gas storage and transportation system. A summary of the problem is presented followed by a discussion of the methodology of its solution. The analysis involves the use of steady-state techniques to simulate the daily operation of the gas system over the course of the season. Daily loading is varied over the period using dimensionless profiles to simulate storage draw-down and determine compressor fuel consumption. At any point in time, the operator may intervene to change the operating strategy; moreover, the simulation will continually check for times at which operator intervention is required. The steady-state analysis utilizes a modified Newton-Raphson method to solve the system of node continuity equations with the substituted element flow equations. This scheme includes logic to accommodate the checking of operating constraints and actions to be taken to stay within these constraints.
This discussion addresses the basic problem of modeling a natural gas system for an extended period of time to determine such effects as: system adequacy to meet the anticipated loading; compressor fuel requirement over the extended time period; and storage injection and withdrawal schedules. Conventional analysis techniques, both steady-state and transient, do not really address this situation. Steady-state analysis, by definition, does not permit parameters to vary with time while traditional transient analysis examines the problem of varying loads over a fairly short period of time (a day for example) to determine the short term capacitance of the system and row it affects the daily operation. When longer periods of study are required, generally two things occur: First, the effects associated with the short term capacitance of the system are minimized, and second, the computational effort required by the transient simulation becomes prohibitive. This effect of exhausting the system capacitance is a good indication that a steady-state technique would be valid to represent a portion of time. The idea of using steady-state analysis to represent a finite time period, and then linking these analyses together through time to perform a study, is not new. In 1976, Orin Flanagan of Arkansas-Louisiana Gas Company, made a presentation to this conference of his work in the area of modeling production field withdrawals using a linked steady-state analysis. At the same meeting, this author presented work of a similar nature being done at Columbia Gas Transmission Company. I am also aware that others, including Union Gas of Canada, have done work on their own models toward the same end.
In examining the characteristics that such a model must have, three rather broad considerations stand out above all others as the real problems which must be faced. First of all, values that are changing over time must be integrated. If this cannot be done, then the entire modeling process breaks down at this point and the use of a transient-based analysis technique is indicated. The second consideration is that of information handling.
