Date of Award

Fall 1998

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)


Civil, Construction, and Environmental Engineering


The focus of this work is on the development of methods to solve certain problems in gas load estimation. One of the problems is to transform historical daily data in one Gas Day format to any other format, depending upon which format is standardized by the gas industries. Gas pipeline companies need to control flow and maintain pressures at different nodes in their pipeline system for efficient distribution of gas. Prediction of pressures at the different nodes will help these companies maintain those pressures at specific values while not violating operational as well as contractual constraints, even under peak conditions. The supply of gas can be scheduled in advance if the demand at different nodes of the pipeline system can be predicted using some method. The contributions of this research are the development of two approaches to modify the historical daily data in current Gas Day format to a Gas Day format which has been accepted as a standard; and a model to predict hourly pressures at different nodes in a pipeline under operating conditions, based on identification methods such as Genetic Algorithm and Gradient Descent. Also, a micro demand model to predict the geographical demand over a service territory is proposed. Three ways of incorporating a priori knowledge to improve prediction of gas consumption, namely reference temperature transformation, load growth adjustment and temperature span effect, are proposed. Use of a priori knowledge in terms of day of week modified temperature to improve prediction of gas load is also proposed.



Restricted Access Item

Having trouble?