TABLE OF CONTENTS Lewis A.Rossman
U.S. Environmental Protection Agency,
National Risk Management Research Laboratory
Cincinnati, OH
Pipe network flow analysis was among the first civil engineering applications programmed for solution on the early commercial mainframe computers in the 1960s. Since that time, advancements in analytical techniques and computing power have enabled us to solve systems with tens of thousands of pipes in seconds using desktop personal computers. This chapter discusses how modern-day computer models are used to analyze the hydraulic and water-quality behavior of distribution systems. It covers how computer models are applied to actual systems, what the internals of the models consist of, and the capabilities and operation of one particular model in the public domain, EPANET. The chapter focuses only on models that analyze successive periods of steady flow through a general arrangement of connected pipes, pumps, valves, and storage facilities. Other, more specialized computer models, such as programs for surge analysis, are not addressed.
The classical pipe-network flow problem asks what the flows and pressures are in a network of pipes subject to a known set of inflows and outflows. Two sets of equations are needed to solve this problem. The first set requires conservation of flow to be satisfied at each pipe junction. The second specifies a nonlinear relation between flow and headloss in each pipe, such as the Hazen-Williams or Darcy-Weisbach equation. Whenever a network contains loops or more than one fixed-head source, these equations form a coupled set of nonlinear...
