Selection of a Representative Long-Term Period for Continuous Model Simulation

Abstract

Continuous-runoff simulation is recognized by USEPA and the United Kingdom's Wastewater Planning Users Group (WAPUG) as the preferred approach for evaluating collection systems and developing long-term control plans for urban sewer systems. Continuous simulations incorporate hydrologic processes that occur in both storm and non-storm conditions. The range of processes that affect collection system flows cannot be simulated using the common event-based simulation approach. Continuous simulation models can simulate storage recovery and loss rates, infiltration capacity recovery, evapotranspiration, and groundwater interaction. Even with the speed of today's computers, it is still often neither desirable nor practical to perform very long-term simulation of collection systems, as simulation times are long and output becomes very unwieldy. For these reasons, many modelers choose to simulate a single representative year, or a period of five or ten years rather than a typical complete 50-year or longer record. The representative year or years used in continuous simulation models is typically selected by identifying a period with nearly average precipitation, or by synthesizing a year's worth of storms based on storm return period statistics. However, this approach is likely to err due to the fallacy of assuming that average precipitation yields average collection system behavior. The methods described in this paper ensure that both precipitation and runoff yield statistics that are representative of the long-term conditions for the study area. A case study for Manchester, New Hampshire is described where representative five-year periods were identified for continuous simulation.