Abstract:
System modeling can be used as a basis of planning and operation decisions. However, model accuracy and uncertainty will impact the confidence in, thus the conservatism built into, model based decisions. Model prediction uncertainty results from uncertainty in model parameters that are determined through calibration or are based upon modeler judgment. Uncertainty in pressure head has been evaluated from the perspective of system reliability (Xu and Goulter (1999) and Bao and Mays (1990) and in design. However, the relative impact of alternative data has not been compared to assess the need for estimating parameters. In addition, water quality within a distribution is strongly affected by the selection of water quality model parameters such as bulk and wall decay coefficients. In addition, since water quality is dominated by advective transport, hydraulic parameters and conditions also impact water quality. This is most clearly seen in the effects of tanks due to flows to and from the storage facility. The hydraulic system representation will also affect the flow distribution and travel times within the system. Many modelers suggest that all pipe models are required to adequately represent the true flow patterns. To date, only anecdotal evidence through analysis of simple networks under deterministic conditions has been presented justifying this conclusion. However, as more pipes are introduced, more uncertain parameters must be defined or calibrated for those components. This paper begins to examine the impact of alternative sources of uncertainty on water quality predictions by examining steady state conditions for a well-documented mid-sized system. The objective is to study the effect of alternative uncertain
