Time Series Analysis of Hourly Domestic Water Demand

by James S. Gidley, Tech-Law Inc, Fairfax, VA, USA,

Abstract: Design flows for water supply and sanitary sewer systems are chosen by rules of thumb in the absence of actual data. This research develops two stochastic models of water demand by analyzing a time series of hourly domestic water demand data, in an effort to move towards risk-based design in water and sanitary sewer systems. The best available data yielded hourly flows over a two week period for nine homes; these data were consolidated using the concept of the composite home. A mixed probability distribution was fitted to the data, having a discrete probability of zero flow and a two-parameter lognormal distribution of nonzero flow. The static model treats each hour as independent of all other hours of the week. The dynamic model (Markov mixture model) includes state transitions from hour to hour. Synthetic sequences of water demands were generated by the two models and were compared to the historical data by calculating statistics for both the historical data and the synthetic sequences.

Subject Headings: Dynamic models | Hydrologic models | Water demand | Data processing | Water supply systems | Mathematical models | Time series analysis

Services: Buy this book/Buy this article


Return to search