Research On Stochastic Model Of Node Water Demand Based On Monte Carlo Method

The accurate establishment of water distribution modeling plays an important role in state analysis of urban water supply system. The accuracy of the water distribution model and the reliability of the results of pipe network state analysis, to a large extent, depend on the accurate degree of the calculation of network nodal demand. The traditional calculation method of node flow distribution is not quite reasonable. Based on the necessity of node demand calculation in probability statistics method, the paper introduced a stochastic model of node demand in detail. With a new perspective on the trail probability statistics of network node demand, the model in a certain extent resolves the unreasonable aspects of calculation the network node demand with traditional method. This thesis mainly included the following:1. Based on the research results on the node demand at home and abroad, this paper analyzed the stochastic character of node demand and variety of node head and pipe flow. The result showed that the random changes in node demand inevitably affected node head and pipe flow change significantly.2. Summarizing traditional methods of nodes demand calculation in the modeling process of water pipe network, and the research status of probability and statistical methods, this paper proposed a stochastic model to evaluate node demand based on the assumption of Poisson Rectangular Pulse, and found that using pulse intensity, pulse duration, the number of pulses and the time pulse occurred to describe the water consumption was more accord with the actual amount. In dealing with the model Monte Carlo method was used, and the results showed that the simulation results and measured results fitted better, and met the engineering requirements better.3. Because different end-use brought different pulses, we divided water consumption into two parts: daytime (6:00 -18:00) and in the evening (18:00 - 6:00 the next day). Using parameters estimation and hypothesis testing theory in probability and statistical to made goodness of fit test on the random variable distribution of node demand model, and determined the distribution form of the water pulse intensity, the pulse duration and other variables.4. With the characteristics of node demand model and the Monte Carlo method, and different distribution forms of the random variables in the model, this paper produced variables random sequences by using the Monte Carlo method and got water pulse in the whole day of single-user and multi-user, translated it into average flow of each period and compared with the measured data. The simulation results were satisfactory.