Water Distribution System Model Automatic Calibration Based On FmGA And Its Applications

Water distribution networks are the key infrastructure of urban development. Digital management of water distribution system has been a main problem of sustainable development of socio-economic in China. The general requirements of 2010 and the long-term objectives in 2020 technology development planning of urban water supply industry are keeping water supply safety and enhancing drinking water quality.To ensure the water supply safety especially water quality safety of drinking water when changing water sources, it is of obvious interest to better understand the general hydraulic and water quality behavior in distribution system. Modeling is one of the tools available for such a task.In accordance with the existing problems that lack of accuracy of the model input, plenty of field tests of model input and key parameters are conducted which includes roughness coefficient of pipes and valves, water demand patterns, pump characteristics, GPS (Global Position System) measurement of elevation, chlorine decay coefficient of bulk water and pipe wall, and AOC (assimilable organic carbon) kinetics. Some of the field test methods were improved which was better to understand the real parameters of the water distribution system. The process and implementation of the water distribution system modeling were introduced. The connection port of hydraulic model and SCADA system was designed to simulate water distribution system dynamiclly.Model calibration is the key part of water system simulation. Model automatic calibration using GA (Genetic Algorithm) is the forefront of current research and future direction of development. Model calibration precision issues were discussed and main problems of the present calibration methods were analyzed, such as parameters uncertainty and search time overwhelming problem that too much time consumed when calibrating large scale models. Corresponding countermeasures were presented, for example, parameters grouping, parallel GA and field tests of initial value of parameters. A novel multi-objective genetic algorithm based on the NSGA-II algorithm, which uses metamodels to determine optimal sampling locations for installing pressure loggers in a water distribution system when parameter uncertainty with FOSM is considered. The new algorithm combines the multi-objective genetic algorithm with adaptive neural networks. After the calibrated models using fmGA were precise enough to be applied to a real distribution system, digital platforms of water distribution systems were established to guide the water supply safety when changing water sources.In the water source change project, simulation results of different working conditions were analyzed. Graphs and tables of the domain elements in the results can be presented clearly when running scenarios using Extended Period Simulations with WNW software. Water consumption in the next few years was forecasted with grey mothod and corresponding water planning schemes and compare indexes were presented. The case study in city M was carried out and corresponding water planning schemes were compared carefully under the condition of changing water sources. Better scheme was recommended for the water supply safety in distribution systems.Water quality safety in distribution system is getting more and more concern. Water quality model was built based on the hydraulic model and field tests of water quality parameters and kinetics. Initial parameter estimation of kb and kw was studied with plenty of field tests. A new model that describes the bulk-mixing and complete mixing behavior within a cross junction is developed. A scaling (mixing) parameter s (0≤s≤1) can be used to combine the results of the bulk-mixing and complete-mixing models for more accurate estimation. Water quality model calibration precision issues were discussed. The scaling (mixing) parameter s is introduced as a parameter to be optimized in water quality two steps model calibration by GA.A multi-source chlorine model was studied as a calibration example to better understand automatic calibration process. Determination of the water supply areas of each source was done according to hydraulic simulation results. Kb is assumeed the same in its respective regions of each water source, and each kb in its respective regions was set up a uniform adjustment factor. Assumed that the orders of residual chlorine decay and limited concentration are known, so that the parameters need to be adjusted in the process of automatic calibration were only kb and node mixing coefficient s. Kw was adjusted with the related H-W C value.The regrowth of heterotrophic microbes (expressed as Heterotrophic Plate Count, HPC) in water distribution systems causes deteriorations in the quality of drinking water. For the relatively common situation where more than one water source supplies a distribution system, current single species water quality models are not able to represent meaningful differences in source water quality. First, a general framework for modeling the reaction and transport of multiple species in drinking water distribution systems is introduced and a bacterial regrowth model is built using EPANET-MSX. Second, the growth potential of heterotrophic microbes in water distribution networks is considered to be correlated with the amount of AOC and HPC can be modeled with AOC concentration indirectly.Heterotrophic microbes were considered to be the least microbial safety in drinking water according to the IWA water safety assessing report. In order to study the kinetics of AOC in full scale water distribution network, effectiveness factor concept was introduced to discuss the kinetics theory of AOC based on a simple reactor. Several assumptions were put forward to analyze the AOC kinetics in full scale water system on bulk reaction and pipe wall reaction. And corresponding kinetic equations were presented. Field test of water samples showed that kinetic equations fit well with AOC variation.AOC model is built based on kinetics study to model the HPC in distribution system indirectly.