Experimental Study On Calibration Of Pipeline Roughness In Water Distribution Networks

With the expansion of the dimension of urban construction, the scale of waterdistribution networks is becoming more and more large, its structure is also moreand more complex, the important status in the urban infrastructure is rising notably.At present, the design, operation, management and maintenance for waterdistribution networks system, water hyd raulic model is the most efficient method.Because of the water distribution networks business charge system and the SCADAsystem put forward applications of water distribution networks system, thestatistical techniques of node demand flow mature, so pipeline roughness becomethe major factor which influences precision of water hydraulic model. At present，the study on pipeline roughness calibration in the water distribution networks,calibration methods are simple, low calibration accuracy, the long calibrationcalculation time. Therefore, the further study on calibration problem of pipelineroughness in the water distribution networks is theoretical and practical value. Thispaper studied on efficient and effective ways of obtaining optimal calibrationparameter values and the uncertainties associated with those values.This paper analyzed the basic theory and method of water hydraulic calculationon the condition of the steady flow, and fina lly chooses Hazen-William coeffic ientas parameter calibrating pipeline roughness coefficient. The paper introduced theinitia l estimate method of pipeline roughness coefficient in the water distribut ionnetworks according to the actual situation. This paper painted pump’ characteristiccurve and value characteristic curve based experimental data of pressure anddifferential pressure which acquired by using pressure sensor and differentialpressure transducer etc. Calculated the sensitivity coefficient index of pipeline inwater distribution networks, and experimented with six kinds of water distribut ionnetworks operation condition. Calibration model of pipeline roughness in the waterdistribution networks is established by the EPANET software based on theexperiments of simulating various operation conditions, which combining thevariable scale genetic algorithm established by the MATLAB software, calculatingsix kind of calibration model of pipeline roughness coefficient, and calibrationresults are analyzed and compared, and concluding that the dynamic range geneticalgorithm grading calibration has the ability of overall and local search, in additionthe efficiency and precision of calibration are high.So far, most of the calibration method of pipeline roughness coefficient in thewater distribution networks based on the optimization model concentrate primarilyon efficient and effective ways of obtaining optimal calibration values. At the same time, very little effort has been made to determine the uncertainties (i.e., errors)associated with those values (and related model predictions). So far, the first-ordersecond moment (FOSM) method is the general method to solve these problems.Even though effic ient computational efficiency, the FOSM method relies on severallimitations hypothesis and requires computationally demanding calculation ofderivatives. To overcome the above limitations, the recently developed shuffledcomplex evolut ion metropo lis (SCEM-UA) global optimization algorithm is linkedto the EPANET hydraulic model and used to solve a least-squares-type calibrationproblem of pipeline roughness coefficient. The methodology is tested and verifiedon the experimental water distribution networks platform case study. Comparedwith the existing optimization algorithm, the main advantages of SCEM-UAalgorithm is that both calibration parameter values and can also determine itsrelated the uncertainties in the calibration models run. In addition, no assumption ofcalibration parameter linear and no calculation of derivative.