A Influência de erros de medição de carga na calibração de fator de atrito em tubulações de redes de distribuição de água através do método transiente inverso e algoritmo genético

Abstract

The study of hydraulic networks for operation purposes or viability analysis for extension or renovation of the same is started the calibration in this context understood as identification parameters, such as friction coefficient, surface roughness and diameter. The Transient Inverse Method in conjunction with genetic algorithm is efficient in this task shows. This method employs the method of characteristics in the solution of the equations of motion for transient flow in networks of pipes and the optimization of solutions is based on Evolutionary Theory and evaluated by an objective function, which in this study is the sum of the difference between the module loads measured and calculated by the model for each set of solutions. Whereas the objective of the development of mathematical models for calibration hypothetical networks is their use in real networks, and that these, the collection of payload data is subject to measurement errors, is due to defects in the equipment or by conditions unfavorable environment or other random effects and taking into account the relevance of friction factors in pipelines, by their relationship to head losses that must be controlled to a great operation of networks, ensuring a continuous supply in quantity and appropriate operating conditions, this work is proposed to verify the influence of the presence of transient load measurement errors in the identification of friction factors in two hypothetical hydraulic networks. They are of different sizes with the number of rings, knots and tubes. Both are each fed by a reservoir. The transient conditions are assigned to a valve maneuver installed in one of the nodes of each network. The load data collection is restricted to 20% of the nodes in each network, one of which is the node where the valve is located. The hydraulic transient observation time is restricted to the valve maneuver time, 20s, and occurs at intervals of 0.1s, resulting in 200 charge records. The permanent condition of networks is initially unknown knowledge about the same is restricted to load in the reservoirs and demands on us as well as pipe diameter, the friction factors are initially stipulated. The determination of the permanent and transient conditions and the identification of the friction factors is performed using a hydraulic model and generate transient loads which are conventionally considered true, then these various steps of receiving systematic and random errors, which generate new burdens and these are considered collected with measurement errors. From these new loads are carried IDs friction factors, which are compared with those obtained considering an ideal case with no measurement errors loads. This comparison is performed using the mean relative error and function great goal. The results show that measurement errors in the identification of interfering friction factors although not possible to draw a relationship between them.