Research On Inversion Method Of Friction Coefficient For Oilfield Water Injection System

The value of friction coefficient of the network system has been changed after long run, but the existence information and the standard can only provide the primal value, which cannot accurately reflect pipeline's actual situation, so the difference between the operating mode and the primal design network is obvious. Therefore, how to estimate friction coefficient according to the limited measured data when computing and analyzing the operating condition of the actual pipe network is worth to be researched. The article uses four methods to calibrate friction coefficient of the oilfield water injection system based on the work of our predecessors.The article proposes the tree law to calibrated friction coefficient of the small scale pipe network. The pipe network has been changed into the tree network which has the tree root, the bough and branch, then adjusted 0, 1, 11 bough of pipe roughness coefficient to serve the calibration purpose, and then the feasibility of this method was confirmed by an deal pipe network.The friction coefficient is calibrated by the node equation set. Sensitivity analysis on the pipe network is done, then some pipes'Hazen-William coefficient of small sensitivity value is taken experiential value, and the others are calibrated by the node equation set. The result can simulate the operating condition of the pipe network better, and the convergence rates of the algorithm is quicker for the middle and small scale pipe network.Friction coefficient is calibrated by the usage of the improved BP algorithm in the small scale pipe network. By simulating the operating condition of the pipe network in different Hazen-William coefficient, the samples that the nodal pressure about the Hazen-William coefficient's changing are obtained. By making use of the improved BP network to establish nonlinear relationship between Hazen-William coefficient and pressure, then actual friction coefficient of the pipe network is forecasted. This method has been confirmed in an ideal pipe network, and then a better forecasting result is obtained.Friction coefficient is calibrated based on the improved particle swarm algorithm by grouping. By collecting the non-correlated data of many kinds of operating conditions, the pipes can be grouped according to the region. Taking Hazen-William coefficient as the decision variable, an actual pipe network has been calibrated by the improved particle swarm algorithm, then the precise optimization result is obtained, and this result could reflect actual operational instance accurately.