| As the natural gas pipeline network systems become increasingly large and complex,the operation and scheduling face unprecedented challenges.Pipeline simulation technology has become a core technique for solving optimization problems in large and complex natural gas pipeline networks.However,its relatively low computational efficiency severely limits the in-depth application of pipeline simulation technology in these extensive and intricate systems.Therefore,research has been conducted in three main aspects: the construction of simulation models for natural gas pipeline networks,optimization of model solving algorithms,and implementation of high-performance computing.The objective is to reduce the computational workload of transient simulation in natural gas pipelines and improve the efficiency of computer resource utilization,ultimately achieving the goal of enhancing the efficiency of transient simulation in natural gas pipeline networks.In the aspect of natural gas pipeline simulation models,compatibility,initial value stability,and computational efficiency of three discrete methods finite difference method,implicit cell centered method,and its linearization method are compared and studied.The advantages of the linearization method in terms of accuracy and efficiency are clarified.Based on this,combined with the requirements of transient simulation tasks for pipeline networks,both nonlinear and linear models for transient simulation of the pipeline network are constructed.The applicability of the two models is analyzed,and the accuracy of the models is validated.In the optimization of transient simulation solution algorithms,considering the characteristic of a large number of linear equations in the nonlinear model of transient simulation for pipeline networks,a fracture dimension reduction solution algorithm is proposed.This algorithm replaces the virtual connection points of inter-station pipeline network models with the inter-station pipeline network model,reducing the dimension of the system of nonlinear equations and improving the efficiency of solving the nonlinear model.To address the low efficiency and poor flexibility of traditional time step adaptive methods in adjusting time steps,a time step adaptive method based on a prediction mechanism is proposed.This approach reduces the transient calculation times in traditional methods while increasing the flexibility of time step adjustments,thereby improving the computational efficiency of the time step adaptation process and reducing the total number of time levels.Taking the transient simulation calculation of the XB natural gas pipeline network system as an example,for a 24-hour simulation task,compared to traditional methods,the improved time step adaptive method reduced the total number of transient simulation executions,the total computational time levels,and the trial calculations of transient equations by 27.00%,24.95%,and 51.81%,respectively.In the implementation of high performance computing(HPC),research has been conducted on sparse matrix compression storage methods for the inter-station pipeline network models in the nonlinear simulation model and the linear simulation model.This approach reduces the space complexity of the coefficient matrix of the linear equation system and improves the efficiency of solving both models.Additionally,a comparison was made between GPU parallel computing and CPU computing for solving sparse linear equation systems in pipeline network simulation.To address the issue of low efficiency in natural gas property calculations,research was conducted on natural gas property update techniques based on multicore parallel computing.The proposed fine-grained parallel update achieved a calculation speedup of 2.74.A matrix calculation method for natural gas based on the BWRS state of equation was established,achieving a calculation speedup of 2.92.Based on the comprehensive research results mentioned above,a rapid transient simulation program for natural gas pipeline networks has been developed.The accuracy and efficiency of this method were verified in a domestic circular natural gas pipeline network system.Compared with the domestic PNS pipeline network simulation software,the maximum pressure deviation in steady-state simulation did not exceed 0.01 MPa,and the relative error of transient simulation pressure was less than 1%.When compared with actual SCADA system measured data,the simulation pressure error did not exceed 1%.In terms of equation solving efficiency,the nonlinear simulation model constructed based on the fracture reduction algorithm significantly reduces the dimension of the equation set,with the dimension being as low as 0.63%.When the pipeline network is 100 times the current size,the computation time for solving the linear simulation model using sparse matrix technology does not exceed 0.24 seconds.In terms of property update efficiency,the optimized natural gas property update calculation efficiency is 3.04 times that of traditional methods. |
PDF Full Text Request