Modeling And Parallelism Simulation Of Pipeline Networks In Chemiacl Processes

Pipeline networks are the major parts of large-scale chemical,petrochemical, petroleum refining and processing plants and various tankfarms. Their safe and efficient operation has a great significance for boththemselves and other pieces of equipment connected by them. The maindifficulties in modeling and solving pipeline networks are thecomplicated topology structure, the nonlinear relationship between flowand pressure in pipelines, the sparse of Jacobi matrix and estimate ofinitial value. Based on the analysis of the previous work, thesis launchedthe following research work of connectivity time-varying pipelinenetworks in chemical process:(1) In order to meet the demand of steady-state analysis method whichmake the independence basic circuit sets as system equations and chordflow as independent variable (Chord flow method), several proceduresare proposed to construct the incidence matrix and fundamental circuitmatrix of a graph directly from those of its parent graph or graphs whenthe topological of the pipeline network changes.(2) According to the characteristics of chemical process, the proposed model and topological analysis procedures are used to establish adynamic solver for a tank farm together with the chord flow method.Finally, the dynamic solver is applied to a tank farm of liquor forverifying the model and procedures.(3) In the Linux cluster system, the preliminary attempt of parallelismdynamic simulation of pipeline network of chemical process has beentaken: the process of several independent subnet steady solving aredistributed into multiple processors, the calculation results of eachprocessors are arrangement unified. The results show that the increasingof computing node can reduce the calculation time significantly.