石油学报 ›› 2011, Vol. 32 ›› Issue (1): 167-170.DOI: 10.7623/syxb201101029

• 石油工程 • 上一篇    下一篇

基于POD的埋地热油管道停输再启动模拟高效算法

张劲军  张文轲  宇  波   

  1. 中国石油大学城市油气输配技术北京市重点实验室  北京  102249
  • 收稿日期:2010-05-24 修回日期:2010-07-22 出版日期:2011-01-25 发布日期:2011-03-23
  • 通讯作者: 张劲军
  • 作者简介:张劲军,男,1962年3月生,1998年获石油大学博士学位,现为中国石油大学(北京)机械与储运工程学院教授,主要从事油气长距离管输技术及易凝高黏原油流变学研究
  • 基金资助:

    国家自然科学基金项目(No.50944030)和北京市优秀博士学位论文指导教师科技项目(YB20081141401)联合资助。

An efficient algorithm of the proper orthogonal decomposition for the restart computation of buried heated crude pipelines

  • Received:2010-05-24 Revised:2010-07-22 Online:2011-01-25 Published:2011-03-23

摘要:

基于特征正交分解(POD)技术,开发了一种进行埋地热油管道正常输送、停输及再启动3个过程计算的高效算法。首先采用有限差分法和有限容积法建立基础样本,然后通过奇异值分解(SVD)算法得到管道停输后的沿线油温、土壤温度场和再启动流量基函数库,并插值得到任意计算参数条件下的谱系数,最后进行重构。该算法可在少量计算条件数值模拟的基础上,快速、准确地求解其他条件下停输再启动过程的相关参数。实际算例表明,POD模拟结果与直接数值解法相比,温度偏差在0.05℃以内,流量偏差小于3%。POD算法计算速度显著提高,单算例平均耗时小于0.1 s,而数值算法单算例平均耗时约20min。

关键词: 埋地热油管道, 停输再启动, 数值模拟, 特征正交分解, 奇异值分解

Abstract:

The present paper proposed an efficient algorithm employing the proper orthogonal decomposition (POD) for the computation of the normal operation,shutdown and restart processes of buried heated pipelines.Firstly,the finite volume algorithm and finite difference algorithm were adopted to generate fundamental samples,and then eigenfunctions of the oil and soil temperature fields after pipeline shutdown and the restart flux were obtained by means of the singular value decomposition (SVD) algorithm.Further,spectral coefficients under the condition of arbitrary computation parameters were calculated by interpolation and finally the relevant physical quantities were rebuilt.By using this new algorithm,pipeline restart parameters under other conditions could be obtained very efficiently and accurately only by means of numerical computations with fewer input data.Case studies demonstrated that the deviations between the POD simulation and the direct numerical computation were less than 0.05℃ for temperature and 3% for the flow rate,respectively.Therefore,the computation efficiency was improved greatly by using the POD algorithm,and the average time consumption for the computation of one-group input data was about 20 min by using the numerical algorithm,while the POD algorithm averagely took only 0.1s or even less time.

Key words: buried heated pipeline, restart after shutdown, numerical simulation, proper orthogonal decomposition(POD), singular value decomposition (SVD)