离心泵网络的节能设计与操作优化:综述与展望

Energy Saving Design and Operational Optimization of Centrifugal Pump Network:A Review

首先对以单泵串并联为基础的离心泵网络的设计与操作优化命题进行解析,提出此类命题的研究框架可按照优化命题的基本结构分类成目标函数、约束条件和求解算法三部分。分析认为目标函数的组成取决于命题背景,等式与不等式约束可按模块化原则从单泵到单泵调整、多泵组合,进而泵网络设计、泵网络操作各级命题研究;而求解算法则聚焦于混合整数非线性规划(mixed-integer nonlinear programming,MINLP)的凸性与非凸性研究。然后依次给出了各部分内容的研究综述,主题集中于优化目标构成、面向建模的泵基本特性方程、泵网络优化设计型问题、泵网络优化操作型问题以及MINLP求解算法。分析表明,当前泵网络设计优化与操作优化的目标主要可以用寿命周期法描述,其中能耗和维护费用占比较大;泵基本方程主要基于出厂特性曲线回归,研究难点聚焦于现场流量缺乏的应对方式和管路对串并联阻力的影响;泵网络优化命题重点讨论工业问题的超结构表达形式并向多类网络的同步设计延伸,操作优化主要研究变负荷控制与操作调度,而MINLP研究重点则集中于相关确定性和不确定性算法开发。最后展望了泵网络的研究前景,实现性能监测和维护调度的智能设备管理和泵网络面向时间与空间二维分布式需求的柔性设计将成为新的研究方向。

Design and operational optimization problems of centrifugal pump network with pumps coupled in series and/or parallel are discussed. Related research framework is categorized into three parts （objective function, constraints and algorithm） according to the procedures of modeling and optimization problem solving. Contents of objective function depend on commercial domain background, and equality and inequality constraints can be grouped following modularizing design principle for single pump, pump coupled in series and/or parallel and design and operation of pump network. Moreover, contents of algorithm used in optimization problems can be classified into two parts depending on the convexity of Mixed-Integer Nonlinear Programming （MINLP）. Recent research on five aspects i.e. construction of objective function, equation oriented pump characteristics, design and operational optimization problems, and corresponding algorithm of MINLP are reviewed. The decomposition of objective function by life cycle cost analysis shows that the proportion of energy consumption and maintenance costs are increasing. Performance equations are interpolated from pump curves provided by pump manufacturer. The challenges come from lack of flowrate measurement and effects of pipe resistance on pumps. Research on design optimization now focuses on super-structure ofindustrial pump network and extends gradually to simultaneous optimization of all networks. Operational optimization focuses on the control of load shifting and pump scheduling. The discussion of algorithm remains in the category of deterministic and heuristic algorithm of MINLP. Finally, a prospect for future research on centrifugal pump network is proposed. Intelligent equipment monitoring, maintenance scheduling and facility management together with flexible design of pump network towards time-space dimensional distribution are discussed as future research areas.