石化企业蒸汽动力系统柔性设计

Synthesis and design of flexible steam power system

蒸汽动力系统的设计工作是一项复杂的、系统的工程,除了力求实现全周期内总费用最小的经济性目标外,还要保证实际运行过程中的各种工况下的操作柔性。针对石化企业蒸汽动力系统变工况的特点,提出了蒸汽动力系统柔性优化设计策略;将确定性变化需求转化成多周期问题,将不确定性变化需求转化成虚拟的多工况问题,并对连续不确定性和离散不确定性变化分别处理;以经济性和可操作性的最优综合为目标,建立蒸汽动力系统柔性优化设计的大型混合整数线性(MILP)模型,并采用广义Benders分解算法进行求解。采用本文提出的柔性优化设计模型和求解方法,对某炼油厂蒸汽动力系统进行了柔性优化设计,并与传统设计方案和经济性最优的设计方案进行对比。与经济性最优的方案对比结果表明,该优化设计方案总费用增加不多,但实现了在各种变化工况下的安全、稳定、柔性操作;而与传统设计方案对比结果表明,不但提高了各种工况下的操作柔性,而且节省了大量的投资和运行费用,即柔性设计方案具备可操作性和经济性的综合最优性。从而证明了本文提出的柔性设计建模和求解方法的合理性和实用性。

Steam power system is an important part of petrochemical industry. Its design and synthesis is a complex and systematic task. Apart from economic objectives, the more important objective of design is operability under all demanding conditions. According to the varying features of the steam power system, a flexible and optimizing design strategy was proposed in this paper. The varying internal and external factors of steam and power were analyzed. The deterministic varying factor was transferred into multiperiod problem and the non-deterministic varying factor was transferred into virtual multi-scenario problem. A mixed integer linear programming （MILP） model was formulated for the flexible and optimizing design of steam power system to find the optimal project according to two objectives of economic efficiency and flexibility. A general Benders decomposition （GBD） algorithm was used to solve the MILP model. In a case study, the presented flexible and optimizing design strategy was applied in a real design project of steam power system for a refinery plant. The traditional design method and economics optimizing method were also applied to design the same case. Three different design schemes were presented and compared. The comparison result indicated that the flexible and optimizing scheme was the best one according to the integrated objective of economic efficiency and flexibility. So the feasibility and practicability of the flexible and optimizing design strategy and solving algorithm were proved.