【作者】 刘永健；

【导师】 袁希钢；

【作者基本信息】 天津大学 ， 化学工程， 2007， 博士

【摘要】 随着水资源和能源的日趋紧缺以及社会对环境保护的日益重视,以提高水资源和能源利用效率及废水最小化排放为目标的化工过程水分配网络和热交换网络合成问题已成为国内外过程系统工程领域研究的热点。然而由于过程系统涉及多种操作单元、过程不确定性以及可能存在多种热交换方式等因素,目前针对连续过程、间歇过程水分配网络合成以及考虑热集成的水分配网络合成的研究还不十分完善。针对存在的问题,本文进行了以下的研究工作:1、通过引入改进的虚拟操作单元,将带有水的损失或水的生成的固定流率用水单元用两个虚拟的操作单元等价代替,建立了用于可同时处理固定流率与固定污染物负荷两类用水操作的连续过程用水网络合成问题的目标设定方法,该方法能够较迅速并准确地确定用水网络问题的最小新鲜水用量及其夹点位置,包括可能存在的多夹点问题。2、基于浓度间隔分析技术,通过引入区域分解和水源分配的策略,建立了用水网络设计的新方法。该方法避免了网络设计因分流、混合、回路识别及断开等比较繁琐的网络设计过程。此外,作为一种启发式的设计方法,该方法还能有效解决数学规划法等求解困难的问题以及黑箱模型问题难于理解的局限。3、基于用于连续过程用水网络的浓度间隔分析技术,通过引入时间间隔,建立了针对单批生产与多批重复生产的间歇过程用水网络合成的系统化方法——时间、浓度间隔分析技术。该方法能够有效地应用于上述用水网络合成问题,且更为快捷和方便。4、基于求解不确定性问题的对称模糊规划方法,提出了针对单组分负荷不确定用水网络合成问题的改进的对称模糊规划模型和用水网络设计的多目标混合整数对称模糊规划模型,同时提出了求解这一多目标问题的主目标设定法策略。算例的求解结果表明,所提模型以及求解策略对负荷不确定的工业用水网络的合成问题有良好的适用性,从而获得优化流程设计。5、针对带有等式约束的非线性规划(NLP)与混合整数非线性规划(MINLP)问题,建立了一种改进的粒子群优化(PSO)算法——简约空间粒子群算法(R-PSO)。应用多个例子对该算法的有效性进行了验证。6、基于分散式废水处理网络的超结构,建立了废水处理网络系统最优化的非线性规划(NLP)模型和混合整数非线性规划(MINLP)模型,并采用新型随机型算法—简约空间粒子群优化(R-PSO)算法求解上述复杂非线性最优化问题。7、通过考虑用水网络特有的直接换热,建立了用水网络与换热网络同时优化问题的“净热量”集成方法,并且结合简约空间的粒子群优化算法,建立了用于考虑能量集成的用水网络合成的演化计算方法。该方法能够充分考虑用水网络系统中存在的直接热交换,进而实现用水系统的最优能量集成并获得较好的热交换网络(所需换热面积较少);同时,作为一种基于演化的优化方法,该方法既能应用于单组分用水系统,也能适应于复杂的非线性多组分系统,且能够实现全系统(用水网络以及热交换网络)的费用最小化。更多还原

【Abstract】 With the increasingly scarcity of water and energy resource and ever-tightening environmental regulations on industrial effluent, the research on synthesis and optimization of water network and heat exchanger network for improving the utilization efficiency of water and energy resource has been becoming more focused, and motivated than ever before. However, such as multiple category operations, uncertainty, multiple heat recoveries model etc. for the synthesis problems are so complex that the currently available methods demonstrate still some major limitations. With the consideration of the above issues, we have performed the following research.1. A strategy for fresh water minimization is developed for continuous water utilization systems with both fixed flowrate (non-mass-transfer-based) and fixed contaiminent (mass-ransfer -based) operations.Through introducing the concept of improved ficitious operation units, the problem of water losing and/or generating for the fixed flowrate operations in continuous water utilization system can be handled easily without tedious graphical or iterative procedure, the pinch point can also be located simultaneously.2. Based on the knowledge of concentration interval analysis technique, a heuristic-based approach is presented to generate water utilization networks.The proposed method has the advantage of capability of considering fixed flowrate operations, the necessity of identifying and breaking loops and the limitation of flow distributing or mixing can be avoided.3. Based on the improved concentration interval analysis technique for continuous process water using system, a time-dependent concentration interval analysis (CIA) method to solve the problems of synthesis of single and repeated batch water-using systems is presented. The proposed method can effectively minimize the freshwater consumption for single or repeated batch operations and can determine a preferable storage tank capacity.4. An improved symmetric fuzzy linear programming (SFLP)model and a multi-objective mixed-integer SFLP model are developed for fresh water minimization of single-component water utilization networks with mass load uncertainty. A main criterion limiting (MCL) solution strategy is further developed for the multi-objective problem. The solution results of the example problems demonstrate that the proposed modeling and solution procedures have a good suitability, and then provide an optimum design for the networks.5. Aiming at the nonlinear programming problem (NLP) and the mixed-integer nonlinear programming problem (MINLP) with equality constraints, an improve particle swarm optimization (PSO) algorithm, i.e. reduced space optimization(R-PSO) algorithm is developed. The advantage of proposed algorithm in applicability and efficiency is demonstrated by several test problems.6. Based on a superstructure representation, a nonlinear programming (NLP) model as well as a mixed-integer nonlinear programming (MINLP) model was proposed for the optimal design of distributed wastewater treatment network system. Furthermore, to overcome the solving difficulty of the corresponding non-convexity problem, a newly proposed random search based optimization method– particle swarm optimization (PSO) is introduced and applied to develop a global optimization method for the wastewater treatment network design.7. By considering of the direct heat recovery that special existing in water using network, a net-heat integration approach for the simultaneous heat and water network integration. And combination with R-PSO algorithms, proposed an evolutionary approach to solve the simultaneous heat and water network optimization problem. By fully taking into account the possibility of the heat recoverie, the proposed approach can realize the optimum energy consumption target and heat exchange network in water using system. Furthermore, as an evolutionary-based approach, the proposed approach can deal with both single and multi-contaminant water-using systems, and can realize the economic trade-off between water network and heat exchanger network.更多还原