【作者】 李保红；

【导师】 姚平经；

【作者基本信息】 大连理工大学 ， 化学工程， 2001， 博士

【摘要】 淡水资源短缺已成为世界关注的焦点。21世纪人类将为水而战的观点也为愈来愈多的人所认同。面对如此严峻的形势，如何在不影响人们生活水平提高的前提下，最大限度的降低淡水的使用量是人们迫切需要解决的问题。对于过程工业来讲，需要做两方面的工作来降低新鲜水的使用量，一方面通过不断的改进工艺，降低过程本身对水的需求，如水冷却器改为空气冷却器、水洗改为干洗等；另一方面就是要通过水的再用、或循环再用、或再生再用以及再生循环等手段来最大限度地提高水的重复利用率。显然，这两个方面的工作对于节约用水都很重要，但在实际工作中，往往优先做好第一方面的工作。在这种背景下，本论文从技术的角度，对过程工业中水分配网络的设计与改造方法进行了研究。在论文的第一部分，研究的是设计型问题。首先将水分配网络分解为用水网络与废水处理网络，分别进行研究。对于用水网络，在补充与系统化网络最优性必要条件的基础上，对基于多杂质最优性必要条件的设计算法进行了大胆尝试。对于废水处理网络，根据对问题本质的理解，提出了一种新的设计方法，算例验证表明，新方法能够克服现有方法的不足，容易地找到更好的解。然后，对水分配网络的同时设计方法进行了研究，提出了新的问题分解策略。在论文的第二部分，研究的是改造型问题。将质量集成的思想应用到水分配网络改造问题的求解上，提出了一种复合式改造方法。上述内容的重要思想有以下几点： 1．针对废水处理网络，提出了一种新的设计方法。这种方法的主要思想是，首先通过保证每一次移去杂质时的单位费用是最低的，得到没有流股混合的初始分布式网络结构；然后依据四个规则对其作进一步的调优。当网络设计的目标是最小化总处理量时，调优的目的是减少处理过程的数目；当网络设计的目标是最小化总费用时，也就是说，可用处理过程的费用函数（形式为AQ~a+BQ，解释见第2点中的规则2）给定，调优的目标不仅是减少处理过程的数目，而且还要使总费用进一步降低。对于多杂质系统，提出先仅考虑一种杂质设计出其没有流股混合的初始分布式网络，接着以流股为单位通过合并得到考虑了多个杂质的没有流股混合的初始分布式网络，最后再依据规则进行结构调优的思想。 2．提出并证明了用于指导分布式废水处理网络设计与调优的四个规则。第一个规则指出，若两个废水流股分别接受相同处理过程的处理之后， 大连理工大学博土学位论文必须进行混合，则将这两个流股混合后再接受同一处理过程的处理，其效果是相同的；第二个规则指出，如果处理过程的费用可以表述成叫‘＋So的形式（式中 A。B为常数，Q是被处理的流量，AQ‘表示设各费，BQ表示操作费），则应根据指数a取值范围的不同来决定是采用分布式处理还是集中式处理：第三个规则给出了有助于人们从宏观上把握问题的重要概念一处理级数；第四个规则指出，如果某一废水流股依次接受两个不同的处理过程的处理且各处理过程的性能指标均为杂质移出率，则这两种处理过程的先后次序不影响处理效果。 3．由于人们认识到用水网络与废水处理网络之间存在相互作用，便试图对它们进行同时设计。目前做法是，先建立包含各种因素的超结构，然后采用数学优化的方法进行求解。但由于问题的复杂性以及MINLP性质，这种方法不能保证所得解的最优性。针对这种情况，我们提出了一种基于概念的设计方法，这种方法先确定目标新鲜水用量，而后将水分配网络分解为用水网络、废水处理网络和再生循环系统三个部分，再对这三个部分分别进行设计 （其中废水处理网络采用本论文提出的新方法），最后对网络结构作进一步的调优或松弛。文献算例验证表明，新的求解策略易于实现，并且十分有效。 4．借鉴质量集成的思想，针对单杂质水分配网络改造问题．提出了一种复合式改造方法。开发这种方法的主要目的是，避开对工艺过程直接采用复杂的MINLP方法，而是依据定性知识使需要优化求解的问题规模降至最小，并且将MINLP问题简化为NLP问题，使人们通过简单的分析、计算迅速地找到近似最优解。该方法在着眼于全局依据定性知识确定废水最小化的目标以及合适的循环结构之后，利用路径图（path diagr。）确定过程废水排放的适宜位置，再基于定性分析确定污染物组成调整的方案，最后对己定的流程结构方案用GAMS软件包进行数学优化，确定出各参数的最优值。运用所提方法对一文献算例进行了求解，得到了更优的结果。更多还原

【Abstract】 The shortage of freshwater causes more and more people’s attention. The opinionthat the war occurred in the tWenty-first century wi1l mostly caused by freshwater resourceis accepted by most peop1e. Faced with such situation, it is an urgent taSk to minimize ourusage of freshwater without blocking the improvement of our living standard. Por processindustry it is required to minimize the requirement for freshwater by carrying on tWo kindsof work. The first one is to improve the production method to reduce water usage, forexampIe, to install air cooler instead of water cooler or change the rinse process to dry-clean process. The second one is to increase the reuse ratio by water reuse or recycle reuseor regenerate reuse or regener8te recycle. Obviously, these tWo options both are importantto reduce the water usage. ln reality the first one has priority. The paper will fOcus on thedesign and retrofit of the water allocation network in process industry for wastewaterminimization. At the first part of this paPer, we study the design method and firstdecompose the water allocation network into two parts, one is water-using network, theother is wastewater treatment netWork. We study the design method of each pall separateIy.As for the water-using netWork, we systemize the optimality conditions of water utilianionsystems in process plant with single contaminant or multiple contaminants and make somecomplements. AdditionalIy, we attempt to develop a design algorithm based on theoptimality conditions of water utilitalion systems with multiple contaminants. As forwastewater treatment netWork, we develop a new design method based on ourunderstanding to the problem’s nature. The new design method can overcome the defectsof present methods and find better solution easily as far as the same problem is concerned.Then the design method for water allocation network is studied, a new solution strategy isproposed. At the second part of this paPer, we deal with the retrofit of water allocationnetwork. The mass integration strategy is applied to the water allocation network retrofitand a hybrid method is developed. The key ideas in this paper are extracted and listedbelowt1. A new method for wastewater treatment network design is put forward. In themethod, a primary distributed network without mixed streams is first obtained byguaranteeing that every unit contaminant mass is removed at the lowest cost wheneach step is taken during the design procedure; then, we evolve the primarynetwork according to four ruIes. When the design objective is to minimize the totaItreated flowrate, those evolutions can reduce the number of treatment processes,. when the design objective is to minimize the total costs, that is, the costfunction(such as AQ’+BQ) of each available treatment process is specified, thoseevolutions can not only reduce the number of treatment processes, but also reducethe total costs. As for multiple contaminants system, the primary distributedsubnetwork considering only one contaminant is designed first. Then, the primarydistributed network considering all contarninants is obtained by merging thosesubnetworks based on each stream. Finally, we evoIve the distributed networkaccording to rules.2. Four rules are proposed and proved to guide the design and evoIution ofwastewater treatment network. The first ruIe is, if two effiuent streams are treatedseparately by the same type of treatment process and have to be mixed immediateIyafter the treatment, the same effect after the treatment can be got by treating themixed effiuent stream. The second rule points oUt, if the costS of a treatmentprocess can be stated as AQ’+BQ(in which A and B are constantS, Q is theflowrate treated by the process, AQ’ represents the capitaI costS and BQ representsthe operating costs), then we shouId decide whether to adopt distributed treatmentor centralized treatment according to the interval of index a. The third ruIe更多还原