DYNAMIC SIMULATION OF A NON-ISOTHERMAL MSMPR CRYSTALLIZER

The effect of including temperature variations in the dynamic simulation of an MSMPR cooling crystallizer is investigated. Power law nucleation and growth kinetic models incorporating an Arrhenius temperature dependence are used. The effects of changes in the input variables feed temperature, feed flowrate and feed concentration on the output variables operating temperature and concentration and moments of the size distribution are investigated and compared with corresponding responses obtained under isothermal conditions. Significant differences are found.     

A NOTE ON THE MINIMUM NUMBER OF UNITS FOR HEAT EXCHANGER NETWORK SYNTHESIS

Although graph theory shows that the minimum number of exchangers (U_min) in a network is usually one less than the number of streams (N - 1), examples have been published, which demonstrate that this limiting case cannot always be achieved. Some of these examples contain a pinch point, for which it is advocated that the ‘N - 1 target’ should be applied on both sides of the pinch.

However, by using a novel arrangement of stream splitting, mixing and by-passes, some of the literature examples can be designed to conform to the N - 1 target.

For networks with similar total areas, those having fewer units will usually be cheaper, so U_min networks often have lower capital costs. Although the networks discussed here have somewhat larger total areas, possible applications are discussed where they may be economically attractive.     

换热网络的同步综合设计

提出了换热网络的一种新的转运模型,以并串模式和非等温混合过程描述换热网络的热力学和经济性的总要求,不使用挟点分离、最小单元数等启发式规则,并允许交叉换热和不同物流间的膜传热系数与温差合理匹配.这一模型与优化模型相结合,自动产生换热网络,确定相应的最优公用工程费用、换热面积、换热器台数与冷热流间的匹配,其可行域由一集线性约束确定,鲁棒性很好,易于求解.该方法克服了换热网络的各种分步骤综合方法的缺陷(包括挟点技术、双温差法及相关数学规划方法).与文献上其它同步优化方法比较,具有模型简单、规模小、可算性强、初始化简易、计算结果可靠的优点.通过对文献中广泛使用的例题的检验,结果优于现有的其它方法.     

SYNTHESIS OF HEAT EXCHANGER NETWORKS WITH DESIGNER IMPOSED CONSTRAINTS

A new sequential, evolutionary design method is proposed for the synthesis of heat exchanger networks with designer-imposed constraints. The design parameters for generating the network are obtained with conventional algorithms by solving a modified trans-shipment formulation of the transportation linear program. Candidate network topologies can then be easily configured by hand computation and the “best” designs are quickly identified even for large problems. The procedure also identifies problems which may have multiple feasible near optimum topolgoies, characterized by minimum energy consumption as well as minimum number of units.