Simultaneous Optimization Of Heat-integrated Water Networks

A large amount of water and energy are consumed in industry. With industrialization developing quickly, water shortage has become one of the most serious problems. Energy is the lifeline of the national economy. The energy demand in China has increased year by year, and China is the second largest energy consumption in the world. Recently, the growing industrial water and energy demands are causing enormous stress on the hydrologic cycle and the atmospheric environment. It is crucial to find a way to reduce water and energy consumptions in chemical industry by heat exchanger network synthesis(HENS), water networks synthesis(WNS) and heatintegrated water networks synthesis(HIWNS). Much work has been done on WNS, while the way to appraise water networks design result has not been highly regarded. The mixed integer nonlinear programming(MINLP) model is generally used for SHENS and HIWNS; however, a good initial point is crucial for the convergence.At first, a new method, named inflection point method(IPM), is proposed for designing single-contaminant WNS. On the basis of mass balance, a few equations are developed to calculate the targets of water networks design, including the minimum freshwater flowrate(FW), the minimum regenerated water flowrate(FR) under the condition of the minimal FW, and the relationship between the freshwater flowrate(FW) and the minimum regenerated water flowrate(FR) accordingly. Furthermore, the operation cost of water networks could be calculated by the relationship between FW and FR, and then the trade-off between cost and freshwater consumption could be determined prior to the networks design. The targets of water networks design determined by IPM are global optimum solutions. Therefore, it could be used to appraise water networks design result obtained via other design methods, for instance water pinch analysis and mathematical programming method.Then, a nonlinear programming(NLP) model is developed for HENS. The MINLP model is generally used for SHENS. Based on MINLP model in the work of Yee and Grossmann(Comput. Chem. Eng. 1990, 14, 1165-1184), we develop a NLP model for SHENS by removing binary variables of the MINLP model. A set of new nonlinear constraints are formulated for removing the binary variables. Approach temperatures for determining the area requirement are incorporated in the objective function; therefore, the temperature approach variables are not necessary. The computational complexity is reduced by decreasing the number of variables. Branch-And-Reduce Optimization Navigator(BARON) solver is used for solving the NLP model with no attempt to obtain a good initial point, which is assumed as the default value(the lower bounds) by GAMS. Twelve examples on different scales containing 4-39 streams are investigated to analyze the performance of the NLP model. The results show that not only the small HEN problems but also the large HEN problems could be solved precisely by the NLP model.At last, on the basis of NLP model for HENS, a NLP model is developed for HIWNS. In MINLP model, binary variables are used for identifying the type of streams such as hot streams and cold streams. A set of new nonlinear constraints is formulated to remove the binary variables. Based on the NLP model for HENS, the number of variables for HIWNS is further reduced. The computational complexity is reduced by decreasing the number of variables. Using BARON solver the NLP model is solved without good initial point strategies. The model is tested on single and multiple contaminant problems. The results show that HIWNS problems could be solved precisely by the NLP model.