| In the industrial process,low-grade waste heat discharged into the environment accounts for more than 50%of the total heat,resulting in significant energy waste.In response to China’s "dual carbon”strategy,which aims to reduce carbon emissions and increase energy efficiency,the recovery of low-grade waste heat is becoming increasingly urgent.Organic Rankine Cycle(ORC)is currently one of the most promising directions for low-temperature thermal power generation,applied to geothermal,solar energy,and industrial waste heat recovery.However,the low thermal efficiency of low-grade waste heat utilization,the limited number of ORC working fluids developed,and no working fluid that can match different kinds of waste heat,lead to long investment payback periods and low market share of ORC.Currently,ORC working fluid screening and parameter design still heavily rely on empirical experience,and there is a significant potential to improve the thermodynamic and economic performance of ORC systems and develop new high-performance environmentally friendly working fluids.It is necessary to identify key parameters that affect the performance of ORC systems,construct corresponding mathematical models and optimize strategies to achieve the optimal design of ORC systems.In this paper,a systematic approach is proposed based on graphical methods,mathematical programming,and machine learning methods for ORC working fluid screening and operational parameter optimization,and extended to the integration optimization design of heat exchange networks(HEN)and ORC with new working fluids,aiming to improve ORC power generation performance and economic benefits.The main contents of this research are as follows:1.To address the problem that temperature-enthalpy diagrams and temperatureentropy diagrams cannot reflect ORC power generation performance,a new exergy diagram method,which can represent the heat exchange process,work process and ORC,is proposed to predict ORC power generation without the need for equation of state and simulation software.The accuracy of power generation prediction using this method is less than 1.9%compared to power generation calculated using equation of state in the literature.Additionally,a system evolution procedure based on this exergy diagram is developed to identify the optimal working fluid and operating parameters for different waste heat conditions,achieving maximum power output.For case 1 with a waste heat of 157℃,using the system evolution procedure,the optimal working fluid is identified as perfluorobutane,and using the Peng-Robinson(PR)equation to simulate the ORC process with this working fluid and corresponding operating parameters,the power generation efficiency is increased by 22.0%compared to power generation using n-pentane in the literature.For case 2 with a waste heat of 240℃ in Fischer-Tropsch system,the system evolution procedure identifies the optimal working fluid as o-xylene,and the power generation efficiency using this working fluid is increased by 9.6%compared to power generation using n-hexane in the literature.2.Aiming at the difficulty of simultaneously optimizing ORC working fluids and operational parameters with economic objectives,a hybrid algorithm framework combining genetic algorithms and deterministic algorithms for ORC design and thermal integration optimization is proposed.The outer layer uses genetic algorithms to optimize ORC working fluid composition,temperature,and pressure,and the REFPROP software is used to calculate the thermodynamic properties of the working fluid.The inner layer uses deterministic algorithms to optimize the heat integration model.This study considers 43 pure substances and 36 binary mixtures as candidate working fluids,and discusses the optimal ORC configurations for waste heat temperatures ranging from 100℃ to 300℃.The results show that a subcritical ORC using binary mixtures has the best economic performance at different waste heat temperatures.For waste heat below 240℃,the mixture of n-pentane and toluene is recommended as the optimal working fluid for economic benefits,while the mixture of cyclopentane and heptane is suggested for waste heat above 240℃.3.To address the problem of large temperature differences in heat exchange processes and the lack of cascade utilization of high-temperature thermal energy caused by the sequential design of HEN and ORC systems,a simultaneous optimization model of HEN-ORC is proposed to embed ORC heat exchange processes into the original HEN,achieving reasonable heat integration and efficient power generation.The model uses two ORC structures:single-loop ORC and dual-loop ORC,and calculates ORC thermodynamic parameters by a polynomial regression model to replace highly nonlinear PR equation and reduce the optimization difficulty.The optimization model simultaneously determines the optimal operating parameters of ORC and the optimal heat transfer structure of HEN-ORC.Case studies show that the proposed simultaneous design model obtains a 24.4%increase in power generation compared to the sequential design,and reduces annual economic costs by 7.0%.4.To address the problem that the highly non-convex nonlinearity of the HENORC model makes it difficult to achieve global optimization,an adaptive partition linearization global optimization algorithm is proposed in this study.This algorithm identifies convex sub-equations in non-convex and nonlinear equations,greatly reducing the decomposition complexity of non-convex expressions.It applies adaptive multi-stage McCormick relaxation,combined with piecewise linearization and firstorder Taylor expansion relaxation,to develop the adaptive linearization global optimization algorithm for this MINLP problem.The algorithm determines initial values of variables by solving the initialization model,obtains the original model lower bound through linear relaxation,finds an upper bound by solving the nonlinear model with fixed binary variables,and iteratively improves the model lower bound using the adaptive partition linearization module.The proposed algorithm is tested on seven HEN and HEN-ORC cases,and the optimization speed is faster than mainstream MINLP global optimization solvers such as Baron,Couenne,and Lindoglobal.5.The current global warming potential(GWP)of working fluids is high,and there is an urgent need to develop new environmentally friendly working fluids.To address these issues,this paper constructs a database containing more than 2,000 hydrocarbon molecules and establishes a group contribution-artificial neural network(GC-ANN)model to predict the critical temperature and thermophysical properties of various hydrogen fluorocarbon(HFO)molecules with carbon chain lengths ranging from C2 to CIO.Furthermore,a new reverse design method from HEN-ORC to new working fluid molecules is proposed,by first analyzing the key thermophysical parameters that affect ORC power generation,including gas-phase heat capacity,liquid-phase heat capacity,and evaporation enthalpy,and proposing quantitative equations expressing the relationship between these key thermophysical parameters and power output.The optimal values of ORC thermophysical parameters and the optimal HEN-ORC structure are then obtained through the HEN-ORC optimization model.Next,the optimized thermophysical parameter values are matched with the predicted values of new working fluid molecules using the GC-ANN model,and the new HFO working fluid with the most suitable thermophysical parameters is identified,achieving the integrated design of the HEN-ORC system and the new molecule.The proposed design method is applied to two case studies,and two environmentally friendly new fluids((?)and(?))are selected as the new working fluids for the two cases,showing better economic efficiency and power efficiency.The total annual cost is reduced by 12.3%,and the power generation efficiency is increased by 4.5%. |
PDF Full Text Request