Thermo-economic Parametric Optimization And Running Control Strategy Of Cascade Orc Cooling System For Low-grade Waste Heat

Plenty of water from chemical industry needs to be cooled down to aim temperature because of the pollution which will lead to healthy and biology problems.Conventional cooling methods taken in industry usually consume a lot of electricity,and lowgrade heat is also wasted.Additionally,many applications in the manufacturing industry reject heat at relatively low temperature except Organic Rankine Cycle,ORC,which has already studied and applied maturely.However,the initial investment of ORC system is really expensive,and it will also consume power when the aim temperature is low enough.In order to balance thermal and economic performance of cooling system,ORC cooling unit and conventional cooling unit are naturally cascaded.This proposed cooling system is named cascade ORC cooling system,where low-grade waste water is firstly cooled down to a middle temperature by ORC cooling unit,and then the middle-temperature waste water is cooled down to aim temperature by conventional cooling unit,and the heat absorbed by organic working fluid is used to generate power.What’s more,in order to improve the performance of the cascade ORC cooling system,two-stage ORC cooling unit including two stage-temperature evaporators is proposed.Thermal model and exchanger model are built to study thermal performance and economic performance of cascade ORC cooling system,respectively,based on first law and second law of thermodynamics.The results show that there exist optimal evaporation temperatures(OETs)to generate maximum net output power or minimum APR,respectively,and the theoretical formula to calculate maximum net output power of two-stage ORC cooling unit is derived.Then,particle swarm optimization(PSO)algorithm is used to optimize OET under different waste heat inlet temperatures and different condensation temperatures.The optimization result shows that the higher the heat source inlet temperature is,the higher the maximum net output power is,and the higher the OET is.The result of polynomial fitting shows that the OET of SCORC cooling system and heat source temperature are cubic,and the high-temperature OET of TCORC cooling system and heat source temperature are also cubic,and the low temperature OET and high-temperature OET of TCORC cooling system have a linear relationship.Additionally,net power generated by TCORC cooling system is more than twice as that of SCORC cooling system.Meanwhile,the economic performance of TCORC cooling system is also better.Therefore,the higher temperature of heat source is,it is more valuable to establish TCORC cooling system.According to previous research,offline steady-state optimization controlling strategy in this paper is adopted,and in order to reduce the optimization time of algorithm,machine learning approach,artificial neural network(ANN),is taken to find the OET generating maximum output net power,and LSTM neural network is taken to forecast the temperature and mass flow of low-grade waste heat 1 hour ahead.Three ANN configurations are used and compared,including BP neural network,cascade BP neural network,and RBF neural network.The result shows that cascade BP neural network shows higher forecasting accuracy when the training samples are more random.And for perfect data set,BP neural network is already good enough.And then,LSTM neural network has high accuracy in predicting the temperature of the heat source,but due to the large fluctuations in the mass flow rate of the heat source,the prediction accuracy is relatively inferior.At last,PID controller is taken to display controlling examples under different inlet heat source temperatures,different condensation temperatures and different pinch point temperature difference.And the response curves show that both SCORC cooling system and TCORC cooling system reached stability within 10 s,which means that the control strategy works well.