چكيده به لاتين
Application of parallel flow condensers (PFC’s) due to their advantages against conventional heat exchangers such as compactness, higher performance, lower weight and reliability has been grown recently in automotive air conditioning (AAC) system. The refrigerant R134a as the most common refrigerant in AAC systems should also be replaced due to its high global warming potential (GWP) according to environmental protocols. Among the proposed refrigerants, R1234yf refrigerant has been gotten more attention because of thermodynamic and operational characteristics similar to R134a. In other hand, because of presence of oil in compressor and circulating throughout the system with refrigerant, investigating the effects of oil on performance of different components such as condenser is crucial. In the present work, the effects of R1234yf-oil mixture on condenser performance are studied experimentally and numerically and the results are compared with those of R134a-oil mixture. To do so, the condenser performance has been simulated using refrigerant at first step and then different proposed models have been expanded to the case of refrigerant-oil mixture.
The condenser performance has been simulated using one dimensional finite element approach which shows a good agreement with experimental results. Then, the finite element method has been coupled with computational fluid dynamics and a new approach for modelling the condenser performance with the ability of predicting the refrigerant mal-distribution in header and also the effects of header geometry on refrigerant mal-distribution was presented. The proposed method causes the modification of the accuracy of the anticipated results in comparison with experimental results. Also, the new geometry of header based on decrement of flat tube protrusion depth, placing the header inlet pipe far from the flat tubes and increasing the header diameter was proposed. Due to importance of foretelling refrigerant pressure drop and also not presence of correlation with high reliability, a new correlation with the ability of considering the effect of inlet refrigerant superheat has been proposed for anticipating the condenser pressure drop which validated against experimental results. The model is then used in a multi-objective optimization procedure to see if it is possible to modify the condenser performance without changing its dimensions and Pareto fronts for one, two, three and four objective optimizations were presented. The feed forward neural network with back propagation learning algorithm has been optimized using genetic algorithm to predict the condenser performance. Then the results of applying the mentioned methods were compared with the results of recurrent neural network and finally this method was proposed for heat exchangers modeling with high accuracy. The unsteady performance of automotive refrigeration system in compressor on-off cycles was investigated due to its importance and a relation for predicting the unsteady condenser performance was presented. Also, using a control strategy for compressor start-up was proposed to decrease the energy consumption. Finally, the condenser performance using refrigerant-oil mixture was studied experimentally and the above modeling methods were expanded to this situation. The condenser pressure drop increment, capacity decrement and not suitable solving of R134a oil in R1234yf at high oil circulation ratios are some of the experimental results of adding oil to refrigeration system. It was found that, the finite element approach has not suitable ability of anticipating the condenser pressure drop using refrigerant-oil mixture (67.5% error) and therefore a new correlation was developed. The recurrent neural network was applied for the condenser performance prediction and was proposed as the most suitable and available method. The unsteady performance of the system and exclusively condenser was also probed in case of circulating refrigerant-oil mixture.
