Thermodynamic analysis of absorption refrigerator system with ejektor performing by alternative refrigerator fluids using artifical neural networks

THERMODYNAMIC ANALYSIS OF ABSORPTION REFRIGERATOR SYSTEM WITH EJECTOR PERFORMING BY ALTERNATIVE REFRIGERATOR FLUIDS USING ARTIFICAL NEURAL NETWORKS (Ph.D. Thesis) MEHMET ÖZALP GAZI UNIVERSITY INSTITUTE OF SCIENCE AND TECHNOLOGY October 2004 ABSTRACT This study specify to three base functions different from literature for absorption refrigeration system: (i) In the system, refrigerant/absorbent fluid couples (methanol/LiBr, methanol/LiCI) that don't damage ozone layer are used, (ii) An ejector in this study is located in front of absorber while in the literature it is commonly located in front of condenser, (iii) Because of containing complex differential equations for the thermophysic characteristics of fluid couples used, the thermodynamic simulation of the system is difficult. For easing this state, Artificial Neural Network (ANN) was used in the determination of thermophysic characteristics. According to the results obtained, in lower generator temperatures the coefficient of performance (COP) and exergy performance (ECOP) of the system have increased in the rate of 49% and 49%, respectively and decreasing circulation ratio (f) in the rate of 57%, the dimensions and the cost of the system was reduced. The using of ANN approach in determining of thermophysic characteristics of the fluid couples is in the acceptable error values (R2 = 0,9998). Science Code : 500.10.15 Key Words : Absorption refrigeration systems, methanol/LiBr, methanol/LiCI, artificial neural networks, ejector Page Number : 173 Adviser : Assistance Professor. Adnan SÖZEN