Prediction of Heat Transfer Coefficient During Condensation of R404a in Helically Coiled Tubes Using Adaptive Neuro-Fuzzy Inference System

Ali Khosravi, Juan Garcia, Tulio Correa, Luiz Machado

Research output: Chapter in Book/Report/Conference proceedingConference contributionProfessional

Abstract

Condenser is an important heat exchanger widely used in refrigeration and air conditioning systems. Heat transfer coefficient (HTC) of two-phase flow of refrigerants is important in order to design a heat exchanger. In this study, Adaptive Neuro-Fuzzy Inference System (ANFIS) is developed to predict the HTC under condensation of R404a in a helically coiled tube. Particle Swarm Optimization (PSO) algorithm and Genetic Algorithm (GA) are used to optimize the ANFIS model. Vapor quality, mass flux of refrigerant, pitch and curvature radius are considered as input variables of the network and HTC is selected as its output variable. The results illustrate that using GA to optimize the ANFIS model improves correlation coefficient with approximately 8% and decreases the value of RMSE around 52% for testing datasets. Also, the ANFIS model optimized with PSO algorithm reports the best performance for the predicted HTC with correlation coefficient and MAPE respectively as 0.9876 and 1.35% for the testing datasets.
Original languageEnglish
Title of host publication17th Brazilian Congress of Thermal Sciences and Engineering
Subtitle of host publicationNovember 25th-28th, 2018, Águas de Lindóia, SP, Brazil
Publication statusPublished - 2018
MoE publication typeD3 Professional conference proceedings
EventBrazilian Congress of Thermal Sciences and Engineering - Aguas de Lindoia, Brazil
Duration: 25 Nov 201828 Nov 2018
Conference number: 17

Conference

ConferenceBrazilian Congress of Thermal Sciences and Engineering
Abbreviated titleENCIT
CountryBrazil
CityAguas de Lindoia
Period25/11/201828/11/2018

Fingerprint Dive into the research topics of 'Prediction of Heat Transfer Coefficient During Condensation of R404a in Helically Coiled Tubes Using Adaptive Neuro-Fuzzy Inference System'. Together they form a unique fingerprint.

Cite this