Artificial intelligence approach for the prediction of heat transfer coefficient in boiling over tube bundles

Abstract

The determination of heat transfer coefficient plays an important role in optimal designing of heat transfer equipments as it directly affects the heat surface area and thereby the weight and cost of the equipment. Thus, prediction of heat transfer coefficient with minimum error reduces the exhaustive experimental work. Therefore, the prime objective of the present work is the application of computational intelligence methods for improving the prediction accuracy of heat transfer coefficient in flow boiling over tube bundles. The adaptive neuro-fuzzy inference system (ANFIS) and artificial neural network (ANN) are applied to predict the flow boiling heat transfer coefficient as output, taking pressure, pitch of the bundle, heat flux, mass flux and vapour quality as input. The performance of different derivatives of neural network such as multilayer perceptron, general regression neural network and radial basis network (RBF) is studied with varying the parameter. The ANFIS is tested with different types and number of membership functions for the prediction of flow boiling heat transfer coefficient. These methods are found to be better for predicting the flow boiling heat transfer coefficient than conventional correlations.

A RBF model for predicting the pool boiling behavior of nanofluids over a horizontal rod heater

“…However, due to complexity of pool boiling process and different influencing factors (including heat flux, pressure, thermo-physical characteristics of liquid, and heater surface conditions), most of these correlations even predict their own test-data within an error of about ±20%. Therefore, for more accurate prediction, utilizing the computational intelligence methods such as artificial neural network (ANN) can be a suitable solution [19]. Lie et al [20] used ANN method to develop a numerical model for predicting the boiling behavior of 30 additives.…”

Soft-ANN based correlation for air-water two-phase flow pressure drop estimation in a vertical mini-channel

“…20 The ANNs are a computational tool to model highly non-linear problems, even without previous knowledge of the physical behavior of the analyzed phenomena. In air-water two-phase flow, ANNs have mainly been applied for prediction of convective heat transfer coefficient, 21 cavitation, 22 and pressure drop, for which the following studies can be mentioned: Al-Naser et al, 23 who proposed a very particular ANN since their inputs were: Water Reynolds Number, Air Reynolds Number, and Pressure Drop Multiplier. Alizadehdakhel et al 24 designed and trained an ANN to predict the total pressure drop of air-water two-phase systems.…”

RCA-PCK: A new structural reliability analysis method based on PC-Kriging and radial centralized adaptive sampling strategy

“…Since the performance function is usually implicit and strongly nonlinear in engineering practices, classical analysis techniques such as first-order reliability methods (FORM), second-order reliability methods (SORM), 1 Monte Carlo simulation (MCS) 2 and variance reduction techniques (IS, SS, LS) 3–5 are generally unacceptable in accuracy and efficiency. To overcome these shortcomings, applying the surrogate models to approximate the true performance function has been rapidly developed in recent decades, such as the response surface method (RSM), 6,7 artificial neural network (ANN), 8,9 support vector machine (SVM), 10,11 Kriging model, 12,13 etc. Among them, the Kriging model, 14,15 as an accurate interpolation method, can not only provide the predicted value of the unsampled sample, but also estimate the prediction variance.…”