Evaluation of performance of an indirect evaporative cooler (IEC) involves solving complex differential and analytical equations. Artificial Neural Networks (ANN) approach provides a simple but powerful tool for predicting the performance of IEC. This paper presents both analytical approach as well as ANN approach in predicting the performance of an IEC. ANN is trained with analytical data using back-propagation learning algorithm with 13 different training algorithms. The logistic sigmoidal function is taken as transfer function. The ANN model is then compared and validated using experimental data from the literature. It was found that the most efficient and most accurate training algorithms were Levenberg-Marquardt (LM) and Bayesian Regularization (BR) back-propagation respectively. After satisfactory training of both the models, the statistical values i.e. R2, RMS, cov, MSE and AIC for the prediction of primary air outlet temperature, ( ) were 0.9999, 0.1786, 1.00, 0.0319 and -3.43 & 0.9999, 0.0546, 0.31, 0.0030 and -5.79 respectively. Similarly, for the prediction of effectiveness ( ) of IEC using the above two models the statistical values were found to be 0.9999, 0.0020, 0.33, 3.8138E-06 and - 12.46 & 0.9999, 0.0004, 0.08, 1.9827E-07 and -15.42 respectively. This tool is highly useful for designers to know apriori the performance characteristics of IEC under a given set of environmental conditions without undergoing complicated analysis of the system. This model can also be very useful for designers to predict the energy savings by an IEC.
