Neuro-genetic optimization of micro compact heat exchanger

Purpose - This paper seeks to introduce an optimization method for maximizing the effectiveness of the micro compact heat exchanger (AM) under various geometrical parameters. Design/methodology/approach - Optimization is realized using the neuro-genetic methodology which combines the application of artificial neural network (ANN) together with genetic algorithms (GA). The analyses are divided into two main sections; the first being the modeling and prediction using finite element method, the second being the neuro-genetic optimization. Initial results obtained from the finite element modeling are utilized for training in ANN. Subsequently, optimization is done using GA, once a well trained ANN is achieved. Findings - ANN accurately predicts the effectiveness of the MHE and compares well with those obtained from the finite element simulation. Optimization shows a significant improvement in the maximum effectiveness of the MHE achievable for the given range of input parameters. Additionally, computational effort has been minimized and simulation time has been drastically reduced. Research limitations/implications - This analysis is valid for constant fluid properties and for steady-state conditions. Additionally, optimization is limited to the range of the trained input parameters. Practical implications - This paper is very useful for practical design of various types of heat exchangers. Originality/value - This paper will be useful for the design of the MHE where its performance can be analyzed for a given range of geometries with minimal effort. This methodology will also be applicable for other types of heat exchangers.