A new hybrid genetic algorithm for optimal design of heat exchanger networks is developed. The mathematical model used in the algorithm is based on an explicit solution of stream temperatures of heat exchanger networks with the stage-wise superstructure. By taking heat transfer areas and heat capacity flow rates as genes in the genetic algorithm, the thermal performance and total cost of arbitrary heat exchanger networks can be calculated explicitly and therefore the individuals (heat exchanger networks) can always be feasible. This characteristic makes the algorithm suitable for large scale synthesis problems. The genetic algorithm is combined with simulated annealing algorithm, local optimizing strategy, structure control strategy and other strategies so that the structural search ability of the algorithm is conspicuously enhanced. The examples presented in this work revealed special search abilities in both structure space and continuous variable space. (C) 2009 Elsevier Ltd. All rights reserved.