Abstract: Heat exchanger network synthesis is a typical mixed integer nonlinear programming problem with non-convex and nonlinear characteristics, whose optimization easily falls into local optimal. To deal with this difficulty, this paper proposes a differential evolution algorithm with competition mechanism, which is further applied in simultaneous synthesis of heat exchanger network. Firstly, Latin hypercube sampling method is used to generate initial individuals such that the diversity in the space of solutions can be ensured. And then, the competition mechanism is introduced to divide the evolutionary population into two groups, named as winner and loser. For the winner group, the opposition-based random search combined with greedy selection is conducted to achieve deeper optimization. For the loser group, it will learn from winner group to improve the quality. Moreover, an adaptive shrinking factor is introduced into the mutation operation of individuals for improving the global and local optimization. Comparing with other methods, the proposed method in this work can attain a better design of heat exchanger network via lower total annual cost, and can efficiently solve the synthesis problems of medium scale heat exchanger network.