The major difficulty in the heat exchanger network synthesis (HENS) is dealing with the simultaneous optimization of large-scale continuous and integer variables. Heuristic algorithms are used in HENS due to their efficient global search ability and the step size (Delta L) constitutes one of the critical concepts in it. In this paper, by analyzing the influence of Delta L in Random Walk Algorithm with Compulsive Evolution (RWCE), it was pronounced that evolution speed got faster when Delta L increased in the early stage and evolution accuracy was higher by Delta L declination, in the late stage. Hence, five new different Delta L adjustment functions were proposed. This case-study concluded that Delta L adjustment functions in an upward parabola declination could maintain high speed in the early stage and improve the accuracy of solutions in late stage respectively. However, during the late stage, the minor Delta L led to the decline of the global search ability and it was difficult to jump out of the local minimum. Furthermore certain individuals in the population were randomly given a relatively large Delta L in late stage. Thus, an integrated RWCE algorithm with Delta L adjustment strategy was presented and demonstrated satisfying global and local search ability. (C) 2017 Elsevier Ltd. All rights reserved.