The residual oil vapor in storage tanks must be replaced before the process of maintenance, cleaning and renovation. In engineering practice, the replacement operation of oil vapor is always implemented empirically, which is inefficient and more prone to accidents. To ensure the safety and efficiency, we employ the numerical method to investigate the replacement process of oil vapor considering coupled factors, and the simulation results are applicable to horizontal tanks with different structures and different capacities. Furthermore, to meet the requirement of real time calculation in engineering, a large amount of numerical simulation results are taken as samples, and trained by neural network, thus the maximum concentration of oil vapor can be predicted quickly in any horizontal tanks. Moreover, to apply the above research results in actual field, a safety evaluation method combining advantages of computer fast prediction and on-site detection is put forward, which conquers the problem that on-site detection is difficult to measure the highest concentration of oil vapor in the tank. The results of this work will provide reference to achieve efficient and safe replacement of oil vapor in horizontal tanks. (C) 2018 Published by Elsevier B.V. on behalf of Institution of Chemical Engineers.