In order to solve the problem of flexible flow shop scheduling with variable processing time (FFSP-VPT), a mathematical model is established, in this paper, based on analysing the processing time selection, defect detection, and rework of jobs with variable processing time. According to the characteristics of FFSP-VPT, a two-stage coding method is designed using ranked over value (ROV) coding. Based on the bat algorithm (BA), a self-adaptive elite bat algorithm (SEBA) is developed as the global optimisation method in this study by using Hamming distance-based elite individual set and adaptive position update. Crossover tests are designed to determine the optimal values of critical parameters in SEBA. Validity of SEBA on solving FFSP-VPT in actual production is verified by comparing the simulation test results of SEBA with those of other algorithms.