There are a lot of achievements of researches on the permutation flow shop scheduling problem since 1970s. A lot of researches focuses on single objective of minimizing makespan (total processing time) or maximum tardiness. However, the obtained schedule from single objective usually cannot fully appropriate in practical production environment. Because huge tardiness would need much cost, even if the makespan was reduced. So, the target of this paper is digging out a set of solutions that consider double objectives, minimizing makespan and maximum tardiness in simultaneous, for decision maker. The decision maker can select a schedule from obtained schedule set based on actually market demands and production environment. This paper proposes a multi objective local search procedure which contains double neighborhood structures. Any movement on a schedule based on proposed neighborhood structures simultaneously takes effect on double objectives. These neighborhood structures are not found in any literature research before. The interaction of double neighborhood structures naturally guides search direction. The proposed multi objective local search procedure integrates with famous multi objective evolutionary algorithm, NSGA-II (Non-Dominated Sorting Genetic Algorithm-II). The experiment results show efficient of our proposed method compared with former algorithm even with same number of individual evaluations.