Abstarct: In this thesis, Optimization of green reentrant flow shop scheduling problem with batch delivery system is investigated in order to minimize the overall costs of tardiness and batch delivery. As we are facing more competitive markets, the need for integrated planning and coordination of production systems in the form of a supply chain network has been felt more than ever in order to reduce the final cost. Production systems cannot be separated from logistics and transportation systems, which is supply chain management. A successful supply chain is one that can provide products efficiently and cost-effectively. For this purpose, in the reentrant flow shop environment, where at least one task must pass through one or more stages more than once, batch delivery limit is defined; Batch delivery is the process of sorting orders and delivering batches using transportation. Therefore, planning an integrated production and distribution supply chain with batch delivery implemented can be effective for a supply chain to achieve its optimal goal, which has not been reviewed in the literature. For this purpose, a new mathematical programming model is presented for the given problem. The first goal of the problem is to minimize the overall cost related to the total tadiness times of the works and batch delivery costs. The second goal is to minimize energy consumption. Mathematical model with the help of single-objective epsilon constraint method is solved with GAMS software. Then a few random problems are evaluated with its help, and the effect of focusing on energy consumption of production scheduling is observed. Then, in order to solve problems with medium and large dimensions, a non-dominant sorting genetic algorithm (NSGA-II) is developed. In order to evaluate the performance of the resulting solutions, convergence metric and the number of Pareto solutions have been used. The results show that comparing the mathematical model with NSGA-II in problems with small dimensions, the mathematical model provides better solutions, but for problems with medium and large dimensions, iteration of 1000 of the mexta-heuristic algorithm works very well and has the ability to create near-optimal Pareto front.