Production, distribution and inventory planning are the most important planning in big companies. Considering environmental issues and traffic rules which prohibit the commute of trucks in cities, manufacturing companies should find a way based on the 2-echelon-inventory-routing problem to reduce their costs. roduction- routing problem tries to integrate decisions about routing and inventory. In most production lot sizing problems with direct traort, distribution with direct traort, and inventory are not evaluated simultaneously. Considering suppliers inventory management system, production- routing problem combines the 2-echelon-inventory-routing problem and lot sizing with direct traort, and explores decisions about production and inventory totally. In this thesis, an integrated model of production- routing in two layers with considering the size of lot is presented. In each time window, multi products can be produced, and the production plan of each product is determined. Also, the amount of products that should be delivered to customers, routs to each customer, and the inventories of Satellite ,plants and customer are determined in two layer. The number and the capacity of vehicles and the capacity of production are supposed to be limited. The objective of proposed model is to minimize the costs of set up, production, inventory and routing in two layers. The findings show proposed model results in significant reduction in costs. Given that this problem is NP-hard, a meta-heuristic algorithm based on VNS algorithm with two heuristic approaches for solving big problem is designed. In metaheuristic algorithm VNS (I), routing problem is solved by algorithm which determine which customers should be visited in each day and then production planning will be solved by a mathematical model. In metaheuristic algorithm VNS (II), first a rout is created by a heuristic algorithm, then visits are determined by a metaheuristic algorithm, and finally production planning will be solved by a mathematical model. To evaluate the performance of the algorithm, the results from VNS algorithm in big problem are compared with the results from a small problem solved with the exact method. The results show appropriated performance of proposed algorithm.