The long range production planning is a starting point of making a good medium and short range scheduling. In addition, the contracts of receiving orders and negotiation with customers can be supported by the effective long range production planning system. So, we focus on the development of a methodology for the long range production planning particularly for the shipbuilding plants. The planning aims to balance the loads among production stages and to meet delivery dates. Since the product mix for a batch determines the load shapes of stages, we select each product mix of batch from the pre-defined feasible product mix cases. Domain constraints and preferences are taken into consideration in this process. This process can be esteemed as a knowledge-based heuristic search because domain knowledge is handled in rules and frames. In order to deal with the variation of load, cumulative load curve, so called S curve, is adopted. To balance load in each stage, the linear programming model is formulated along with the cumulative load curve for evaluation purpose. A prototype system DAS-LPP is developed encompassing the knowledge-based heuristic and linear programming model. To implement DAS-LPP, a tool UNIK, programming language C++, and linear programming solver MINOS were employed.