Abstract: In general project management practice, the expected objectives and resource allocation are delivered from the project plans at the tactical level to the plans at execution level. However, the optimization approaches of tactical project planning are scarce while there are lots of existing approaches focusing on the project scheduling problems at the execution level. In this paper, a new optimization method of project planning at strategic level is studied in view of the actual situation of product development project management. Based on the workload estimation of the work package, a mixed integer programming model for the tactical project planning problem is addressed, with the dual objectives of minimizing the resource level and the make-span of multiple projects under various constraints such as resource, precedence relationship among work packages, etc. An improved bi-objective genetic algorithm is developed in the framework of the fast non-dominated sorting genetic algorithm (NSGA-Ⅱ). There are two significant improvements based on the basic NSGA-Ⅱ in the algorithm. Firstly, aiming at the characteristics of the problem, a decoding algorithm based on resource smoothing is designed. Secondly, referring to the key feature of NSGA-Ⅲ, the calculation method of congestion density is improved and a new calculation method is proposed. The performance of the algorithm is verified by comparision of the improved algorithm with the basic NSGA-Ⅱ for solving a real project case. The effectiveness of the appraoch proposed in this paper is shown and illustrated based on the analysis of the results calculated by the presented algorithm,since it can provide the project decision maker with more comprehensive support.The research will extend the research of project planning and scheduling to the macro level of project management.

Key words: project management, tactical project planning, multi-objective optimization, genetic algorithm