A new robot manipulator simulation software (RobosDyn) is being developed based on the MCAE (Mechanical Computer Aided Engineering) framework developed in the previous research. This framework is developed as a plug-in component structure with standardized MCAE data. The robot manipulator simulation software enables users to easily and conveniently to create a virtual manipulator model from DH (Denait-Hartenberg) parameters, to create and analyze an inverse kinematic analysis model that can obtain the position speed of a joint and velocity manipulability of end-effector, and to create and analyze a dynamic analysis model that can obtain the drive torque of a joint. In this paper, an optimization process is developed to optimize DH parameters of the virtual manipulator model in the robot manipulator simulation software. The design variables are DH parameters and the performance index is rotational work of all joints. A genetic algorithm to search global optimum solutions is used in the optimization process.