In effort to reduce environmental impact of the aerospace industry, More Electric Aircraft (MEA) concepts with electrical systems for fuel pumping, wing ice protection, environmental control systems and aircraft actuation are becoming more and more widely researched. The replacement of hydraulic actuators by motor drives for flight control surfaces is particularly attractive for maintainability, reduction in operating costs and to eliminate the hydraulic fluid. High power density of aerospace motor drives is a key factor in the successful realization of these concepts. An integrated system design approach offer optimization opportunities for further improvements in power density however the challenge lies in its multi-disciplinary modelling and the handling of numerous optimization variables or constraints that are discrete and non-linear in nature. A 4-level modelling paradigm has been proposed by multiple authors to represent a motor drive. Within this paradigm, the entire system is modelled at different levels separated by modelling details and dynamic frequencies. This paper first describes a quick motor drive sizing procedure with an emphasis of developing sizing functions of the main weight contributors within a motor drive for use in optimization. The motor drive topology detailed in this paper is a 12-slot 10-pole surface permanent magnet machine, with a voltage sourced back-to-back 2-level converter and an extruded fin cooling system. Based on the 4-level modelling paradigm, the paper then proposes a systematic multi-level method for applying random heuristic optimization. In this method, sequential optimizations are performed separately at different levels, only applying design variables and constraints that are within the relevant dynamic frequencies of the level. Genetic Algorithm was chosen as the random heuristic method using penalty factors to handle non-linear constraints. Performance verification of the resulting design is finally carried out for the optimal motor drive designs using different multi-physics software.