Procedural content generation (PCG) is a growing field of interest in the domain of computational intelligence as it relates to games. There are ever increasing examples and applications of PCG that have been studied in academic contexts. Player expectations of the amount of content in games increase as computers and video game consoles are capable of using more content, and automation of content creation becomes more desirable. While many means of procedural content generation using some form of search algorithm have been tried and tested, we examine evolutionary algorithms as a means to generate content, where it has not frequently been used before. We examine the generation of vehicles, specifically spaceships, within two dimensional game simulations. These simulations are based upon a simple Newtonian physics system with different physical rules, representing games such as Lunar Lander or Asteroids, and evolve linear vectors of real numbers that act as vehicle genotypes by encoding placement of components to a vehicle point mass, with a form defined by the placement of each component. We use simple 1-ply lookahead controllers, simple rule-based controllers, and MCTS-based controllers as means to test and therefore indirectly guide the evolution of vehicle designs. We are able to demonstrate that evolutionary algorithms can be used to generate effective vehicle designs, suitable for use by the same controller as used for testing, for simple tasks without much issue. We also show that there are some factors of a problem environment that impact the demands and the conditions affecting vehicle design evolution more than others, such as velocity loss factors and the topology of the game world used. It is also evident that the use of different controllers to test vehicles causes different designs to emerge based on the strengths of said controllers.