Designing mechanical complex systems like helicopters requires more and more information about the subsystems found within. The number of models required to describe their internal behavior also tends to grow drastically. To overcome the challenging task of combining multiple physical domains within a global representation, we adapt the Model Driven Engineering concepts to multiphysics simulation. First applied to software development, this approach is based on metamodeling, a method well suited to establish the generic definition of modeling formalisms and thus useful to ease the comprehension and enhance the design of complex systems. The study focuses on two modeling techniques, the Bond Graph and Energetic Macroscopic Representation (EMR), usually employed to describe the energetic exchanges occurring within the system. Based on the metamodels proposed for each methods, this contribution introduces a model transformation able to automatically generate an EMR model from a bond graph. The implementation is realized under Eclipse Modeling Framework (EMF) and based on ATLAS Transformation Language (ATL). The transformation is validated and analyzed through its application on an industrial use case, here the design of a helicopter flight control subsystem.