In the context of more and more drastic noise regulation and increasing customers demand for lower noise annoyance, acoustic shields become essential for a wide range of vehicles. Due to reduced development time, acoustic design must start in the early stage of industrial projects, requiring precise and reactive prediction tools. The most widely used computation methods perform a numerical resolution of Helmholtz equation with a spatial discretization into Finite Elements or Boundary Elements. These methods are efficient in the low frequency range, but they reach their limits at higher frequencies, due to high computational cost, very precise mesh required, and high sensitivity to geometry and frequency. Then Ray Tracing techniques may be an alternative in some cases, but diffused reflection is generally ignored and convergence is not always reached, observation points receiving too few rays. The method proposed here is also based on a light/sound analogy, but the resolution is based on Boundary Elements. Lamberts law is used to model diffuse reflection. Absorbing materials are characterized by diffuse absorption coefficients, widely used in the industry. The frequency based meshing criterion is relaxed: only the description of the geometry must be considered. The resolution can be made every third-octave if the evolution of acoustic properties is smooth enough. The phase information is not kept. Meshing and computation time are drastically reduced. This energy method is dedicated to industrial acoustic issues in the mid and high frequency range, preferably with complex geometries, broadband and distributed sources. The method is therefore particularly suited for the prediction of the noise radiation of thermal engine in its environment. The acoustic resolution is carried out by a specific program and relies on view factors computed with thermal radiation software. Successful comparisons with measurements are presented, concerning the engine compartment of trucks. Successful comparisons with measurements are presented, concerning both external and internal noise predictions for trucks.