Diesel engine operates on higher temperature and pressure compare to gasoline engine system. Owing to its operating principle, exhaust system of diesel engine contains rubber hoses which are composed of various hyper-elastic materials. First, metal parts of exhaust system are modelled and correlated using modal test. Also, metal and rubber parts are assembled on finite element (FE) and validated through experiments. Furthermore, the non-linearity and hyperelastic properties of rubbers are modelled using various strain energy models from the tensile test of rubbers. Due to non-linear property and comparably small elastic modulus of rubber components, we found out that exhaust assembly consisting rubber appears to have high number of modal densities in general frequency range. This phenomenon increases complexity on FE model and result computational cost. Therefore, we applied component mode synthesis (CMS) on rubber parts to reduce computing time. Moreover, bellows and inside of EGR cooler are simply modelled by shell elements. Reduced models turn up to be highly efficient than original model while solving vibrational problems. The reduced models are validated by vibrational experiments (Natural frequency, mode shape, frequency response function) and we obtained highly correlated FE models.