In recent years, stricter emission regulations for internal combustion engines have been implemented, including controls on evaporative fuel vapors from motorcycle fuel systems. To comply with these regulations, motorcycles are increasingly adopting evaporative emission control (EVAP) systems equipped with carbon canisters. The carbon canister adsorbs fuel vapors while the vehicle is stationary, preventing them from being released into the atmosphere. During engine operation, the stored vapors are purged back into the engine by the vacuum in the intake manifold, thereby regenerating the canister. EVAP system development must ensure compliance with emission standards while minimizing any negative impact on engine performance. As regulations are expected to become stricter in the future, there is increasing demand for high-performing canisters and more effective purge systems. This highlights the need for more efficient development methods.
The aim of this study is to enhance the efficiency of EVAP system development by utilizing model-based development (MBD) with CAE technologies and machine learning-based surrogate models. Typically, there is a trade-off between simulation time and model accuracy in CAE, but surrogate models can significantly reduce computation time. By integrating CAE and surrogate models throughout the development process, a practical and efficient approach to developing EVAP systems is proposed.