In this proposal, the piezoelectric energy harvester, which harvests wasted mechanical vibration energy, and the thermoelectric generator, which harvests human body thermal energy, are studied. To design and optimize these energy harvesters, computer-aided development processes have become state-of-the-art because of their low cost and high efficiency. However, the capacity of today’s computers is not always sufficient to meet the demands of industrial simulations. The complex geometry and multi-physics effects in energy harvesting devices have led to rather time-consuming simulations. To improve the efficiency of the simulations using existing computers, the methodology of model order reduction (MOR) and parametric model order reduction (pMOR) methods are applied to generate reduced order models, which guarantee computational efficiency, e.g., for the system-level simulations by replacing the original full-scale model with the reduced but still accurate surrogate.