Comparative analysis of iterative approaches for incorporating learning-by-doing into the energy system models

Abstract

Appropriate treatment of technological changes has been an important issue in the field of energy system modeling. Although some mixed integer programming-based formulations have been introduced to incorporate learning-by-doing endogenously, they require high computational effort. Therefore, many practitioners have not considered the technological changes endogenously. Recently, some studies have suggested iterative approaches to incorporate learning-by-doing indirectly. This study provides a comparative analysis among the most famous mixed integer programming-based formulation and iterative approaches. We also propose a revised iterative approach that can overcome the cons partially. Lastly, as a numerical study, we apply the previously suggested methods and our proposed method to analyze two renewable energy policies, carbon taxation and subsidy, in the Korean electricity sector. This numerical study illustrates the results of our comparative analysis. The iterative approaches can be approximately 5–23 times more computationally efficient compared to the revised formulation. In addition, the required total carbon tax or total subsidy in scenarios using different iterative approaches are 10%–50% lower than no learning scenarios. The practical implications of this study are the correct approaches that would aid in determining the accurate futuristic scenarios. This will lead to the timely and effective implementation of relevant policies.