Estimating Energy Demand for Electric Motorcycle Battery Swap Stations Using Logistic Regression and Monte Carlo Simulation

The growing adoption of electric motorcycles poses new challenges, particularly in establishing efficient battery stations that adequately meet demand. To address this challenge, we developed an approach aimed at estimating the battery swap demand for electric motorcycles across multiple demand sites. Our proposed approach involves three steps. In the first step, we use socioeconomic variables to estimate the population using a logistic regression approach. Next, we employ Monte Carlo simulation to predict the penetration rate in regions adopting electric vehicles. The final step involves using the population estimates and the penetration rate to estimate battery swap demand at each point. The approach was then applied to the case of Jakarta, the capital city of Indonesia, which consists of five regions and 44 demand districts based on data from the year 2023. The results indicate the penetration rates for each region as follows: 0.47 for Kepulauan Seribu and Jakarta Utara, 1.59 for Jakarta Pusat, 1.30 for Jakarta Barat, 0.91 for Jakarta Selatan, and 1.73 for Jakarta Timur. This study's estimation of battery swap demand, reflected in a mean square error (MSE) of 172.21, a root mean square error (RMSE) of 13.12. and a mean absolute error (MAE) of 9.95, underscores its effectiveness in leveraging individual socioeconomic factors. By incorporating these factors, the approach improves the accuracy of demand forecasts, allowing for localized predictions of battery swap demands across various urban environments. The findings of this study hold significant for policy formulation and infrastructure planning in the transportation sector.