Peramalan Beban Puncak Berdasarkan Triwulan Menggunakan Metode Interpolasi dan Koefisien Energi (Studi Kasus: PT. PLN UIP2B Jawa Madura Bali)

Abstract

The electricity demand in the Java-Madura-Bali System has experienced consistent growth, increasing from 179.813 GWh in 2017 to 209.062 GWh by the end of 2022. However, in 2020 and 2021, anomalies occurred due to significant restrictions, resulting in shifts in energy consumption patterns, a decline in peak loads, and an annual energy consumption drop, which led to a 40% reserve margin in 2020. These exceptional circumstances prompted the load planning unit to develop plans adaptable to the ever-changing situation in the Java-Madura-Bali System. Following the Ministry of Energy and Mineral Resources' Grid Code Scheduling and Dispatch Code (SDC) 6.2, PT. PLN UIP2B Java Madura Bali bears the responsibility to ensure that the Mean Absolute Percentage Error (MAPE) of load forecasts does not exceed 5% of the actual realization. Therefore, this study aims to develop the Energy Coefficient Method by adding load filtering in daily coefficient formation and grouping data by quarters at each planning stage, resulting in the Energy Coefficient Method with Quarterly Filtering (MKEFT) designed to meet SDC 6.2 requirements. Furthermore, this research includes a comparative analysis with the Exponential Triple Smoothing method, commonly recommended for electricity load forecasting, featuring three levels of data smoothing. The MAPE results indicate that MKEFT effectively meets SDC 6.2 standards, with 2.372% weekly and 3.465% daily MAPE results for energy consumption growth scenarios. Meanwhile, peak load growth scenarios yield 2.419% weekly and 3.343% daily MAPE results in 2022.