Medium term load forecasting of 33 kV line loading: a case study of Ota 132/33 kV sub-station Ogun State, Nigeria

Abstract

Peak load forecasting plays a pivotal role in the efficient operation and planning of power systems, influencing decision-making processes for resource allocation and infrastructure development. The Ota 132/33 kV substation in Nigeria is facing increasing demand due to rapid industrialization and urbanization. This has strained the substation's infrastructure, leading to issues like transformer overloading, voltage fluctuations, and power outages. The area's trade center status and numerous industries further stress transformers, increasing wear and tear and potentially jeopardizing their reliability. This study delves into the realm of 33 kV feeders at the OTA Transmission Substation, aiming to unravel the intricacies of peak load patterns and provide a forecast for the next five years. Leveraging historical data spanning from 2018 to 2022, sourced from OTA Transmission Substation was used to forecast from 2023 to 2027. The research employs the Auto-Regressive Integrated Moving Average (ARIMA) model to discern trends and project future peak loads. Performance metrics, including Mean Absolute Error, Mean Absolute Percentage Error, Root Mean Square Error, and R-squared, are meticulously evaluated to assess the robustness of the forecasting model. The findings shed light on the unique characteristics of each feeder, with Sumo, Amje, and Idiroko having a better predictive accuracy performance with minimal errors, while Sango, FSM, and Estate show a moderate level of predictive accuracy probably due to the presence of little nuance in their data set. Where the Sango 33 kV feeder displayed an upward trend of 18.91MW in 2023 and 19.34 MW in 2027. Sumo 33 kV feeder exhibits a decline trend from 3.21 MW (2023) to 2.23 MW in 2027. FSM 33 kV feeder shows a fluctuation pattern while Amje 33 kV feeder indicates a highly stable trend of 22.29 MW all over. Idiroko 33 kV feeder shows a steadily increasing trend of 16.09 MW (2024) to 16.25 MW in 2027. The Estate 33 kV feeder on the other hand depicts a relatively stable pattern. This study not only contributes to the localized understanding of peak load dynamics but also serves as a template for similar investigations in other power distribution networks and unveils other alternative data science-based models for future researchers.