An Empirical Reliability Assessment and Forecast of the Auchi Power
Distribution Network, Edo State, Nigeria.
Umahon Ovbiagele¹*, Odiaise Friday², H.E. Amhenrior³
¹,3Department of Electrical and Electronic Engineering, Edo State University, Uzairue, Edo State,
Nigeria.
²Department of Electrical and Electronic Engineering, University of Benin, Benin-City, Edo State,
Nigeria.
*Corresponding Author
Received: 11 November 2025; Accepted: 18 November 2025; Published: 26 November 2025
ABSTRACT
This study presents an empirical reliability assessment and forecast of the Auchi power distribution network in
Nigeria, addressing the scarcity of granular, data-driven analyses in the sector. Utilizing a case study research
design, the study analyzed actual 2023 operational data for three 11kV feeders—Auchi Town, Jattu, and Auchi
GRA—obtained from the Benin Electricity Distribution Company (BEDC). The methodology involved a
quantitative, two-phase approach: first, computing standard reliability indices (SAIDI, SAIFI, CAIDI, ASAI)
based on IEEE Standard 1366, and second, employing the Facebook Prophet time-series model to forecast
ASAI values from 2024 to 2035. The empirical results for 2023 revealed critically low and variable reliability,
with the Jattu feeder, for instance, recording an ASAI of 0.1614 in February, indicating power was available
only 16.14% of the time. The forecast revealed starkly divergent feeder trajectories: stagnation for Auchi
Town, seasonal variation for Jattu, and consistent improvement for Auchi GRA. These findings provide crucial
evidence of significant service disparity and underscore the urgent need for feeder-specific investment and
policy interventions. The study demonstrates a replicable framework combining reliability indices and
predictive modeling to guide targeted maintenance and planning in similar contexts.
Keywords: Reliability Indices, Predictive Modeling, Power Distribution
LITERATURE REVIEW
Predictive modeling has become a cornerstone of modern power system management, enabling a transition
from reactive maintenance to proactive reliability management. Globally, machine learning techniques—
ranging from decision trees and neural networks to advanced methods such as Long Short-Term Memory
(LSTM) networks— have demonstrated significant effectiveness in forecasting outages and optimizing
maintenance operations. The performance of these predictive models is commonly evaluated using
standardized reliability indices such as the System Average Interruption Duration Index (SAIDI) and the
System Average Interruption Frequency Index (SAIFI), which serve as essential benchmarks for assessing
system reliability in various regions (Folarin et al., 2017; Kumar et al., 2018; Hashemi, 2021).
The Nigerian power sector, characterized by persistent challenges such as inadequate infrastructure and
frequent interruptions in supply (Dahunsi et al., 2022), stands to benefit substantially from the adoption of
data-driven analytical tools. Predictive analytics provides a structured framework for prioritizing maintenance
activities, optimizing network planning, and enhancing decision-making for improved reliability performance
(Miroslaw Parol et al.., 2022; Parol et al., 2022). Despite these benefits, a clear limitation persists between the
potential of predictive modeling and its practical implementation in Nigeria. Most existing studies remain
Page 2599
