Abstract
Industrial machinery fleets in developing economies face unique reliability challenges due to operational environments and maintenance constraints. A systematic methodology for forecasting their reliability is required for proactive asset management and capital planning. This article presents a methodological framework for evaluating fleet reliability and develops a bespoke time-series forecasting model to predict future system performance, enabling data-driven maintenance and replacement strategies. A hybrid methodology integrates reliability-centred maintenance analysis with statistical forecasting. The core forecasting model is a seasonal autoregressive integrated moving average (SARIMA) process, formalised as $\phi(B)\Phi(B^s)\nabla^d\nablas^D yt = \theta(B)\Theta(B^s)\epsilont$, where $\epsilont$ is white noise. Model parameters were estimated using maximum likelihood, with forecast uncertainty quantified via 95% prediction intervals. The methodological application demonstrates a clear downward trend in aggregate fleet reliability, with a forecasted decline of approximately 15 percentage points over the forecast horizon. Model diagnostics indicated robust standard errors, and the SARIMA(1,1,1)(0,1,1)_12 specification provided the best fit to the historical data pattern. The proposed integrated methodology provides a technically sound framework for fleet reliability assessment and forecasting. It successfully captures the temporal dynamics of system degradation, offering a practical tool for engineers and asset managers. Implement the methodology with quarterly data updates to recalibrate forecasts. Future work should integrate real-time sensor data into the model and explore machine learning extensions for non-linear patterns. reliability engineering, time-series analysis, fleet management, predictive maintenance, infrastructure asset management This paper provides a novel, integrated methodological framework that combines reliability analysis with formal statistical forecasting, specifically tailored for industrial machinery in a developing economy context, and yields a directly implementable forecasting tool.