An Integrated Data-Driven Framework for Diagnosing Production Collapse and Optimizing Manufacturing Recovery in a Nigerian Brewery

Chijike Macdonald Ogbu

Department of Mechanical Engineering, Federal University of Technology, Owerri, Imo State, Nigeria.

Basil Eberechukwu Okafor

Department of Mechanical Engineering, Federal University of Technology, Owerri, Imo State, Nigeria.

Osita Obineche Obiukwu

Department of Mechanical Engineering, Federal University of Technology, Owerri, Imo State, Nigeria.

Victor Uchechukwu Opara

Department of Mechanical Engineering, Federal University of Technology, Owerri, Imo State, Nigeria.

Daniel Arinze Ekpechi *

Department of Mechanical Engineering, Federal University of Technology, Owerri, Imo State, Nigeria.

*Author to whom correspondence should be addressed.


Abstract

Production collapse remains one of the most significant issues limiting the competitiveness and sustainability of manufacturing industries, particularly in developing economies where ineffective maintenance practices and ageing equipment constrain operational performance. This study investigates the root causes of production collapse at Premier Breweries Ltd., Nigeria, between 1997 and 2005, and develops an integrated data-driven framework for production recovery. Historical monthly maintenance and production datasets were analysed using descriptive statistics, correlation analysis, Ordinary Least Squares (OLS) regression, Random Forest (RF) machine learning (ML), and scenario-based optimisation. The analysis identified downtime hours, spare-parts unavailability, and breakdown frequency as the primary operational drivers, collectively accounting for 68.70% of the observed production decline, whereas political and financial variables exhibited comparatively limited influence. Regression analysis demonstrated that each additional equipment breakdown reduced production by approximately 112,800 units, while every added downtime hour resulted in a mean loss of 10,580 units. The optimised RF model achieved robust predictive performance, recording a mean absolute percentage error (MAPE) of 12.61% and a coefficient of determination (R2) of 0.7850, outperforming conventional statistical modelling in capturing complex nonlinear interactions among operational variables. Furthermore, scenario-based optimisation suggested that improving spare-parts availability, capping monthly downtime at 60 hours, and minimising breakdown frequency could increase production from 2.79 million to 4.16 million units (a 49.10% capacity recovery). This study presents a scalable decision-support framework integrating ML, maintenance optimisation, and reliability engineering to strengthen long-term operational resilience, optimise production efficiency, and enhance equipment reliability in brewery and process manufacturing systems.

Keywords: Production collapse, brewery operations, machine learning, reliability engineering, predictive maintenance, scenario optimization, manufacturing optimization


How to Cite

Ogbu, Chijike Macdonald, Basil Eberechukwu Okafor, Osita Obineche Obiukwu, Victor Uchechukwu Opara, and Daniel Arinze Ekpechi. 2026. “An Integrated Data-Driven Framework for Diagnosing Production Collapse and Optimizing Manufacturing Recovery in a Nigerian Brewery”. Asian Journal of Current Research 11 (4):396-420. https://doi.org/10.56557/ajocr/2026/v11i411411.

Downloads

Download data is not yet available.