Study goals
Analyze the reliability of the sludge mechanical dewatering process in a wastewater treatment plant, identifying failure modes and proposing actions to optimize performance, availability, and final sludge quality.
Relevance / originality
The study addresses a critical process in treatment plants, combining FMEA and Machine Learning to predict failures and propose improvements, contributing to operational efficiency and environmental sustainability.
Methodology / approach
Uses quantitative failure analysis, functional mapping, FMEA application, and Machine Learning algorithms (Decision Tree and KNN) to predict failures and propose corrective and preventive actions.
Main results
Identified critical components such as plate filters, pumps, and mixers. Decision Tree achieved 88% accuracy in failure prediction. Physicochemical parameters were most influential in process reliability.
Theoretical / methodological contributions
Integrates traditional reliability techniques with artificial intelligence, demonstrating effectiveness in failure prioritization and anomaly prediction, expanding methodological scope in maintenance engineering.
Social / management contributions
Proposes predictive maintenance strategies and operational adjustments that enhance reliability, reduce costs, and promote sustainability, aligned with legal requirements and asset management best practices.