Predictive maintenance of pumps in civil infrastructure: State-of-the-art, challenges and future directions

Predictive maintenance (PdM) is a technique that employs data-driven analysis to detect anomalous working conditions and predict future failure risks of assets. Despite wide applications in the manufacturing and oil and gas industries, the application of PdM in infrastructure facilities, such as wastewater treatment plants, is scarce. Recent advent of information and communication technologies and artificial intelligence presents a great opportunity to enhance the practice in infrastructure maintenance by integrating PdM techniques. This study aims to investigate the potentials and challenges of integrating emerging technologies in the PdM of pumps. A quantitative review of the literature was conducted to identify primary research themes and knowledge domains. A qualitative review was conducted to assess their potentials for realizing PdM in pump maintenance. Findings from this research are expected to point out key technical and practical challenges and future research directions. [Display omitted] •Highlights pump fault diagnosis heavily relies on quality monitoring data and analytics•Limited types of monitoring data lead to false-negatives in dynamic conditions.•Deep learning methods can handle multivariate and low-quality, high-volume data.•A BIM-enabled DT framework can enhance the performance of deep learning methods.•Further research is required to integrate DT, BIM, and machine learning techniques.