Predictive Maintenance and Anomaly Detection for Building Management Systems

01/09/2020

Overview

THE EARLY ANOMALY DETECTION SYSTEM FOR BMS was developed as an intelligent monitoring and analysis solution for Building Management System data in a large-scale facilities management environment. The solution was designed to help operational and engineering teams move beyond reactive alarm monitoring toward earlier detection of equipment anomalies, plant inefficiencies and emerging maintenance risks. It combined graphing, reporting, anomaly detection, cost monitoring and error diagnostics in a single web-based application.

The challenge

The operating environment involved remote monitoring of approximately 190 sites and around 60,000 BMS data points, with data captured every minute. This produced an extremely large volume of information each day, making manual analysis impractical. Traditional monitoring focused mainly on alarms and setpoints, which typically indicate that equipment is already approaching failure. The opportunity was to create a solution that could detect abnormal behavior earlier, support proactive intervention, and improve plant performance visibility.

My role

I was responsible for the specification, conceptual design and project management of the project. My role focused on translating the business need into a workable solution, shaping the overall concept, and driving stakeholder alignment rather than on software programming. I discussed, presented and demonstrated both the concept and the final solution to senior management and clients, engaged closely with key stakeholders including the contact center and engineering departments, participated in software testing, and supervised implementation of the solution in the contact center.

The solution

THE EARLY ANOMALY DETECTION SYSTEM FOR BMS was conceived as a web application that would analyze BMS sensor and equipment performance data, identify normal operating patterns, and flag deviations that could indicate developing faults or inefficiencies. It was designed to provide users with clear visual trends, anomaly reporting and operational insight, while also supporting integration with existing alarm and dispatch processes. The solution's broader value proposition included early anomaly detection, predictive maintenance support, plant optimization and more informed operational response.

The concept also aligned with the Bidvest Facilities Management BCC operating model and the intended integration with systems such as SAM and the FIC App, enabling anomalies to be escalated into operational workflows for action and follow-through.

Approach

My focus was on defining the solution and ensuring it could work in a real operational setting. This included:

  • specifying the business and operational requirements
  • shaping the conceptual design of the platform
  • aligning the solution with contact center and engineering workflows
  • engaging stakeholders across multiple functions
  • presenting the concept and business value to decision-makers and clients
  • supporting testing and validating that the solution met operational needs
  • supervising implementation in the contact center environment

Outcome

THE EARLY ANOMALY DETECTION SYSTEM FOR BMS created a framework for earlier identification of equipment issues and inefficiencies, helping shift the organization from reactive alarm handling toward proactive maintenance decision-making. The solution was designed to detect anomalies before they escalated into alarms, improve visibility of plant behavior, and support faster, more informed operational responses. It also strengthened the link between data analysis and action by aligning anomaly detection with existing monitoring and dispatch processes.

Business value

The project demonstrated how large-scale operational data could be translated into practical business value through better visibility, earlier intervention and stronger coordination between technical and operational teams. It supported:

  • earlier detection of developing faults
  • improved predictive maintenance capability
  • reduced risk of downtime
  • more effective alarm and incident management
  • better plant optimization opportunities
  • clearer communication of technical insights to operational users and leadership 
Share