Case Study: Predictive maintenance platform for HVAC systems
A cloud-based predictive maintenance solution for a U.S. real estate operator managing dozens of commercial buildings. The platform integrates IoT sensors, Azure Anomaly Detector, and the UpKeep CMMS system to forecast HVAC failures, reduce emergency repairs, and shift maintenance from reactive to predictive mode.
Project details:
About the Client:
The Client is a U.S. property management company based in North Carolina that operates a network of approximately forty commercial and retail buildings. The company manages leasing, facility operations, and maintenance of outdated HVAC systems across all properties.
Location: USA, North Carolina
Industry: Real estate & facility management
Team size: 7 specialists
Project duration: 8+ months
Business сhallenge
The company was losing 15% of its HVAC maintenance budget on urgent repairs of HVAC systems. Moreover, the Client got constant complaints from tenants, but a full upgrade was economically unfeasible.
Thus, the company wanted to reduce HVAC downtime and emergency repair costs through predictive maintenance.
Additional requirements:
- coordinate with our HVAC-IoT partner responsible for sensor installation;
- compliance with building-level network security policies.
Our solution
We developed a cloud-based predictive maintenance platform for the HVAC. Sensors capture vibration, ΔT, pressure drop, and electrical load; data flows to Azure Anomaly Detector. Detected issues automatically create tickets in UpKeep CMMS and notify engineers. A SumatoSoft web dashboard shows fleet health, trends, and work order status – closing the loop from data to action.
Core analytics layer
At the heart of the platform is a streaming analytics pipeline. Gateways push time-series data from vibration, ΔT, ΔP and electrical sensors to Microsoft Azure, where we clean, normalize and aggregate it before sending features into Azure Anomaly Detector. The pipeline correlates anomalies across units, writes results to Azure Data Lake Storage and triggers CMMS tickets in near real time.
Additional features:
- automated anomaly detection and failure prediction powered by Azure Anomaly Detector;
- continuous telemetry streaming through Azure IoT Hub, Azure Stream Analytics;
- automatic work order creation and status synchronization in UpKeep CMMS;
- network isolation, per-device certificates, TLS encryption, and egress-only traffic policy;
- scalable modular architecture supporting additional buildings and sensor types;
- unified web dashboard displaying equipment status, performance trends, and maintenance history.
Business value
Before:
- HVAC issues were detected only after tenant complaints or complete system failures.
- 15% of the HVAC maintenance budget was consumed by urgent repairs.
- Maintenance tickets were created manually in UpKeep CMMS, which slowed reaction time.
- Engineers relied on manual and route-based inspections, which took budget and time, but didn’t bring the desired results.
After:
- Early anomaly detection provides clear signals hours or days before failures occur.
- The predictive maintenance cuts emergency repair costs by 45%.
- Automatically generated maintenance tickets in UpKeep accelerate engineer response, reducing downtime by 40%.
- Continuous telemetry and analytics replaced most routine manual inspections with remote, real-time visibility into the condition of each HVAC unit.
