Preventive maintenance of HVAC systems

The client was tackling increased losses from perishable shrinkage due to malfunctioning HVAC systems. Though their facilities had an existing installed E2 energy management system that generated alarms, the store personnel often did not have the expertise or time to analyze the alarms and respond appropriately. They wanted a system that could help them proactively maintain their HVAC systems without having to increase their workforce. The new system had to predict a failure in advance and trigger an alarm at least two days before it for timely and economical maintenance. They partnered with SLK to co-create this solution.

The SLK team identified the ten most critical alarms for HVAC systems and built a system to predict failures two days in advance. The system was built on the Microsoft Azure Services Platform with end-to-end solution ownership. The platform uses transmitter and sensor log data to identify constraints and influencing factors and has inbuilt algorithms to predict faults before an alarm triggers. This predictive analysis is done through Supervised Learning Multi-Class classification model. Connecting over two million devices, the system highlights 10+ critical alerts and factors on real-time intuitive dashboards.