A leading retailer implements proactive fault detection in refrigeration systems to reduce energy

TCS helped a leading retailer improve fault detection in their refrigeration systems. We developed a  first of its kind refrigeration monitoring solution that detects and identifies faults in supermarket refrigeration systems, even in the absence of comprehensive onboard sensor information, using data logs from an ensemble of stores.

The customer

The client is a leading retail chain in the United States offering home goods, clothing, electronics, groceries, and other products.

Business scenario

Refrigerated systems are used in supermarkets to maintain perishable goods at the required temperature. Any fault in refrigeration systems can further push up energy consumption levels. A leading retailer wanted to ensure proactive detection and rehabilitation of faults to reduce energy consumption and avoid service disruption to customers in the stores.


We have extensive expertise in remote monitoring. We used this to provide the retail major with an innovative approach adopted for this initiative that would meet specific business needs.

TCS Solution

TCS developed an energy model that leverages the energy consumption patterns of refrigeration systems and identifies even the smallest irregularity in the energy signal and helps maximize the likelihood of detecting an anomaly. If the actual energy consumption of a system deviates significantly from the value predicted by the model, it is flagged as a potential fault in the system. The TCS team used the magnitude and the direction of the deviation in energy from the expected value to perform root cause analysis of the fault. Additionally, information on non-energy parameters (such as pressure and temperature at various points of the refrigeration system) helped narrow down possible causes and sub classify the root cause of the anomaly.

Key benefits

With our solution, the retailer can monitor refrigeration systems without complex and expensive instrumentation. The solution also enables the identification of the individual component that is at fault. Faults can be identified 3-21 days (depending on the fault type) before they are detected by store employees with the false positives rate limited to less than 0.2%.

Reach Us.