The Challenge
‍The public transport station’s refrigeration plant and relevant equipment in the transport station were aging quickly due to frequent daily usage. A fault prediction and equipment maintenance system was urgently needed in order to ensure an efficient operation for the public transport facility.


‍Our SolutionsCarnot provided unsupervised Machine Learning based Fault Pattern Detection in this project, in which detected an anomaly for one specific chiller. Under the Anomaly Matrix view, our software diagnosed that the chiller water flow rate was much lower than normal. Thus, this issue was proactively addressed and resolved by Carnot. Through the ability to predict and detect malfunctions at an early stage, Carnot helped our client to contain potential issues from turning into severe system failure. Preventive maintenance was also in place to avoid costly events


‍ResultsCarnot provided a web-based dashboard for the BMS operators to view and monitor performance diagnostic functions and proposed actions for fault detection, in which our client gained the benefits of learning and getting familiar with our advanced analytics platform. Now that optimization has been completed, the client has successfully saved over 11% of energy consumption annually, which reduced electricity cost for more than HKD120,000 (USD15,500) yearly. The 11% operational cost savings was achieved without additional retro-fits. Thus, it only took 1.5 years to cover the initial costs. Besides economic benefits, occupant enjoy higher comfort level with more consistent air conditioning control. Maintenance schedule is carried out in a routine manner for the reporting of several aspects, including an overall energy profile, achieved energy savings and fault prediction, savings forecast, status of AI model, and the means for further enhancement. Carnot continues with the ongoing maintenance and commissioning for the public facility, meanwhile in discussions with our partner for new project opportunities. Since the results in this project have been fairly satisfying, we are optimistic that our software can be widely used in other public transportation facilities.
‍