EINSTEIN R&D Project Investigates Automated Fault Detection and Diagnostics (AFDD)

By Daniel Coakley on Thursday 9 June 2016

Up to 20% of the total energy used in developing countries is consumed within HVAC systems with between 15-50% of this consumption being attributed to faulty operation. A UK survey conducted in 2000 which found prompt detection and diagnosis of HVAC faults can reduce the average plant consumption by more than 10-35%. Similarly, analysis of VAV systems found considerable HVAC energy savings could be made through the adoption of Automated Fault Detection and Diagnostics (AFDD).

Many approaches to HVAC AFDD have been developed, but the commercial viability of many of these techniques still needs to be thoroughly investigated. A history of known ‘tagged’ instances of faults within the data-set is essential for assessing and comparing the frequency of false alarms or detection success rates. IES has found that historical system data is often unavailable and inadequate for this purpose. Therefore, simulation of faults using real test plant or software provides a promising alternative.

The following paper was recently published by IES R&D in conjunction with the EINSTEIN project. The Simulation Enhanced Integrated Systems for Model-based Intelligent Control(s) Project (EINSTEIN) is a European funded project which aims to develop and deploy a prototype building operation optimisation framework. The paper proposes a scheme for the procurement and preparation of synthetic BEMS data in which faults are present using the IESVE.

“Synthesis and Refinement of Artificial HVAC Sensor Data Intended for Supervised Learning in Data-Driven AFDD Techniques.”

David McCabe presented this paper at the 9th International Conference Improving Energy Efficiency in Commercial Buildings and Smart Communities (IEECB&SC’16) in Frankfurt on 16th March 2016. His slides can be viewed below: