Skip to Main content Skip to Navigation
New interface

Méthode pour la détection de défauts des systèmes énergétiques : couplage expertise et méthode de réduction de dimensions

Abstract : The building field accounts for around 41% of the primary energy consumption. As the international directives aim to reduce the energy consumption and greenhouse gas emissions, they count on this sector. So they work toward the creation and refurbishment of buildings and a more energy efficient system. Among the building energy consumption, 50% come from the systems use for indoor comfort as the heating, ventilation, and air conditioning (HVAC). However, the real building consumption data analysis highlights a gap between the expected consumption of the design phase and the measured consumption. These differences can be explained in part by the poor operation of the setup systems. These system failures induce energy waste between 25% and 50% but can be significantly reduced by a good follow-up. These systems’ bad operations are called systems faults. These faults can take different forms such as sensors drift, breakdown, setting, or maintenance problems. Building commissioning or fault detection and diagnosis are very effective to reduce the systems’ unwanted consumption. Thus they are a major lever to cut down the energy consumption in the building field. These methods allow to detect system faults apparition to alert the maintenance and operation staff. This thesis aims to develop a fault detection and diagnosis method that allow to removing some scientific and technical obstacle whose inhibit the diffusion of these tools. The hybrid fault detection method developed is knowledge and data-based. This coupling form an effective and adaptable tool, that allows the monitoring and the system state estimation in real-time thanks to continuous expert rules treatment. The results are presented under the form of the system operation map obtain thanks to the dimension reduction method. This tool is tested on a ventilation system for a new building and an existing building cases. For the new building case, numerical system operation data are generated for the map construction. Then the fault detection is validated thanks to experimental data from the same system measured in a half-controlled environment. The experimental data represent the nominal and faulty operation of the system. They are projected on the map in order to check their fault detection ability. The existing building case is tested on real measured data of a few years. These studies show the efficiency of the method to quickly detects the system faults
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, December 16, 2021 - 12:02:15 PM
Last modification on : Wednesday, April 6, 2022 - 9:55:09 AM
Long-term archiving on: : Thursday, March 17, 2022 - 6:42:18 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03483106, version 1



Hugo Geoffroy. Méthode pour la détection de défauts des systèmes énergétiques : couplage expertise et méthode de réduction de dimensions. Génie civil. Université Savoie Mont Blanc, 2020. Français. ⟨NNT : 2020CHAMA051⟩. ⟨tel-03483106⟩



Record views


Files downloads