Master's Thesis Laura Burgaud


Development of a generic data-driven tool for optimization and fault detection of Combined Heat and Power Units

Overview of the structure of the tool for optimization and fault detection Copyright: EBC Overview of the structure of the tool for optimization and fault detection

Thanks to their efficient use of fuel compared to separate production of heat and electricity, Combined Heat and Power (CHP) units are a key technology to help reduce primary energy consumption and gas emissions. Cogeneration installations being complex, an automatic monitoring and, where necessary, optimization of the system is required, especially to detect and predict faulty or inefficient operation. Scientific research has been carried out on single units, however they do not allow for a scaling analysis. A generalisation would save time and cut costs on the future optimization of units. The goal of this thesis is therefore to develop a generic tool for fault detection and diagnosis (FDD) on cogeneration systems including their heating system integration. This work explores the selection and programming of the fault detection method and the validation through tests on timeseries from simulations. The analysis of the system operation is conducted through the selection of scalable key performance indicators (KPIs). The results of these calculations are evaluated by a fuzzy rule-based fault detection algorithm that triggers alarms if a faulty behaviour is observed. Finally, the quality of the fault detection is tested on timeseries from a Modelica model of a CHP unit in a heating loop, on which several faults are triggered. The FDD tool is included in the already existing aedifion platform as part of the analytics framework.