« Development of simplified RC models with physically deducible parameters using the time-constant concept for the calculation of the heating load demand in single detached dwellings and districts »
Buildings are one of the main energy sinks in European countries. They consume more than 40% of the total energy and they are responsible for almost 36% of carbon emission. The 2030 climate and energy framework include at least 40% cuts in greenhouse gas emissions and 32.5% improvement in renewables and energy efficiency compared to projections of future energy consumption based on the current criteria. In this context, the European Union introduces a research and innovation program to achieve a low-carbon and climate-resilient future. This program asks for further development, demonstration and validation of breakthrough technologies for decarbonization in buildings and districts as one of the major categories.
In European countries 50% to 70% of total used energy in buildings is consumed for heat generation. Due to the importance of residential buildings for energy consumption and CO2 emission, different tools and software are developed to simulate and to determine energy needs at a district level. Most of the available tools calculate the model parameters deterministically, which constraints the model precision due to available information and it also can cause problems for the post-validation of results. On the other hand, data-driven models require extensive measurements and datasets and the lack of infrastructures for providing and accessing to such data confront designers with difficulties to expand the application of data-driven models for a large number of buildings in a district.
In this context, we introduce grey-box models as an alternative approach for generating models of buildings and districts. For this purpose, a hierarchy of simplified thermal models is developed in order to justify the simplest structure that can represent the thermal behaviour of a simulated/real building with relatively good precision. The main features of the selected model are to predict the indoor temperature with a root mean square error of less than 1 °C and to determine maximum heat power and total heat demand with less than 25% of error compared to simulation results in TRNSYS and measurements of a real building case study.
Moreover, four main geometric features of a building including the surface area, width-to-depth ratio, windows-to-floor ratio, and orientation angle in heavy and light structured buildings are studied to understand how effectively the geometric features act on the developed model. Afterwards, the grey-box model is developed based on the RC model approach but instead of using deterministic or estimated parameters in the model, it is proposed to use the UA-values and the time constants of the building to calculate model’s parameters. At this stage of research, different time constants in two of the developed RC models are evaluated from statistical analysis of estimated parameters of a large number of fictitious buildings in TRNSYS. Using the developed technique in this research expands RC model applications to model single detached dwellings and districts when there is not enough information to make
detailed models of a district. Moreover, the proposed approach in this thesis binds the determination of parameters in a specific grey-box model with the obtained results from statistical analysis of estimated parameters of the same model from the thermal performance of a large number of buildings. Estimated thermal resistances and capacitances are obtained from time constants and distribution ratios of thermal resistances in a specific building model for determination of parameters.
The proposed approach has been achieved from the simulation of more than 400 simulated building in TRNSYS and it has been validated with the twin house case in Holzkirchen in Munich. Finally, a grey-box model of a small neighbourhood for Pic-au-Vent eco-district in Tournai is generated by means of the proposed approach and the achieved results with this approach are quite comparable with TRNSYS simulations and data-driven models. It can be concluded, the proposed approach in this research can be considered as an alternative technique for the energy assessment of buildings and districts in the early stage of design in order to make fast and accurate model from a limited amount of information about buildings structure and geometry or when detailed information of a building or district is not accessible.