A Study About Residential Building Stock Energy Modelling

Energy is fundamental for social development and has a significant strategic position in the national economy.As one of the most important building types and the main living environment for households,the residential building accounted for 58.5%of the total floor area in China,its operating energy consumption has reached 48%of the total building energy consumption.China is paying great attention to the global issue of climate change,reducing carbon emissions and delaying climate change had been placed at an important position in the country's long-term development strategy.However,the indoor thermal environment in the hot summer and cold winter zone is relatively poor.With the economic growth and the improvement of people's living standards,the residents in the hot summer and cold winter zone are desperate to improve the indoor thermal condition,which is likely to put tremendous pressure on national energy conservation and emission reduction targets.In order to calculate the residential building stock space heating and cooling energy consumption for both the current stage and various future development scenarios.As well as for seeking and exploring suitable energy saving and emission reduction measures for the residential building stock space heating and cooling energy consumption.It is necessary to develop a stable and reliable residential building stock energy consumption models.A large number of scholars have studied and analyzed the energy consumption of residential buildings in hot summer and cold winter zone,especially the heating and cooling energy consumption through computer simulation and field investigation.They also analyzed and compared building energy consumption under different building design parameters by computer simulation.However,research in building stock energy consumption is still lacking.The existing researches only studied a limited number of residential buildings,so it is almost impossible to obtain the actual energy consumption of the residential buildings stock or calculate the total energy consumption of the study area based on them.Currently,researches about the impact of residential building renovation measures are limited to the study of one single building,so the effects of different renovation measures applied to all residential buildings in the studied stock are not analyzed or discussed.However,in reality,the residential buildings energy conservation and emission reduction policies generally covered a quite wide area.The analysis of the impacts of energy conservation and emission reduction policies on individual building or individual household cannot represent the residential building stock in the studied area.Therefore,by taking Chongqing,a municipality in the hot summer and cold winter zone as a case study.In this thesis,regional residential building energy consumption stock models are developed.Those models can estimate the heating and cooling energy consumption of all residential buildings in the regional scale and evaluate the stock results of different energy conservation and emission reduction policies in the study area.At the same time,due to the inherent complexity and the high cost of modeling and calculation,this research also explores the feasibility of using machine-learning methods instead of detailed building energy computer simulation software to predict building energy consumption.This thesis first presented a study about the energy consumption of residential buildings in the survey area with a total area of 3.4 square kilometers in Yuzhong District of Chongqing City by using the building-by-building aggregation approach.The bottom-up building energy consumption stock model was built by collecting detailed building information through field survey,building locating and modeling with the help of an online map,and a stock energy simulation software to analyze the buildings one by one.The calculation results of the building energy consumption stock model show that the heating energy usage intensity of 321 residential buildings constructed before2001 is in the range of 16.5 to 25.5 kWh/m~2,with a median of 19.2 kWh/m~2;The range for cooling energy usage intensity is between 14.8 and 33.4 kWh/m~2,with a median of20.4 kWh/m~2.Secondly,based on the shape characteristics of residential buildings(including building height,aspect ratio,and compactness ratio),the cluster analysis was used to generate the residential archetype buildings for the stock energy consumption calculation.The result shows the residential archetype buildings generated by selecting the building height and the building aspect ratio as the clustering factors and used K-medoids clustering technique are the most accurate in calculating the stock energy consumption.Compared with the energy consumption results of the residential stock calculated via building-by-building aggragation approach,the relative error for space heating and cooling energy consumption is only 1.55%.Then,this thesis uses the archetype approach to construct a bottom-up engineering building stock energy model for studying the heating and cooling energy consumption of Chongqing urban residential buildings and its corresponding carbon emissions.By combining the statistical data to defining the archetype buildings,simulating the energy consumption of the archetype buildings,calculating the proportion of the residential building floor area completed in different construction ages and considering weather adjustment to energy intensities,a building stock energy model development method had been proposed.The building stock energy model can be used to analyze heating and cooling energy consumption as well as corresponding carbon emissions at both the current stage and future scenarios.Moreover,the energy conservation and emission reduction effects of different renovation measures applied to urban residential buildings in Chongqing were also evaluated.In order to simplify the process of detailed modeling and calculation using building energy simulation software,the building thermal-physical characteristics and occupant behavior are selected as predictors while machine learning methods are used to predict the annual heating and cooling load intensities of residential buildings.Based on the annual heating and cooling load intensities database generated by the EnergyPlus energy simulation software,five machine learning regression models,including linear kernel support vector regression,polynomial kernel support vector regression,Gaussian radial-basis function kernel support vector regression,linear regression,and artificial neural network were tested and compared for their prediction performance.The Gaussian radial-basis function kernel support vector regression model has the best prediction performance compares to all other models,as it has shorter training time and higher accuracy.