Proceedings of BS2015: 14th Conference of International Building Performance Simulation Association, Hyderabad, India, Dec. 7-9, 2015.

Wei Tian1,2, Lai Wei1,2, Pieter de Wilde3, Song Yang1,2, QingXin Meng1

1 College of Mechanical Engineering, Tianjin University of Science and Technology, Tianjin,China
2 Tianjin Key Laboratory of Integrated Design and On-line Monitoring for Light Industry & Food Machinery and Equipment, Tianjin 300222, China
3 School of Architecture, Design and Environment, Plymouth University, Plymouth, UK

As energy data for buildings at urban scales becomes more widely available, it is possible to conduct spatial analysis for these energy data in order to understand and subsequently better model spatial patterns of energy use. A fundamental concept for spatial phenomena is spatial autocorrelation to explore how building energy data may be related to its location and neighbourhood.

This study uses London as a case study to apply global and local spatial autocorrelation analysis for domestic electricity and gas at two different scales. The results, using Moran’s I statistics for determining the degree of spatial autocorrelation, indicate that there are positive global spatial autocorrelations for both electricity and gas use in terms of per person or per household at two spatial scales. This means that spatial areas with similar energy use intensity would cluster together in London. Furthermore, the Moran’s I statistics from bivariate analysis indicate that there is an apparent positive spatial autocorrelation
between gas use in one area and electricity use in its neighbouring areas.

As building energy data in urban areas is becoming increasingly available, there is an increasing amount of research on patterns of urban building energy use (Taylor et al., 2014; Caputo et al., 2013; Choudhary, 2012). Various methods have been used in detecting patterns of building energy in urban environment (Tian et al., 2014; Choudhary and Tian, 2014).

The research methods can be categorized into exploratory data analysis and model-based inference analysis from a statistical perspective (Tian and Choudhary, 2012; Howard et al., 2012; Anselin, 2012). This exploration and the resulting understanding is a prerequisite for urban simulation efforts.

A model-based inference analysis is often involved in creating an engineering-based or statistical energy
model to investigate the relationship between building energy and explanatory variables in urban settings (Tian et al., 2014; Bourdic and Salat, 2012).

In contrast, explanatory data analysis is used to summarize the main characterises of a data set using both numerical and visual methods (Fischer and Getis, 2010). This research concentrates on the latter analysis. In the context of urban building energy assessment, GIS (geographical information system) can play an important role in exploring the trends of energy consumption for urban buildings, often called ESDA (exploratory spatial data analysis). It should be emphasized that it is only a preliminary step in ESDA to map the data in a GIS environment. Spatial autocorrelation is a central concept in the field of
studying spatial phenomena (Fischer and Getis, 2010). For spatial data, a variable may be related to its location; such a relation is called spatial autocorrelation or spatial dependence.

For urban building analysis, these approaches can be used to relate energy use for buildings to the buildings’ location or neighbourhood. The reasons for this spatial autocorrelation may include similarities between resident behaviours in one local
neighbourhood, local urban microclimate conditions, similar income and expense in one small area, or the same building technology being used in nearby areas.

It is necessary to understand these spatial patterns of building energy in order to make informed decisions on energy saving measures implemented in urban areas. However, there is only very limited research on this topic. Tian et al. conducted preliminary exploratory spatial analysis on energy use in London (Tian et al., 2014).

This paper implements spatial autocorrelation analysis to identify spatial patterns (clustering or dispersion) of energy use in urban areas. This study uses London as a case study to demonstrate the application of global and local spatial autocorrelation methods for domestic electricity and gas use. Both univariate and bivariate spatial analysis are conducted to explore the clusters of energy use for urban buildings. The results of spatial analysis may depend on the spatial scale used. Hence, two spatial scales are used in this research to provide robust results that will be explained in detail in section of
“Data and Methods”.

This paper is structured as follows. The energy used for exploratory spatial data analysis is introduced first. Then two types of statistical methods are briefly presented: spatial weight matrices and (global/local) spatial autocorrelation. The results are discussed for ...