A systematic procedure to study the influence of occupant behavior on building energy consumption

Abstract

Efforts have been devoted to the identification of the impacts of occupant behavior on building energy consumption. Various factors influence building energy consumption at the same time, leading to the lack of precision when identifying the individual effects of occupant behavior. This paper reports the development of a new methodology for examining the influences of occupant behavior on building energy consumption; the method is based on a basic data mining technique (cluster analysis). To deal with data inconsistencies, min–max normalization is performed as a data preprocessing step before clustering. Grey relational grades, a measure of relevancy between two factors, are used as weighted coefficients of different attributes in cluster analysis. To demonstrate the applicability of the proposed method, the method was applied to a set of residential buildings’ measurement data. The results show that the method facilitates the evaluation of building energy-saving potential by improving the behavior of building occupants, and provides multifaceted insights into building energy end-use patterns associated with the occupant behavior. The results obtained could help prioritize efforts at modification of occupant behavior in order to reduce building energy consumption, and help improve modeling of occupant behavior in numerical simulation.

Introduction

The identification of major determinants of building energy consumption, together with a thorough understanding of the impacts of the identified determinants on energy consumption patterns, could assist in achieving the goal of improving building energy performance and reducing greenhouse gas emissions due to the building energy consumption. In general, the factor influencing the total building energy consumption can be divided into seven categories:

• Climate (e.g., outdoor air temperature, solar radiation, wind velocity, etc.),

• Building-related characteristics (e.g., type, area, orientation, etc.)

• User-related characteristics, except for social and economic factors (e.g., user presence, etc.),

• Building services systems and operation (e.g., space cooling/heating, hot water supplying, etc.),

• Building occupants’ behavior and activities,

• Social and economic factors (e.g., degree of education, energy cost, etc.), and

• Indoor environmental quality required.

Among these seven factors, social and economic factors will partly determine the occupant attitude toward energy consumption, and building occupants will embody such impact on their daily activities and behavior, thereby influencing building energy consumption. At the same time, indoor environment quality could be regarded as being basically decided by building occupants, thereby influencing building energy consumption. In essence, these two categories of factors which represent occupants’ influences affect building energy consumption indirectly. Therefore, their influences on building energy consumption are already contained within the effects of occupant behavior, and there is no need to take them into consideration when identifying the effects of influencing factors.

The separate and combined influences of the first four factors on building energy consumption can be identified via simulation. With a variety of parameter settings, current simulation software is robust in respect to simulating different situations based upon these four factors. However, it is difficult to completely identify the influences of occupant behavior and activities through simulation due to users’ behavior diversity and complexity; current simulation tools can only imitate behavior patterns in a rigid way. In recent years several models have been established to integrate the influence of building occupant behavior into building simulation programs [1], [2], [3], [4]. However, these models focus only on typical activities such as the control of sun-shading devices, while realistic building user-behavior patterns are more complicated.

A number of studies [5], [6], [7] suggest that, in order to obtain the full effects of user behavior, one possible approach is to extract corresponding useful information from real measured data, since such data already contains the full effects. For example, Yu et al. [7] proposed a decision tree method for building energy demand modeling, and applied this method to the historical data on Japanese residential buildings. The generated model has a flowchart-like tree structure, enabling users to quickly extract useful information on the influence factors of building energy consumption. Such model along with derived information could benefit the improvement of building energy performance greatly. Generally, the previous studies on the effects of occupant behavior can be divided into two categories. The first category focuses on the effects of building user presence on building energy consumption. For example, Emery and Kippenhan [8] reported a survey on the effects of occupant presence upon home energy usage in four nearly identical houses. The four houses were divided into two pairs, and the building envelope of one pair was constructed with improved thermal resistance. One of each pair of houses was left unoccupied, while the other was occupied by university student families. Researchers compared the first heating season's (1987–88) total energy consumption of the occupied and unoccupied houses (i.e., the sum of heating, lighting, and appliances). They found that the presence of occupants increased the total energy consumption of both occupied houses, and the house with the improved building envelope had a smaller increase. The second category focuses on the effects of actions occupants took to influence energy consumption. For example, Ouyang and Hokao [9] investigated energy-saving potential by improving user behavior in 124 households in China. In this study, these houses were divided into two groups: one was educated to promote energy-conscious behavior and put corresponding energy-saving measures into effect in July 2008, while the other was required to keep behavior intact. Comparisons were made between monthly household electricity uses in July 2007 and July 2008 for both groups. Researchers found that, on the average, effective promotion of energy-conscious behavior could reduce household electricity consumption by more than 10%. Evidently, comparative analyses on measured data were conducted in these studies to identify the effects of user behavior. However, the limitations of this method are significant. First, apart from user behavior, the other four influencing factors also contribute to the variation in building energy consumption simultaneously, while this method is unable to adequately remove the effects of those four factors and identify the influences of occupant behavior. Although in these studies some measures were implemented to remove the impact of those factors, such as using nearly identical housing characteristics and taking energy data in other years with similar climatic conditions as a reference, the effects of these measures are questionable since even a slight difference in some building parameters (e.g., heat loss coefficient) and weather parameters (e.g., annual average outdoor air temperature) would result in remarkable fluctuations in the building energy consumption. Second, in real building databases, buildings are usually described by a mixture of variable types such as numerical variable, categorical variable (e.g., residential building types are divided into detached and apartment), and ordinal variable (e.g., buildings are rated as platinum, gold, and silver). Such data of mixed variable types is difficult to process by statistical methods that are normally utilized in comparative analyses. This also adds the difficulty of distinguishing between building-related effects and user-related effects. Third, with regard to comparative analyses, buildings are usually classified into different groups to simplify research. Such classification is commonly based on building-related parameters, such as floor area. For example, if building floor area ranges from 100 m² to 400 m², it can be replaced by small, medium, and large corresponding to the intervals [100, 200], [200, 300], and [300, 400], respectively. Accordingly, all the buildings are classified into three groups, i.e., small buildings, medium buildings, and large buildings; and further study can be performed on each group. In this process, the partition of building-related parameters is normally decided by considerations of convenience and intuition. Why should 200 m² and 300 m² be the interval between each group? Hence, a more rational classification method for grouping buildings is required.

Moreover, buildings are commonly represented by various typical parameters at the same time, such as building age and floor area. All these parameters may be divided into different levels, such as low and high, for simplicity. In order to perform a comprehensive investigation, the sample size (i.e., number of buildings) necessary for research should be determined by the combination of different levels of all parameters. For example, suppose seven typical parameters are selected for representation and each are stratified into 3 levels (e.g. small, medium, and large). In terms of combinatorial theory, it can be calculated that at least 3⁷ = 2187 buildings should be investigated for comparison, which may be quite impractical.

The main purpose of this paper is to develop a methodology for identifying the effects of occupant behavior on the building energy consumption through data analysis, thereby evaluating the energy saving potential by improving user behavior and providing deep insights into the building energy consumption patterns.

This paper is organized as follows: Section 2 introduces the proposed methodology. Section 3 describes the results of applying this method to a set of field measurement data and discusses the related work. Section 4 concludes the paper.