Urban form,building characteristics,and residential electricity consumption: A case study in Tainan City

Abstract

Which urban form factor most affects household electricity consumption? This study investigated the relationships between urban density, community layout, and land use factors and household electricity consumption simultaneously, along with building characteristics and demographic indicators. The study site involved 231 communities located in the former provincial area of Tainan City, Taiwan. Due to the area’s subtropical climate, air conditioning accounts for approximately 40% of the total yearly household electricity consumption. Of the urban form factors examined, greater population density was most strongly associated with lower household electricity consumption, followed respectively by greater urban canyon narrowness, or higher height to width ratios, and greater percentages of vacant space and building land use. Notably, both urban canyons and building land use percentages were associated with decreased consumption only after increasing past threshold levels, specifically a 1.5 height to width ratio and 40.7%, respectively. In addition, building characteristics, namely smaller household living areas and greater building age, were most strongly connected with lower household electricity consumption. In contrast, larger household living areas were linked with decreased household electricity consumption/floor area, revealing the importance of lower energy intensities of sizable scales. Of the demographic indicators studied, higher percentages of older adults were associated with lower household electricity consumption. Concerning urban form, the findings suggest that to reduce residential energy usage in a subtropical climate, buildings should be clustered to maximize the inter-building shadows resulting from narrower urban canyons, while simultaneously increasing non-built land use percentages in the adjacent areas.

Keywords

East Asia land use residential space sustainability urban form

Introduction

How much does urban form influence residential electricity consumption? With urban and regional policies increasingly targeting the reduction of greenhouse gas emissions in recent years, researchers have begun examining urban planning policies that affect energy usage in buildings. The majority of these studies have focused on urban density (Ko, 2013; Mitchell, 2005), with researchers analyzing how housing size, dwelling type, and residential density are related to energy consumption (Andrews, 2008; Ewing and Rong, 2008; Kaza, 2010). Other studies have examined how community layout features, such as urban canyons and building orientation, and also different land uses affect residential energy use. Most of the community layout studies have concentrated on relatively small sections of cities, using either real data (Lin et al., 2004) or, more frequently, computer simulations involving hypothetical city blocks (e.g. Allegrini et al., 2012; Futcher et al., 2013; Pisello et al., 2012; Strømann-Andersen and Sattrup, 2011; Stupka and Kennedy, 2010; Wong et al., 2011) and replicated city blocks (e.g. Chen and Wong, 2006; Liao and Jhou, 2013; Ratti et al., 2005). Additionally, the land use studies have primarily documented the beneficial impacts of city parks and green spaces on building energy usage (Ca et al., 1998; Wong et al., 2011) and urban temperatures (Bowler et al., 2010; Huang et al., 2008; Steeneveld et al., 2011).

A large number of studies have also been conducted at the site scale, concentrating on how building features and occupant demographics influence residential energy consumption (Pacheco et al., 2012). These studies have been conducted using both survey data and computer simulations to uncover relationships between these factors and consumption (Ko, 2013; Pacheco et al., 2012; Zhao and Magoulès, 2012).

However, to the best of our knowledge, there has not been a study that has simultaneously examined how urban density, community layout, land use, building, and demographics features affect household electricity consumption. Especially lacking is a study that uses real-world data representing community layout factors in a large urban region.

Study site

The study site is located in Tainan City, which is located along the southwestern coast of the country of Taiwan. The average household electricity consumption of the 231 communities located within the former provincial Tainan City boundaries were investigated (see Figure 1). These communities constitute the smallest administrative areas for which government data could be obtained. The communities located farther from the center of the city, primarily to the north and east, were not included in this study. These latter communities are more rural and less developed, and their municipal land regulations differ from those more centrally located. Tainan City has a subtropical climate characterized by mild, dry winters and hot, humid summers, with a dry season running from October to March and a rainy season from April to September (Tainan City Guide, 2015; Tourism Bureau of Tainan City Government, 2016).

Figure 1.

Average household electricity consumption of each of the 231 communities within the former provincial district boundaries of Tainan City, for the year 2010 (Environmental Protection Administration, 2010). The right figure depicts the location of the provincial portion of Tainan City.

In the country of Taiwan as a whole, 90% of the households have at least one air-conditioner and on average there are at least two air-conditioners in every household (Huang, 2016; Statistica, 2016). Almost all of these residential air-conditioners are single-room units regardless of income levels, with central air units primarily found only in commercial buildings. In addition, due to the relatively warm winters, most homes do not have central heating. In the year 2015, the national average monthly household electricity consumption peaked in August at about 400 kWh/month, while the lowest consumption of about 230 kWh/month was in December, a 42.5% reduction (see Figure 2; Taiwan Power Company, 2016). These air-conditioning percentages and numbers, and electricity consumption seasonal differences may be even greater in Tainan City, due to this city’s hotter, southern location. For example, a study conducted in Tainan City found that air-conditioner usage was responsible for the largest percentage of residential electricity consumption during the six month air-conditioning season (Lai and Wang, 2011). In cities with similar subtropical climates, such as Hong Kong, air-conditioning has been found to be the single largest electricity consuming item in households, accounting for 37% to 46% of the annual consumption (Lam, 1996; Tso and Yau, 2003).

Figure 2.

National average monthly household electricity bill and the corresponding average monthly household electricity consumption, from February 2015 to January 2016 (Taiwan Power Company, 2016).

Urban form

A number of studies have been conducted on the relationships between urban density, community layout, and land use on energy consumption. Higher density cities and those with more multi-family housing have consistently been associated with lower residential energy consumption per household (Ewing and Rong, 2008; Kaza, 2010; Ko, 2013; Pitt, 2013). Concerning community layout, much of this research has involved investigating urban canyons and building orientation. Urban canyons are defined geometrically as the height to width (H/W) ratio of the space between adjacent buildings (Strømann-Andersen and Sattrup, 2011; see Figure 3). For any building orientation, cooling demands have been found to be lower in narrow urban canyon configurations than for buildings in wide urban canyons. In narrower canyons, the entrance of solar shortwave radiation is decreased during the day due to inter-building shadowing (Allegrini et al., 2012; Pisello et al., 2012; Strømann-Andersen and Sattrup, 2011). In fact, the shadowing effect of buildings has been found to be much greater than urban heat island (UHI) effects (Futcher et al., 2013; Yang et al., 2011). Related studies have been conducted in tropical climates that examined outdoor temperatures. Using field measurements or computer simulations, these studies have revealed that narrower urban canyons (Emmanuel et al., 2007; Krüger et al., 2011) and taller buildings (Kakon et al., 2010) were associated with lower outdoor temperatures and greater thermal comfort.

Figure 3.

An urban canyon in Tainan City that is neither wide nor narrow, but where the average heights (H) of the buildings along the street are approximately the same as the street width (W).

In addition, higher albedo surfaces in street canyons have been found to have only a minimal effect on cooling demands, because these surfaces also lead to an increase of radiation entrapment (Allegrini et al., 2012; Pisello et al., 2012; Yang et al., 2012). However, such surfaces have been shown to affect the availability and distribution of daylight, and the associated energy use for artificial lighting in buildings (Strømann-Andersen and Sattrup, 2011).

Building orientation research has revealed that the best orientation for reducing energy consumption depends on the local climate. In the subtropical climate experienced in Tainan City, Taiwan, studies have found that west-facing housing used more electricity than those facing other directions, followed by housing facing east, south, and north, respectively (Lai and Wang, 2011; Lin et al., 2004).

The most important connection between land use and building energy consumption concerns the presence of green areas, such as city parks and gardens. Research has consistently revealed that nearby green areas can reduce building energy consumption (Ca et al., 1998; Wong et al., 2011) and that such areas can lower the surrounding urban air temperature from 0.5 km up to over 1 km away (Chen and Wong, 2006; Lin et al., 2015; Upmanis et al., 1998).

Building characteristics, demographic factors, and occupant behavior

Researchers have discovered that the energy performance of buildings with similar functions can display a 20-fold variation. The most important factors involve building design features (2.5× variation or about 30.6% of the total variation), followed by systems efficiency (2× variation or about 23.1%) and occupant behavior (2× variation). These researchers have also speculated that urban form (e.g. density, community layout, land use) may be largely responsible for the remaining 2× variation that has yet to be completely explained, and have documented through computer simulation that such features can affect energy use by almost 10% (Baker and Steemers, 2000; Ratti et al., 2005).

Important building characteristics linked with greater energy consumption include larger household living areas, older buildings, and detached dwellings (e.g. single-family homes) as compared with multi-family dwellings (e.g. townhouses, apartments, condominiums; Kaza, 2010; Min et al., 2010; Nelson et al., 2012; Stupka and Kennedy, 2010; Yun and Steemers, 2011). The average household energy utilized per floor area, though, has been found to decrease in residential homes as floor areas increase. The reason for this involves what might be termed the lower energy intensities of sizable (larger) scales (LEISS), with similar household appliances being used in homes that differ in size (Kaza, 2010; Lai and Wang, 2011; Nelson et al., 2012; Touchie et al., 2013). Concerning systems efficiency, ownership of air-conditioners and clothes dryers (Chen et al., 2010; Min et al., 2010; Tso and Yau, 2003, 2007), central as compared with individual room air conditioners (Nelson et al., 2012), and electric water heaters as compared with natural gas or solar (Chen et al., 2010; Hu et al., 2013) have been associated with greater electricity and overall energy usage.

Last, demographic factors and occupant behavior have been shown to influence levels of energy consumption. For example, higher income, larger family size, and larger numbers of non-adult family members have been consistently connected with greater energy consumption (Druckman and Jackson, 2008; Hu et al., 2013; Tso and Yau, 2003; Vassileva et al., 2012; Yun and Steemers, 2011). In addition, energy-saving attitudes and behaviors of building occupants have been found to affect energy consumption (Bartusch et al., 2012; Faruqui et al., 2010; Kang et al., 2012; Kim et al., 2013; Ouyang and Hokao, 2009; Teŕes-Zubiaga et al., 2013), with some researchers concluding that such attitudes and behaviors are the second most dominant factor in determining household energy use following climate (Yun and Steemers, 2011).

Lack of corresponding government data

With regard to urban form, building characteristics, demographic factors, and occupant behavior, government information concerning albedo surfaces, types of dwelling (e.g. single-family homes, townhouses, apartments, condominiums), ownership of different types of home appliances and heating and cooling systems, and the energy-saving behaviors of household occupants in each community was not available. Concerning the validity of the results of this study, it was hoped that the large number of households considered would randomly distribute all such potentially confounding variables across the 231 communities examined.

Objectives

The main objectives of this study, which can be considered a type of ecological study in which groups are studied rather than individuals, are to first identify associations between household electricity consumption and urban form factors, building characteristics, and demographic indicators. The following gaps in the current understanding of urban form were investigated:

Which urban form factor has the strongest relationship with energy usage? Studies have primarily examined urban density, community layout, or land use factors separately without considering all three simultaneously.

Is household electricity consumption on a citywide scale more dependent upon urban form factors or building characteristics? Most studies have either surveyed or computer simulated only small sections of a city.

At different urban densities, are different urban form factors and buildings characteristics more strongly related to household electricity consumption?

Second, the insights gained by investigating these associations will be used to formulate preliminary hypotheses that can be investigated in future studies.

Method

Dependent variables

Two dependent variables were examined separately in this study, namely average household electricity consumption and average household electricity consumption/floor area. The average household electricity consumptions for each of the 231 communities in the study site for the year 2010 were obtained by selecting residential cases only from the Environmental Protection Administration (2010) website. The overall average for these communities was 4661 kWh/year, with a minimum and maximum of 298 to 7260 kWh/year, respectively, and a standard deviation of 1208 kWh/year. The year 2010 was chosen because this was an official census year, which resulted in the most recent complete database in the past five years.

The second dependent variable, the average household electricity consumption/floor area for each of the 231 communities for the year 2010, was calculated by dividing the average household electricity consumption of each community by its corresponding average household living area (floor area). The data concerning the average household living area for each community were obtained from the Ministry of the Interior (2011). The overall average of these communities was 107.2 kWh/m²/year, with a minimum and maximum of 6.4 to 198.9 kWh/m²/year, respectively, and a standard deviation of 27.9 kWh/m²/year.

Explanatory variables

The explanatory variables were categorized as urban form factors, building characteristics, and demographic indicators. These data were obtained from the Taiwan national and Tainan City government databases, as referenced in Figure 4. Averages rather than the median values corresponding to each community in the study site were the only information available. The validity and uncertainties of this information and the derived variables are unknown. Box plots displaying the data distribution of these variables, other than building orientation, are shown in Figure 4. Building orientation is displayed in an orientation rose in Figure 5.

Figure 4.

Box plots of the explanatory variables involving urban from factors, building characteristics, and demographic indicators corresponding to the 231 communities.

Figure 5.

Orientation rose displaying the average building orientations, for the four cardinal and four intermediate directions, of all of the residential buildings within the 231 communities.

The urban form factors consisted of three subcategories, namely, urban density, community layout, and land use. Urban density was measured with population (people/hectare), residential (people/hectare of residential area), and building densities (buildings/hectare), and floor area ratio (FAR), which is the ratio of the total area on all floors of a given residential building to the size of the land on which this building is constructed. The community layout variables included urban canyons and building orientation. The land use variables were comprised of the percentages of the land in each community occupied by buildings of any type, housing, commercial buildings, green areas (e.g. agriculture, forests, parks, gardens), water, and vacant spaces (e.g. vacant land, abandoned military property with no buildings or trees, wasteland).

The residential building characteristics studied included household living area (floor area), building age, building height, and the percentage of buildings made out of brick in each community. Approximately 90% of the homes in Tainan City are built out of reinforced concrete, with the rest being built primarily out of brick. Most of the reinforced concrete homes consist of concrete frames and brick infill walls, with the newer homes consisting of both concrete frames and concrete infill walls (Hsieh and Forster, 2006; Yin, 2000). Last, the demographic indicators examined included average household income (after-tax household income), household size (number of people per household), and percentage of older adults (percentage of people over 65).

The values of some of the urban form factors, specifically urban canyons, building orientation, and the land use percentages, and one of the building characteristics, namely building height, were calculated with the ArcGIS 10.0 software. The average urban canyon values for each community were estimated as follows:

Step 1. For each building in the community, the corresponding street width (W) was estimated by subtracting the distance between the centroid of each building’s footprint to the nearest property line, from the distance between this centroid to the nearest street centerline. This value was then doubled.

Step 2. Each building height (H) was divided by the corresponding street width (W). Building height was approximated by multiplying the total number of floors, provided on a GIS map, by 3 m, which is the average height of one floor in Taiwan’s residential buildings.

Step 3. A weighted arithmetic mean of these H/W values was calculated for each community, weighted by the corresponding area of each building’s footprint.

This estimation of urban canyon values assumed that the distances from the front of each building to its front property line and building depths were equal, and that there was a building on the opposite side of the street. The estimate used in Step 1 resulted in negative distances and abnormally large distances for 22.3% of the 92,352 residential buildings located in the study site. These distances and corresponding buildings were removed from the database before proceeding to Step 2.

Building orientations were estimated as follows:

Step 1. For each building in the community, the direction of the line from the centroid of each building’s footprint to the nearest street centerline was determined. This direction was assumed to be perpendicular to the front façade of a building.

Step 2. A weighted arithmetic mean of these directions was calculated for each community, weighted by the corresponding total floor area of each building.

Step 3. The percentages of these directions falling into each of the eight cardinal and ordinal directions for a given community were calculated.

Statistical analyses

First, multiple linear regression was used to determine the urban form factors, building characteristics, and demographic indicators significantly associated with the average household electricity consumption and average household electricity consumption/floor area of each community. Multiple linear regression was utilized to gain insights into the relative strengths of the associations between multiple explanatory variables and these two dependent variables. Polynomial regression models, a special type of multiple linear regression, were used to better capture the relationships between these explanatory variables and these two consumptions.

Second, the elasticity and marginal effects of each significant explanatory variable with respect to both types of consumption were calculated, to provide additional information concerning the relative strengths of the connections between the explanatory and dependent variables. These elasticities and marginal effects were averaged across the 231 communities and also subgroups of these communities, based on whether the population densities were below or above the median population density. The elasticity formulas utilized in these calculations are listed in Appendix 1.

Results

Urban form factors

The regression results revealed that one or more variables from each of the three explanatory variable categories were related to the first dependent variable, household electricity consumption, and that the explanatory variables connected with the second dependent variable, household electricity consumption/floor area, were a subset of the former (see Table 1). Within the urban form factors category, greater population density was related to household electricity consumption. This finding is in agreement with the great majority of studies, which have revealed that higher density residential communities are associated with lower household energy usage, for example Ko (2013), Ewing and Rong (2008), Kang et al. (2012), and Pitt (2013). Household electricity consumption/floor area, however, was neither associated with population density nor any of the other variables used to measure urban density.

Table 1.

Household electricity consumption, Panel A, and household electricity consumption/floor area, Panel B, regressed onto the urban form factors, building characteristics, and demographic indicators explanatory variables.

Category	Variable	Model used	Total model
Category	Variable	Model used	Β₁ (X)	t-value	Β₂ (X²)	t-value (VIF^a)
Panel A: Household electricity consumption
Urban form factors
Urban density	Population density	Quadratic	–		−0.02***	−5.23 (2.12)
Community layout	Urban canyons	Quadratic	1825.21**	2.90	−616.90*	−2.36 (11.49) (9.99)
Land use	Buildings	Quadratic	4236.39**	2.70	−5200.32*	−1.99 (19.23) (16.81)
	Vacant spaces	Quadratic	–		−2541.27*	−2.44 (1.18)
Building characteristics	Household living area Building age	Linear Linear	28.10**	3.20		(1.81)
Building characteristics	Household living area Building age	Linear Linear	−37.05**	−3.38		(1.60)
Demographic indicators	Older adults	Linear	−4299.58**	−2.71		(2.11)
Model summary						n = 231
F-value						17.33***
Adjusted R²						.39
Panel B: Household electricity consumption/floor area
Urban form factors
Community layout	Urban canyons	Quadratic	40.36*	2.49	−14.10*	−2.10 (11.54) (9.98)
Land use	Buildings	Quadratic	80.34*	2.00	−196.72**	−3.01 (18.89) (16.08)
	Vacant spaces	Quadratic	−45.31**	−2.68		(1.32)
Building characteristics	Household living area	Linear	−1.21***	−6.23		(1.35)
Model summary						n = 231
F-value						13.58***
Adjusted R²						.25

The high variance inflation values (VIF) for the quadratic models are caused by the inclusion of powers of the corresponding variables and does not adversely affect the results of the other explanatory variables.

p < .05; **p < .01; ***p < .001.

The urban canyon quadratic model revealed that both household electricity consumption and household electricity consumption/floor area increased with the narrowing of urban canyons up to the threshold ratios of 1.5 and 1.4, respectively, after which such consumptions decreased with increasing magnitude. To the best of our knowledge, this finding has not been reported in past studies. One explanation focuses on the shadow effects of buildings (Allegrini et al., 2012; Futcher et al., 2013; Pisello et al., 2012; Strømann-Andersen and Sattrup, 2011; Yang et al., 2011). Perhaps only in narrower urban canyons (larger H/W ratios) do these inter-building shadows become significant in lowering building cooling demands. Moreover, in agreement with the findings of other studies, urban canyon narrowness was found to be more important than building orientation or the presence of vegetation in reducing household energy usage (Allegrini et al., 2012; Pisello et al., 2012; Strømann-Andersen and Sattrup, 2011). Both building orientation and green areas were not associated with either dependent variable in our study.

Similarly, the building land use quadratic model indicated that both household electricity consumption and household electricity consumption/floor area increased until the percentage of land occupied by buildings reached the threshold percentages of 40.7% and 20.4%, respectively, after which such consumptions decreased with increasing magnitude. Narrower urban canyons were associated with larger building land use percentages in the communities studied (see Table 2). This association suggests that as building land use percentages continue to grow, the corresponding urban canyons become narrower, resulting in greater inter-building shadows, a lowering of building cooling needs, and a consequent reduction in household electricity consumption.

Table 2.

Correlations among the explanatory variables, including population density, urban canyons, and the land use, building characteristics, and demographic indicators, used in this study.

	Population density	Urban canyons	Buildings	Housing	Commercial buildings	Green areas	Water	Vacant space	Household living area	Building age	Building height	Brick material	Household income	Household size
Urban canyons	.45**
Buildings	.59**	.39**
Housing	.71**	.42**	.86**
Commercial buildings	.14*	.25**	.39**	.31**
Green areas	−.43**	−.40**	−.67**	−.68**	−.35**
Water	−.30**	−.20**	−.39**	−.40**	−.27**	.21**
Vacant space	−.18**	−.15*	−.26**	−.27**	−.19**	−.02	.02
Household living area	−.37**	−.16*	−.47**	−.37**	−.18**	.33**	.26**	.04
Building age	−.03	−.08	.35**	.20**	.20**	−.11	−.15*	−.05	−.37**
Building height	.42**	.85**	.39**	.39**	.33**	−.44**	−.17**	−.17*	−.17**	−.09
Brick material	−.22**	−.55**	−.28**	−.35**	−.26**	.41**	.07	.15*	−.23**	.26**	−.58**
Household income	.07	.29**	−.07	.07	.09	−.24**	−.16*	−.06	.15*	−.44**	.30**	−.39**
Household size	−.37**	−.35**	−.52**	−.42**	−.31**	.61**	.34**	−.08	.61**	−.11	−.36**	.03	−.16*
Older adults	.03	−.11	.38**	.17*	.14*	−.21**	−.23**	.15*	−.53**	.55**	−.12	.44**	−.38**	−.57**

p < .05; **p < .01; ***p < .001.

Building characteristics

The elasticity results revealed that building characteristics, including smaller household living areas and greater building age, were found to be more elastic and thus more strongly associated with lower household electricity consumption than urban form factors and demographic indicators (see the overall column in Panel A of Table 3). The importance of building characteristics with regard to energy use was previously found by research conducted on the smaller scale of city blocks (Baker and Steemers, 2000; Ratti et al., 2005). Smaller household living areas have also been associated with both lesser cooling and heating demands in a number of studies (Chen et al., 2010; Holden and Norland, 2005; Hu et al., 2013; Kaza, 2010; Ko, 2013; Ouyang and Hokao, 2009; Tso and Yau, 2007; Vassileva et al., 2012).

Table 3.

Elasticity and marginal analyses of the significant regression explanatory variables at different population densities corresponding to the dependent variables household electricity consumption, Panel A, and household electricity consumption/floor area, Panel B.

Category	Variable	Overall		Community^a
		Overall		Lower density		Higher density
		Slope	Elasticity	Slope	Elasticity	Slope	Elasticity
Panel A: Household electricity consumption
Urban form factors
Urban density	Population density	−4.02	−0.20	−01.76	−0.03	−6.31	−0.37
Community layout	Urban canyons	820.01	0.13	1003.41	0.15	635.01	0.11
Land use	Buildings	1290.89	0.01	2549.41	0.05	21.43	−0.04
	Vacant space	−432.26	−0.05	−631.13	−0.10	−231.67	−0.005
Building characteristics	Household living area	28.10	0.30	28.10	0.34	28.10	0.27
Building characteristics	Building age	−37.05	−0.25	−37.05	−0.25	−37.05	−0.25
Demographic indicators	Older adults	−4299.58	−0.15	−4299.58	−0.16	−4299.58	−0.13
Panel B: Household electricity consumption/floor area
Urban form factors
Community layout	Urban canyons	17.38	0.12	21.57	0.15	13.15	0.09
Land use	Buildings	−31.08	−0.20	16.52	−0.05	−79.11	−0.36
	Vacant space	−45.31	−0.06	−45.31	−1.00	−45.31	−0.02
Building characteristics	Household living area	−1.21	−0.59	−1.21	− 0.71	−1.21	−0.48

The communities were divided into two groups based on the median population density of the 231 communities used in this study, namely 129.5 people/ha.

The result of older homes being associated with lower household electricity consumption was, however, not expected. Many studies have concluded that newer buildings are more energy efficient than older buildings (Bartusch et al., 2012; Chen et al., 2010; Holden and Norland, 2005; U.S. Energy Information Administration, 2013). Nonetheless, there are two possible explanations for this contrary finding. The older homes in this study were found to have smaller household living areas as compared with those of the newer ones and to be occupied by residents with lower household income (see Table 2). Some researchers have concluded that regarding energy use, the size of the household living area was more important than the age of the housing (Holden and Norland, 2005). In addition, lower income residents have been found by many studies to be associated with lower household energy consumption (Chen and Ng, 2013; Druckman and Jackson, 2008; Min et al., 2010; Santamouris et al., 2007; Teŕes-Zubiaga et al., 2013; Vassileva et al., 2012; Yun and Steemers, 2011).

Concerning household electricity consumption/floor area, changes in household living area also had the strongest connection with this second dependent variable (see the overall column in Panel B of Table 3). However, increases in household living area were related to decreases in such consumption. This result is supported by the findings of other studies and demonstrates the importance of LEISS associated with larger homes (Kaza, 2010; Lai and Wang, 2011; Nelson et al., 2012; Touchie et al., 2013). In addition, no connections were found between household electricity consumption/floor area and any of the other building characteristics that this study examined.

Demographic indicators

A higher percentage of family members comprised of older adults in a given community was positively associated with lower household electricity consumption, but was not associated with household electricity consumption/floor area (see Table 1). Greater percentages of older adults were associated with smaller household living areas, older buildings, and lower household income (see Table 2). These relationships may explain why older adults were an explanatory predictor of household electricity consumption.

On the other hand, in addition to older adults, population density and building age were not found to be significant predictors of household electricity consumption/floor area. One can speculate that the influence of LEISS may be partly responsible, because household living area was negatively correlated with all three of these explanatory variables (see Table 2).

Lower and higher community population densities

At different urban population densities, different urban form factors and building characteristics were found to be more strongly related to household electricity consumption and household electricity consumption/floor area. At lower population densities, variables reflecting the beginning growth of built environments, specifically reductions in vacant spaces and increases in household floor area, were more closely associated with increases in household electricity consumption as compared with higher density communities (see Table 3). A similar relationship concerning reductions in vacant spaces was found for household electricity consumption/floor area. In contrast, increases in household living area were connected with greater decreases in household electricity consumption/floor area.

At higher population densities, variables representing even more densely built environments, specifically greater population density and building land use percentage, were connected with greater decreases in household electricity consumption, as compared with lower density communities. For household electricity consumption/floor area, a similar relationship was found for increases in building land use percentage (see Table 3).

Discussion

This ecological study examined real-world data corresponding to a citywide scale, and comprehensively examined the connections between a number of urban form factors, building characteristics, and demographic indicators with both household electricity consumption and household electricity consumption/floor area. The results provide valuable insights into both the three questions posed in the introduction and the hypothetical questions related to the findings that only further research can answer.

Summary of the results

First, of the urban form factors examined, greater population density was most strongly associated with the reduction of household electricity consumption. This was followed, respectively, by greater urban canyon narrowness after a threshold level was surpassed, and greater percentages of vacant spaces and building land use, after the latter surpassed a threshold level. Household electricity consumption/floor area was also associated with these latter three factors, and threshold levels involving urban canyons and building land use were also found.

Second, building characteristics, specifically smaller household living areas, had the strongest connection with lower electricity consumption, followed by greater building age. The importance of LEISS was also revealed, with increases in household living area being associated with decreases in household electricity consumption/floor area.

Third, with regard to urban population densities, the findings suggest that at the higher levels of population densities found in the communities studied, even greater densities could lead to an overall reduction in average household electricity consumption. On the other hand, at lower population densities, perhaps non-built areas should be safeguarded and larger household living areas should be restricted to reduce average household electricity consumption. Hence, different policies concerning these factors may be needed at different urban densities.

Hypothetical questions for further studies

Hypothesis #1: Population density was not related to household electricity consumption/floor area, because the effects of greater density are countered by reductions of LEISS associated with larger household living areas.

Most researchers agree that higher density residential communities are associated with lower household energy usage (Ewing and Rong, 2008; Kaza, 2010; Ko, 2013; Pitt, 2013). With regard to the urban form factors examined in this study, the results suggest that urban planners should concentrate first on developing policies that increase urban density. However, the underlying reasons why population density was not found to be connected with household electricity consumption/floor area need to be better understood. Are the reductions in LEISS due to the smaller household living areas of greater population density the primary reason for this lack of connection? Are other factors also involved? In addition, further studies are needed to more precisely determine the optimal population densities that correspond to both lower household electricity consumption and household electricity consumption/floor area in subtropical climates and other parts of the world. These studies will also need to address the effects that higher population density can have on transportation, industrial, and commercial energy demands, the UHI phenomenon, environmental issues involving air and water quality, and the conservation of natural environments, and the physical health and psychological well-being of the inhabitants.

Hypothesis #2: Urban canyons and building land use percentages must both first pass threshold values before they are associated with a reduction in both household electricity consumption and household electricity consumption/floor area. Below these thresholds, the effects of LEISS associated with larger household living areas are more important, with inter-building shadows not coming into play until the urban canyons become narrower.

Narrower urban canyons and larger percentages of land occupied by buildings of all types were associated with both lower household electricity consumption and lower household electricity consumption/floor area, but only after each surpassed respective threshold levels. In fact, even narrower urban canyons than those that currently exist in Tainan City may be warranted. Recent simulation studies investigating the outdoor thermal comfort of cities located in tropical climates have concluded that buildings as tall as 8 to 10 stories (Kakon et al., 2010) or even 100 m in height (Ndetto and Matzarakis, 2013) would provide optimal comfort.

Below these thresholds, the effects of LEISS are perhaps more important as household living areas become smaller with increasing population density, resulting in an initial increase in consumption until these threshold levels are reached. Additional studies must be conducted to better understand the impacts of LEISS, urban canyons, and built areas at both lower and higher population densities on residential energy usage in subtropical climates. These studies must also take into account the local effects that urban canyons have on wind, air and noise quality, and sunlight penetration.

Hypothesis #3: Clustering buildings will lower overall community household electricity consumption, because such clustering will result in greater population density, narrower urban canyons, and larger non-built areas.

The findings of this study suggest that urban planners in subtropical climates should concentrate on clustering buildings to reduce household electricity consumption and household electricity consumption/floor area. If properly designed, such clustering would result in higher population densities, narrower urban canyons, and greater building land use percentages in the built areas, as well as greater percentages of vacant space land use in the adjacent non-built areas. All four of these urban form factors were found to be associated with decreased household electricity consumption, and the latter three with reduced household electricity consumption/floor area. The fact that both greater percentages of vacant space and building land use were connected with lower household electricity consumption may on the surface appear contradictory. In addition, there may be just one or many clusters depending on the size and geographical layout of the city. Moreover, although not a significant urban form factor in this study, larger urban parks with more natural areas should be planned for in the non-built areas. Such parks, if large enough, have been found to have a cooling effect on the air temperatures that reach into the neighboring built areas (Lin et al., 2015; Spronken-Smith and Oke, 1998; Upmanis et al., 1998). Before specific policy recommendations can be made, however, additional studies will be needed to determine the optimal sizes and distribution of these clusters, with regard to decreasing household electricity and overall community and regional energy consumptions.

Limitations

An important limitation of this study centers on the relatively small sample size involving 231 communities, which resulted in the use of simplified statistical analyses. The main disadvantages of ecological studies involve finding associations at the group level that may not exist at the individual level. Nonetheless, the merits of using these simplified models include the uncovering of potentially valuable insights and the development of additional questions concerning household energy usage. These questions may spur further studies that incorporate a larger real-world database, a survey of individual households, or more detailed and comprehensive computer simulations.

Conclusions

In this study, urban form factors, building characteristics, and demographic indicators were confirmed to have simultaneous, significant connections with household electricity consumption. Our investigation used real-world data involving a large area of a city with a subtropical climate, where air-conditioning usage can be responsible for the greatest percentage of yearly household electricity usage. The results revealed that the most important urban form factors involved population density and urban canyons, followed by greater percentages of vacant space and building land use. In addition, building characteristics, namely household living area and building age, were found to be more closely linked to electricity consumption than any of the urban form factors or demographic indicators investigated. The importance of LEISS was also revealed by the connection found between larger household living areas and reductions in household electricity consumption/floor area.

Furthermore, the findings suggest that at the higher levels of urban densities found in the communities studied, even greater densities could lead to an overall reduction in average household electricity consumption. In fact, only after surpassing respective threshold levels were both narrower urban canyons and greater building land use percentages associated with lower electricity usage. Apparently, only with narrower urban canyons and at greater building densities do the cooling benefits of inter-building shadows become effective in lowering a community’s average household electricity consumption.

The results also led to the posing of three hypotheses involving population density, economies of scale, inter-building shadows, and the clustering of buildings. Additional research will be required to more fully understand these hypotheses before urban policies can be established regarding the reduction of residential energy usage. Notwithstanding the deficiencies of this study, the findings of this study nonetheless suggest that an urban model that emphasizes the clustering of buildings in the built areas and that provides larger percentages of non-built land use in the adjoining areas should be considered. Both building characteristics and urban form factors must be incorporated into future urban planning policies that are created to produce more sustainable cities.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by funding from both the Research Center for Energy Technology and Strategy (RCETS) and the Department of Urban Planning at the National Cheng Kung University, Tainan, Taiwan.

Yen-Jong Chen is a distinguished professor and former Chair of the Department of Urban Planning, National Cheng Kung University (NCKU), Taiwan. Since 2012, he has also been the Vice Director of the Research Center for Energy Technology and Strategy (RCETS) of NCKU and is engaged in renewable energy application and strategies. His research interests include urban and housing economics, transportation planning, and environmental spatial planning. He has served as a developing project reviewing committee member of the Urban Planning and Regional Planning for the Taiwan government for more than 20 years. For more information, please refer to .

Rodney H Matsuoka is currently a project-appointed assistant researcher in the Department of Urban Planning and the Research Center for Energy Technology and Strategy (RCETS) at the National Cheng Kung University, Taiwan. His research explores the human needs and preferences that underlie successful landscape designs, with a focus on the psychological, social, and health benefits provided by contact with natural features in the outdoor environment. More recently, he has begun to investigate the relationships between urban form and energy consumption, along with other renewable energy issues in the urban sector.

Tzu-Min Liang is currently at the Institute for Physical Planning & Information and graduated from the master’s degree program of the Department of Urban Planning at the National Cheng Kung University, Taiwan. Her research interests are centered on urban form and urban planning analysis utilizing GIS data. She is involved with the development of a comprehensive national land use plan and the creation of a nationwide real estate information database.

Appendix 1

References

Allegrini

Dorer

Carmeliet

(2012) Influence of the urban microclimate in street canyons on the energy demand for space cooling and heating of buildings. Energy and Buildings 55: 823–832.

Andrews

(2008) Greenhouse gas emissions along the rural-urban gradient. Journal of Environmental Planning and Management 51: 847–870.

Baker

Steemers

(2000) Energy and Environment in Architecture: A Technical Design Guide, London: Taylor & Francis Group.

Bartusch

Odlare

Wallin

et al. (2012) Exploring variance in residential electricity consumption: Household features and building properties. Applied Energy 92: 637–643.

Bowler

Buyung-Ali

Knight

et al. (2010) Urban greening to cool towns and cities: A systematic review of the empirical evidence. Landscape and Urban Planning 97: 147–155.

Asaeda

Abu

(1998) Reductions in air conditioning energy caused by a nearby park. Energy and Buildings 29: 83–92.

Chen

(2013) Simulation of the effect of downtown greenery on thermal comfort in subtropical climate using PET index: A case study in Hong Kong. Architectural Science Review 56: 297–305.

Chen

Yoshino

(2010) Statistical analyses on summer energy consumption characteristics of residential buildings in some cities of China. Energy and Buildings 42: 136–146.

Chen

Wong

(2006) Thermal benefits of city parks. Energy and Buildings 38: 105–120.

10.

Druckman

Jackson

(2008) Household energy consumption in the UK: A highly geographically and socio-economically disaggregated model. Energy Policy 36: 3177–3192.

11.

Emmanuel

Rosenlund

Johansson

(2007) Urban shading - a design option for the tropics? A study in Colombo, Sri Lanka. International Journal of Climatology 27: 1995–2004.

12.

Environmental Protection Administration. Ecolife cooler – carbon emissions reduction resolutions. 2010. . Available at: http://ecolife.epa.gov.tw/Cooler/English/eng_statistic_power.aspx (accessed 9 January 2016).

13.

Ewing

Rong

(2008) The impact of urban form on U.S. residential energy use. Housing Policy Debate 19: 1–30.

14.

Faruqui

Sergici

Sharif

(2010) The impact of informational feedback on energy consumption - A survey of the experimental evidence. Energy 35: 1598–1608.

15.

Futcher

Kershaw

Mills

(2013) Urban form and function as building performance parameters. Building and Environment 62: 112–123.

16.

Holden

Norland

(2005) Three challenges for the compact city as a sustainable urban form: Household consumption of energy and transport in eight residential areas in the greater Oslo region. Urban Studies 42: 2145–2166.

17.

Hsieh

H-Y

Forster

(2006) Residential construction quality and production levels in Taiwan. Engineering, Construction and Architectural Management 13: 502–520.

18.

Yoshino

Jiang

(2013) Analysis on urban residential energy consumption of hot summer & cold winter zone in China. Sustainable Cities and Society 6: 85–91.

19.

Huang

Zhao

et al. (2008) A fieldwork study on the diurnal changes of urban microclimate in four types of ground cover and urban heat island of Nanjing, China. Building and Environment 43: 7–17.

20.

Huang

P-C.

(2016) Solving the nation’s energy crisis. Taipei Times 6.

21.

Kakon

Nobuo

Kojima

et al. (2010) Assessment of thermal comfort in respect to building height in a high-density city in the tropics. American Journal of Engineering and Applied Sciences 3: 545–551.

22.

Kang

Cho

Kim

(2012) The energy-saving effects of apartment residents’ awareness and behavior. Energy and Buildings 46: 112–122.

23.

Kaza

(2010) Understanding the spectrum of residential energy consumption: A quantile regression approach. Energy Policy 38: 6574–6585.

24.

Kim

Cho

Kim

(2013) Energy use of households in apartment complexes with different service life. Energy and Buildings 66: 591–598.

25.

(2013) Urban form and residential energy use: A review of design principles and research findings. Journal of Planning Literature 28: 327–351.

26.

Krüger

Minella

Rasia

(2011) Impact of urban geometry on outdoor thermal comfort and air quality from field measurements in Curitiba, Brazil. Building and Environment 46: 621–634.

27.

Lai

C-M

Wang

Y-H

(2011) Energy-saving potential of building envelope designs in residential houses in Taiwan. Energies 4: 2061–2076.

28.

Lam

(1996) An analysis of residential sector energy use in Hong Kong. Energy 21: 1–8.

29.

Liao H-C and Jhou ST (2013) Taiwan’s severe energy security challenges. Available at: http://www.brookings.edu/research/opinions/2013/09/12-taiwan-energy-security-liao (accessed 9 January 2016).

30.

Lin

H-T

Wang

J-C

Shen

J-L

et al. (2004) Electricity analysis of town houses in Tainan City’s residential blocks. Journal of City and Planning 31: 215–237.

31.

Lin

Chang

et al. (2015) Calculating cooling extents of green parks using remote sensing: Method and test. Landscape and Urban Planning 134: 66–75.

32.

Min

Hausfather

Lin

(2010) A high-resolution statistical model of residential energy end use characteristics for the United States. Journal of Industrial Ecology 14: 791–807.

33.

Ministry of the Interior, Taiwan. Real estate information platform. 2011. . Available at: http://www.moi.gov.tw/stat/english/survey.asp.

34.

Mitchell

(2005) Urban Development, Form and Energy Use in Buildings: A Review for the SOLUTIONS Project, Swindon, UK: Engineering and Physical Sciences Research Council.

35.

Ndetto

Matzarakis

(2013) Effects of urban configuration on human thermal conditions in a typical tropical African coastal city. Advances in Meteorology 2013. Available at: https://www.hindawi.com/journals/amete/2013/549096/.

36.

Nelson

McHale

Peterson

(2012) Influences of landscape and lifestyle on home energy consumption. Urban Ecosystems 15: 773–793.

37.

Ouyang

Hokao

(2009) Energy-saving potential by improving occupants’ behavior in urban residential sector in Hangzhou City, China. Energy and Buildings 41: 711–720.

38.

Pacheco

Ordóñez

Martínez

(2012) Energy efficient design of building: A review. Renewable and Sustainable Energy Reviews 16: 3559–3573.

39.

Pisello

Taylor

et al. (2012) Inter-building effect: Simulating the impact of a network of buildings on the accuracy of building energy performance predictions. Building and Environment 58: 37–45.

40.

Pitt

(2013) Evaluating the greenhouse gas reduction benefits of compact housing development. Journal of Environmental Planning and Management 56: 588–606.

41.

Ratti

Baker

Steemers

(2005) Energy consumption and urban texture. Energy and Buildings 37: 762–776.

42.

Santamouris

Kapsis

Korres

et al. (2007) On the relation between the energy and social characteristics of the residential sector. Energy and Buildings 39: 893–905.

43.

Spronken-Smith

Oke

(1998) The thermal regime of urban parks in two cities with different summer climates. International Journal of Remote Sensing 19: 2085–2104.

44.

Statistica (2016) Household penetration of air conditioners in Taiwan from 2004 to 2013. Available at: http://www.statista.com/statistics/324744/taiwan-share-of-households-with-air-conditioners/ (accessed 9 January 2016).

45.

Steeneveld

Koopmans

Heusinkveld

et al. (2011) Quantifying urban heat island effects and human comfort for cities of variable size and urban morphology in the Netherlands. Journal of Geophysical Research 116: D20129.

46.

Strømann-Andersen

Sattrup

(2011) The urban canyon and building energy use: Urban density versus daylight and passive solar gains. Energy and Buildings 43: 2011–2020.

47.

Stupka R and Kennedy C (2010) Impact of neighborhood density on building energy demand and potential supply via the urban metabolism. In: 2010 ACEEE summer study on energy efficiency in buildings, Pacific Grove, CA, 15–20 August 2010, pp.239–252. Washington, D.C.: American Council for an Energy-Efficient Economy.

48.

Tainan City Guide (2015) About Tainan City. Available at: https://tainancity.wordpress.com/about-tainan-city/ (accessed 9 January 2016).

49.

Taiwan Power Company (2016) Taiwan Power Company annual report. Available at: http://www.taipower.com.tw/content/new_info/new_info-d14.aspx?LinkID=14 (accessed 9 January 2016).

50.

Teŕes-Zubiaga

Martín

Erkoreka

et al. (2013) Field assessment of thermal behaviour of social housing apartments in Bilbao, Northern Spain. Energy and Buildings 67: 118–135.

51.

Touchie

Binkley

Pressnail

(2013) Correlating energy consumption with multi-unit residential building characteristics in the city of Toronto. Energy and Buildings 66: 648–656.

52.

Tourism Bureau of Tainan City Government (2016) About Tainan. Available at: http://www.twtainan.net/en-us/Statics/AboutTainan (accessed 9 January 2016).

53.

Tso

GKF

Yau

KKW

(2003) A study of domestic energy usage patterns in Hong Kong. Energy 28: 1671–1682.

54.

Tso

GKF

Yau

KKW

(2007) Predicting electricity energy consumption: A comparison of regression analysis, decision tree and neural networks. Energy 32: 1761–1768.

55.

U.S. Energy Information Administration (2013) New U.S. homes are 30% larger but consume about as much energy as older homes. Available at: http://www.eia.gov/todayinenergy/detail.cfm?id=9951 (accessed 9 January 2016).

56.

Upmanis

Eliasson

Lindqvist

(1998) The influence of green areas on nocturnal temperatures in a high latitude city (Göteborg, Sweden). International Journal of Climatology 18: 681–700.

57.

Vassileva

Wallin

Dahlquist

(2012) Analytical comparison between electricity consumption and behavioral characteristics of Swedish households in rented apartments. Applied Energy 90: 182–188.

58.

Wong

Jusuf

Syafii

et al. (2011) Evaluation of the impact of the surrounding urban morphology on building energy consumption. Solar Energy 85: 57–71.

59.

Yang

Lau

SSY

Qian

(2011) Urban design to lower summertime outdoor temperatures: An empirical study on high-rise housing in Shanghai. Building and Environment 46: 769–785.

60.

Yang

(2012) Predicting and understanding temporal 3D exterior surface temperature distribution in an ideal courtyard. Building and Environment 57: 38–48.

61.

Yin

(2000) Building performance in the Taiwan earthquake: A review of Nantou County. Report for the Bureau of Engineering, March. Los Angeles: Department of Public Works, City of Los Angeles.

62.

Yun

Steemers

(2011) Behavioural, physical and socio-economic factors in household cooling energy consumption. Applied Energy 88: 2191–2200.

63.

Zhao

H-X

Magoulès

(2012) A review on the prediction of building energy consumption. Renewable and Sustainable Energy Reviews 16: 3586–3592.