Abstract
The effect of urban form on residential energy use has attracted much research, but it may be difficult to grasp the conclusions of that research because of inconsistencies in scope and methods employed. This article reviews the literature on how urban form affects residential energy use, particularly energy for space-conditioning (heating and cooling). Climate-responsive design principles are examined first and linked to research on how several factors affect residential energy use: housing type, density (physical compactness and dwelling unit density), community layout (street orientation and building configuration), and planting and other surface coverage. The research on each of these factors is summarized under three categories: experiments, simulation modeling, and statistical analysis of empirical data. Finally, implications for future research are discussed and suggestions for planning are made.
Introduction
Does the physical form of our cities affect how much energy we use in the home? This is not a simple question, and indeed has sparked a range of controversial answers. Many planners would agree that urban form affects transportation energy use, and there is an extensive body of research to support them, but few studies have investigated whether urban form significantly affects the energy performance of buildings. Additionally, research tends to be divided between (i) building/site scale design guidelines and simulation research pursued by architects and building scientists and (ii) community/city scale empirical approaches used by planners. This article strives to combine these disparate research agendas into a more comprehensive approach to reducing the energy use of residential buildings. This is critical due to the large amount of energy used by buildings and the resulting substantial contribution to carbon dioxide emissions (Andrews 2008).
In developed countries such as the United States and the European Union, buildings consume more energy than either industry or transportation (Pérez-Lombard, Ortiz, and Pout 2008). In the United States, buildings are the largest source of energy consumption and account for 41 percent of the total energy consumed (22 percent residential and 19 percent commercial), compared with 31 percent for industrial use and 28 percent for transportation (US Energy Information Administration [EIA] 2012a). Nationwide, buildings account for 1.5 times the energy consumption of transportation. However, in places where public transportation is more widely available and used, transportation energy use is lower and buildings represent an even greater share of the total energy consumed. For example, in the City of London, buildings use 2.2 times the energy than transportation systems (Mitchell 2005; Steemers 2003). With this in mind, more aggressive efforts are needed to improve the energy performance of buildings, minimize their consumption of energy resources, and promote better understanding of how our buildings actually consume energy.
Urban form elements that affect building energy use examined in this article include housing size and type, density (both physical compactness, dwelling units, and population), community layout (street orientation and building configuration), planting (trees and other vegetation), and surface coverage (pervious or impervious cover). Other variables affecting building energy use are nonspatial and include changing building design, improving the efficiency of heating, ventilation, and air-conditioning (HVAC) systems, and changing people’s behavior (Figure 1; Ratti, Baker, and Steemers 2005). From their modeling and field studies, Baker and Steemers (2000) conclude that energy consumption is affected due to (i) building design by a factor of 2.5 times; (ii) system efficiency by a factor of 2 times; and (iii) occupant behavior by a factor of 2 times. If a building is poorly designed, equipped with an inefficient mechanical system, and occupied by energy-wasting occupants, it could consume more energy by a factor of 10 times (2.5 × 2 × 2 = 10; Baker and Steemers 2000).
Building energy use can be explained as a function of urban form, building design, energy system efficiency, and occupant behavior (Ratti, Baker, and Steemers 2005). Urban forms affect urban microclimate, which influences space-heating and cooling energy demand.
With these multiple independent variables in play, determining the effect of urban form on energy use has been somewhat elusive (Lantsberg 2005; Mitchell 2005; Ratti, Baker, and Steemers 2005). Ratti, Baker, and Steemers (2005) argue for the contribution of an urban form factor, though they suggest that urban density and geometry affect energy consumption only within a range of approximately 10 percent of total building energy use. Nevertheless, even if urban form has a relatively small impact on overall energy use in buildings, the long-term impact across thousands or millions of buildings can add up to a substantial difference.
Beginning in the late 1970s and early 1980s, researchers studied aspects of urban form related to planning, such as new climate responsive and passive solar neighborhood designs, and related to architecture and landscape architecture, such as building design and vegetation planting guidelines. In the ensuing decades, researchers ran simulations focusing on the effect of one or two variables (e.g., house size, type, street layout, and trees) on various elements of microclimate (e.g., solar access, wind flow, and air and surface temperature), as well as on thermal comfort and building energy use. Recently, with the advent of more urban and regional policies focusing on reducing greenhouse gas emissions (GHG), researchers have turned their attention toward planning-related variables (mostly related to urban density) that impact energy use in buildings. Using government statistics from the US Residential Energy Consumption Survey (RECS, the only existing large-scale data set on energy consumption in the United States), statistical studies have begun to look into energy policies and planning. For example, Ewing and Rong’s (2008) study uses multivariate statistical analysis to look at how urban form variables (housing size, housing type, and density) affect energy consumption for heating and cooling. This study received a great deal of attention, including some critical reviews. Randolph (2008) and Staley (2008) issued comments criticizing (i) the legitimacy of the link between the complex data set, the methods used, and the conclusions; (ii) the lack of attention to impacts from energy pricing and new energy-efficient technologies; and (iii) the use of complex statistical analysis rather than an engineering simulation model that could control other variables.
Although both architectural and planning approaches have helped to conserve energy in buildings and neighborhoods, the two fields often lack agreement and communication on this topic. Architectural and landscape architectural studies have focused on developing design guidelines for passive solar homes and for planting, but rarely on measuring the effects of energy-conserving layouts and community design on building energy performance. On the other hand, planners have investigated how urban form affects residential energy use at a macro scale, but have rarely been aware of the effect of varying architectural designs. This article aims to create a common ground for design-based and planning-based research on urban forms that reduce residential energy use by examining the variables of house size, housing type, density, community layout, planting, and surface coverage.
Scope of This Review
This article aims to provide guidance for new research into the effect of various urban form variables on building energy use, specifically residential buildings. First, the article browses through a spectrum of design principles and links them with research findings. The bibliography of climate-responsive design, layout, and planning for housing is updated and gaps in the literature are identified. The intent of the article is to make this research accessible to a wider audience and therefore provide plain language definitions of many specialized terms or jargon used by various researchers. This study includes only residential buildings, in order to limit the variation that energy operation and occupant behavior would cause in different types of buildings.
Improving the energy efficiency of residential homes is a high priority for the US federal government. Under the American Recovery and Reinvestment Act of 2009, the Department of Energy’s Weatherization Assistance Program (WAP) received $5 billion for weatherizing nearly 600,000 homes. “Weatherizing” consists of cost-effective measures to reduce energy use (e.g., insulation, improvements to heating and cooling systems or electrical systems, and replacing appliances with lower-energy models), and the WAP affects existing residential and multifamily housing with low-income residents. Compared to the enormous attention that building weatherization has received, energy-saving measures for neighborhood planning and design are rare. A more comprehensive program will have to go beyond weatherizing individual buildings to making entire communities more energy-efficient, using systematic strategies based on research findings.
This article reviews the literature that shows a relationship between urban form variables and residential energy use, especially for space-conditioning (heating and cooling), focusing on the effects of house size, type, density, community layout, planting, and surface coverage. The literature reviewed includes: theories and guidelines that cover climate-responsive housing form, layout, and planning at the site and community scale (beyond the individual building scale) and research (experiments, simulation models, and statistical analyses) that shows the effects of physical urban form variables on space-heating/space-cooling energy use. research on the effect of urban design on microclimate alone (without addressing energy consumption); research on the effects of socioeconomic and demographic variables on energy use (but not including physical urban form predictors); research on energy-conserving architectural designs, building-scale weatherization, retrofits, and occupant responses (e.g., the well-known Princeton’s experiments at Twin Rivers; Socolow 1978a, 1978b).
This article better focuses this review by excluding:
This study includes only peer-reviewed journal articles in the literature list in the Appendix. Although some articles are referred to investigate the above three excluded topics, and/or some nonpeer-reviewed studies, those articles are not included in the literature list.
Design Principles
Urban planning and design factors affect microclimates such as daylight, solar radiation (causing heat gain), wind flow (either shelter from wind or desired ventilation), and local temperature (including the urban heat island). These microclimates in turn affect residential energy use (Figure 2). A neighborhood’s density is strongly associated with its housing size and housing type, and therefore is associated with that neighborhood’s solar access, natural ventilation, and urban heat island effect. Street layout design typically determines street orientation and building configurations that are critical for solar access and wind flows. Open space planning, tree planting, and surface coverage also have a significant influence on urban microclimate, particularly on solar access and the urban heat island effect.
Flowchart of how urban form variables impact urban microclimate and residential space-conditioning energy use.
Since 1963, when Olgyay (1963) published his seminal book on architectural principles of bioclimatic building design and site planning, various professional and academic groups (landscape architects, architects, planners, and scientists) have published guidelines for energy-conserving site planning and design. Beyond building design, these studies considered how other factors could improve urban microclimates and save energy, including housing type, street layout, building configuration, housing density, and planting (Table 1).
Chronological Review of Selected Classic Studies of Energy-Conserving Site Layout, Community Design, and Landscape Planning.a
Note: aThis list includes selected studies that specifically focus on energy-conserving site planning and design. A number of seminal studies on energy-efficient building design or general site planning are excluded from this list.
General Principles of Climate-Responsive Design by Four Different Climate Regions Based on the American Institute of Architects Research Corporation (1978), Erley and Jaffe (1979), and Golany (1996).
Hot–dry (e.g., southwestern United States): Solar control should be the key strategy to mitigate high day temperature with excessive dryness. Compact urban form and tree planting on the east and west of the building can reduce heat transfer and block solar radiation. Deciduous trees to the south of a building can block the sun in summer but allow the sun through in winter.
Hot–humid (e.g., southeastern United States): Focus on natural ventilation in order to alleviate excessive humidity as well as heat. A dispersed urban form with wide streets can promote wind flows.
Cold–dry (e.g., Great Basin area): Protecting buildings from cold winds is most important, so compact and clustered urban form with tree planting for windbreaks is recommended.
Cold–humid (e.g., northeastern United States): This region experiences extreme cold and wind in winter, but high precipitation and humidity in summer. Multifamily housing with a mix of open and enclosed urban form and mixed tree planting (deciduous trees to the south and evergreens to the north) would reduce heat transfer and block cold wind in winter, but allow some natural ventilation in summer.
Review of Research by Urban Form Variable
In this section, previous studies are reviewed by urban form variable: housing type and size, density, community layout (including street orientation and building configuration), and planting and surface coverage. In each variable, research is grouped by different methods, mainly simulation and statistical analysis. Studies using experiments are reviewed only for planting and surface coverage because experimental studies rarely exist for the other urban form variables.
The major advantage of simulation is that researchers can easily control a complex system—climate, form, construction, HVAC, occupancy, and energy price—to obtain specific results. The results provide spatially and temporally finer scale estimates of energy use, such as hourly load energy use at the individual building scale (Heiple and Sailor 2008). In addition, the results from algorithms such as DOE-2 and American Society of Heating, Refrigerating, and Air-Conditioning Engineers have been validated with measured values (Diamond, Cappiello, and Hunn 1986; Meldem and Winkelmann 1998). With these advantages, building energy simulation is hugely favored by architects, building scientists, and mechanical engineers.
Still, the outputs of those engineering models are only simulations of reality, not reality itself. Due to the lack of publicly available detailed data regarding model inputs, researchers make many assumptions with little reliability, adding uncertainty to the model (Kavgic et al. 2010). Occupant behavior is a striking example. A real-world occupant’s behavior could vary widely and significantly affect the output (Swan and Ugursal 2009). Many simulations for highly energy-efficient buildings fail to show accurate energy use because they neglect the “rebound effect” where occupants actually use more energy in energy-efficient settings. In this regard, simulation models may be more useful as an analysis tool for building design by comparing how different archetypes and building components (e.g., HVAC equipment, insulation, windows) perform relative to each other, rather than as a method for obtaining absolute figures of energy use.
Statistical models allow researchers to infer a relationship using empirical data. Regressing energy-billing data as a function of the characteristics of a house and its occupants can provide a powerful estimate of energy use. Using surveys, researchers can assess how one variable affects energy use as they control for other variables. With this advantage of recognizing patterns of the real world, many planners prefer an empirical approach over simulation research.
However, statistical models of empirical data can have disadvantages, particularly that it can be challenging to obtain a wide range of disaggregated data (Owens 1986). Researchers are often unable to secure energy-billing data, due to concerns over the privacy of customers. In addition, conducting a survey of occupants and housing characteristics requires significant effort. These drawbacks have made it difficult for researchers to investigate the independent effects of different housing types on energy use through a statistical/empirical approach.
Housing Type and Size
The ratio of a building’s surface area to volume (S/V ratio) is most relevant to heat transfer into or out of a building. The type of housing—such as single-family detached or multifamily—is intrinsically linked to this S/V ratio. While multifamily units and single-family attached units share walls internally, the walls and roof of a single-family detached house are all external and are susceptible to outdoor temperature. For a given building volume, a single-family detached unit has more surface area and a higher S/V ratio than a multifamily unit, so it more readily loses or gains heat and thus consumes more energy.
According to the 2009 RECS, single-family houses in the United States use more energy than multifamily houses, per household and per person (Table 3). Although a single-family house (attached or detached) uses less energy per square foot than a multifamily building, a single-family house typically has more residents and more area per resident, and therefore uses more energy as a unit. The average household in all types of single-family housing uses roughly twice as much energy as the average household in all multifamily housing (103.6 vs. 55.8 million British thermal unit [Btu]/year). Each person in a single-family house also consumes 1.4 times more energy than a person in a multifamily unit (37.7 vs. 27.2 million Btu/year). Although energy consumption for all housing types has decreased since 1978, the gap in energy use between single-family detached and multifamily housing has remained relatively constant through 2005 (Kaza 2010) and 2009 as well (US EIA 2012b). As shown in Table 3, the energy used by an average household in single-family detached housing (105.7 million Btu/year) is 2.3 times that of an average household in multifamily housing with five or more units (46.3 million Btu/year). The reason this gap has not closed, despite new building codes for single-family houses, may be due to the average floor area of new single-family detached houses, which increased by 39 percent to 2,900 ft2 between 1980 and 2005, whereas that of multifamily units remained the same (Kaza 2010).
Residential Energy Consumption in 2009, by Housing Type.
Source: 2009 RECS Survey Data Consumption and Expenditures, accessible from: http://www.eia.gov/consumption/residential/data/2009/index.cfm?view=consump-tion (Release Date: December 14, 2012, accessed March 16, 2013).
Housing size matters in energy consumption. In theory, larger houses require more energy for space-conditioning; assuming a fixed height per floor, a larger floor area means a greater larger volume to be heated or cooled. In addition to floor area, the ceiling height of new homes has been increasing—in the 1970s only 17 percent of homes had higher ceilings than traditional 8-ft ceilings, while that number increased to 52 percent in homes built in the 2000s (US EIA 2012c). These results indicate that the average “volume” to be heated and cooled in US residential buildings has significantly increased in the last two decades. Furthermore, many large homes are located in the areas with extreme temperatures (e.g., the Midwest, Northeast, and South), which leads to more space-heating and space-cooling energy consumption (US EIA 2012c).
The effect of housing type and size can be reduced by improving energy efficiency. The 2009 RECS reports that average residential energy consumption per household in the United States has been decreasing over the last thirty years in spite of the increase in housing size, mainly due to improvements in energy efficiency for space-conditioning and major appliances (US EIA 2012d). This result is comparable with Holden and Norland’s (2005) study in Oslo, Norway, which finds that the adoption of new building codes stressing energy efficiency reduces the difference in energy use among different housing types.
Simulation Modeling
Simulation tools for modeling building energy use have evolved as hypotheses evolved, testing variations in housing size and type, as well as other building characteristics. The classic 1975 study from the British Building Research Establishment showed that space-heating for a detached house could be three times greater than for an intermediate flat (a unit in a multifamily building) of an equivalent size—a difference similar to that between a poorly insulated unit and one with medium insulation standards (BRE 1975). Today, many models to simulate a building’s energy use, such as ResFen, Energy10, Ecotect, eQUEST, Energy Plus, and DOE-2, are available, with a range of complexity.
Statistical Analysis
Since the late 1970s, the RECS has provided nationwide cross-sectional data on physical characteristics of the housing unit, demographic attributes of a household, heating and cooling equipment, and fuel types. This data set has been useful for some empirical studies, but it does not provide enough detail for researchers to determine the causes of a huge variation in energy use (Hirst, Goeltz, and Carney 1982; Kaza 2010; Randolph 2008). Other variables, such as house age, occupants’ income, and ownership (whether the occupant owns the unit), can provide indirect clues to the building design, efficiency of HVAC systems, and occupant behavior. However, the RECS is not practical for investigating the effect of spatial factors (urban form) on energy use, as it does not provide any spatial characteristics of disaggregated housing units, save for aggregate census regions (Min, Hausfather, and Lin 2010).
Most empirical studies generally agree that larger or single-family houses use more energy than smaller or multifamily units. Analyzing the 1978 National Interim Energy Consumption Survey with ordinary least squares regression, Hirst, Goeltz, and Carney (1982) argue that floor area is a key determinant of space-heating and total residential energy use, but they see a huge variation in energy use per unit of floor area. Using a multiple regression model with their own survey data, Holden and Norland (2005) note a statistically significant difference in energy use among different housing types in Oslo, Norway, but there was less difference in energy use between single-family and multifamily homes built after 1980, since new building codes had been adopted. With newer statistical methods, Ewing and Rong (2008) use a hierarchical model to account for the shared characteristics of households in the same place. Using RECS data, they argue that a household in a multifamily home consumes 54 percent less heating energy and 26 percent less cooling energy than one in a single-family detached home. By creating a high-resolution statistical model of residential energy use for the entire United States, Min, Hausfather, and Lin (2010) argue that housing size plays a role in residential heating and cooling energy use, along with other factors such as climate, heating equipment, fuel, and age of building. Kaza (2010) uses quantile regression to overcome the high variability in the RECS data set, and concludes that moving a household from a single-family detached house to a multifamily apartment would reduce the heated area by more than 100 m2 in all but the 10th and 90th percentile, while the cooling energy savings would be equivalent to reducing the cooled area by 40–70 m2. He also concludes that housing size matters, but housing type has a more nuanced effect on space-conditioning (Kaza 2010).
As advanced as these statistical methods may be, it can be misleading to state absolute figures unless the study uses a perfect data set with numerous relevant variables (Randolph 2008). Some research questions the use of a complex statistical method when a simple simulation would suffice. However, empirical research is valuable to observe how energy use relates to building options in a real-world environment. It can indicate how significant each variable is on energy use, evaluate how effective new energy policies have been, and validate the results from simulation research. The effectiveness of the statistical/empirical approach will grow, as richer data sets become available.
Density
Density can be described in two different ways: population density (or dwelling unit density) from the planning side and compactness from the physical or architectural side. In general, population density is correlated with housing size and type, as single-family houses in suburbs tend to be larger in floor area than multifamily houses in center cities, resulting in lower population density in suburbs (Ewing and Rong 2008; Kaza 2010). The greatest benefit of higher population density is that multifamily housing allows lower residential energy consumption per household and per capita. Pitt (2012a) aggregates housing units from the 2000 US Census and the 2005–2009 American Community Survey into 2005 RECS data to determine the impact of aggressive implementation of compact housing, such as requiring 75 percent of new units to be multifamily housing. In his case study of Blacksburg, Virginia, the implementation of compact housing could result in a 35.5 percent GHG emissions reduction compared to the baseline scenario of primarily low-density housing, without considering additional energy conservation actions. In a similar study on ten Virginia Census-designated metropolitan regions, Pitt (2012b) finds that the most aggressive scenarios of compact housing had potential GHG emissions reduction of approximately 23 percent over the most conservative scenarios.
In addition to the lower energy consumption of multifamily housing, higher-density development helps alleviate the urban heat island effect. Although dense development emits more heat per area of building footprint than sprawling development, suburban development (with its larger lot sizes) has a higher urban heat island effect per household than dense development. Based on studies of the association between urban form and thermal efficiency in Atlanta, Georgia, Stone, Hess, and Frumkin (2010), Stone and Rodgers (2001), and Stone and Norman (2006) report that lower-density development where forest is replaced with lawn has a lower albedo and releases more radiant heat energy. This, in turn, contributes to the urban heat island effect and exacerbates extreme heat events. Needless to say, higher-density neighborhoods also have the advantage of promoting better public transit and walkability than sprawling neighborhoods, see many examples from the Center for Neighborhood Technology (www.CNT.org), Congress for New Urbanism (www.CNU.org), and Calthorpe Associates (www.calthorpe.com).
Compactness describes how tightly spaced buildings stand on a site, including the width of the streets, the distances between buildings, and the height of the buildings themselves. Many researchers use the term aspect ratio, or the ratio of building height to street width. In a compact urban setting with high lot coverage and tall buildings, the aspect ratio can be the key influence on residential energy use.
Existing studies have investigated how aspect ratio affects microclimates, including characteristics of solar access (such as the shading of one building by another), wind flow, and outdoor comfort. For example, wide streets (low aspect ratio) provide more space for solar access, so air temperature is higher (Givoni 1998; Sharlin and Hoffman 1984), especially in winter when the sun is low in the sky (Ali-Toudert and Mayer 2006; Arnfield 1990). In addition, wide streets promote natural ventilation. In hot climates, this ventilation is useful for cooling, but the greater solar exposure often offsets the benefit of ventilation, and free air flow along wide streets can aggravate problems with windblown dust in hot and dry regions (Golany 1996). A narrow street oriented in line with prevailing winds can create a wind tunnel or “street canyon” effect. Here the wind tunnel ventilates the street itself, but not the buildings if the buildings have high wind speeds (and therefore the same air pressure) on both sides (Givoni 1998). The best arrangement for a hot–dry region would include narrow winding streets aligned with the prevailing wind and compactly spaced buildings with staggered heights. This combination diverts strong wind to the streets and promotes natural ventilation (Aggarwal 2006), while allowing buildings to shade one another (Hough 1995; Minne 1988).
However, in cold regions, a compact form can lead to higher demand for heating, as the buildings block solar access (Steemers 2003). In modern cities where buildings themselves generate excessive heat, a compact form minimizes heat loss and exacerbates the urban heat island (Krishan et al. 2001). Compact form has a negative effect not only on passive solar conditioning but also on opportunities for on-site (rooftop) solar energy generation, as one building’s roof could be shaded by another. To maximize solar energy generation, the buildings’ vertical and horizontal layout should be carefully considered (Cheng et al. 2006; Hui 2001; Steemers 2003). Researchers are examining the “solar envelope” in order to ensure solar access (passive or active) and improve energy efficiency in dense urban settings (Knowles 2003; Knowles and Berry 1980; Morello and Ratti 2009).
Simulation Modeling
The overall effect of aspect ratio on energy use (especially demand for space-heating/space-cooling) is complicated and has not been quantified, through either simulation or statistical/empirical analysis. Steemers (2003) argues that the solar potential for housing in dense environments is reduced mainly due to obstructions from neighboring buildings. Using the lighting and thermal (LT) model (Baker and Steemers 2000), Steemers reports that a passive solar house whose south facade has an obstruction of 30° would use 22 percent more energy for space-heating compared to a house with an unobstructed facade. The drawback of lower solar access (greater shading) in a compact form may be offset, however, by a reduction in heat loss to the atmosphere (Steemers 2003). Using the LT method for existing cities (London, Toulouse, and Berlin), Ratti, Baker, and Steemers (2005) argue that variations in urban density and geometry can affect energy use by about 10 percent, which is still considerable. Central London, for example, is more compact and requires less energy than central Berlin. Continued study on various urban forms, in order to discover a wider range of ways to reduce demand for energy, is warranted.
Statistical Analysis
The impact of density on residential energy use received much attention in the last decade, mostly through empirical studies in planning. However, studies have shown that density affects transportation energy use much more than residential energy use (Holden and Norland 2005; Kahn 2000; Lariviere and Lafrance 1999; Norman, MacLean, and Kennedy 2006; Brownstone and Golob 2009).
The effect of density on residential energy use is a complicated topic and still controversial. Lariviere and Lafrance (1999) report that higher-density cities use slightly less electricity per capita than low-density cities in their study of Canadian cities. They predict that if a city of 1,000 inhabitants per km2 (equivalent to Boise, Idaho) increased its population density by three times (equivalent to Baltimore, Maryland), the electricity use per capita would be reduced by only 7 percent. However, their study did not include residential gas consumption, which is the predominant fuel for heating during the cold Canadian winter, and one could imagine that a denser city would use dramatically less gas per capita. Holden and Norland (2005) also argue that residents in dense areas use less energy than those in less dense areas in Oslo, Norway. However, their results are limited, as they may have failed to separate density from its strong correlation with housing type. For the United States, using the 1993 RECS data, Kahn (2000) finds no significant difference in residential energy use between suburban areas and central cities. Using the quantile approach with updated RECS data, Kaza (2010) also concludes that neighborhood density itself (a figure only indirectly reflected in the self-reported “urban/rural” classification) appears not to reduce energy use. Only apartments in large blocks have substantially different energy-consumption profiles from single-family detached homes. However, these findings by Kahn and Kaza are limited, in that they distinguish only between “city” and “suburban” housing, rather than analyzing energy use at different densities along a continuum.
Recent research findings support the notion that density indirectly affects residential energy use through other intermediate variables such as housing type, size, ownership, and income (Ewing and Rong 2008; Kaza 2010). Ewing and Rong (2008) note that the choice of housing type is strongly related to urban form; multifamily housing is seven times more prevalent in compact urban areas than in sprawling ones. Using path analysis, Ewing and Rong conclude that residents in sprawling counties tend to live in large, single-family detached homes, and the higher residential energy use is a function of both the housing type and the low density. An average residential unit in a compact county consumes 20 percent less energy (expressed in Btu per year) than one in a sprawling county, mostly due to the differences in housing type and size. This energy reduction of 20 percent in compact settings is consistent with the simple engineering calculation conducted by Randolph (2008).
Community Layout
The position of a building on its site—including its orientation and spacing from other buildings—is critical to ensuring that the building receives an appropriate amount of sunlight, especially for groups of buildings. In northern latitudes, orienting a building within 10° to 30° of true south is generally the most desirable position to maximize its solar access (Goulding, Lewis, and Steemers 1991; Holtz 1990; Littlefair et al. 2000). North-facing buildings receive the least sun, while an east or west orientation is problematic because buildings gain too much direct heat in the morning or in the late afternoon. For example, in London, changing a building’s orientation from true south to true west would reduce solar warming overall, requiring 16 percent more energy for space-heating in a passive solar house or 9 percent more in a conventional house (Steemers 2003). Since realizing that orientation could have a significant effect, researchers studied energy-conserving guidelines for site planning, including street and building layouts for solar communities, especially in the late 1970s and early 1980s (Brown 1985; Center for Landscape Architectural Education and Research 1978; Erley and Jaffe 1979; Hammond, Zanetto, and Adams 1981; Robinette 1983).
Many solar neighborhood design guidelines generally support an east–west street orientation that allows north–south lots and therefore accommodates more south-facing buildings (Figure 3). Street orientation usually determines how buildings are oriented, especially where those buildings cover most of the buildable land (Edminster 2009). However, there can be variation in building orientation depending on house siting, lot size, and shape. For example, an east–west lot (on a north–south street) or an irregular lot (such as on a cul-de-sac) can also accommodate a south-facing house, if the house is not rigidly aligned to the lot boundary (Hammond, Zanetto, and Adams 1981; Littlefair et al. 2000; Thayer 1981) or if the lot size and shape allows the longer axis of the building to be oriented south. However, it is rare to site a building independent of its property boundary in order to improve solar access, especially in urban settings where the ratio between building and property coverage is usually maximized.
East–west street orientation maximizes south-facing homes, which allows for solar access. Created by the author, based on Randolph and Masters (2008, 271).
Besides the effect on solar access, proper street orientation can alleviate the effect of extreme climate. In hot climates with mild winters, a south-facing building (on an east–west street) may be undesirable unless the streets are narrow enough to minimize sun penetration between buildings. For example, Littlefair et al. (2000) argue that in Athens (38°N), the aspect ratio should be at least 4 (buildings at least four times as tall as the street is wide), and in Rome (42°N) 3.5 is the best aspect ratio. Street orientation can also protect buildings from cold wind, and thereby reduce the energy required for heating. If streets are oriented perpendicular to the wind, the primary air flow is above the buildings rather than against their walls (Givoni 1998).
Building configuration—the vertical and horizontal arrangements—is another important factor affecting solar access. Compagnon (2004) used three-dimensional (3D) models to show how to achieve higher passive and active solar potential with optimal layout designs, even in denser environments. Among his four hypothetical designs for denser neighborhoods, two of them (a striped configuration with uniform building heights, and a stepped block configuration with variable heights), had the greatest solar potential. Cheng et al. (2006) used 3D simulation to test a range of built forms and densities with three variables: (i) uniform or random horizontal and vertical layouts (Figure 4), (ii) plot ratio, and (iii) site coverage. They concluded that building configurations with more horizontal and vertical randomness, less site coverage, and more open space would enhance daylight performance and solar potential (Cheng et al. 2006).
Examples of different horizontal and vertical building layouts. Created by the author, based on Cheng et al. (2006).
Simulation Modeling
Even though many design guidebooks emphasize the importance of site layout, only a few simulation studies have been published in peer-reviewed journals. Through a simulation of a typical single-family house in Québec City, Paradis, Faucher, and Nguyen (1983) argued that an optimal street orientation (20° east of south) could reduce a house peak heating load in winter by 24–70 percent (with the maximum reduction occurring on a windy day) and could also reduce annual household energy use by 16.5 percent (as the energy saved in winter would be offset slightly by the greater need for cooling in summer). Littlefair (1998) and Littlefair et al. (2000) cited a study by NBA Tectonics’ (1988) on how site layout affected the passive solar performance of low- and medium-density housing in the United Kingdom. Here passive solar housing with an appropriate orientation could reduce the space-heating load by 11 percent, compared to only about half that in dwellings on nonoptimal street layouts. Randolph and Masters (2008) modeled a 1,500-ft2 house in Blacksburg, Virginia, and found that a simple sun-tempered design with maximized south-facing windows (40 percent south facing, or 100 ft2 of a total 250 ft2) and an east–west roof peak (typical for an east–west street orientation) could reduce heating demand by 20 percent over a house oriented north–south or a house with shading but no solar gain.
There appears to be no empirical research on how street orientation affects residential energy use.
Planting and Surface Coverage
Planting trees is widely known as an effective way to reduce a building’s energy demand. If properly sited, trees can keep a building cooler (and reduce cooling loads) by shading the building facade and the ground surface, cooling the air by evapotranspiration, and modifying air flow to promote natural ventilation. Trees can keep a building warmer (and reduce heating loads) by blocking cold winds. Trees and other plants can also moderate the urban heat island effect by reducing the temperature of the air and of surfaces in general. With so many anticipated ways to reduce energy use, researchers have studied the effects of trees much more than other urban form variables.
The right type of trees must be planted in the right location to have the greatest effect. In Sacramento, California (a hot–dry region), trees planted to the west of houses provide benefits that are three times higher than the average for all trees planted through the city’s shade tree program (annual benefits of $120 vs. $39 in 1998 dollars; Hildebrandt and Sarkovich 1998). Selecting and placing the proper tree, based on its growth rate and crown shape, can improve seasonal solar access and wind patterns, and maximize energy savings (Heisler 1986b).
However, trees do not always save energy. Planting an unsuitable type of tree or planting trees in an undesirable location can lead to using more energy for heating and cooling (Láveme and Lewis 1995), due to blocking desirable sunlight or ventilation. Thayer, Zanetto, and Maeda (1983) found a significant net increase in the annual energy costs of a solar house if trees are placed directly south of the house.
Considering all of these varied effects, planting trees appears to have a positive net effect on saving space-conditioning energy. If trees are blocking cold winds, this effect saves more energy than the increase in heating from the shade (Simpson 1998; Simpson and McPherson 1998). The highest windshield effect comes from a combination of deciduous and evergreen trees, planted with a moderate to high density in an arrangement perpendicular to the wind (Hammond, Zanetto, and Adams 1981; Jaffe and Erley 1979). Heisler (1986b) found that proper tree planting around a conventional detached house can save up to 25 percent in annual energy costs, across a range of climate, house, and tree conditions. In Chicago, increasing tree cover by 10 percent—planting about three trees per lot—is estimated to reduce annual heating and cooling costs by $50–$90 per dwelling unit (McPherson et al. 1997). In the 1970s and the early 1980s, energy-conserving planting guidelines were developed to provide optimal solar access or shading and wind blocking for a variety of climates, including specific instructions on tree types, size, and distance and orientation to the house (Erley and Jaffe 1979; McClenon and Robinette 1977; McPherson 1984; Moffat and Schiler 1981).
Among the various ways in which trees can save energy, the cooling effect of tree shade and evapotranspiration has received the most attention. For most climates, placing deciduous trees on the west side (or southwest side in regions with hot summers and cold winters, as long as these trees are lightly twigged) of houses is highly recommended because they provide shade in summer but allow solar radiation to pass through during winter. Under clear skies, a midsized sugar maple on a south-facing wall provides shade that reduces irradiance by about 80 percent when in leaf and by nearly 40 percent when leafless (Heisler 1986a). Tree shading is even more important in hot climates. Parker (1987) reports that walls shaded by shrubs are 24–29°F cooler than uncovered walls during periods of direct sunlight and are significantly cooler even at other times without direct sun. Perhaps surprisingly, the indirect energy savings from evapotranspiration may be even three to four times greater than the cooling effect of direct shade (Huang et al. 1987). Supported by these research findings, various cities have instituted tree-planting programs since 1990, such as the Sacramento Municipal Utility District’s shade tree program. Planting trees has proven to be cost effective from the perspective of energy efficiency, especially in hot–dry regions like Sacramento (Hildebrandt and Sarkovich 1998; McPherson and Rowntree 1993; McPherson and Simpson 2003). The effect of trees on reducing energy use has been intensively studied through experiments, simulations, and empirical research, but it is challenging to compare the results of various studies because of differences in the experimental settings or the assumptions of each model, tree sizes and types (deciduous or coniferous), condition of the buildings, and climate.
Compared to tree planting, surface coverage—which encompasses a range of variables, from permeability to color (especially albedo, or lightness of color)—has received much less attention in simulations or statistical analyses of its direct energy saving, yet it appears to be an important factor. Indirectly, numerous studies on the urban heat island effect reveal how surface coverage affects urban thermal efficiency (mainly the tendency to retain heat). In general, daytime urban thermal efficiency is strongly associated with land use and land cover; areas with impervious cover (mainly urban or industrial) are warm, and areas with plants or water bodies are cool (Roth, Oke, and Emery 1989). Stone and his colleagues (Stone and Rodgers 2001; Stone and Norman 2006) provided a new perspective on the urban heat island effect from their studies of surface coverage and warming in Atlanta, Georgia. Using high-resolution thermal imagery and path analysis, they looked at how impervious cover, pervious cover (e.g., lawns), or tree canopy cover affects net thermal emission and surface warming. Both studies argued that exposed surfaces (unshaded by trees or buildings) on large lots in lower-density developed properties appear to matter most in excess surface warming. While the flux density of energy (watts per square meter) from lawn surfaces is lower than that of impervious cover, the total area of lawns is greater—resulting, in many cases, in a greater total flux of energy (watts) from pervious materials per parcel (Stone and Norman 2006). Their conclusions show a different perspective from the conventional assumption that higher-density development contributes to a greater urban heat island effect than does lower-density development (Hoyano 1984). Stone and Norman (2006) suggested that planners promote higher-density development with smaller lots, less lawn, less impervious cover, and more shade through the proper configuration of trees and buildings.
The color of surfaces may also be as important as the physical character or arrangement of surfaces. Many studies from the Lawrence Berkeley National Laboratory’s (LBNL) Urban Heat Island Group suggested that increasing the albedo (reflectivity) of roofs, streets, and other impervious surfaces—that is, making them lighter in color—is a cost-effective approach to reducing energy use, for both existing and new construction (Akbari et al. 1992; Akbari, Pomerantz, and Taha 2001; Akbari and Taha 1992; Rosenfeld et al. 1998).
Experiments
Several experiments have shown that trees can reduce the demand for cooling energy by 25–80 percent in a variety of climates, house conditions, and vegetation settings. In Tucson, Arizona, an experiment by McPherson, Simpson, and Livingston (1989) on three similar 1/4-scale model buildings with various landscaping—turf, rock mulch with a foundation planting of shrubs, and rock mulch with no plants—showed that the house with no vegetation consumed 25 percent more electricity for air-conditioning (cooling) than the house surrounded with turf and 27 percent more than the house with shrubs. In Sacramento, California, Akbari et al. (1997) found that a house with eight large and eight small shade trees used 30 percent less cooling energy, due to tree shade, than a house without trees. Through the energy analysis of an insulated mobile home in Miami, Florida, Parker (1983) showed that proper landscaping can reduce energy use for air-conditioning by at least half during warm summer days (5.56 to 2.28 kWh in the morning and 8.65 to 3.67 kWh for afternoon peak hours). From research in Beauregard, Alabama, during April to September 2008, Laband and Sophocleus (2009) concluded that a building in dense shade uses 62 percent less electricity for cooling to 72°F than a building in full sunlight. DeWalle, Heisler, and Jacobs (1983) report even greater energy savings: 75 percent less air-conditioning, seasonally averaged, for a mobile home in a forest site in central Pennsylvania.
There have been fewer experiments on using trees as a windbreak to save energy on heating, but those experiments have reported a range of savings from 3 percent to 40 percent in the northern states of North Dakota, Pennsylvania, and New Jersey (Heisler 1986b; McPherson and Rowntree 1993). The majority of energy saved comes from reducing the infiltration of cold air into the building. For example, in central Pennsylvania, a small mobile home located one tree height (3 m) from a windbreak of small trees uses up to 18 percent less energy for winter space-heating (DeWalle and Heisler 1983). Depending on the direction of the wind and the location of the windbreak, a windbreak of trees might also cast shade on a building, making it colder and thereby increasing the demand for space-heating. DeWalle, Heisler, and Jacobs (1983) report that a deciduous windbreak reduces the demand for heating energy by only 8 percent, while heating energy demand rises 12 percent due to the shade of a dense pine forest. Overall, the effect of tree windbreaks (and the counter effect of shading) has not been thoroughly examined across various climates with experimental settings, compared to the abundance of experiments on how trees can reduce cooling energy.
Although these experiments provide meaningful real-world results to help develop strategies for landscape design, the results are far from comprehensive, because the settings are less controlled and the sample sizes are small (usually one or two options in addition to the control, and one sample to represent each category). It is difficult to consider these studies as a thorough representation of each case, as only a few experiments have been conducted for each condition of climate and housing type. Performing more experiments and simulation studies would help address this lack of representation. Instead of comparing experimental results from various settings, not all of them relevant to a particular case, it is more useful to see whether an experiment validates other types of studies (such as simulations) in a similar setting.
Simulation Modeling
In addition to experiments, simulation is an effective approach to examining the effect of trees by controlling other variables such as tree configuration, housing conditions, and climate. Building energy analysis tools such as DOE-2, Shadow Pattern Simulator (SPS), and MICROPAS have been used to simulate the effect of trees on energy use. DOE-2, public domain software developed by James J. Hirsch and Associates in collaboration with LBNL, simulates the hourly energy performance of a building depending on climate, building envelope, equipment use, and occupant behavior. The user can account for the trees by altering weather files that incorporate the trees’ effect on microclimate and can modify the building description file to include the surrounding tree canopy. MICROPAS is a commercial building analysis tool that requires shading data from SPS in order to incorporate the effect of tree shade. The “lookup table approach” of MICROPAS allows a user to interpolate the results from a typical configuration to match a desired specific case, and it can also be used to scale up the analysis from the site to a broader scale, such as the neighborhood or even larger (Simpson 2002).
The Urban Heat Island Group at LBNL is one of the most active research groups examining the effect of vegetation and surface coverage (especially albedo) on microclimates and building energy use, ranging from a single building up to a global scale. This group has produced several noteworthy simulations. Using the program DOE-2.1C, which considers a tree canopy to be “exterior building shade” with a determined geometry and transmissivity, researchers have estimated the changes in solar gain, wind speed, and evapotranspiration, and calculated the change in the energy used by a building. Huang et al. (1987) found that trees have a significant impact on reducing cooling loads, regardless of their placement. They calculated that adding 25 percent to the tree cover around a building, even if the trees are not optimally located, could save 40 percent of the annual cooling energy use for an average house in Sacramento, and about 25 percent in Phoenix and in Lake Charles. With optimum tree placement, they found the savings would be further increased to more than 50 percent in Sacramento and 33 percent in the other two cities. In four Canadian cities—Toronto, Edmonton, Montreal, and Vancouver—Akbari and Taha (1992) used a similar approach to investigate the effects of trees and white surface on energy use for heating and cooling. For Toronto, increasing the vegetative cover of a neighborhood by 30 percent (about three trees per house) and increasing the albedo of the houses by 20 percent (from moderate-dark to medium-light color) would reduce heating energy by approximately 10 percent in urban houses and 20 percent in rural houses, and reduce cooling energy by even more (40 percent and 30 percent). The total annual savings are greater in rural areas ($60–$400 in 1992 dollars) than in urban areas ($30–$180 in 1992 dollars).
Rosenfeld et al. (1998) used a similar approach for Los Angeles and reported that not only is tree shade most effective in reducing building energy use, but the indirect savings from mitigating the urban heat island effect through evaporative cooling is also significant. He stated that the reduction in demand for air-conditioning (cooling) at peak periods is about 1.5 GW in Los Angeles (more than 15 percent of the city’s existing demand for air-conditioning energy) and 25 GW in the entire United States, yielding potential annual benefits of about US $5 billion by the year 2015 (i.e., after seventeen years of implementing the study’s recommendations). Akbari, Pomerantz, and Taha (2001) estimated that about 20 percent of the national cooling demand could be avoided by implementing a large-scale “cool communities” program that includes cooler, higher albedo surfaces (roofs and pavement) and urban trees. Employing the results from Rosenfeld et al. and other previous studies, Akbari (2002) estimated how the direct effects of shade trees and indirect effects of community cooling would lead to reduced emissions of CO2.
McPherson and Simpson of the USDA Forest Service have undertaken many simulations using SPS and MICROPAS to investigate the impact of trees on space-conditioning energy use (both heating and cooling). Assuming a typical one-story house in four cities—Madison, Wisconsin (a cold–humid climate); Salt Lake City, Utah (cold–dry); Tucson, Arizona (hot–dry); and Miami, Florida (hot–humid)—McPherson, Herrington, and Heisler (1988) simulated the effects of vegetation on heating and cooling energy use via two paths: irradiance reduction (shading) and wind reduction (windbreak). As one might expect, irradiance reduction has a negative impact in cold climates but a positive one in hot climates, while wind reduction has a positive effect in cold climates but a negative one in hot climates. In the two cold cities, Madison and Salt Lake City, dense shade from conifers would increase annual heating costs by as much as $128 (21 percent) and $115 (24 percent) in 1988 dollars, respectively, but the shade from leafless deciduous trees would be less significant. Reducing wind speeds by 50 percent would reduce annual heating costs by 11 percent in Madison and 9 percent in Salt Lake City. These effects would be reversed in cities with a hot climate. Reducing wind speeds by 50 percent would impede ventilation and raise the annual cooling costs by 23 percent in Tucson and by 17 percent in Miami, but providing dense shade on all surfaces in these cities would reduce annual space-cooling costs by $155–$249 in 1988 dollars (53–61 percent) and peak cooling loads by 32–49 percent. Shade on the roof and the west wall had the greatest impact on reducing cooling loads, while shade on the south and east walls had the greatest impact on reducing heating loads. These results suggest that designers can have the greatest impact by planting trees to shade the roof and west wall while minimizing winter shade on the south and east walls.
However, designers should not generalize the landscaping strategies, as the interactions between shade and wind reductions are complex, especially in temperate climates. In another simulation, Simpson and McPherson (1996) continued to examine how tree orientation affects residential energy use for air-conditioning (cooling) and heating over a range of climate zones and degrees of building insulation in California. They found that the shading benefits from a south side tree were greater than the heating drawbacks in winter in almost all places in California, save for those places with little demand for summer air-conditioning. For all climate zones and insulation levels considered, a tree shading a west wall showed the largest savings, for both annual (kilowatt hour) and peak (kilowatt) energy use. The next largest savings were from trees on the southwest (annual and peak) or east side (annual only). Three trees—two on the west and one on the east side—reduced annual energy use for cooling by 10–50 percent (200–600 kWh) and peak electrical use by up to 23 percent (0.7 kW).
Researchers have used simulations of tree orientation to assess the cost-effectiveness of urban tree-planting programs in California (Hildebrandt and Sarkovich 1998; McPherson and Simpson 2003; Figure 5). Simpson and McPherson (1998) evaluated the effectiveness of the Sacramento Shade program by investigating the trade-off between cooling benefits (welcome shade in summer) and heating penalties (unwelcome shade in winter) from a random sample of 254 residential properties selected from the 20,123 program participants in 1991–1993. Averaged over all homes, the program added 3.1 trees per property, and each tree reduced cooling energy use by 7.1 percent annually (a savings of $15.25) and by 2.3 percent peak, but increased the annual heating load by 1.9 percent ($5.25), for a net shade-related savings of $10.00 in 1998 dollars. On the other hand, when considering reduced wind speed, they found an annual cooling penalty of $2.80 per tree and a heating savings of $6.80 per tree, yielding a net wind-related savings of $4.00 per tree in 1998 dollars. The total annual savings (shade- and wind-related) was therefore $14.00 per tree, or $43.00 per property in 1998 dollars.
Typical effects of trees planted around a building (Northern hemisphere).
Since the late 1990s, McPherson and Simpson have been enlarging the scale of their analyses to measure the regional impact of urban trees on climate and energy savings. Simpson (1998) extended the results from previous simulation studies that assessed energy savings from trees on a single building, to the regional scale of Sacramento County, California. This simulation divided the county into Sub-Regional Assessment Districts (SubRADs), and all of the data for each variable (energy use, number of buildings, building ages, tree cover, and tree density) were combined for each SubRAD to estimate the total impact on space-conditioning energy use. Aiming to develop a simpler method to simulate large numbers of houses, Simpson (2002) also developed a “lookup table” that included energy savings for typical tree types (based on their size and shape), relations of trees to buildings (distance and direction from buildings), and the frequency of trees at those locations. He tested this method by comparing results from the “lookup table” to detailed simulations of 178 homes in Sacramento.
Statistical Analysis
Whereas many active research efforts into tree-related energy savings have relied on simulation, few studies have used actual energy-billing data. The largest barrier in using energy-billing data is the difficulty of accessing data in a disaggregated form (by individual household or dwelling unit) due to the confidentiality of private utility customers. In order to acquire these data, researchers must get approval from sample households or use indirect methods to protect customer information (such as aggregation or employing pseudo-addresses). Furthermore, it is extremely challenging to control for other factors that affect energy use (e.g., occupant behavior), as their reporting records are most often not based on individual properties. Obtaining enough reliable information is uncertain even when collecting information through surveys.
In spite of these challenges, some empirical research has detected patterns of how vegetation impacts energy use. Láveme and Lewis (1995) first looked at the effect of vegetation density on energy use for space-conditioning (both gas and electricity) by analyzing 101 single-family homes in Ann Arbor, Michigan. They characterized the vegetation at each house as one of the three levels—low, medium, or high strata—and reported that the differences in energy use among the three strata are noticeable but not statistically significant. In addition, they surveyed individual homeowners to control factors relating to appliances, structures, and behavior, but they were uncertain about the reliability of information on the most influential factors. Pandit and Laband (2010a, 2010b) conducted a large-scale empirical study for Auburn, Alabama. They found that in addition to shade coverage (physical extent of shade), the density of the shade has a statistically significant effect in reducing summertime electricity consumption, as compared to no shade. However, they found morning shade in wintertime increases electricity consumption.
Like simulation research, most empirical studies have focused on how trees or other plants reduce a building’s cooling energy load during the summer. Evaluating the effectiveness of various energy conservation measures in Phoenix, Arizona, Clark and Berry (1995) included tree plantings (three trees on average) as one treatment to reduce cooling energy. Their regression analysis showed that although trees do reduce energy use, the effect did not appear to be statistically significant. Jensen, Boulton, and Harper (2003) used remote sensing techniques to measure the leaf area index (LAI) of the urban forest in Terre Haute, Indiana, and investigated the relationship between LAI and household electricity use during the summer. Using a simple scatter plot, they reported an inverse relationship between LAI and summertime energy use—less energy used by households that had more trees—but this correlation was not statistically significant, and they did not employ any other controlling variables in their study. Donovan and Butry (2009), also interested in summertime energy use, examined the relative impact of different tree configurations and the size of tree cover. Using a regression model on 460 houses in Sacramento, they concluded that trees planted in the west quadrant (within 20-, 40-, and 60-ft buffers) and the south quadrant (within 20- and 40-ft buffers) significantly reduced summertime electricity use, but tree cover in the north quadrant (within a 20-ft buffer) increased electricity use, likely due to those trees reducing wind flows and release of heat at night, or their configuration creating more demand for lighting.
Summary
This article reviewed the literature on how urban form variables affect residential energy use, particularly energy for space-conditioning (heating and cooling), and focused on the effects of house size, housing type, density, community layout, and planting and surface coverage. Based on previous studies using experiments, simulation models, and statistical analyses, summaries of findings and suggestions for future research in each research area are provided, as follows:
Housing type and size
Housing type (single-family attached or detached, and multifamily of varying numbers of units) does affect energy consumption for space-conditioning. Despite new building codes for single-family houses, single-family housing in the United States uses more energy than multifamily housing per household and per person, as the floor area of new single-family detached houses has increased since 1978 and has outpaced the improvements in efficiency.
Housing type is strongly correlated with housing size and neighborhood density. Single-family houses in suburbs are generally larger than urban multifamily homes and require more heating and cooling energy use. Single-family houses in lower-density developments also contribute to a greater urban heat island effect per household than multifamily housing in compact developments, as a result of the larger exposed surface area on larger lots.
Density
Higher-density development consisting of multifamily housing leads to comprehensive energy savings in cities due to lower space-conditioning energy use per household and per capita, reduced urban heat island effects per household, and less energy used for vehicle travel. Compact urban form affects microclimate in several ways, including solar access, natural ventilation, heat transfer, and urban heat island effect. Given the trade-offs among these effects in different climates, future research must evaluate various options for density more thoroughly, in order to find urban design strategies that produce net benefits for each climate.
The pure impact of density on energy use is still unclear. Researchers must recognize the limitations of energy data sets for statistical analysis and design their studies carefully. Obtaining richer data sets at a disaggregated level of geographic information would help close this gap in the literature. Assessing more various types of “density,” encompassing both compactness and population density, will help researchers incorporate real-world variables into their models.
Community layout
Solar neighborhood guidelines suggest that an east–west street orientation is generally desirable because the lots are oriented north–south, which accommodates more south-facing buildings to maximize solar access in a neighborhood. However, few studies have actually proved this guideline.
More empirical and simulation studies are necessary to quantify how certain variables (street orientation, horizontal and vertical building layouts, site coverage) affect residential energy use. Researchers can also conduct quasi experiments (experiments occurring in a “real” setting, not as rigorously controlled by researchers as those performed in laboratories) to compare residential energy across different densities and layouts in real-world environments. Future research should focus on larger scale site layout designs (not individual buildings) that conserve energy and where political entities require actual developments to follow such layouts.
Planting and surface coverage
Many experiments, simulations, and statistical analyses show that tree planting is an effective means to reduce demand for cooling and heating energy in multiple ways: controlling solar access, evapotranspiration, natural ventilation, and moderating the urban heat island effect. Although trees may increase heating loads by creating unwanted shade, tree planting appears to have a positive net effect on saving space-conditioning energy.
The energy-saving effect of trees, especially for cooling, has been intensively proved and implemented in several municipal tree-planting programs. Compared to experiments and simulations, much less empirical research has been conducted, due to limited access to energy-billing data.
Reducing exposed surface area (both impervious and pervious, such as lawn cover) through higher-density development, reducing lot sizes, and increasing tree canopy cover is recommended to reduce urban heat island effects. Increasing the albedo of an area, including using lighter colored materials for hard surfaces (roofs and pavement), can be a cost-effective way to reduce energy use in both existing and new construction.
Discussion and Conclusion
Energy-efficient site design and neighborhood planning guidelines were developed during the late 1970s and early 1980s. Some studies have evaluated the effectiveness of these guidelines, but not enough for a thorough evaluation. The research articles published in peer-reviewed journals since the late 1970s have clearly shown that the scale of studies is increasing. Beyond a single building, researchers are also interested in the effects of outdoor elements, such as building configuration, neighborhood density, neighborhood layout, and planting (both on individual lots and at the regional scale). While similar studies continue to be performed, not enough research has been done to clearly and accurately understand the full impact of urban form elements on energy efficiency.
Many research findings have supported the hypotheses that urban form variables across scales—housing type, density (compactness and population/dwelling unit density), street orientation, building configuration, and planting—affect the energy use for space-conditioning in a house. Due to the complex trade-offs across urban forms and climates, it is unlikely that one urban form is universally ideal for reducing such energy use. Although this study has described some of these trade-offs in terms of design principles, not all have been rigorously evaluated through empirical research.
With regard to the amount or popularity of research, this study finds a large variation among the urban form variables. The energy-related benefits of trees, especially on reducing cooling energy use through shading, are the most intensively studied. Recent studies on trees are more comprehensive and include other environmental and economic benefits, such as improving air quality and increasing property value, using a cost–benefit analysis or a life cycle assessment. On the other hand, the effect of community layout is the least studied. The variation in the intensity of research seems related to how broadly the particular design principle has been (or can be) implemented. With reinforcement from research, several municipalities have implemented citywide tree-planting programs (such as Sacramento, Los Angeles, Denver, New York, and Seattle). However, it is rare to find a residential development that specifically focuses on energy saving through a passive solar community layout (such as street orientation and building configuration), except for a few symbolic communities developed during an energy crisis (e.g., Village Homes, in Davis, California, which started its development in 1975).
As for the three methods of research evaluated in this study, simulation studies have been more popular than experiments or statistical analysis for studying the effect of urban form variables on space-conditioning energy use. Simulation provides easy control of variables and does not require collecting significant quantities of data. Empirical studies are becoming more popular; for example, several studies in urban planning, forestry, and economics have used large-scale statistical analysis to show patterns of energy savings or penalties similar to what previous simulations have shown. Experiments have largely been limited to evaluating the effect of trees.
When reading each study, readers must understand its results within the context of the methods used, geographic region, climate, and other specific conditions. Caution should be used when attempting to generalize or compare the results from one study to another. However, the majority of research shows that residences use less energy per household and per capita in compact development (attached housing at higher densities) than in sprawling development (detached housing at lower densities). More importantly, a compact urban form allows for dramatically less energy use in transportation and therefore contributes to a larger framework of urban sustainability. With careful design and planning, especially to ensure solar access and accommodate on-site generation of solar energy, compact development should come to be considered the most desirable urban form in general.
Future research must pursue a more interdisciplinary approach, in order to craft and implement practical energy-conserving designs, plans, and policies at a variety of scales. Beyond the energy efficiency of individual buildings, research should extend to the neighborhood scale or beyond to achieve greater energy savings. New quasi experiments on energy use at the neighborhood scale, as well as simulations and empirical studies that include urban form variables, demographic data, and disaggregated billing data, will reveal better ways to reduce residential energy demand. Future research should focus on the development of comprehensive methodologies to examine issues such as (i) the energy-related trade-offs among design strategies and (ii) the cost–benefit analysis and life cycle analysis of different scenarios of urban form. Future research should also go beyond the residential scope, to study the effect of urban form on energy consumption in commercial buildings (e.g., using the Commercial Energy Building Consumption Survey), as well as on-site generation of renewable energy, with a goal of achieving zero net energy performance in entire neighborhoods and cities.
This literature review has practical implications for planning energy-efficient neighborhoods:
Increase urban density (through multifamily and attached housing), reduce the sizes of units and lots, and weatherize existing units. These are proven ways to reduce the energy used for residential space-conditioning.
Provide more energy-related incentives and regulations for community design and planning. Form-based codes can be implemented to reduce the energy used for space-conditioning at the neighborhood and city scale. Rezoning to allow higher floor area ratios and open area ratios, maximizing north–south lots (with south-facing houses) through east–west street orientation, and allowing a modified siting for houses on their lots (for better solar access) can be effective tools for encouraging energy-efficient neighborhoods.
Plant more trees through municipal tree-planting programs, particularly as a way to reduce the demand for cooling energy. Given current financial shortages in many cities, priorities should be set that are appropriately tailored to the particular area. In hot–dry regions, for example, planting trees on the west side of a building must be encouraged first, due to the maximum benefit per cost they yield, followed by trees on the east side to gain additional useful shade.
Reduce the urban heat island effect through higher-density development with a minimum of exposed surfaces. Change counterproductive subdivision regulations, including lot-size restrictions, front yard setback requirements, and ordinances on tree planting and building configurations, in order to encourage smaller lots, smaller lawns, less impervious surfacing, and more shade. Use cooler surfaces, such as light-colored roofs, streets, and other impervious surfaces, as a cost-effective way to reduce energy use around both existing and new construction.
In addition to the numerous efforts in transportation planning already underway, integrate further strategies for reducing space-conditioning energy, such as community layout and tree planting, into comprehensive planning tools (e.g., PLACE3S), green development initiatives such as Leadership in Energy and Environmental Design in Neighborhood Development, or other form-based community codes.
Set planning goals for long-term density to achieve the best net energy savings in the residential, commercial, and transportation sectors. Base these density settings on comprehensive research findings that assess the life cycle costs of weatherization, tree planting, vehicles, and construction and maintenance of different types of urban forms.
This article reviews key climate-responsive urban design principles and describes research findings related to the impact of urban form variables on residential energy consumption. It also demonstrates the importance of energy efficiency technologies and the collaborative role of planners and designers in reducing the energy use of residential buildings. Using an interdisciplinary approach with smart planning, design, and technology, the communities of the twenty-first century will play an essential role in overcoming the challenges of global climate change while ensuring a high quality of life for their citizens.
Footnotes
Appendix:
The List of Peer-Reviewed Research Evaluating the Effect of Urban Form Variables on Residential Space-Conditioning Energy Use.
Note: EXP = experiment; SIM = simulation modeling; STAT = statistical analysis.
Study
Method
X or Input
Y or Output
Study Area/Scale
Notes/Major Findings
Housing type, density, and community layout
Hirst, Goeltz, and Carney (1982)
STAT
Demographic, energy-related structure, heating equipment and appliances, fuel price
Household energy consumption (space-heating fuel and electricity)
4,081 individual households in the United States
Pioneer attempt to use disaggregated data (1978 US National Interim Energy Consumption Survey); floor area is a key determinant of energy use for space-conditioning
Paradis, Faucher, and Nguyen (1983)
SIM
Street orientation
Heating and cooling loads
A typical house in Québec city, Canada
Optimal street orientation (20° east of south) could reduce the maximum instantaneous heating load by 24–70 percent and could reduce annual household energy use by 16.5 percent
Lariviere and Lafrance (1999)
STAT
Urban density, climate, and socioeconomic variables
Annual city electricity consumption per inhabitant
The forty-five most populous cities of Québec, Canada
The effect of population density on electricity consumption appears less significant than its effect on gasoline use
Kahn (2000)
STAT
Location (central city or not), demographic, climate
Household energy consumption (space-heating fuel and electricity)
7,040 individual households in the United States
Attempt to investigate the impact of suburbanization; used 1993 US Residential Energy Consumption Survey (RECS); suburbanites do not consume more energy than their central city counterparts
Steemers (2003)
SIM
Variables that represent urban density (compactness)
Changes in space-heating; annual energy use per area
Office buildings in London, United Kingdom
Pioneer simulation study to address the trade-off impacts of urban density on buildings and transport
Holden and Norland (2005)
STAT
Land use (including housing type and density), sociodemographic factors
Household energy consumption
120 households in each of the eight residential areas in the Greater Oslo Region
Addressed the challenges of compact urban form; residents in single-family housing use about 50 percent more energy than residents in multifamily housing. However, the difference in energy use between single-family housing and multifamily housing is reduced after 1980
Ratti, Baker, and Steemers (2005)
SIM
Variables that represent urban geometry (compactness)
Average energy consumption
Central areas in London, Toulouse, and Berlin
Used lighting and thermal (LT) model; the variation of energy use in different urban density and geometry was about 10 percent
Ewing and Rong (2008)
STAT
Demographic, energy-related structure, climate, neighborhood density, and fuel price
Household energy consumption (heating and cooling)
3,737 housing units from cities, towns, and suburbs in the United States
Used 2001 RECS, used path and hierarchical model: the impact of density (county sprawl index, Ewing et al., 2003) on house size, type, urban temperature, and residential energy use
Kaza (2010)
STAT
Demographic, energy-related structure, climate, neighborhood density, fuel price
Household energy consumption (heating and cooling)
4,382 individual households in the United States
Used 2005 RECS, used quantile regression; housing type matters more for space-conditioning than housing size. Some, not all, types of multifamily housing offer almost as much savings as reducing housing area by 100 m2, compared to single-family houses
Min, Hausfather, and Lin (2010)
STAT
Demographic, climate, housing characteristics, fuel
Household energy consumption (heating and cooling)
United States
Using 2005 RECS, estimated space-heating and space-cooling, water heating, and appliance energy end uses, fuel used, and carbon emissions at a zip code-level resolution for the entire United States; housing size was included as “number of rooms” and it showed a statistically significant impact on both heating and cooling energy use
Vegetation and surface coverage
Parker (1983)
EXP
Planting/no planting
Average rate of energy consumption
An insulated mobile home in Miami, Florida
Planting can reduce cooling energy by more than 50 percent during warm summer days
Thayer, Zanetto, and Maeda (1983)
SIM
Continuous row of street trees (three deciduous and one evergreen) south of the dwelling
Annual energy cost
Three test houses (solar, conventional, and solar retrofitted), Sacramento, California
Significant net increase in annual energy costs when street trees are placed in the zone directly south of a solar house
Rudie and Dewers (1984)
STAT
Tree shade on roofs
Summer electricity use
113 residents in College Station, Texas
The amount of shade, roof color, and wall color were significant determinants of residential energy consumption; shade had the greatest effect on reducing energy demand
DeWalle, Heisler, and Jacobs (1983)
EXP/STAT
Deciduous grove/pine forest
Cooling and heating energy use, air infiltration
A mobile home in Central Pennsylvania
In summer, one deciduous stand reduced cooling energy needs by 75 percent less than the open. In winter, one deciduous stand saves heating energy needs by 8 percent, but and with greater heating energy from shading in a dense pine forest. Percentage changes caused by the forest on reducing summer cooling and winter air infiltration were statistically significant
DeWalle and Heisler (1983)
EXP/STAT
Windbreak (a single row of white pine trees)
Heating energy use, wind velocity, air infiltration
A mobile home in Central Pennsylvania
Placing a small mobile home at about one tree height (3 m) from a windbreak results in reducing winter space-heating up to 18 percent
Huang et al. (1987)
SIM
Tree canopy
Summer cooling energy saving
One-story prototype house in Sacramento, Phoenix, Lake Charles, Los Angeles
Used DOE-2.1C. DOE-2.1C considers tree canopy to be “exterior building shade” and calculates potential energy savings using an estimated reduction of solar gain, wind speed, and evapotranspiration; trees have a significant impact on reducing cooling loads even if not optimally located
McPherson, Herrington, and Heisler (1988)
SIM
Irradiance and wind reduction due to vegetation
Heating and cooling costs
Typical one-story ranch home in Madison, WI; Salt Lake City, UT; Tucson, AZ; and Miami, FL
Used Shadow Pattern Simulator (SPS) and MICROPAS. Examined the effects of vegetation on heating and cooling energy use through two paths: irradiance reduction and wind reduction
McPherson, Simpson, and Livingston (1989)
EXP
Landscaping: turf, rock mulch with a foundation planting of shrubs, and rock mulch with no plants
Cooling energy use
1/4-scale model buildings in Tucson, Arizona
The house with no vegetation consumed 25 percent more electricity for air-conditioning than the house surrounded with turf and 27 percent more electricity than the house with shrubs
Akbari and Taha (1992)
SIM
Vegetative cover and white surface
Heating and cooling energy saving
Row house, detached one-story and two-story houses in Toronto, Edmonton, Montreal, and Vancouver, Canada
Using DOE-2.1D, they calculate potential energy savings using an estimated reduction of solar gain, wind speed, and evapotranspiration
Clark and Berry (1995)
STAT
“[P]lanting large trees (three trees on average) to shade sunstruck sides of houses” and other cooling energy-saving treatments
Weather-normalized residential electricity use
148 residents in Phoenix, Arizona
A study of evaluating the effectiveness of various energy-conservation measures; through regression analysis, they report that tree shading reduces energy use, but the effect is not statistically significant at a level of 10 percent
Láveme and Lewis (1995)
STAT
Vegetation density with three distinct levels (low, medium, and high strata)
Residential space-conditioning energy use (both gas and electricity use)
101 single-family homes in Ann Arbor, Michigan
First empirical study that investigated the impact of vegetation on use of both gas and electricity; differences in energy use among strata were noticeable but not statistically significant
Simpson and McPherson (1996)
SIM
Tree shade with various configurations
Residential air-conditioning (cooling) and heating energy use
Eleven climate zones in California
Examined the relative effects of different tree orientations on residential air-conditioning (cooling) and heating energy use for a range of building insulation levels and climate zones in California
Akbari et al. (1997)
EXP
Shade trees/no trees
Cooling energy use
Two homes in Sacramento, California
Tested one home with eight large and eight small shade trees, and the other with no trees, and found a 30 percent reduction of the cooling energy use due to the tree shade
Rosenfeld et al. (1998)
SIM
Cool surfaces (cool roofs and cool pavements) and urban trees
Residential air-conditioning (AC/cooling) bills and smog concentration
Los Angeles, California
Tree shade is most effective in reducing building energy use, but the savings due to mitigating the urban heat island effect through evaporative cooling are also significant
Simpson (1998)
SIM
Tree cover, tree density, number of buildings, building vintages
Space-conditioning energy use
Sacramento, California
Extended results from previous simulation studies that assessed tree impacts on energy savings in a single building to a regional scale; Sub-Regional Assessment District (SubRAD) was used as a unit of analysis
Simpson and McPherson (1998)
SIM
Tree shade
Cooling and heating energy load
254 residential properties in Sacramento, California
Evaluated the effectiveness of the Sacramento Shade program by investigating the trade-off between heating penalties and cooling energy savings from 1991 to 1993
Akbari, Pomerantz, and Taha (2001)
SIM
Cool surfaces (cool roofs and cool pavements) and urban trees
Cooling energy use and smog concentration
Eleven metropolitan statistical areas in the United States
Used DOE-2.1E and reported that about 20 percent of the national cooling demand can be avoided by implementing a large-scale “cool communities” program that includes cool surfaces (roofs and pavements) and urban trees
Akbari (2002)
SIM
Shade trees
Carbon sequestration
Los Angeles, California
Estimated reduction of CO2 emissions due to the direct effects of shade trees in reducing building energy use, as well as indirect effects of community cooling
Simpson (2002)
SIM
Tree types, configuration, and frequency
Cooling and heating energy use
178 homes in Sacramento, California
Aiming to develop a more simplified method for simulating large numbers of houses, developed a lookup table that includes energy savings for typical tree types (e.g., size and shape) and locations around buildings (tree location by distance and direction from buildings), combined with the frequency of trees at those locations
McPherson and Simpson (2003)
SIM
Tree canopy cover
Cooling and heating energy load
Eleven California regions
Evaluated the cost-effectiveness of tree shade programs in eleven California regions
Jensen, Boulton, and Harper (2003)
STAT
Urban forest leaf area index (LAI)
Summer household electricity use
118 sampling points in Terre Haute, Indiana
A scatter plot showed the inverse relationship between LAI and summertime energy use, but the correlation was not statistically significant
Donovan and Butry (2009)
STAT
Tree configuration and crown area, dwelling characteristics
Summer household electricity use
460 houses in Sacramento, California
First examined the relative impact of different tree configuration and size of tree cover. Trees planted in the west quadrant (within buffers of 20, 40, and 60 ft) and the south quadrant (within buffers of 20 and 40 ft) significantly reduced summertime electricity use; in contrast, tree cover in the north quadrant (within a buffer of 20 ft) increased electricity use
Laband and Sophocleus (2009)
EXP
Full shade/full sun
Electricity consumption for air-conditioning
Two buildings in Beauregard, Alabama
A building situated in dense shade uses 2.6 times less electricity for cooling (about 62 percent energy saving) to 72°F during April to September in 2008, than one exposed with full sun
Pandit and Laband (2010a, 2010b)
STAT
Tree shade, weather, dwelling characteristics, occupant behaviors
Summer household electricity use
160 residents in Auburn, Alabama
Shade on the average house reduced electricity usage by 4.8 percent, as compared to a house with no shade
Pandit and Laband (2010b)
STAT
Tree shade, weather, dwelling characteristics
Summer and winter household electricity use
160 residents in Auburn, Alabama
Denser shade, and/or shade with larger coverage, provides a statistically significant reduction in summertime electricity consumption as compared to no shade; morning shade in wintertime increases electricity consumption
Acknowledgments
The author deeply thank Professors John Radke, Edward Arens, and Louise Mozingo from the College of Environmental Design at UC Berkeley for their expertise and guidance. The author also thank the editor and three anonymous reviewers for their constructive comments.
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 research was partially supported by the Beatrix Farrand Dissertation Grant from the University of California at Berkeley and the Eloise Gerry Fellowship from the Sigma Delta Epsilon, Graduate Women in Science (SDE/GWIS).