Cities are responsible for over 75% of energy consumption (Grubler et al. 2012) and 80% of CO2 emissions globally (Lombardi et al. 2017). A recent International Energy Agency (IEA) estimate suggested that residential and commercial energy consumption accounts for approximately 37% of global energy use (IEA 2020). In response to a growing number of ambitious commitments from local and regional governments, a convergence of transdisciplinary research toward reducing energy and emissions in urban areas has emerged (Gober & Gober 2018; Hughes et al. 2018; Masson et al. 2014). At the interface for climate action, many cities already have aspirational goals and sophisticated toolkits to address climate change, particularly mitigating greenhouse gas (GHG) emissions.
There is a proliferation of modeling methods for measuring energy use and GHG emissions from the individual building scale (Ahmad et al. 2018; Fumo 2014; Harish & Kumar 2016; Hensen & Lamberts 2011; Ryan & Sanquist 2012) to large complex urban neighborhoods (Ferrando et al. 2020; Johari et al. 2020). However, regardless of its scale and resolution, energy modelling is a complex and varied field, with individual models of different scopes, methodologies and objectives (for a review, see Hall & Buckley 2016; Hong et al. 2020). Generally, models encapsulate the relationships between energy supply and/or demand factors to understand the impact of potential interventions at different scales. While there is a range of models that combine disciplines and methodologies, there are two main categories of energy modelling not often well reconciled: city-scale energy–economy models and building-scale design–planning models. Energy–economy models understand energy output in relation to high-level economic variables, market behavior, and sectoral relationships. Due to their scale and resolution, energy–economy models are primarily used to model and assess provincial or federal policy strategies. In some cases, they contain explicit representations of energy-using technologies. At a much finer resolution, design–planning models encompass a physical building energy model (BEM) or urban building energy model (UBEM), a category of energy simulators that are often spatially explicit and heavily reliant on engineering methodologies to simulate energy usage (Reinhart & Cerezo Davila 2016). Both BEM and UBEM are common tools among architects, engineers, and planners to assess the energy impacts of design decisions. Although they vary conceptually and methodologically, energy–economy and design–planning models capture intertwined drivers of energy and emissions at different scales.
Energy issues are both spatial and aspatial. For a residential example, the carbon intensity of electricity used for home heating is aspatial, while factors such as the positioning of buildings, how they relate to other buildings (e.g. shading), the weather, and consequently energy use are inherently spatial. In most cases, energy–economy models are aspatial, operating from top down, at global, national, and subnational (state, province, region, and city) levels to estimate the energy use and emissions reduction from climate-energy policies. These models simulate how market and individual behavioral factors change over time in response to GHG-related policy decisions. They particularly focus on the turnover (i.e. retirement) of capital stock, i.e. infrastructure, buildings, industrial plants and equipment, electricity generation and other energy supply equipment, vehicles and other transport equipment, and appliances and other household durable goods. This methodology simulates supply and demand between economic sectors by bringing them into equilibrium, known as ‘clearing’ the market (Bhattacharyya 1996). An example is UKENVI, a 25-sector model parameterised on data from the UK (Allan et al. 2007). Importantly top-down energy–economy models take into account previous market behavior to predict how energy usage will respond to future policy experiments (Nakata 2004). While they can have considerable regional detail, they are typically aspatial and require specific geographic contexts. This is challenging for urban designers and planners to assess spatially explicit urban changes in energy–economy simulations.
From bottom-up, building-scaled energy models develop building archetypes based on physical building parameters to simulate proposed structure performances common to the engineering and architecture disciplines (Sokol et al. 2017). For example, the Oak Ridge National Laboratory (DOE/Oak Ridge National Laboratory 2019) was able to model every single building at a 15-minute interval in Chattanooga, US. Such a model explores the components of future experiments using a detailed building construction and operating system represented in the model (Nakata 2004). That said, these methods require highly detailed parameters about the proposed buildings and extensive computing power, thus they have limited their utility in future urban design and planning when the detailed information is not yet available or quickly overwhelms the computing systems. At the intermediate scale, the UBEM consolidates details while scaling up building energy modeling to the neighborhood or city with generalised building archetypes (e.g. a Shoebox model; Dogan & Reinhart 2017). Examples of a bottom-up model are MESSAGE and MAED-2. MESSAGE models energy systems as a network where energy flows from primary resources through to various end-use technologies (Gritsevskyi & Nakićenovi 2000). Its goal is to optimise the evolution of the energy system under assumptions about future experiments and technological attributes such as unit size, efficiency, costs, and emissions. MAED-2 determines energy demand within end-use categories through mathematical relationships with social, economic, and technological factors that affect each end use. Recently, Dogan et al. (2016) and Dogan & Reinhart (2017) made significant progress in UBEM, improving the authenticity and accuracy of neighborhood-scale simulation. Quan et al. (2015) also stressed the importance of recognising spatial urban context in UBEM by implementing a geographic information system (GIS) in the urban energy performance coefficient (EPC) calculator. Overall, these methods use simplified building archetypes to generalise energy performance data, allowing models to operate over large areas with relatively irresolute data (Reinhart et al. 2016). However, as they do not directly take into account economic factors and market behavior, they often ignore the market dynamics in adopting new technologies.
In reality, regardless of the modeling approach used, solutions that work in one city or neighborhood may not be suitable in another; those for large metropolitan areas may not apply in small communities. Exactly which policy instruments, such as densification, higher building energy standards, and fuel switching, bring the most emissions-reduction outcomes is not well understood. Policy instruments employed by states, provinces, and municipalities are both spatial (e.g. infilling) and aspatial (e.g. incentivising fuel switching and retrofitting) strategies. Many policies target sectors and implementation strategies separately, overlooking the potential interrelationship among those policy instruments. As a result, strategies in one sector may affect initiatives in another in positive or negative ways, introducing great uncertainties in assessing policy interventions (Anderson et al. 1996). Energy models have improved tremendously over the past decade (for reviews, see Johari et al. 2020; and Reinhart & Cerezo Davila 2016) and are able to undertake timely, highly complex simulation tasks at the individual building scale (e.g. DOE/Oak Ridge National Laboratory 2019). However, what is often neglected is the ability to integrate and reflect higher level energy–economy models with the physical energy-simulation process. Many higher level models often use aggregated non-spatial data, creating knowledge gaps and technical challenges among planners, energy modelers, and policy-makers. Therefore, high-level economy models do not capture the effect of spatial nuances on building energy use and emissions in the built environment, while UBEMs cannot endogenously assess policy and market impacts such as incentives, taxes, or many common regulatory policies. This creates a unique research opportunity seeking a workflow that facilitates the integration between energy–economy models and spatially explicit UBEMs.
Ultimately, such workflow must have sufficient spatial awareness, yet also being highly flexible for various types of communities and neighborhoods. Planners require models that can iteratively assess the outcomes of parcel, block or neighborhood scale. Building energy usage depends on economic, behavioral, and physical considerations, from fuel prices to household behavior to mechanical efficiency, and research papers have been written examining energy issues from a wide variety of disciplinary lenses (Sovacool 2014). Decisions on energy usage and supply are multiscale and multifaceted, including individual household and national government. Therefore, such integrated workflow also needs to reflect the disciplinary complexity, scale of energy solutions, and be adaptive to communities of varying sizes, resources, policy goals, and resources.
There are synergies in linking high-level energy–economy models and UBEMs (Bourdic et al. 2012) to reflect dynamics sensitive to both planners and energy managers. This paper develops a hybrid modelling workflow for integrating these two types of energy models. Extending the work of Salter et al. (2020), the ‘sandbox’ approach and hybrid model are tested at the scale of a neighborhood. Specifically, this workflow integrates a city-level version of the CIMS energy–economy simulation model developed by Rivers & Jaccard (2006) with an urban modeling interface (UMI) (Reinhart et al. 2013). This paper demonstrates in detail the methodology using a case study urban form ‘sandbox’ model in Vancouver, BC, Canada.
The following section describes the energy–economy model (CIMS) (Section 2.1), UBEM UMI (Section 2.2), the integration of CIMS and UMI, key modelling assumptions, and the construction of a neighborhood sandbox model (Sections 2.4–2.6). The application of this workflow (Figure 1) is then demonstrated as a case study in Vancouver (Section 3).
CIMS is a technologically explicit model, meaning it keeps track of vintages of capital stocks of different efficiencies as well as other vintage-specific characteristics. It is also behavioral realistic in that it incorporates consumer preferences based on empirical research to better simulate households and market decisions. CIMS combines technological detail elements found in traditional ‘bottom-up’ optimisation models with elements of microeconomic decision-making found in ‘top-down’ models as well as macroeconomic feedbacks in provincial and national applications of the model (Rivers & Jaccard 2006). CIMS residential and commercial building sectors simulate the turnover of building shells (e.g. walls and windows) and heating, ventilation and air-conditioning (HVAC) equipment (e.g. water heaters, furnaces, fridges, and lighting) over time. In these simulations, the model incorporates behavioral realism by considering not only financial costs but also the non-financial costs that consumers and firms face when purchasing new technology and equipment (e.g. differences in the perceived risk or quality for new low- or zero-emission technologies). The sector can operate at provincial, regional, or city scales. CIMS has been used to model the effect of the urban land-use and transportation network policy (Jaccard et al. 2019), end-use technology efficiency, and carbon prices from municipal to national (Jaccard et al. 2019; Murphy et al. 2016).
Urban modelling interface (UMI) is an open-source UBEM developed in MIT’s Sustainable Design Lab, which conducts dynamic building simulations based on the US Department of Energy’s EnergyPlus simulation engine (Reinhart et al. 2016). UMI establishes sets of building archetypes (i.e. building templates) tied to generalised energy performance data, making it possible to model building energy data over large areas with relatively irresolute data (Reinhart & Davila 2016). The methodology in the present study uses the UMI template system (see the supplemental data online) to distribute 2.5D models representative of the building stock across the sandbox. UMI then employs radiance-based solar radiation modeling to calculate the incident radiation for each building surface.
The benefit of using UMI in this work is apparent. First, the use of Autozoner (Dogan et al. 2016) in UMI enables multi-zone building simulation, offering improved results than conventional single-zone models that often fail to represent the complex shading effects found at a neighborhood scale (Dogan & Reinhart 2017). Second, also known as the ‘shoebox’ model, UMI’s building templates are able to characterise the construction types, HVAC systems, fuel types, and local climate data of a give urban location. The concept of using abstract building archetypal models allows the UMI to automate the simulation process at the neighborhood scale (Dogan & Reinhart 2017). Third, UMI’s ability to efficiently accommodate a large volume of buildings enables the scale of this work, with the possibility to further extend to a larger area. For all building templates and model parameters, see the supplemental data online.
Salter et al. (2020) first introduced the ‘sandbox’ method at a block scale (400 × 400 m) to simulate urban form and building retrofit policy changes. There is a need to further this approach and use detailed, specific data on existing building technologies at the neighborhood scale. Differences in building technologies between communities depend on energy infrastructure, the relative costs of fuels, and climate. Furthermore, there are limited data that could outline a relationship between building variables, such as type or age, and the technologies within them, beyond broad categories such as commercial or residential buildings. In fact, such a relationship may not be particularly well defined or even exist at all. Upgrades or retrofits may happen many times over the course of a building’s lifetime as a result of mechanical failure, changes in ownership, adaptive reuse, or the availability of financial incentives to do so, which may complicate any clear-cut relationships existing at the time of construction. In smaller communities, the availability of certain technologies may also play a role.
In the urban design domain, a neighborhood consists of a composite of characteristics describing land-use patterns, mobility, and urban design and, to a certain extent, aggregated from repetitive patterns. At least in the context of North America countries, a city is essentially a tessellation of repeated patterns that are unique and undifferentiated in aspects such as block size, building types, street layouts, and neighborhood configurations (Lozano-Pérez 1990). Urban form pattern types simplify reality in that the ‘unnecessary’ real-world variables and structures are intentionally excluded enabling a more comparative, iterative, and systematic modelling process (Bosselmann 1998). Those patterns are common in academic research as a means to model cities (Gil et al. 2012).
The sandbox model provides a common ground, enabling a multi-scale model that links spatial and non-spatial parameters in energy–economy modeling and UBEM. This multi-scale model allows for an increasingly realistic representation of the integrated processes that influence energy-usage and emissions output, such as fuel choice, building design, mechanical systems, urban form, and behavior (Ratti et al. 2005). Following methods informed by Rode et al. (2014) and Salter et al. (2020), a sandbox model is created for an area of 1600 × 1600 m. The sandbox typifies a certain urban form, such as population density, parcel density, street patterns, block sizes, parcel sizes, and land-use proportions derived from the real neighborhood. Building types are assigned to each parcel, based on BC Assessment data, including building type, age, use, occupancy, construction type and date, and fuel sources to each parcel of land (BC Assessment 2019). The model exists in 2D ArcGIS and 2.5D Rhino for purposes of conducting UBEM. All building data are associated with the 2.5D model (Figure 2).
An integrated model combines the strengths of both modeling approaches, allowing for a wider range of interventions to be tested, and a nuanced understanding of the effects on urban energy and emissions. This work argues that downscaling coarse resolution energy models to the level of the finer scale models is feasible using Markov chains models (MCMs), a mathematical model that governs the change between states using probabilities. Based on market shares of HVAC and fuel options informed by CIMS, the MCM disaggregates the output of the energy–economy model CIMS to individual buildings probabilistically over several time steps (e.g. five years), simulating the spatial effect of policies on urban energy and emissions into the future. This method outlines future directions for building integrated, comprehensive energy models.
This paper uses the commercial and residential building sectors of the CIMS energy–economy model. Because Vancouver’s energy consumption has a minimal impact on overall national and subnational energy demand and prices, energy prices in these sectors are set exogenously rather than being calculated through CIMS’ macro-economic and energy supply-and-demand feedbacks. Similarly, floor space and building types are defined exogenously in both sectors. In addition to its appropriate energy–economy model framework, the CIMS model is well suited to this work because it produces detailed outputs of technology market shares over time that can be disaggregated spatially in a UBEM. The model operates within a Canadian context, and therefore includes technologies common in the Canadian market as well as behavioral parameters calibrated to this specific context with the UBEM model.
Energy and Materials Research Group (EMRG) researchers have detailed the methodology of CIMS (Jaccard et al. 2019; Murphy et al. 2016; Rivers & Jaccard 2006). This section presents an overview and key assumptions of the CIMS model. CIMS simulates the turnover of building shells and equipment over time through retirements and new acquisitions. In each five-year period, a portion of the existing building stock is retired according to an age-dependent function. As new capital stocks are required, technologies compete for market share based at several levels of decision-making. For example, at one level there is competition among different space-heating technologies, which are nested inside a higher level competition between different building shells. These competitions are calculated through the CIMS market share algorithm:
where the market share (MS) of technology j is dependent on its life-cycle cost relative to the life-cycle costs of all other K competing technologies. CIMS does not base this comparison of life-cycle costs on financial costs alone: it also includes non-financial or ‘intangible’ costs (i) that reflect technology-specific preferences (Rivers & Jaccard 2006). Financial costs are comprised of the technology’s capital cost (CC), maintenance and operating cost (MC), and energy cost (EC). Capital costs are annualised using a revealed private discount rate (r) specific to the group of technologies being competed. Last, a v parameter is used to represent market heterogeneity, influencing the relationship between technology life-cycle costs and market share.
CIMS’ behavioral realism comes from its three behavioral parameters, r, i, and v. Without these three parameters, CIMS would allocate market shares based on financial costs alone instead of accounting for household and market behavior. The revealed private discount rate, r, is the weighted average time preference of decision makers for a given energy service demand (Rivers & Jaccard 2006). While many engineering models contain social discount rates in their financial cost calculations, the inclusion of revealed private discount rates that can vary between energy sectors and services aims to improve the behavioral realism of the model.
The i parameter accounts for all non-financial costs or benefits associated with the use of a technology or energy service relative to a baseline conventional technology. Common examples of intangible costs include the inconvenience or the perceived risks of adopting a new technology, such as a ground-source heat pump or an ultra-efficient building shell. It is critical to consider intangible costs when assessing policies because individuals often do not make choices solely on financial cost minimisation. For example, if financial costs are the only costs that influenced consumer preferences, then it would be likely that everyone in a city would take public transit as their preferred mode of travel for long-distance trips. However, many people drive personal vehicles, take taxis or Ubers, or carpool with a friend or colleague, even though these are more expensive modes of transport. While taking public transit may be less expensive than driving a personal vehicle or taking a taxi, it entails additional intangible costs such as wait times, less comfort, and unpredictable schedules.
Finally, the market heterogeneity parameter, v, accounts for variation in consumer preferences–different individuals may choose different technologies when presented with identical options. The inclusion of a market heterogeneity parameter allows CIMS to divide the market share among competing technologies in a way that better reflects the diverse choices of consumers and businesses. The level of CIMS’ behavioral realism, as well as the level of confidence the researcher can have in the model, is dependent on the accuracy of r, i, and v. Therefore, it is important for these parameters to be grounded in empirical evidence. Over the past two decades, researchers have estimated the behavioral parameters in CIMS using revealed and stated preference techniques, literature review, and expert judgement.
User-defined CIMS residential inputs include population growth, gross domestic product (GDP) growth, people per dwelling projections, and dwelling type projections (e.g. single-family detached, low-rise multi-unit building, etc.). Similarly, for the commercial sector, future GDP growth and breakdowns of commercial building types drive the growth of commercial floorspace projections. CIMS technology database is sourced from public statistic agencies, energy utilities, literature reviews, industry associations, equipment manufacturers, and surveys of sector experts. Government policies are another type of user-defined input. CIMS can assess regulations on end-use technologies, fuels, subsidies, and taxes.
The sandbox described above serves as a base from which to iterate series of urban-form experiments with the future growth of the community to understand the energy implications of these strategies. A series of what-if experiments are developed including urban form, land-use pattern, and dwelling mix at a neighborhood scale, reflecting climate policy interventions (for details of Vancouver Case Study, see Section 3). Each experiment is grounded in local census data and existing building stock representing future population projections, possible policy interventions, and market conditions for future decades. The data used to build and establish the model allow the sandbox to reflect the conditions of the community it represents. Further input data structure the growth of the community. Fundamentally, population growth projections moderate how many dwellings are necessary to add to the model across experiments. Each experiment has a set of land-use, building, and transportation policies. The assigned spatial land-use policy dictates where new development is located within the sandbox. Dwelling mixes split new development into building types that reflect the planned or potential growth patterns of the community. Further building policies determine the energy efficiency of new construction. Transportation policies inform where and what types of new transportation infrastructure will be added to the community.
Overall, the sandbox reflects the new growth with sets of calculations, including a suite of metrics meaningful for urban planners to examine the success of each policy interventions at various scales. First, the data for each building allow for the calculation of its energy use and GHG emissions. Second, the physical model serves as a measured visualisation itself, showing how the look and feel of the community would change under different sets of policies.
An MCM is a process defined by the Markov property, whereby the future is conditional only on the present, not the past (Gagniuc 2017). The model environment is composed of a set of states and a matrix of transition probabilities between those states; in this matrix, the probability of moving to a new state (i.e. a new building technology or fuel source) is dependent upon the current state. The structure of the model environment is dependent upon the transition probabilities, which gives the model certain properties regarding the way states relate to each other. Change in the MCM can function in either discrete or continuous time. MCM methodologies have been used to simulate occupant behavior in buildings (Rysanek & Choudhary 2015) and it performs a similar role in this workflow.
In this work, an MCM is established to assign building design and technology change over time and across individual parcels of land. The MCM is parameterised so that the rate of technology adoption and building change across the simulated building stock of the UBEM matches closely against the regional predictions drawn by CIMS. In principle, the MCM is an emulator that seeks to disaggregate CIMS to the individual building scale. For example, consider that each building in the sandbox is an agent that undergoes its own Markov chain process. The likelihood of the building undertaking a renovation or being demolished is conditional on its specific physical and socioeconomic characteristics. If, in a particular year, it is chosen to undergo a renovation, the likelihood of adopting a technology such as a heat pump may also be conditional on the property’s physical characteristics, such as the age of the existing heating/cooling.
This MCM process can carry on until all salient new physical and economic characteristics of the property are defined. This chain can also repeat in subsequent years and elucidate the evolution of the property. Although CIMS does not model the decisions at the individual building scale, transitional probabilities, such as the relationship between heat pump adoption and dwelling characteristics, can be produced by CIMS. Thus, in the adopted MCM, its design and parameters are established so that it mimics, at the individual building scale, the consumer actions predicted by CIMS at the regional scale.
This emulation process occurs in three major steps: baseline establishment, retirement, and assigning change. The baseline establishment step assigns initial parameters to each property in the sandbox so that the sandbox is representative of the real-world modelled by CIMS. This includes assigning building type, age, and systems (BC Assessment 2019). The retirement phase establishes the likelihood of future changes occurring in the sandbox as a function of age, of either the property or the components within it. Each component of a building system, such as the building envelope/shell itself, has a lifespan, thus the retirement of a certain technology is conditional on the age of the component and its lifespan (see Equation 2 used by CIMS and MCM integration). Once retired, a technology needs to be replaced with a new component as determined by the technology market share from CIMS.
where Runyear is the base year for CIMS (i.e. 2000); Lifespank is the lifespan of technology k; and Basestockk is the basestock of technology k.
Baseline establishment sets up the baseline year (2020) of the sandbox, assigning building systems technologies to all the buildings in the model informed by measured energy data for the study community. First, CIMS initialises with data from the actual community and a baseline policy experiment is set up that includes federal and provincial energy policies and building code standards. The energy usage within the CIMS is validated against historic data for the study community (British Columbia Community Energy & Emissions Inventory 2012).
The purpose of the baseline establishment is to assign each type of building system technologies successively, from heating, water heating, large appliances, cooling, lighting, and small appliances to each building in the base year informed by the technology market shares from CIMS. The resultant MCM is an absorbing MCM where all states are transient, as the chain passes through once in order to assign a specific technology and then moves on to the next set of technologies. The chain begins with heating technology and terminates with small appliances.
This process controls the evolution of the building components in the sandbox according to the policies from CIMS. In CIMS, the retirement of building technology stocks (e.g. furnaces, lighting, etc.) creates demand that must be filled in each five-year time step by new technology stocks. The retirement process is driven by the age and average lifespan of the stocks of technologies. The adoption of technologies in the model is derived from the CIMS market share algorithm, driving technological change over time based on MCM. Capital and intangible costs for each technology are influenced by declining capital cost and declining intangible cost functions. The declining capital cost function links a technology’s capital cost in future periods to cumulative production, reflecting economies of scale and economies of learning that lower capital costs. The declining intangible cost function links a technology’s intangible cost in a given period with its market share in the previous period. This ‘neighbor effect’ reflects how consumer perceptions of risk decline as a new technology gains market share. CIMS reports on total and new market share for each technology in each five-year model time step.
As the sandbox has more detailed individual building-age information than CIMS, the retirement outputs in each year require an additional disaggregate process using change MCM. The decay function built into CIMS is converted into probabilities for each age of technology: the probability of a technology retiring increases rapidly as it nears its lifespan, reaching 50% that year, and 95% five years thereafter (Figure 3).
Accordingly, there exists a separate MCM for each building component, as each has different retirement probabilities due to the varying lifespan and stock building age. The states in the change MCM are therefore dependent on two variables: technology type and its age. These MCMs are time inhomogeneous, meaning that the transition probabilities change with each time step, given the increasing age of the technology and the changing market shares from CIMS. Figure 3 represents a simplified example of the change MCM for heating technology: the chain begins in the state corresponding to the characteristics of the current technology. In each time step the system’s current state has a certain probability of retiring or remaining with the same technology. If it moves to retirement, a new heating technology is chosen independently of which state the system was in previously. If the technology is not retired, it moves to a state with the same technology attribute, but a more advanced age.
Energy-use models are initially built in UMI for all existing building stocks in 2020 (Salter et al. 2020). Studies have demonstrated its capability for simulating building energy use at the neighborhood scale. Using the US DOE’s EnergyPlus as the energy simulation engine, the initial UMI simulation required two sets of critical inputs: building templates and a local weather profile. A total of 34 building templates (see the supplemental data online) are developed to represent the existing building stock (Salter et al. 2020). The BC Building Code (2018) is primarily used to simulate building envelope requirements by building type (window, wall, roof, floor, etc.) and Vancouver’s energy modelling guidelines (City of Vancouver 2018) are used to model building descriptors (occupant schedules, ventilation rate, infiltration rate, etc.). To account for local climate conditions, Vancouver weather data are acquired from EnergyPlus weather data (EnergyPlus 2005), including dry bulb temperature, dew point temperature, extraterrestrial horizontal radiation, global horizontal radiation, wind direction/speed, and total sky cover. In the subsequent experiment, the model uses a constant climate profile between 2020 and 2050.
For all future what-if experiments, the retirement process described above identifies buildings that are eligible for a complete teardown and reconstruction. New buildings constructed after 2020 are modelled to achieve the highest BC Energy Step Code (Energy Step Code Council 2017) for that building type. As a performance-based building code, the BC Energy Step Code incentivises new constructions to achieve higher efficiency by setting clear and measurable targets to builders. Compared with a traditional prescriptive building code, the Step Code enables much greater flexibility and encourages builders to make energy-efficient buildings with all possible technologies and fuel sources. Following the Step Code performance guideline (Energy Step Code Council 2017), energy-use values (i.e. total energy-use intensity–TEUI) are assigned to each new building based on their floor area and primary use. Because the Step Code is a performance-based metric, individual energy values for the HVAC systems are not computed for new buildings in this case.
The aim of the case study is to contextualise the proposed model workflow and, ultimately, to evaluate how different energy and emission-reduction policies may interact with urban-form changes on building energy use and emissions. In the province of British Columbia (BC), Canada, 55% of all greenhouse emissions originate in the built environment–the cities, towns, and villages where approximately 86% of the province’s population lives and works, and the building sector represents 41% of the total provincial GHG emissions (Light House Sustainable Building Centre 2014). Province-wide, improving built environment energy and emissions performance is a complex undertaking. There are over 160 BC municipalities of widely varying size, land-use mix, density, physical diversity, geography, and climate. Within them are hundreds of thousands of buildings (over 1 million residential buildings alone) of even more diverse and variable purpose, size, construction type, and vintage. This diversity profoundly impacts the energy and emissions intensity of BC communities as well as the proportions of energy emissions attributable to building operations (ranging from 23% to 51% of BC’s community emissions inventories) and to transportation demand (ranging from 42% to 66% of BC’s community emissions inventories) (Burch et al. 2014). Performance differences are attributable to the interaction of many built-environment factors, including the planning and regulation of land use and transportation as well as the design, engineering, construction, and operation of buildings. Vancouver is a city representative of a large, high-growth metropolitan area. As its growth is dependent on its political and geographic boundaries, therefore infill redevelopment is a common growth mechanism across the city.
The proposed experiments test different growth and dwelling mix strategies (Table 1) based on local census data, existing road networks, parcel characteristics, and the existing building stock (Figure 1 and Tables 1 and 2). Population growth rates reflects projections for the city by BC Statistics (2019). Future housing mixes are informed by the City of Vancouver’s plans and policies (City of Vancouver 2018). This case study assumes a 10% population increase per decade from 2020 to 2040, and three urban-form policy experiments, including dispersed, commercial corridor, and transportation-oriented development (TOD). Compared with the dispersed experiment where no particular constraints are placed on allocations of new dwelling units, both commercial corridor and TOD experiments prioritise urban densification and place new dwelling units either on the commercial corridor or in a TOD (Figure 4). A CIMS model established for the City of Vancouver is supplied by researchers at EMRG to establish the baseline and future dwelling mix and market conditions in all three experiments.
|DISPERSED||COMMERCIAL CORRIDOR||TRANSIT-ORIENTED DEVELOPMENT (TOD)|
|Development areas||No concentration||Concentrated along commercial corridors||Concentrated around a major transit center|
|New dwellings||Emphasise low-rise gentle infill||Emphasise low-rise infill||Emphasise mid- and high-rise multi-family|
|CURRENT POLICY||RENEWABLE ENERGY EMISSIONS|
|Building shell||Current GHG and TEDI requirements according to building codes||Increasingly stringent regulations for GHG intensity and TEDI based on Vancouver’s Zero Emissions Building Policy|
|Space heating||Current GHG and TEDI requirements according to building codes||Replacement of gas-fired mechanicals at the end of their lifespan with heat pumps. Increased energy efficiency regulations for gas-fired furnaces|
|Water heating||Current energy efficiency regulations||Enhanced energy efficiency regulations for gas-fired water heaters|
|Appliances||Current energy efficiency regulations||Enhanced energy efficiency regulations|
Two energy policy experiments are assessed (Table 2). The current policy experiments represent a future where existing energy-efficiency and emissions policy at the local, provincial, and federal levels remain at the same level of stringency and no new policies are added. The renewable energy emission (EE) experiment assumes that bioenergy would be costly in the future, in line with current expectations, and implements technology policies that aim to reduce GHG intensity and increase energy efficiency. Table 2 details the GHG intensity and energy-efficiency policies applied to critical building components in each experiment.
For the Vancouver case study, the intermediary step is to distribute the CIMS market shares across the building shells in the sandbox. This is necessary to infer the transition probabilities of the MCM assuming that technologies are distributed evenly, with a slight skew towards the oldest or newest buildings for the least and most efficient technologies, respectively. The baseline calibration MCM takes the results of the current policy experiment and downscales them to the baseline year (2020) for the sandbox. One current limitation of this work is that building technologies are neither linked to each other nor to building shells. Therefore, there is an intermediary step that adjusts the total market shares according to the building type and age combinations that exist in the sandbox. Probabilities are adjusted such that a greater share of the new, more efficient technologies in 2020 are allocated to the newer buildings, while a greater share of the older, least efficient technologies are allocated to the older buildings. For detailed probabilities used by MCM, see Appendix A. MCM then takes the market shares of the new technologies and allocates them accordingly in each successive time step (i.e. 2020, 2030, and 2040). As building technology stocks in the sandbox model reach their lifespans, their probability of retiring and being replaced increases while new building technologies are assigned with the lowest probability of retirement.
As the policy experiment results in increased market shares of the most efficient building technologies (Figure 6) in each CIMS time step, the transition matrix in the MCM is updated to reflect a higher probability of transitioning to these technologies. For space and water heating, for example, this reflects a shift to electrically powered technologies as well as increased energy efficiency. A similar procedure is conducted for shell retrofit for all standing buildings in every iteration (i.e. experiment). A shell retrofit in this paper involves upgrades on walls, windows, roofs, doors, and thermosets. On average, 5% of the total building stock (excluding new buildings after 2020) receives at least one aspect of shell retrofit. Total energy values of retrofitted buildings are updated subsequently whenever there is a change in either technology or shell retrofit.
Figure 5 summarises the energy and emissions results of each urban form experiment in 2040. While the dispersed and corridor experiments perform similarly, the TOD experiment clearly represents a reduction in energy and GHG equivalent emissions. In the dispersed urban form experiment, the technology policy experiment accounted for a 23% reduction in energy consumption by 2040. In the corridor experiment, it only accounts for a 20% reduction; and in the TOD experiment, it is 28%. The technology retrofit only accounts for a 1–2% reduction in emissions in each experiment by 2040, however. The greater influence of energy policy in the TOD experiment may be due to the greater number of existing dwellings from 2020 that remain in 2040; since the new dwellings added in this experiment are high-rise apartments, less land area is needed to house the increasing population.
Figure 7 distributes the total emission (tCO2e) by building type as well as by building technologies. It is expected that single-family dwelling units (SFD) had the greatest share of energy use and emission overheat and domestic hot water (DHW). Figure 7 also reveals that fuel switch (e.g. electrification) in TOD, particularly in cooling, lighting, and plug load (CLP), has led to the lowest energy consumption and overall emission among all experiments with an energy-use intensity (EUI) of 0.42 GJ/m2.
The complexity of energy issues necessitates models that account for a multitude of spatial and aspatial determinants. Conventional energy–economy models work with economic drivers and dynamics at a variety of scales and resolutions, while UBEMs take a physics-based approach to determining energy and emission using specific building design characteristics and technological components. UBEMs are often explicitly spatial, providing a detailed medium for understanding the spatial dimensions of energy demand. The detail in such models means they are often accurate with respect to individual building energy use, and therefore in understanding the energy and emission impact of specific changes to the building. However, these models cannot holistically assess various local and senior government policies in energy use and emissions over the long term.
In the late 1970s, the academic literature discovered the performance discrepancies introduced by non-technical factors (Socolow 1978). The uncertainties of economic and market-driven factors in UBEMs have been one of the main challenges for building energy modelers. For example, Rysanek & Choudhary (2013) suggested that economic influences such as government incentives and investment return often cannot be modelled as methodically in a parametrically driven UBEM. As a result, despite the level of sophistication and resolution of the simulation, UBEMs fail to recognise the financial and social influences over new buildings and/or retrofitting existing building stocks. Many have attempted to bridge such a gap in energy modelling. Ma et al. (2012) used an economic model predictive control (MPC) to account for the daily electricity costs while optimising HVAC energy load. A life-cycle cost analysis (LCCA) was also proposed to evaluate the finical possibilities of adopting certain building technologies in retrofit (Ruparathna et al. 2017). Efforts have also been made from the energy–economy modelling perspective. Recently, Jaccard et al. (2019) integrated an urban land-use model with CIMS to reconcile the spatial dimension in accessing city-level policy interventions.
The major contribution of this research is the originality of the integrated modelling approach, as well as in its attempt of quantifying the co-benefits of multi-scale effects of energy and emission-reduction policies. This paper remains a first attempt to open up the further exploration of the connections between multi-scale energy modeling approaches. More specifically, the results from the Vancouver case study illustrate the combined results of a technology policy regulating a shift towards electrification and a spatial urban form densification policy (i.e. dispersed, corridor, TOD). They suggest that in certain growth-management strategies, such as TOD, technology policy focusing on HVAC is relatively more effective for reducing emissions than when integrated with other urban form strategies. This is likely because a greater number of existing homes remain as a result of concentrating new growth in mid- and high-rise apartment buildings. Therefore, policies focused on energy and emissions standards for new buildings will be relatively more effective in areas with a growth-management strategy that necessitates the redevelopment of many existing buildings, or in fact, a real estate market that encourages it.
This paper also demonstrates the potential of the GIS-integrated MCM as a way of disaggregating aspatial economic parameters and market shares of building technologies to individual building shells for calculating energy use and emissions in a UBEM. It bridges energy–economy models and UBEMs to illustrate how (combinations of) policies affect the adoption of certain building technologies, in turn affecting energy and emissions at a neighborhood scale. The MCM methodology operationalises transition probabilities using detailed market research and surveys to understand the effect of energy and emissions policies over time. It is well suited for the level of detail in the CIMS outputs and could potentially be used with outputs at even greater levels of aggregation.
However, the methodology is not without its drawbacks. First, urban form affects both operational and embodied energy use and emission throughout the life span of a building. Our method, although accounting for detailed retrofit experiments, does not include embodied emissions from building demolition, construction, and maintenance. Second, comparisons among proposed urban-form experiments are solely evaluated based on their energy and emission performances, excluding any monetary variables other intangible costs such as occupant behavior (Yu et al. 2011) to implement certain policy interventions (e.g. electrification fuel switch), all of which will in turn affect the market uptake rates in a real-world scenario. Third, the current size of the sandbox (i.e. 1600 × 1600 m) limits the model’s ability to incorporate mobility-related energy and emission-reduction policy which operates at a greater scale involving variables (e.g. trip mode and demand) that might be affected by urban form changes way beyond 1600 m. Lastly, although this work demonstrates the potential of fusing GIS and MCM to downscale aggregated energy models to a neighborhood scale, additional data linking specific building technologies to shells of a certain age and building type will further make this method more locally applicable and realistic. Additionally, the MCM is developed in response to the particular structure of the CIMS energy–economy model for this project. Other energy–economy models or communities that lack robust market and economic data that do not include bottom-up or time-series components may not interact as well with the MCM. Currently, this project is exploring opportunities to improve the workflow, specifically on these aforementioned limitations, while making it accessible for other communities within BC.
This paper presented a new methodology for integrating downscaled simulation of an energy–economy model to a spatially explicit community-level urban building energy model (UBEM) that ultimately enables one to iteratively test cross-scale what-if policy interventions. In addition, by using a Markov chain-derived probability model, this work modeled the evolution of individual buildings (e.g. a heating, ventilation and air-conditioning (HVAC) system and the building shell) in a neighborhood over time. This allows an exploration of the interactions between building technology policies and different urban growth-management strategies to understand their effects on reducing building energy and emissions. Findings from a Vancouver case study suggest that this methodology is a simple, yet effective, way to downscale technology shares from an aspatial economic model to individual buildings, assessing energy and emission outcomes with combined spatial urban growth-management strategies.
|High-efficiency natural gas furnace||43%||0%||0%|
|Very high-efficiency natural gas furnace||42%||0%||0%|
|Air-source heat pump||15%||100%||100%|
|Standard electric water heater||29%||30%||16.5%|
|High-efficiency electric water heater||8.5%||19%||28.5%|
|Tankless natural gas water heater||32.5%||0%||0%|
|Natural gas water heater with solar||2%||0%||0%|
|Standard large appliances||42%||14%||13%|
|High-efficiency large appliances||58%||86%||87%|
|Very high-efficiency large appliances||0%||0%||0%|
|High-efficiency small appliances||24%||0%||0%|
|Very high-efficiency small appliances||76%||100%||100%|
|Compact fluorescent light||23.5%||0%||0%|
|Light-emitting diode (LED)||52%||100%||100%|
|Ground-source heat pump||46.5%||100%||100%|
|Very high-efficiency natural gas furnace||23.5%||0%||0%|
|High-efficiency natural gas furnace||23.5%||0%||0%|
|Standard electric water heater||4%||0%||0%|
|Electric gas water heater with solar||13.5%||26%||26%|
|High-efficiency electric water heater||4%||0%||0%|
|Standard natural gas water heater||17%||0%||0%|
|High-efficiency natural gas water heater||23.5%||0%||0%|
|Standard large appliances||0%||0%||0%|
|High-efficiency large appliances||94%||94%||94%|
|Very high-efficiency large appliances||6%||6%||6%|
|Standard efficiency small appliances||100%||100%||100%|
|Medium efficiency lighting||88%||88%||88%|
|Very high-efficiency lighting||6%||6%||6%|
The authors have no competing interests to declare.
The authors thank the Pacific Institute for Climate Solutions (PICS) for funding this research (grant number PICS 36170–50280). They also thank the Environmental Systems Research Institute (ESRI) Canada and the Natural Sciences and Engineering Research Council (NSERC) for funding portions of the tool development and related research.
UMI building templates (formatted as *.json files) can be accessed at: https://doi.org/10.5334/bc.71.s1
Ahmad, T., Chen, H., Guo, Y., & Wang, J. (2018). A comprehensive overview on the data driven and large scale based approaches for forecasting of building energy demand: A review. Energy and Buildings, 165, 301–320. DOI: https://doi.org/10.1016/j.enbuild.2018.01.017
Allan, G., Hanley, N., McGregor, P., Swales, K., & Turner, K. (2007). The impact of increased efficiency in the industrial use of energy: A computable general equilibrium analysis for the United Kingdom. Energy Economics, 29(4), 779–798. DOI: https://doi.org/10.1016/j.eneco.2006.12.006
Anderson, W. P., Kanaroglou, P. S., & Miller, E. J. (1996). Urban form, energy and the environment: A review of issues, evidence and policy. Urban Studies, 33(1). DOI: https://doi.org/10.1080/00420989650012095
BC Assessment. (2019). BC Assessment—Independent, uniform and efficient property assessment. https://www.bcassessment.ca/
BC Building Code. (2018). BC building code. https://free.bcpublications.ca/civix/content/public/bcbc2018/?xsl=/templates/browse.xsl&xsl=/templates/browse.xsl
BC Statistics. (2019). pop_bc_annual_estimates. https://www2.gov.bc.ca/assets/gov/data/statistics/people-population-community/population/pop_bc_annual_estimates.csv
Bhattacharyya, S. C. (1996). Applied general equilibrium models for energy studies: A survey. Energy Economics, 18(3), 145–164. DOI: https://doi.org/10.1016/0140-9883(96)00013-8
Bosselmann, P. (1998). Representation of places: Reality and realism in city design. Choice Reviews Online, 36(02), 36-0733–36-0733. DOI: https://doi.org/10.5860/CHOICE.36-0733
Bourdic, L., Salat, S., & Nowacki, C. (2012). Assessing cities: A new system of cross-scale spatial indicators. Building Research & Information, 40, 592–605. DOI: https://doi.org/10.1080/09613218.2012.703488
British Columbia Community Energy & Emissions Inventory. (2012). Community energy & emissions inventory. https://www2.gov.bc.ca/gov/content/environment/climate-change/data/ceei
Burch, S., Herbert, Y., & Robinson, J. (2014). Meeting the climate change challenge: A scan of greenhouse gas emissions in BC communities. Local Environment: International Journal of Justice and Sustainability, 20(11), 1290–1308. DOI: https://doi.org/10.1080/13549839.2014.902370
City of Vancouver. (2018). City of Vancouver land use and development policies and guidelines: Chinatown HA-1 Design Policies. Planning, Urban Design and Sustainability Department, City of Vancouver. https://guidelines.vancouver.ca/C018.pdf
DOE/Oak Ridge National Laboratory. (2019). Modeling every building in America starts with Chattanooga. https://www.sciencedaily.com/releases/2019/11/191113201333.htm
Dogan, T., & Reinhart, C. (2017). Shoeboxer: An algorithm for abstracted rapid multi-zone urban building energy model generation and simulation. Energy and Buildings, 140, 140–153. DOI: https://doi.org/10.1016/j.enbuild.2017.01.030
Dogan, T., Reinhart, C., & Michalatos, P. (2016). Autozoner: An algorithm for automatic thermal zoning of buildings with unknown interior space definitions. Journal of Building Performance Simulation, 9(2), 176–189. DOI: https://doi.org/10.1080/19401493.2015.1006527
Energy Step Code Council. (2017). BC Energy Step Code: A best practices guide for local governments. https://www2.gov.bc.ca/assets/gov/farming-natural-resources-and-industry/construction-industry/building-codes-and-standards/guides/bcenergystepcode_guide_v1.pdf
EnergyPlus. (2005). Weather data by location | EnergyPlus. https://energyplus.net/weather-location/north_and_central_america_wmo_region_4/CAN/BC/CAN_BC_Vancouver.718920_CWEC
Ferrando, M., Causone, F., Hong, T., & Chen, Y. (2020). Urban building energy modeling (UBEM) tools: A state-of-the-art review of bottom-up physics-based approaches. Sustainable Cities and Society, 62(July), 102408. DOI: https://doi.org/10.1016/j.scs.2020.102408
Fumo, N. (2014). A review on the basics of building energy estimation. Renewable and Sustainable Energy Reviews, 31, 53–60. DOI: https://doi.org/10.1016/j.rser.2013.11.040
Gagniuc, P. A. (2017). Markov chains: From theory to implementation and experimentation. Wiley. DOI: https://doi.org/10.1002/9781119387596
Gil, J., Beirão, J. N., Montenegro, N., & Duarte, J. P. (2012). On the discovery of urban typologies: Data mining the many dimensions of urban form. Urban Morphology, 16(1), 27–40. https://www.researchgate.net/profile/Nuno_Montenegro/publication/256895610_On_the_discovery_of_urban_typologies_Data_mining_the_many_dimensions_of_urban_form/links/00b7d52a6fa83339a1000000/On-the-discovery-of-urban-typologies-Data-mining-the-many-dimensio
Gober, P., & Gober, P. (2018). Urban climate adaptation. In Building resilience for uncertain water futures (pp. 149–162). Springer. DOI: https://doi.org/10.1007/978-3-319-71234-5_7
Gritsevskyi, A., & Nakićenovi, N. (2000). Modeling uncertainty of induced technological change. Energy Policy, 28(13), 907–921. DOI: https://doi.org/10.1016/S0301-4215(00)00082-3
Grubler, A., Bai, X., Buettner, T., Dhakal, S., Fisk, D. J., Ichinose, T., … Sathaye, J. (2012). Urban energy systems. In Global energy assessment (GEA) (pp. 1307–1400). Cambridge University Press. DOI: https://doi.org/10.1017/CBO9780511793677.024
Hall, L. M., & Buckley, A. R. (2016). A review of energy systems models in the UK: Prevalent usage and categorisation. Applied Energy, 169, 607–628. DOI: https://doi.org/10.1016/j.apenergy.2016.02.044
Harish, V. S. K. V., & Kumar, A. (2016). A review on modeling and simulation of building energy systems. Renewable and Sustainable Energy Reviews, 56, 1272–1292. DOI: https://doi.org/10.1016/j.rser.2015.12.040
Hensen, J. L. M., & Lamberts, R. (Eds.). (2011). Building performance simulation for design and operation. Taylor & Francis/Informa. DOI: https://doi.org/10.4324/9780203891612
Hong, T., Chen, Y., Luo, X., Luo, N., & Lee, S. H. (2020). Ten questions on urban building energy modeling. Building and Environment, 168(August), 106508. DOI: https://doi.org/10.1016/j.buildenv.2019.106508
Hughes, S., Chu, E. K., & Mason, S. G. (2018). Climate change in cities: Innovations in multi-level governance. Springer. http://www.springer.com/series/14773
IEA. (2020). World energy balances: Overview. International Energy Agency (IEA). https://www.iea.org/reports/world-energy-balances-overview
Jaccard, M., Murphy, R., Zuehlke, B., & Braglewicz, M. (2019). Cities and greenhouse gas reduction: Policy makers or policy takers? Journal of Energy Policy, 134, 110875. DOI: https://doi.org/10.1016/j.enpol.2019.07.011
Johari, F., Peronato, G., Sadeghian, P., Zhao, X., & Widén, J. (2020). Urban building energy modeling: State of the art and future prospects. Renewable and Sustainable Energy Reviews, 128(April), 109902. DOI: https://doi.org/10.1016/j.rser.2020.109902
Lombardi, M., Laiola, E., Tricase, C., & Rana, R. (2017). Assessing the urban carbon footprint: An overview. Environmental Impact Assessment Review, 66, 43–52. DOI: https://doi.org/10.1016/j.eiar.2017.06.005
Lozano-Pérez, T. (1990). Spatial planning: A configuration space approach. In Autonomous robot vehicles (pp. 259–271). Springer. DOI: https://doi.org/10.1007/978-1-4613-8997-2_20
Ma, J., Qin, J., Salsbury, T., & Xu, P. (2012). Demand reduction in building energy systems based on economic model predictive control. Chemical Engineering Science, 67(1), 92–100. DOI: https://doi.org/10.1016/j.ces.2011.07.052
Masson, V., Marchadier, C., Adolphe, L., Aguejdad, R., Avner, P., Bonhomme, M., Bretagne, G., Briottet, X., Bueno, B., de Munck, C., Doukari, O., Hallegatte, S., Hidalgo, J., Houet, T., Le Bras, J., Lemonsu, A., Long, N., Moine, M.-P., Morel, T., Nolorgues, L., Pigeon, G., Salagnac, J.-L., Viguié, V., & Zibouche, K. (2014). Adapting cities to climate change: A systemic modelling approach. Urban Climate, 10(P2), 407–429. DOI: https://doi.org/10.1016/j.uclim.2014.03.004
Murphy, R., Boyd, K., & Jaccard, M. (2016). Evaluation of actions and policies to reduce urban GHG emissions using multiple criteria: A contribution towards energy efficiency in British Columbia’s built environment. Pacific Institute for Climate Solutions, University of Victoria. https://pics.uvic.ca/sites/default/files/uploads/publications/Murphy%20etal%20Urban%20Energy%20Literature%20PICS%20August%202016.pdf
Nakata, T. (2004). Energy–economic models and the environment. Progress in Energy and Combustion Science, 30(4), 417–475. DOI: https://doi.org/10.1016/j.pecs.2004.03.001
Quan, S. J., Li, Q., Augenbroe, G., Brown, J., & Yang, P. P. J. (2015). Urban data and building energy modeling: A GIS-based urban building energy modeling system using the urban-EPC engine. Lecture Notes in Geoinformation and Cartography, 213, 447–469. DOI: https://doi.org/10.1007/978-3-319-18368-8_24
Ratti, C., Baker, N., & Steemers, K. (2005). Energy consumption and urban texture. Energy and Buildings, 37(7), 762–776. DOI: https://doi.org/10.1016/j.enbuild.2004.10.010
Reinhart, C. F., & Cerezo Davila, C. (2016). Urban building energy modeling—A review of a nascent field. Building and Environment, 97, 196–202. DOI: https://doi.org/10.1016/j.buildenv.2015.12.001
Reinhart, C. F., Cerezo Davila, C., & Davila, C. C. (2016). Engineering advance urban building energy modeling: A review of a nascent field. Building and Environment, 97, 196–202. DOI: https://doi.org/10.1016/j.buildenv.2015.12.001
Reinhart, C. F., Dogan, T., Jakubiec, A., Rakha, T., & Sang, A. (2013). UMI—An urban simulation environment for building energy use, daylighting and walkability 2 3. Paper presented at the 13th Conference of International Building Performance Simulation Association, Chambéry, France (pp. 476–483). http://web.mit.edu/SustainableDesignLab/publications/umi_introduction.pdf
Rivers, N., & Jaccard, M. (2006). Useful models for simulating policies to induce technological change. Energy Policy, 34(15), 2038–2047. DOI: https://doi.org/10.1016/j.enpol.2005.02.003
Rode, P., Keim, C., Robazza, G., Viejo, P., & Schofield, J. (2014). Cities and energy: Urban morphology and residential heat-energy demand. Environment and Planning B: Planning and Design, 41(1), 138–162. DOI: https://doi.org/10.1068/b39065
Ruparathna, R., Hewage, K., & Sadiq, R. (2017). Economic evaluation of building energy retrofits: A fuzzy based approach. Energy and Buildings, 139, 395–406. DOI: https://doi.org/10.1016/j.enbuild.2017.01.031
Ryan, E. M., & Sanquist, T. F. (2012). Validation of building energy modeling tools under idealized and realistic conditions. Energy and Buildings, 47, 375–382. DOI: https://doi.org/10.1016/j.enbuild.2011.12.020
Rysanek, A. M., & Choudhary, R. (2013). Optimum building energy retrofits under technical and economic uncertainty. Energy and Buildings, 57, 324–337. DOI: https://doi.org/10.1016/j.enbuild.2012.10.027
Rysanek, A. M., & Choudhary, R. (2015). DELORES—An open-source tool for stochastic prediction of occupant services demand. Journal of Building Performance Simulation, 8(2), 97–118. DOI: https://doi.org/10.1080/19401493.2014.888595
Salter, J., Lu, Y., Kim, J. C., Kellett, R., Girling, C., Inomata, F., & Krahn, A. (2020). Iterative ‘what-if’ neighborhood simulation: energy and emissions impacts. Buildings and Cities, 1(1), 293–307. DOI: https://doi.org/10.5334/bc.51
Sokol, J., Cerezo Davila, C., & Reinhart, C. F. (2017). Validation of a Bayesian-based method for defining residential archetypes in urban building energy models. Energy and Buildings, 134, 11–24. DOI: https://doi.org/10.1016/j.enbuild.2016.10.050
Sovacool, B. K. (2014). What are we doing here? Analyzing fifteen years of energy scholarship and proposing a social science research agenda. Energy Research and Social Science, 1, 1–29. DOI: https://doi.org/10.1016/j.erss.2014.02.003
Yu, Z., Fung, B. C. M., Haghighat, F., Yoshino, H., & Morofsky, E. (2011). A systematic procedure to study the influence of occupant behavior on building energy consumption. Energy and Buildings, 43(6), 1409–1417. DOI: https://doi.org/10.1016/j.enbuild.2011.02.002