Occupant perceptions of building information model-based energy visualizations in eco-feedback systems

Applied Energy 221 (2018) 220–228

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Occupant perceptions of building information model-based energy visualizations in eco-feedback systems

T

⁎

Abigail Franciscoa, Hanh Truongb, Ardalan Khosrowpourb, John E. Taylora, , Neda Mohammadia a b

School of Civil and Environmental Engineering, Georgia Institute of Technology, 790 Atlantic Dr NW, Atlanta, GA 30332, United States Department of Civil and Environmental Engineering, Virginia Tech, 200 Patton Hall, Blacksburg, VA 24061, United States

H I GH L IG H T S

visualizations have been used to represent energy consumption. • Different the effectiveness of such techniques varies across studies. • However, introduce a new eco-feedback information representation method. • We method integrates energy information into a building information model. • The • We validate the method and find users understand and prefer 2D visualizations.

A R T I C LE I N FO

A B S T R A C T

Keywords: Building information modeling Eco-feedback Energy efficiency Information representation Perceptions Visualization

While technology advancements are improving the energy efficiency of buildings, occupant behavior remains a critical factor in ensuring the effectiveness of such enhancements. To this end, numerous eco-feedback systems have been developed to reduce building energy use through influencing occupants' behaviors during building operations. Information representation is a critical component in eco-feedback systems, affecting the users' interpretation, engagement, and motivation to reduce energy consumption. Many studies have focused on using different charts and technical units or abstract and artistic visualizations to represent energy consumption. However, the effectiveness of such techniques varies across studies. Recent research emphasizes the need to integrate information representation strategies that balance numeric and aesthetic appeal. Concurrently, studies have called for increased adoption of a Building Information Model (BIM) during a building's operations phase to improve facility management. In this paper, we introduce a new eco-feedback information representation method that combines numeric and aesthetic appeal through leveraging spatial and color-coding techniques in BIM. The BIM-integrated energy visualization approach developed in this paper uses the Revit Application Program Interface (API) and allows users to visualize and compare energy consumption values in 2D and 3D views of a multi-family building through a color-coding scheme in an as-built BIM. The method is validated through a user survey that quantitatively and qualitatively assesses the proposed 2D and 3D BIM eco-feedback compared to more traditional bar chart based eco-feedback. Our findings suggest that 2D spatial, color-coded eco-feedback provides the optimal information representation, as it is easy to understand, while evoking engaging and motivating responses from users. This study advances our understanding of eco-feedback information representation while expanding BIM applications during building operations. These are important steps to address the human dimension of energy efficiency in the built environment.

1. Introduction The efficient use of energy resources has gained global attention as scarce resources are being depleted and greenhouse gas emissions continue to rise. In the United States, approximately 39% of the total energy consumption and its associated CO2 emissions are attributed to

⁎

building-related activities [1]. The percentage of CO2 emissions from buildings in the U.S. has been projected to increase by 1.8% per year [2]. In order to address this, many approaches have been introduced to improve energy efficiency in buildings, such as: building retrofits, utilizing energy efficient equipment, and implementing energy efficient technologies recommended by various energy rating organizations.

Corresponding author. E-mail address: [email protected] (J.E. Taylor).

https://doi.org/10.1016/j.apenergy.2018.03.132 Received 29 December 2017; Received in revised form 26 March 2018; Accepted 27 March 2018 0306-2619/ © 2018 Elsevier Ltd. All rights reserved.

Applied Energy 221 (2018) 220–228

A. Francisco et al.

2. Background

However, these approaches cannot be fully exploited if they do not consider the impact of building occupants’ behavior. It has been demonstrated that occupants’ energy consumption behavior is critical to building energy efficiency [3–6], providing the potential to offset up to 7.4% of the associated CO2 emissions [7]. Recently, researchers have examined occupant-based energy efficiency approaches (e.g., eco-feedback systems) that can encourage building occupants to conserve energy [3,7–9]. Eco-feedback systems provide feedback on individual or group behaviors with a goal of reducing environmental impact. The research is based in part on the premise that people lack awareness and understanding of how their everyday consumption behaviors affect the environment. The feedback system may bridge this gap by automatically providing energy consumption feedback on building-related activities by sensing these activities through computerized means [10,11]. Through informative feedback, eco-feedback systems can improve users’ awareness of their own energy use impacts, and positively reinforce or promote change of users’ behaviors [12]. Although there are numerous factors that impact the effectiveness of an eco-feedback system in promoting user awareness and behavior change, two critical features are information content and information representation [3,11,13,14]. Information content includes the energy consumption data of an individual and/or group of individuals. Information representations of eco-feedback data are diverse and include units such as direct kilowatt hour units, monetary units, or CO2 emission units [13,14]. Additionally, different representation techniques, such as tables, line charts, and bar charts, have been examined based on descriptive and injunctive norms. Researchers continue to explore ways to create more effective and impactful information representation techniques in eco-feedback systems. One recent promising trend is the integration of energy use information into a building information model (BIM). Researchers have begun to examine the effect of building-based visualization in the design of eco-feedback systems by using BIM [15–17]. In the current practice, BIM is a widely used form of information storage and visualization in the architecture, engineering, and construction (AEC) industry, which allows users to build accurate computer-generated models [18]. These models can incorporate simulation and visualization tools to support energy-efficient design decision making [19]. Most previous studies have focused on using BIM for energy efficiency to facilitate better design integration in the early design phase [20–22]. However, research on the potential of using BIM for energy efficiency and energy use visualization in the operational phase of a building's lifecycle is in its infancy, and studies lack consideration of users’ impact beyond facility management [23,24]. Integrating energy use information into BIM to reflect the actual operating conditions of buildings will provide many benefits to not only building operators, but also building occupants. For example, BIMs displaying energy use that is accessible to occupants can enable an occupant-operator feedback loop. Studies have indicated such communication exchanges can increase occupant trust and understanding of energy information [25]. As BIM adoption increases in the AEC industry, an increasing number of new buildings have BIM as a part of the required project documentation and deliverables. These models could be employed as an energy use visualization framework in eco-feedback systems. This would help occupants easily and intuitively determine their energy use levels if a comprehensive color-based energy use information visualization were established. It is thus necessary to establish a user-friendly communication method that does not require a priori knowledge of energy units or values, which may result in improved energy conservation outcomes when eco-feedback systems are implemented. In this paper, we develop a technique to integrate measured energy consumption data into an as-built BIM and empirically validate our approach to examine its potential in driving behavior change.

2.1. Eco-feedback system overview There are numerous eco-feedback studies on residential and commercial buildings investigating the effectiveness of underlying system components such as information representation methods, psychological motivators, and interface design on motivating occupants to reduce energy consumption. However, some components, such as information content and information representation in eco-feedback systems, have been found to be critical factors heavily affecting the efficacy of these systems [10,13,14], particularly if consumers have an interest in conservation [26]. Prominent examples of information content include appliance-specific feedback and social comparison feedback. In a review of ecofeedback studies, Fischer [3] found both of these features are likely to promote energy conservation behaviors. Appliance-specific feedback can improve the relevancy of eco-feedback by connecting the user with the energy use impact of their interactions with a particular appliance. This enables occupants to relate eco-feedback information to specific activities [27] and can improve the sense of control a person feels they have over changing their energy consumption [3]. In addition, normative comparison is a type of social comparison feedback method in which an individual or a group is compared to a norm (i.e., a reference value or benchmark), thereby using social norms as a motivation to encourage the conservation of energy. Studies have demonstrated that the normative comparison component of eco-feedback systems can drive energy efficient behavior from users through competition and public perception [3,11], and increase user responses to feedback system notifications [28]. Notably, the predominant visualization strategies adopted in normative eco-feedback deployments have generally been limited to bar and line charts, in both research studies [10,11,29,30] and industry applications [31,32]. Information representation techniques in eco-feedback are diverse and have explored a variety of energy units. Jain et al. [13] investigated the effect of environmental externality units (i.e., numbers of trees needed to offset CO2 emissions) feedback versus energy unit (i.e., W or Wh) feedback, and concluded that participants who received feedback with environmental externality units saved more than participants receiving direct energy units relative to a control group. Furthermore, Asensio and Delmas [14] implemented a health-based feedback motivator to drive behavior changes for occupants in an 8 month-long study of 118 family apartments in Los Angeles. The study determined that the energy consumption of the group who received health-based feedback (i.e., the effect of energy use associated CO2 emissions on childrens’ health) was reduced by 8%. The savings increased to 19% for families with children. Despite the effectiveness of the above approaches, the information was presented to users by numerical units and various bar and line charts, which studies have found much of the population has difficulty understanding [33–35]. Karjalainen [34] conducted 14 interviews involving people from different educational levels and ages in Finland to examine occupants’ interpretation of various energy use representation methods. This study determined that participants encountered difficulties in interpreting various kinds of charts and scientific units, and tended to understand monetary unit representation. However, monetary representations in eco-feedback have not been an effective approach to increase occupant energy efficiency in some studies [14,36]. Moreover, Piccolo et al. [35] determined in a cultural analysis study that 70% of the population (from 15 to 64 years old) lacked essential skills to understand the charts and energy units in Brazil. In addition, Rodgers and Bartram [33] reported that in a study of 23 participants, around half had difficulties in comprehending energy units or were confused by traditional feedback visualizations including bar graphs and tables. Recognizing limited user understanding of energy information 221

Applied Energy 221 (2018) 220–228

A. Francisco et al.

literature, there is an opportunity to evaluate the efficacy of eco-feedback systems enabled in existing building information platforms, such as BIM. In addition, such applications should be validated through statistical methods with a control group. However, to address these opportunities, we first need methods to integrate energy consumption information into BIM, as well as methods to automate the representation of color-coded energy use information in BIM. Our research focuses on developing a novel color-based normative comparison representation integrated with BIM that can be employed in eco-feedback systems.

representations through technical charts and numerical units, several studies developed novel information representation approaches by integrating virtual aids such as images, avatars, and arts to the interface design of eco-feedback systems to explore if these are more effective at promoting behavior change. Grevet et al. [37] introduced an interface design by combining social comparison of energy use and mock-up images of the surrounding area. The research found study participants using this design reported more energy-saving actions compared to participants who were only given individual feedback with bar and line charts instead of mock-up images. However, this study was limited by a small sample size and short duration, and results were not statistically significant. Rodgers and Bartram [33] tested three energy displays featuring abstract art energy representations. Participant responses varied, and while researchers found the displays to be intuitive and a promising eco-feedback approach, many participants expressed the desire for numerical data to also be included on the display. In addition, Orland et al. [38] developed and deployed an application in an office environment with virtual pets, whose appearance represented a user’s real-time energy efficiency. Across the six-month intervention 13% total energy savings were realized. However, researchers reported the application development was an extremely complex and time-intensive undertaking, and participants expressed they felt the interface was “kind of silly”. In summary, the results of information representation research are not entirely consistent, and in response, multiple studies reflect it is necessary to integrate techniques that balance diverse perspectives, such as incorporating both numeric and aesthetic appeal [33,39].

2.3. BIM for energy efficiency BIM consists of a number of functions, components, and parametric objects which allow users to build accurate computer-generated building models [18]. Although prior research addressed the importance of using BIM to facilitate energy conservation during design [20–22], there are limited studies focused on using BIM for energy conservation during the operations phase. Energy consumption during the operations phase is critical, as it typically accounts for approximately 80–95% of the total energy consumed during a building’s lifecycle [44]. In addition, while technical, informational, and organizational barriers currently limit BIM adoption in existing operational buildings, the potential benefits of expanding BIM use during operations are immense [45]. Improved project management, risk reduction, on-site tracking, and automated monitoring are a few examples of the many advantages of using BIM during operations and deconstruction phases. A few studies have explored energy management applications that use BIM during the operations phase. Mousa et al. [46] proposed an approach to convert sensor data into carbon emissions values, and integrate the results into BIM using color-based techniques. However, the proposed approach has not yet been implemented or validated. Costa and colleagues [23] examined the applicability of an integrated toolkit that links BIM and the monitoring, analysis, and optimization of energy in a building. More recently, Gerrish et al. [24] also implemented an approach to connect building sensor data and visualization in BIM. Researchers also conducted interviews with facility managers to identify barriers and future opportunities regarding the visualizations. However, both of these studies produced technical charts and were intended for facility managers; importantly, their methods lack information representation techniques that drive occupant engagement and energy efficient behavior change. We still lack a method to link energy consumption into a BIM in a visually intuitive, user-friendly manner. In this paper, we adopt an asbuilt BIM and develop a method that integrates energy consumption data into BIM using a color-coding approach. Our method facilitates broader use of BIM during a building’s operations phase, leverages existing platforms to enable eco-feedback, and enhances the representation of energy use information for eco-feedback systems through numeric and aesthetic appeal. Moreover, we validated our approach through user surveys. The underlying algorithms, programming, and validation are discussed in the following sections.

2.2. Color-coding and spatial eco-feedback Color-coding and spatial feedback are two information representation techniques that are understudied in eco-feedback research, with promising potential to engage diverse audiences across artistic and numeric preferences. In other research domains, color-coding has been shown to provide effective learning and interpretation due to increased retention and transfer performance. Kalyuga et al. [40] demonstrated that color-coded formats reduced working memory load by reducing cognitive search processes. The results of Keller et al. [41] extended these findings, showing that color-coded information visualization increases comprehension. Similarly, cognitive science fields have found spatial displays also facilitate information searching and organization, freeing up greater space for cognitive processing [42]. Color-coding and spatial information representation have been applied in an eco-feedback study conducted by Bonino et al. [43], in which color-based energy feedback combined with explicit numbers were spatially visualized on a 2D building floor plan. This design aimed to balance artistic and technical appeal, and the paper found 71.77% of respondents reported the color-based energy representation helped them understand how much energy the building consumed. While this study’s experimental setup concentrated on other aspects of eco-feedback, it nevertheless demonstrated the promising success of energy use represented through color-coded, spatial information. To further explore the efficacy of such techniques, statistical validation and comparison to a control group is necessary. Furthermore, in [43] the 2D building floor plan was a mock-up design created by researchers. While this is feasible for a single building, deploying this system at scale is impractical. As mentioned, other eco-feedback studies have been limited by the time-consuming and complex nature of developing ecofeedback applications [38], and this can impede wider eco-feedback adoption. In construction domains, BIMs are commonly used to spatially visualize building layouts and data for various stakeholders. As industry-standard platforms already exist that spatially communicate building information, there is a missed opportunity to leverage these platforms to more efficiently and automatically generate eco-feedback systems. Based on the findings and shortcomings in the eco-feedback

3. Methodology We developed a Revit add-in using the Revit Application Programming Interface (API) to integrate a test-bed building’s energy consumption into the model through an automated color-coding approach. The developed add-in, BIM-EN (BIM-energy), interprets the energy consumption data captured through smart meters and generates a color-coded visualization of the energy use levels in an as-built model. In the following subsections, the three main steps implemented to operationalize this method are explained.

222

Applied Energy 221 (2018) 220–228

A. Francisco et al.

can also be used to draw on electrical equipment (e.g., a television, a microwave, or a bulb) to track their energy use data. Therefore, the developed method can be refined for different purposes in information representation or tracking energy use data from a macro-scale, such as an apartment in a building, to a micro-scale, such as a microwave in a kitchen. The following paragraph details the procedure for implementing the Split Faces technique to integrate the energy consumption data with the building elements through color coding. For a specific apartment, a number of split-faces are drawn on the main elements (e.g. wall, floor, and ceiling) that are chosen based on the apartment geometry and level of visibility in 2D and 3D views. Accordingly, the number of split faces increases for an apartment with a complicated geometry with several surrounding walls, doors, windows, and a stepped floor. This technique splits the selected face of an element and provides various regions of application for different materials. The split-faces are required to be drawn in each targeted apartment for color-coding. In order to colorcode these split face elements, a set of pre-defined Paint materials with various colors are required. In the Revit API, each apartment is defined by a unique ID that is a BuiltInParameter used to locate a specific apartment among others in a BIM. In addition, the Paint material used to color the split-faced elements is represented by a material ID in the Revit API, and is connected to a BuiltInParameter. This provides a convenient means to change the Paint material based on the input BuiltInParameter (i.e., the apartment ID).

3.1. Energy data collection An approach that was compatible with the data collection setup described by Jain et al. [10] was developed to capture the energy consumption of each apartment in a building. The energy data collected consisted of electricity use, and was measured using current transducers in a test-bed apartment building in New York City. Each current transducer monitored the electrical current flow for each individual apartment. Onset Computing HOBO U30 data loggers were used to record the Root Mean Square (RMS) amperage for each apartment. By multiplying the RMS amperage value by 110 V, the apparent power was then calculated and stored in a Structured Query Language (SQL) database. The stored data contained the daily energy use information for each apartment along with the associated unique ID to identify the apartment. 3.2. As-built model color-coding procedure Autodesk Revit 2014 was used to construct an as-built model for the test-bed building. To color-code each apartment unit, a color spectrum was defined corresponding to different energy use levels. To compute the energy use levels, first the Energy Use Intensity (EUI) of each unit was computed by dividing each unit’s daily energy use by the unit’s area. Next, energy use levels for each day were computed by applying the maximum and minimum EUI to the following equation,

Ei = Emin + i ⎛ ⎝

Emax−Emin ⎞ 8 ⎠

3.3. BIM-energy integration (1) We developed the BIM-EN add-in to integrate the energy consumption data and color-coded energy use levels into the as-built BIM. We developed the add-in using the Revit API and C#. In the following paragraphs, the steps required to color-code and display energy use levels in the BIM are explained in detail. First, input data was organized in a matrix form and stored in a text file to be read and processed using C#. The input data contains the apartment IDs, material IDs, and the daily energy use for a specific month. The material ID holds the color of the split-faced elements discussed in Section 3.2. The procedures to obtain the input data file are illustrated in Fig. 1. Next, a date range selection feature was created to allow BIM-EN users to select a specific date range from the input data for which they would like to generate a visualization. The selected EUI data are then averaged for each apartment over the requested date range and stored. The selected data are classified into one of the nine different energy use levels as defined by Eq. (1). The energy use levels and input data file described above serve as inputs in the BIM-EN workflow. To color-code each apartment the apartment ID is retrieved from that as-built model, which is a unique BuiltInParameter. BIM-EN then matches the retrieved ID with the apartment ID from the input data file to get the corresponding material ID, EUI, and energy use level data. The color of the material assigned to the apartment is then accessed

where Emax is the maximum EUI in the selected data, and Emin is the minimum EUI in the selected data. Table 1 outlines the color scheme consisting of nine colors, where each energy use level corresponds to a color. Ei represents the EUI for level i, which varies from 0 to 8, enabling normative comparison of EUI by apartment unit. The green, yellow, and red represent the low (E0), medium (E4), and high (E8) energy use levels, respectively. In Revit, various techniques were tested for color-coded visualization in BIM where a variable is mapped to the desired color spectrum. First, the Revit Color Scheme function was used to color and apply fill patterns to the apartments in the BIM. This technique was simple; however, during certain time periods when the EUI of an apartment was substantially different from others, one or more energy use levels were missing from the color spectrum in Table 1. In this event, Revit randomly re-assigned a new color spectrum. This is a drawback, as it would make it difficult for users visualizing this system to track their energy levels across time. Another Revit technique was tested that involved changing the color of a floor-finish element (e.g., adding a ceiling element with its height equal to zero created a floor-finish element). This technique overcame the drawback of the Color Scheme technique described above, and enabled the color spectrum to be viewed in 2D and 3D. However, this procedure did not enable visualization of the color of walls in the 3D view. In BIMs, walls are normally painted using different colors, and it is difficult for the users to distinguish the color spectrum of energy use levels when the spectrum is mixed with the wall’s color. The use of Split Faces can overcome the drawbacks of the aforementioned techniques. The Split Faces technique was applied on the floors and walls to show the energy use data of an apartment with a single color. We found the split face technique to be flexible and that it Table 1 Color spectrum setting.

Fig. 1. Process to obtain input data. 223

Applied Energy 221 (2018) 220–228

A. Francisco et al.

a. There will be a difference between how engaging 2D BIM-EN and 3D BIM-EN are (Pair 1). b. 2D BIM-EN will be more engaging compared to the energy bar graph (Pair 2). c. 3D BIM-EN will be more engaging compared to the energy bar graph (Pair 3).

through a BuiltInParameter called and changed to a specified color as illustrated in Table 1. In addition to displaying the energy use level color, the apartment name and corresponding EUI value is displayed in the visualization when the user clicks on an apartment. This enables users to simultaneously gather information through color-based, spatial, and numerical means, thus balancing numeric and aesthetic appeal.

Hypothesis 3. A difference exists between how intuitive 2D BIMEN, 3D BIM-EN, and the energy bar graph are.

3.4. Validation

a. There will be a difference between how intuitive 2D BIM-EN and 3D BIM-EN are (Pair 1). b. 2D BIM-EN will be more engaging compared to the energy bar graph (Pair 2). c. 3D BIM-EN will be more engaging compared to the energy bar graph (Pair 3).

In this paper, we developed a method to visualize energy consumption data in an as-built model during the operations phase of a building. The developed method provides an effective visualization, which can potentially enhance eco-feedback systems by providing a color-based, spatial comparison of energy that balances aesthetic and numeric appeal. In order to validate the method, BIM-EN was tested for accuracy and underwent user evaluation. The following sections detail the validation approach.

Hypothesis 4. A difference exists between how motivating to reduce energy consumption 2D BIM-EN, 3D BIM-EN, and the energy bar graph are.

3.4.1. BIM-EN accuracy To test the functional accuracy of BIM-EN, we created a set of data using pseudo energy consumption values for two separate days. The area of each apartment was extracted and the energy values were calculated, ranked, and assigned to the corresponding colors manually before importing the data into our software. Thus, we created a benchmark that enabled us to assess the robustness of our add-in after automated color-coding of the 2D and 3D models.

a. There will be a difference between how motivating to reduce energy consumption 2D BIM-EN and 3D BIM-EN are (Pair 1). b. 2D BIM-EN will be more motivating to reduce energy consumption compared to the energy bar graph (Pair 2). c. 3D BIM-EN will be more motivating to reduce energy consumption compared to the energy bar graph (Pair 3). Because for each main hypothesis we are comparing all 3 ecofeedback types, the Friedman Test is first employed to detect if a difference exists between the three eco-feedback tools. The Friedman Test is a non-parametric test where the dependent variable is ordinal, such as Likert-type scale data. Using the Friedman Test, a p-value below 0.05 indicates statistical significance. If a statistically significant difference is detected, the Wilcoxon Signed-Rank Test is used to compare eco-feedback tool pairs. Because multiple hypothesis tests are calculated, Bonferroni correction was used, and thus a p-value below 0.0167 indicates statistical significance for the Wilcoxon Signed-Rank Test. Survey participants were recruited on Amazon’s Mechanical Turk to assess BIM-EN through an online survey. The online survey was approved by Georgia Tech’s Institutional Review Board (IRB # H17235). The survey consisted of three sections, and aligned with the user survey format in [43]: (1) familiarization of 2D BIM-EN, 3D BIM-EN, and bar graph eco-feedback tools; (2) Likert-type scale rating of eco-feedback tools; and (3) collection of demographic information. The first section consisted of six questions; its purpose was for participants to gain exposure to and interact with each eco-feedback tool. As an example, they were shown the 2D BIM-EN and asked to identify the top three rooms consuming the most energy. The same questions were asked for the 3D BIM-EN and bar chart. In both the 2D and 3D BIM-EN tools, participants could understand levels of energy consumption through the colorcoding, or by hovering over the apartment to view the energy use numerically. The second section of the survey gathered data to fulfill both objectives of the experiment. First, to establish if our findings were consistent with the results pertaining to color-coding in [43], participants were asked if the colors helped them understand an apartment’s energy consumption. Next, users were asked to rank each tool based on the four eco-feedback characteristics above: easy to understand, engaging, intuitive, and motivating to reduce energy use. For each tool, user rankings were selected based on a 5-point Likert-type scale (i.e., strongly agree, agree, neutral, disagree, and strongly disagree). Finally, in the third section demographic information was collected to understand how representative the survey population was compared to the general population.

3.4.2. BIM-EN user evaluation Testing user perceptions of BIM-EN is critical to assess its potential to be an effective eco-feedback system. As previously mentioned, Bonino el al. [43] conducted an initial evaluation of 2D spatial, colorcoded eco-feedback displays through user surveys. The method developed in this paper builds on Bonino et al. by introducing a control group and statistical analyses into the experiment to extend the robustness of such evaluations. In addition, BIM-EN outputs spatial, color-coded energy use in 2D and 3D views. This study evaluates user perception of both 2D and 3D spatial eco-feedback, in comparison to a control group. Information visualization studies in general have found little difference in user performance when perceiving information in 2D and 3D figures, and some studies have recommended 2D or 3D approaches based on specific applications [47,48]. We created a web-based user survey, with two objectives: (1) establish whether our findings are consistent with the results in [43]; and (2) extend those findings by determining if there are statistical differences in the effectiveness of 2D and 3D spatial, color-coded eco-feedback displays compared to a control group. A bar graph served as the control visualization, which is a standard means of presenting energy use in many eco-feedback studies and utility bills [10,29,31,33,49]. To quantify eco-feedback system effectiveness, four characteristics were identified in the literature as central to determine eco-feedback potential: easy to understand [3,50], engaging [10,51], intuitive [52], and motivating to reduce energy consumption [53]. The hypotheses below were tested to evaluate user perception of the BIM-EN approach: Hypothesis 1. A difference exists between how easy it is to understand 2D BIM-EN, 3D BIM-EN, and the energy bar graph. a. There will be a difference between how easy it is to understand 2D BIM-EN and 3D BIM-EN (Pair 1). b. 2D BIM-EN will be easier to understand compared to the energy bar graph (Pair 2). c. 3D BIM-EN will be easier to understand compared to the energy bar graph (Pair 3). Hypothesis 2. A difference exists between how engaging 2D BIMEN, 3D BIM-EN, and the energy bar graph are. 224

Applied Energy 221 (2018) 220–228

A. Francisco et al.

Fig. 2. 3D BIM-EN with mouse hovered over apartment D (left); 2D BIM-EN with mouse hovered over apartment E (right)

4. Results

Table 2 Survey participant average accuracy breakdown by eco-feedback type.

Fig. 2 provides a sample of the results of the 2D and 3D BIM-EN visualization effort. Red colors indicate higher energy consumption values and green colors indicate lower energy consumption values. Users can also click on an apartment to view the numerical energy consumption.

Eco-feedback type

Average participant accuracy

Bar graph 2D BIM-EN 3D BIM-EN

83.3% 83.1% 74.9%

Table 3 Do the red, yellow, and orange colors help with understanding how much energy an apartment consumes?

4.1. BIM-EN accuracy Comparison of the manually generated and BIM-EN results using the pseudo-data revealed the resulting numerical and color-coded information were accurate in both the 2D and 3D BIM-EN visualizations. This demonstrates that the implemented BIM-EN method accurately processes and integrates energy data into an as-built BIM. 4.2. BIM-EN user evaluation

Response

Bonino et al. [43]

This study

Yes A bit No

71.77% 25.40% 2.82%

78.4% 17.5% 4.1%

Results in the first phase of the survey revealed that, on average, participants answered the 2D BIM-EN and bar graph questions with approximately the same level of accuracy, while 3D BIM-EN questions were answered with slightly lower accuracy (Table 2). Table 3 compares the results pertaining to color-coding helpfulness in [43] with our study’s results. Furthermore, analysis of the Likert-type scale ratings is shown in Table 4. Significance values below the threshold p < 0.05 were evident across all four eco-feedback characteristics, allowing us to reject the null hypothesis, and move forward with the Wilcoxon SignedRank Test for each category. Notably, the p-values for the engaging and motivating characteristics were well below 0.05, while easy to understand and intuitive had p-values just below 0.05. For Hypothesis 1(a), a p-value of 0.0148 provides evidence to conclude a difference exists between how easy it is to understand 2D BIM-EN and 3D BIM-EN (Pair 1). Alternatively, results for Hypotheses 2(a), 3(a), and 4(a) all show p-values above the threshold, and therefore there is evidence that there is no difference between 2D BIM-EN and 3D BIM-EN concerning the engaging, intuitive, and motivating

200 people participated in the survey. Fig. 3 displays the survey participant age and education levels collected through demographic questions. Additionally, all participants resided in the United States. Most participants reported they directly paid for their utility use at home (82%) and had no previous experience with eco-feedback tools (94%). 29 participants were eliminated from the analysis due to unreliable results. 2 participants completed the survey in less than 60 s, far below the average time to take the survey, and responses were viewed as not reliable. An additional 27 participants misinterpreted the color scale by interpreting red to be the lowest energy use level and green to be the highest. Users who interpreted the scale this way were identified based on their responses in Section 1 of the survey. As this misinterpretation can be overcome via more explicit scale labeling, these responses were removed to eliminate this inverted understanding from influencing the results. In Appendix A, the user survey employed can be found as an e-component. The survey was generated using Qualtrics software, Copyright ©2017 Qualtrics.

Fig. 3. Survey participant (A) age and (B) education demographics. 225

Applied Energy 221 (2018) 220–228

A. Francisco et al.

survey population in [43] was larger, the majority (76%) were from academia and participants skewed to younger populations, with a mean age of 23. In our study, survey participants came from more diverse educational backgrounds and age groups, and results may be more reflective of the general population. Furthermore, the present study also extends findings from [43] by evaluating the potential of spatial, color-coded eco-feedback across four characteristics found to be critical for eco-feedback success: easy to understand [3,50], engaging [10,51], intuitive [52], and motivating to reduce energy consumption [53]. Rejecting the null hypothesis for Hypothesis 1 provides evidence that differences exist when comparing the 2D BIM-EN, 3D BIM-EN, and bar graph eco-feedback, in terms of user ease of understanding. This enabled us to progress to Hypotheses 1(a), 1(b), and 1(c). In Hypothesis 1(a), there was enough evidence to reject the null hypothesis, indicating that users found there is a difference between how easy it is to understand 2D BIM-EN and 3D BIMEN. Upon further examination, we found 2D BIM-EN was statistically easier to understand compared to 3D BIM-EN (p = 0.0074). There was not enough evidence to reject the null of Hypothesis 1(c), and upon further investigation we found users thought it was also easier to understand the bar graph compared to 3D BIM-EN (p = 0.0043). Furthermore, in Hypothesis 1(b) there was no evidence to reject the null hypothesis; however, we found there was no statistical difference between how easy it was to understand 2D BIM-EN and the bar chart. Similar trends are reflected in the first section of the survey, where users answered questions on 2D BIM-EN and bar graph eco-feedback with similar rates of accuracy (83.1% and 83.3%, respectively), while questions with 3D BIM-EN had slightly lower accuracy rates (74.9%). This contributes to the information representation studies [47], by adding empirical evidence that for building visualization applications, information layered onto 2D building visualizations is easier to understand compared to information integrated into 3D building models. Next, rejecting the null hypothesis for Hypotheses 2 provides evidence that differences exist when comparing the 2D BIM-EN, 3D BIMEN, and bar graph eco-feedback, in terms of perceived engagement. In Hypothesis 2(a), by failing to reject the null hypothesis we conclude users do not perceive a difference between how engaging the 2D and 3D BIM-EN eco-feedback are. In addition, rejecting the null hypothesis in both Hypothesis 2(b) and 2(c) leads us to conclude users perceive both 2D and 3D BIM-EN to be more engaging than the traditional energy bar chart. This is of important interest as qualitative and quantitative ecofeedback studies have emphasized user engagement should be a central focus of eco-feedback designs and have shown that eco-feedback user engagement is linked to reductions in energy consumption [10,51]. The null hypothesis for Hypothesis 3 was rejected, and through Hypothesis 3(a) there is enough statistical evidence to suggest there is no difference between how intuitive users perceive the 2D and 3D BIMEN. There was not enough statistical evidence to reject the null hypothesis for Hypothesis 3(b) and 3(c). Thus, we conclude there is no difference between how intuitive users perceive 2D BIM-EN and a bar graph, and 3D BIM-EN and a bar graph. Finally, there was enough statistical evidence to reject the null hypothesis for Hypothesis 4. Moreover, by failing to reject the null for Hypotheses 4(a) and rejecting the null for Hypotheses 4(b) and 4(c), we conclude users perceive there is no difference between 2D BIM-EN and 3D BIM-EN in motivating to reduce energy use (Hypothesis 4(a)), while 2D BIM-EN and 3D BIM-EN are both more motivating to reduce energy use compared to a typical energy bar chart (Hypotheses 4 (b) and 4(c)). This has implications as many eco-feedback systems and utility programs use bar charts to communicate energy consumption to users [10,29,31,33,49]. An essential goal of eco-feedback studies is to motivate users to improve behaviors to reduce energy consumption [53], and these results suggest spatial, color-coded information may be more effective than energy bar charts. The results of the statistical analyses indicate that bar graph energy feedback and 2D BIM-EN are the easiest to understand, and 2D and 3D

Table 4 Hypothesis test results. Hypothesis

H1. Easy to understand H2. Engaging H3. Intuitive H4. Motivating

Friedman test (p < 0.05) All eco-feedback types

Wilcoxon signed-rank test (p < 0.0167) (a) Pair 1

(b) Pair 2

(c) Pair 3

0.02568*

0.015†

0.913

0.996

2.2e−16* 0.03785* 2.793e−05*

0.234 0.256 0.455

1.138e−12† 0.032 9.192e−04†

7.192e−13† 0.148 4.375e−04†

* p < 0.05. † p < 0.0167 (Bonferroni p-value correction).

characteristics. Comparing 2D BIM-EN and the bar graph (Pair 2), Hypotheses 1(b) and 3(b) results are both insignificant, and there is therefore not enough evidence to reject the null hypothesis. Hypotheses 2(b) and 4(b) are both statistically significant, offering enough evidence to reject the null hypotheses for 2(b) and 4(b). Similar results exist when comparing 3D BIM-EN and the bar graph (Pair 3). For Hypotheses 1(c) and 3(c) p-values are above the threshold for significance. Hypotheses 2(c) and 4(c) are well below the threshold, rejecting the null hypotheses for both 2(c) and 4(c).

5. Discussion Information representation in eco-feedback systems has been recognized as one of the most significant factors impacting users’ energy conservation [3]. Many studies have focused on improving the effectiveness of information representation by investigating different representation units like direct energy units, environmental or monetary units, or integrating various virtual aids like images, art, or avatars [13,29,33,38]. Conflicting results concerning user preferences has led to several studies calling for information representation techniques that integrate numeric and aesthetic appeal [33,39]. Bonino et al. [43] answered this call by developing a spatial, color-coded eco-feedback system on a 2D floor plan, and demonstrated promising results with 77.77% of participants finding the color-based visualization helpful for understanding energy consumption. The present study builds on Bonino et al. in three primary ways. First, the aforementioned study did not include a statistical evaluation or a control group, which are two key factors influencing the robustness of eco-feedback studies [54]. We validate spatial, color-coded eco-feedback through robust methods by introducing a control group to compare the performance of spatial ecofeedback with a standard way of viewing energy (i.e., the bar graph), and applying statistical methods in the analysis. Second, Bonino et al. considered eco-feedback in only 2D spatial views, and the potential of 3D spatial views has not yet been studied for energy applications [47,48]. Our experiment tests the performance of eco-feedback in both 2D and 3D views. Third, the developed system in Bonino et al. was comprised of mock-up images and is limited in its ability to efficiently scale to other non-similar buildings. This is of concern because other studies have found novel eco-feedback systems to be limited by how time-intensive they were to create [38]. We leverage an existing technology that is already heavily adopted by industry today, BIM [18], in a novel way to improve the potential of the proposed eco-feedback system to scale for other real-world applications. This contribution expands the application of BIM during a building’s operational phase, which is an emerging area of interest to the current BIM research community [45]. The eco-feedback information representation techniques developed through our method were empirically evaluated to deepen our understanding of user perceptions of this prototype. The results of this study agree with those in [43], as a similar number of users found colorcoding to be helpful with understanding energy consumption. While the 226

Applied Energy 221 (2018) 220–228

A. Francisco et al.

eco-feedback systems in promoting energy conservation.

BIM-EN are more engaging and motivating to reduce energy use compared to the bar graph representation. Qualitative feedback reflects these findings. In a voluntary, open-ended response section, participants wrote commentary for the 2D BIM-EN (N = 38), 3D BIM-EN (N = 44), and bar chart (N = 33) eco-feedback tools. Predominant themes showed participants found the energy use bar chart to be the most familiar type of feedback (N = 11), while others agreed with the familiarity but thought it was boring (N = 9). One respondent adequately summed up bar graph feedback by, “it’s boring, but it gets the job done”. Alternatively, 3D BIM-EN had a varied response. While some commented this view was the most attractive and easiest to understand (N = 14), many questioned whether it was necessary to view the information in 3D (N = 6) or felt it detracted from readability (N = 7). The 2D BIM-EN responses were more consistent, with the predominant comments reflecting its combination of being easy to read and engaging (N = 15). As an example, a participant summarized 2D BIM-EN was, “very user friendly, eye catching, and easy to process”. In summary, quantitative and qualitative evidence direct us to suggest the 2D BIM-EN method is optimal for eco-feedback information representation, as it is easy to understand, while evoking engaging and motivating responses. This extends the findings of Bonino et al. [43] by (a) introducing a control group and statistical validation to improve the robustness of the study findings; (b) expanding evaluations of spatial, color-coded feedback to include both 2D and 3D views; and (c) leveraging a technology widely adopted by industry—BIM—to improve the potential for scalability of this eco-feedback method in real-world applications. This study opens several avenues for new eco-feedback research and industry applications. Future research studies could extend these results by implementing this BIM-based energy use visualization method in practice to explore the impact BIM-EN eco-feedback has on users’ energy consumption behaviors. Notably, future studies should be wary of assuming users will understand the red-green color spectrum intuitively. In industry, color-coded, spatial models can be exported and integrated into other environments such as a website or eco-feedback software. Our approach has flexible applications, as we used the split faces technique on the floors and walls to show the energy use data, which can also be drawn on equipment (e.g., a television, a microwave, or a refrigerator). Thus, future applications could leverage the BIM-EN visualization method to color-code not only building elements, but also plugged-in appliances in BIM. In previous studies, appliance-specific feedback has been found to evoke more energy conscious behaviors from occupants [3,27]. Finally, the method developed in this paper can be applied to enhance the occupant-operator feedback loop across a broad spectrum of building performance and resource management issues facility managers face, including, water, gas, comfort, indoor environmental quality, and building maintenance feedback systems. Such a visualization method not only can be implemented to provide feedback with the necessary level of detail, but also can facilitate facility manager decision making through faster interpretation and intuitive understanding of building energy statuses. There are some limitations to this approach that should be addressed in future research. The energy use values are not imported into the as-built model, which means that the BIM does not become a “dynamic” BIM [16]. However, storing the collected energy use data in external files reduces the size and complexity of the BIM database. This technique is more efficient when the collected data are expansive, as in this study. Moreover, to visualize energy use levels that correspond to expected colors, a step of making split faces and including the paint material for each apartment in Revit is required. In addition, user evaluations of this method are based on their initial perceptions of BIMEN during the survey. Actual levels of eco-feedback understanding, engagement, intuitiveness, and motivation may vary or change throughout time during real-world eco-feedback deployments. Nonetheless, this eco-feedback prototype and initial validation provides an essential first step to understanding the potential of spatial, color-coded

6. Conclusions In most eco-feedback systems research, energy consumption information has been represented to users using different charts and technical units, which are not easily comprehensible to many users [33–35,43], or using abstract visualizations, which are difficult to relate to energy consumption for users [29,33,38]. The method developed in this paper provides a novel information representation approach that enables spatial energy consumption visualization using different colors in an as-built Building Information Model (BIM). User evaluation results allow us to infer that users perceive 2D and 3D BIM-EN (BIM-energy) to be more engaging and motivating to reduce energy consumption compared to an energy bar chart. In addition, 2D BIM-EN feedback was as easy to understand as the bar chart. These color-coded models create an aesthetically interesting user interface and may increase user engagement and interpretation of energy data, potentially impacting energy consumption behaviors. Overall, the integration of a color-based normative comparison energy visualization into BIM represents an integral step toward deepening our understanding of the impact of eco-feedback information representation on fostering pro-environmental behaviors. Furthermore, the method presented demonstrates an application of BIM during a building’s operations phase, which has been shown to be an important tool for optimized facility management. Future research can extend these results by implementing the proposed approach in an occupied building to determine the actual impact on energy consumption. In addition, the above method is flexible in that it can adapt to incorporate data streams beyond energy, such as water consumption, room temperatures, or indoor air quality. Integrating these data streams will improve the visibility of building performance and can promote better management of building resource consumption, comfort, and health. However, to realize the full benefits of such feedback systems, it is essential to research information representation techniques that best encourage user understanding, engagement, and behavior change. A deepened understanding of user interactions with BIM-integrated building performance feedback systems has the potential to address the key drivers of energy efficiency as related to occupant behavior in our increasingly connected buildings, surrounding communities, and cities at large. This is critical if we are to meet our energy consumption reduction targets and achieve a low-carbon future. Acknowledgments This material is based upon work supported by the National Science Foundation under Grant No. 1142379. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors would like to also thank the Vietnam Education Foundation for providing funding to support the efforts of the second author. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apenergy.2018.03.132. References [1] USDOE. US Department of Energy, Buildings Energy Data Book 2010. < http:// buildingsdatabook.eren.doe.gov/ > . [2] USGBC. Buildings and Climate Change Document 2014. [3] Fischer C. Feedback on household electricity consumption: a tool for saving energy? Energy Effic 2008;1:79–104. http://dx.doi.org/10.1007/s12053-008-9009-7. [4] Abrahamse W, Steg L, Vlek C, Rothengatter T. The effect of tailored information, goal setting, and tailored feedback on household energy use, energy-related

227

Applied Energy 221 (2018) 220–228

A. Francisco et al.

behaviors, and behavioral antecedents. J Environ Psychol 2007;27:265–76. [5] Menassa CC, Kamat VR, Lee S, Azar E, Feng C, Anderson K. Coupling distributed energy simulation and occupancy models for comprehensive building energy consumption analysis. Comput Civ Eng ASCE;2013:275–82. [6] Clevenger CM, Haymaker JR, Jalili M. Demonstrating the impact of the occupant on building performance. J Comput Civ Eng 2013;28:99–102. [7] Dietz T, Gardner GT, Gilligan J, Stern PC, Vandenbergh MP. Household actions can provide a behavioral wedge to rapidly reduce US carbon emissions. Proc Natl Acad Sci 2009;106:18452–6. [8] Petersen JE, Shunturov V, Janda K, Platt G, Weinberger K. Dormitory residents redyce electricity consumption when exposed to real-time visual feedback and incentives. Int J Sustain High Educ 2007;8:16–33. http://dx.doi.org/10.1108/ 14676370710717562. [9] Anderson K, Song K, Lee S, Lee H, Park M. Energy consumption in households while unoccupied: evidence from dormitories. Energy Build 2015;87:335–41. [10] Jain RK, Taylor JE, Peschiera G. Assessing eco-feedback interface usage and design to drive energy efficiency in buildings. Energy Build 2012;48:8–17. http://dx.doi. org/10.1016/j.enbuild.2011.12.033. [11] Gulbinas R, Jain RK, Taylor JE. BizWatts: a modular socio-technical energy management system for empowering commercial building occupants to conserve energy. Appl Energy 2014;136:1076–84. http://dx.doi.org/10.1016/j.apenergy.2014. 07.034. [12] Karlin B, Ford R, Squiers C. Energy feedback technology: a review and taxonomy of products and platforms. Energy Effic 2014;7:377–99. http://dx.doi.org/10.1007/ s12053-013-9227-5. [13] Jain RK, Taylor JE, Culligan PJ. Investigating the impact eco-feedback information representation has on building occupant energy consumption behavior and savings. Energy Build 2013;64:408–14. [14] Asensio OI, Delmas MA. Nonprice incentives and energy conservation. Proc Natl Acad Sci 2015;112:E510–5. [15] Muthumanickam A, Jain R, Taylor J, Bulbul T. Development of a novel BIM-energy use ontology. Constr Res Congr 2014;2014:140–9. http://dx.doi.org/10.1061/ 9780784413517.176. [16] Chen J, Bulbul T, Taylor JE, Olgun G. A case study of embedding real-time infrastructure sensor data to BIM. Constr. Res. Congr Constr. a Glob. Netw ASCE 2014;2014:269–78. [17] Eguaras-Martínez M, Vidaurre-Arbizu M, Martín-Gómez C. Simulation and evaluation of building information modeling in a real pilot site. Appl Energy 2014;114:475–84. [18] Eastman C, Teicholz P, Sacks R, Liston K. A Guide to Building Information Modeling for Owners, Managers, Designers, Engineers and Contractors: BIM Handbook. New York: Wiley, NJ; 2011. [19] Crawley DB, Lawrie LK, Winkelmann FC, Buhl WF, Huang YJ, Pedersen CO, et al. EnergyPlus: creating a new-generation building energy simulation program. Energy Build 2001;33:319–31. http://dx.doi.org/10.1016/S0378-7788(00)00114-6. [20] Bazjanac V. IFC BIM-based methodology for semi-automated building energy performance simulation. Lawrence Berkeley Natl Lab 2008. [21] Jalaei F, Jrade A. Integrating BIM with green building certification system, energy analysis, and cost estimating tools to conceptually design sustainable buildings. Constr. Res. Congr. 2014 Constr. Glob Netw. 2014:140–9. [22] Reeves T, Olbina S, Issa R. Validation of building energy modeling tools: Ecotect™, Green Building Studio™ and IES < VE > ™. Simul. Conf. (WSC), Proc. 2012 Winter, IEEE; 2012, p. 1–12. [23] Costa A, Keane MM, Torrens JI, Corry E. Building operation and energy performance: monitoring, analysis and optimisation toolkit. Appl Energy 2013;101:310–6. [24] Gerrish T, Ruikar K, Cook M, Johnson M, Phillip M, Lowry C. BIM application to building energy performance visualisation and management: challenges and potential. Energy Build 2017;144:218–28. [25] Simcock N, MacGregor S, Catney P, Dobson A, Ormerod M, Robinson Z, et al. Factors influencing perceptions of domestic energy information: content, source and process. Energy Policy 2014;65:455–64. http://dx.doi.org/10.1016/j.enpol. 2013.10.038. [26] Vassileva I, Dahlquist E, Wallin F, Campillo J. Energy consumption feedback devices’ impact evaluation on domestic energy use. Appl Energy 2013;106:314–20. http://dx.doi.org/10.1016/j.apenergy.2013.01.059. [27] Ellegård K, Palm J. Visualizing energy consumption activities as a tool for making everyday life more sustainable. Appl Energy 2011;88:1920–6. http://dx.doi.org/10. 1016/j.apenergy.2010.11.019. [28] Khosrowpour A, Xie Y, Taylor JE, Hong Y. One size does not fit all: establishing the

[29]

[30]

[31] [32] [33] [34] [35]

[36]

[37]

[38]

[39] [40] [41]

[42] [43] [44] [45] [46] [47] [48]

[49] [50]

[51]

[52]

[53]

[54]

228

need for targeted eco-feedback. Appl Energy 2016;184:523–30. http://dx.doi.org/ 10.1016/j.apenergy.2016.10.036. Boomsma C, Goodhew J, Pahl S, Jones RV. The feasibility of saving energy in challenging organisational contexts: testing energy visualisation in a social services office in the United Kingdom. Energy Res Soc Sci 2016;15:58–74. http://dx.doi. org/10.1016/j.erss.2016.02.004. Anderson K, Song K, Lee SH, Krupka E, Lee H, Park M. Longitudinal analysis of normative energy use feedback on dormitory occupants. Appl Energy 2017;189:623–39. http://dx.doi.org/10.1016/j.apenergy.2016.12.086. My Home Energy Report. Duke Energy 2017. < https://www.duke-energy.com/ home/products/my-home-energy-report > [accessed December 6, 2017]. Oracle. Oracle Utilities Opower Home Energy Reports: Customer Service Guide; 2018. Rodgers J, Bartram L. Exploring ambient and artistic visualization for residential energy use feedback. IEEE Trans Vis Comput Graph 2011;17:2489–97. Karjalainen S. Consumer preferences for feedback on household electricity consumption. Energy Build 2011;43:458–67. Piccolo LSG, Scharl A, Baranauskas C. Design of eco-feedback technology to motivate sustainable behavior: cultural aspects in a Brazilian context. Conf-IRM, Austria; 2012. p. 1–13. Wood G, Newborough M. Energy-use information transfer for intelligent homes: enabling energy conservation with central and local displays. Energy Build 2007;39:495–503. Grevet C, Mankoff J, Anderson SD. Design and evaluation of a social visualization aimed at encouraging sustainable behavior. 43rd Hawaii Int. Conf. Syst. Sci., IEEE; 2010, p. 1–8. doi: 10.1109/HICSS.2010.135. Orland B, Ram N, Lang D, Houser K, Kling N, Coccia M. Saving energy in an office environment: a serious game intervention. Energy Build 2014;74:43–52. http://dx. doi.org/10.1016/j.enbuild.2014.01.036. Riche Y, Dodge J, Metoyer R. Studying always-on electricity feedback in the home. CHI 2010 Home Eco Behav 2010:1995–8. Kalyuga S, Chandler P, Sweller J. Managing split-attention and redundancy in multimedia instruction. Appl Cogn Psychol 1999;13:351–71. Keller T, Gerjets P, Scheiter K, Garsoffky B. Information visualizations for knowledge acquisition: the impact of dimensionality and color coding. Comput Human Behav 2006;22:43–65. http://dx.doi.org/10.1016/j.chb.2005.01.006. Hegarty M. The cognitive science of visual-spatial displays: implications for design. Top Cogn Sci 2011;3:446–74. Bonino D, Corno F, De Russis L. Home energy consumption feedback: a user survey. Energy Build 2012;47:383–93. http://dx.doi.org/10.1016/j.enbuild.2011.12.017. Cole RJ, Kernan PC. Life-cycle energy use in office buildings. Build Environ 1996;31:307–17. Volk R, Stengel J, Schultmann F. Building Information Modeling (BIM) for existing buildings – literature review and future needs. Autom Constr 2013;38:109–27. Mousa M, Luo X, McCabe B. Utilizing BIM and carbon estimating methods for meaningful data representation. Procedia Eng 2016;145:1242–9. Shah P, Hoeffner J. Review of graph comprehension research: implications for instruction. Educ Psychol Rev 2002;14:47–69. Sebrechts MM, Cugini JJ V, Laskowski SSJ, Vasilakis J, Miller MS. Visualization of search results: a comparative evaluation of text, 2D, and 3D interfaces. In: Proc 22nd Annu Int ACM SIGIR Conf Res Dev Inf Retr 1999:3–10. doi: 10.1145/312624. 312634. Putting You In Control. Florida Power Light Co 2017. < https://www.fpl.com/ smart-meters/control.html > (accessed June 12, 2017). Sernhed K, Pyrko J, Abaravicius J. Bill me this way! -customer preferences regarding electricity bills in Sweden. In: Proc 2003 Summer Study Eur Counc an Energy Effic Econ 2003:1147–50. Buchanan K, Russo R, Anderson B. The question of energy reduction: the problem(s) with feedback. Energy Policy 2015;77:89–96. http://dx.doi.org/10.1016/j.enpol. 2014.12.008. Froehlich J, Findlater L, Ostergren M, Ramanathan S, Peterson J, Wragg I, et al. The design and evaluation of prototype eco-feedback displays for fixture-level water usage data. CHI ’12 Proc SIGCHI Conf Hum Factors Comput Syst 2012:2367–76. doi: 10.1145/2207676.2208397. Froehlich J, Findlater L, Landay J. The design of eco-feedback technology. Proc SIGCHI Conf Hum Factors Comput Syst (CHI ’10) 2010:1999–2008. doi: 10.1145/ 1753326.1753629. Delmas MA, Fischlein M, Asensio OI. Information strategies and energy conservation behavior: a meta-analysis of experimental studies from 1975 to 2012. Energy Policy 2013;61:729–39. http://dx.doi.org/10.1016/j.enpol.2013.05.109.