Pathway III: Energy Efficient Building Design Decisions
Energy efficient building design decisions pathway provides new design methods and tools to designers and engineers, who make early-stage design decisions, which have the long lasting impact on buildings’ energy footprint. Current design methods are based on assumptions regarding the number of occupants, their movement, and behavior, as represented in architect’s graphic standards and client briefs, which are usually not validated and unspecific. To improve the design process, we are investigating how the behavior of occupants, building systems and operations of existing buildings can be brought back to the design stage for shaping early and critical energy-aware decisions.
Benchmarking Immersive Virtual Environments to Physical Virtual Environments:
We hypothesized that IVEs could be used to provide occupant feedback to the building designers and engineers and also could provide a platform where we study the impact of user-built environment interactions on energy consumption in buildings. To ensure that the data collected and analyzed in such environments represent physical environments adequately, we designed a study, through which we tested whether IVEs are adequate representations of physical environments and measured user performance in such environments. Specifically, we investigated user performance on a set of everyday office-related activities (e.g., reading text and identifying objects in an office environment) and benchmarked the participants' performance in a similar physical environment. We also measured the sense of presence within an IVE through a set of questionnaires. We used a typical physical office room for the physical conditions and modeled it digitally and brought it to IVEs for the virtual conditions. We created two conditions for both the virtual and physical environments with similar settings: a dark and a bright office room. In order to compare occupants’ performance between virtual and physical office rooms, we recruited 112 participants and measured three parameters both in the virtual and physical office. These parameters were (1) user performance when given simple office related tasks (e.g., reading speed, comprehension); (2) user perception of color recognition and object identification; and (3) user's sense of immersion and presence. In order to assess the IVE's effects on the end- users' performance, the speed and accuracy of the performed tasks were recorded and compared to the physical office environment with very similar settings.
Results:
By analyzing the experimental data from 112 participants, we found that there exists a difference between participants' performances between dark and bright conditions in the physical environment, as well as in the IVE; yet this difference is almost equal between the two environments, physical and virtual. The questionnaire data showed that the participants felt a strong sense of presence within an IVE. The questionnaire data also showed that the participants' ability to control the events happening in the IVE was strongly similar to those in the physical environment and their interaction within the IVE felt moderately natural. These results indicate that IVE is a satisfactory representation of the physical environment based on the performance measures and questionnaire data and there are no significant differences in performances on the measures of everyday office-related tasks in a physical environment and in an IVE. This study thus provided evidence that IVEs can be an effective tool to study user behavior and measure user performance.
Aggregating opinions to Design Energy-Efficient Buildings:
We are exploring a multi-agent team that collaborates in designing energy-efficient buildings (Pathway III). We use the H.D.S. Beagle system, developed by PI David Gerber, as a base. Beagle is a MDO (multi-disciplinary design optimization) software framework that assists users in the early stage design of buildings. It incorporates an optimization methodology that combines parametric modeling with multi-objective optimization through an integrated platform for enabling rapid iteration and trade-off analysis across the domains of design, energy use intensity, and finance. We model each Beagle as an agent, and explore the quality of the solutions obtained by aggregating the opinions of multiple Beagle runs. It was recently shown in the multi-agent systems literature a model that predicts that a team of diverse agents can overcome a uniform team made of copies of the best agent. We study if such model applies in the early stage design architecture domain, and evaluate a team made of diverse agents, composed by Beagle agents
with different options, and a team made of copies of the best agent, composed by Beagle agents with the same options.
Beagle is a system based on a Genetic Algorithm (GA) framework. It performs a multi-objective optimization, across the domains of design, energy use intensity, and finance. Each solution in Beagle is represented by a set of parameters. Giving values to each parameter creates a certain early stage design for a building. We aggregate the opinions of a set of Beagle agents, in order to explore if we can obtain better solutions. We explore two main ideas: (i) a system where each agent optimizes for a different factor of the multiobjective optimization framework; (ii) a system where all agents optimize for all factors, but each agent is initialized with different options for the GA algorithm.
Studies have shown that occupant behavior has a significant impact on a building's overall performance and energy consumption. In order to understand occupants' lighting-use behavior, we investigated the influence of manual and semi-automatic control systems on lighting-use in a single- occupancy office space. 114 participants were recruited and randomly assigned to one of the four conditions that varied in available lighting control options in an immersive virtual environment. They were asked to adjust the room's lighting by choosing one of the following lighting control options: (Group 1) manual control system for artificial lights and interior shades; (Group 2) same options as condition 1 and a semi-automatic control system for the shades; (Group 3) same options as condition 1 and a semi-automatic control system for the artificial lights; and (Group 4) same options as condition 1 and semi-automatic control systems for both the artificial lights and shades.
Understanding the Links between Personal Control and Occupant's Lighting Related Behavior
Results:
The results from 114 participants showed that the participants use natural light more in order to increase the lighting levels in the room when a remote option is available to open the shades compared to when this remote option is not available. The analysis also showed that, when a remote control option is added to semi- automatically turn on the artificial lights in addition to being able to semi-automatically open the shades (Group 4), the participants were less likely to use natural light to increase the lighting levels of the room. Therefore, we concluded that having a remote control option to control the shades might be a more effective strategy in order to motivate end-users to increase the lighting levels by using natural light rather than artificial light sources. It is important to also note that when the remote control option was added to only turn the lights on (Group 3), the analysis showed that the participants were directionally, but not significantly, more motivated to use the artificial light; this could be due to the fact that participants chose to use the artificial light slightly more in the control condition (Group 1). Additionally, the follow-up questionnaire suggests that participants chose the remote control options more in Groups 2 and 3 since it was an ‘easier’ and more convenient option, which shows that such features could potentially not only be integrated to existing buildings but could also be part of the design alternatives during the design phase of future office buildings.
Publications:
Heydarian A, Carneiro J P, Gerber D, Becerik-Gerber B. (2014) “Towards Measuring the Impact of Personal Control on Energy Use Through the Use of Immersive Virtual Environments,” The 30th International Symposium on Automation and Robotics in Construction, July 9-11, 2014, Sydney, Australia
Heydarian A, Carneiro J P, Gerber D, Becerik-Gerber B (2015). Immersive Virtual Environments, Understanding the Impact of Design Features and Occupant Choice upon Lighting for Building Performance. Journal of Building and Environment. 89 217.
Results:
We executed experiments with one parametric design model. We found that aggregating the opinions of three agents that optimize for a single factor led to solutions that were considered as top solutions when ranked together with the ones provided by a GA that optimizes for all the three factors of the multi-objective optimization problem. The top solutions of the multi-objective GA and the solutions of the team were Paretoranked together as 1st rank solutions. We also evaluated a team of 4 agents, where all of them optimize for all the 3 factors of the multi-objective optimization problem. Each agent was initialized with different options for the GA. First we rank together all the solutions of these agents. We found out that one of them had the highest number of top ranked solutions. We evaluate two teams: diverse, composed by one copy of each agent; and uniform, composed by four copies of the best performing agent (the one with the highest number of top ranked solutions). We found that a diverse team was able to propose a higher number of top ranked solutions, outperforming the number of top ranked solutions found by each individual agent and by the uniform team. Each individual agent could find from 6 to 7 top ranked solutions,
Publications:
Heydarian A, Gerber D, Carneiro J P, Becerik-Gerber B, Hayes T, Wood W. (2014) “Immersive Virtual Environments: Experiments on Impacting Design and Human Building Interaction,” CAADRIA2014, 19th International Conference on the Association of Computer-Aided Architectural Design Research in Asia, May 14-17, 2014, Kyoto, Japan
Heydarian A, Carneiro J P, Gerber D, Becerik-Gerber B, Hayes T, Wood W (2015). Immersive Virtual Environments versus Physical Built Environments: A Benchmarking Study for Building Design and User-Built Environment Explorations. Automation in Construction. 54 116.
Analyzing Agent Teams for Design Problems:
Design imposes a novel problem for the multi-agent study of social choice: using a team of voting agents, maximize the number of optimal solutions; allowing a user to then take an aesthetical choice. For maximum applicability, we envision agents that are queried for a single opinion at each decision point of design, and hence multiple solutions are obtained by multiple voting iterations. We developed the first model for such problems, where we study the potential of diverse teams composed by different agents, and uniform teams composed by multiple copies of the best available agent. Besides our theoretical study, we also analyze the performance of these different teams on synthetic experiments and on experiments with Beagle, an actual design system for the early design of energy efficient buildings.
We aim at: (1) develop a theoretical model for agent teams that vote at each decision point of a design problem; (2) use our model to obtain insights into the best team compositions for design problems. In particular, we are interested in studying diverse teams, composed by different agents, and uniform teams, composed by multiple copies of the best agent; (3) develop synthetic experiments, in order to further study our model, and obtain more realistic insights --- for example, what happens in bounded computational time; (4) run experiments in an actual, complex design system. We use the Beagle system to run experiments in architecture design, where we study different teams of agents that vote to design energy-efficient buildings. We analyze the performance of diverse and uniform teams for three building design problems.
In our theoretical study we showed that: (1) uniform teams are in general suboptimal, and converge to a unique solution; (2) diverse teams are optimal as long as the team's size grows carefully; (3) the minimum optimal team size is constant with high probability; and (4) the worst case for teams is a prime number of optimal actions. In our synthetic experiments we showed that: (1) uniform teams really decrease in performance as the team size grows; (2) diverse teams increase in performance as the team size grows; however, performance converges and does not improve much after a certain number of agents; and (3) performance of diverse teams further increases by increasing the number of voting iterations, but with diminishing returns. Finally, in our experiments in architectural design we showed in an actual system that teams of design agents are really able to obtain a much larger number of optimal solutions than the individual agents (in our case, agents running the Beagle system, a genetic algorithm system for building design). This was confirmed across three different building design problems, where the agents searched for optimized solutions for the design of a building according to cost, project requirements, and energy-efficiency.
Results: Coming soon...
Design of a Virtual Feedback Approach for Energy Efficient Building Design and Operations:
Early design and engineering decisions greatly impact buildings’ energy consumption. However, current design methods are based on assumptions regarding occupant behavior and preferences, as represented in designers’ graphic standards, which are usually not validated and unspecific. Currently, to the best of our knowledge, the impact of building design on occupants’ energy relevant behavior is also not well understood. We are investigating how the behavior, performance, and preference of building occupants, building systems and operations of existing buildings can be brought back to the design stage for shaping early and critical energy-aware decisions (Pathway III). In addition, we are investigating if design features and/or operational strategies have any impact on building occupants’ energy relevant behavior in office buildings (PathwayII).
Results: Coming soon...
Publications: Coming soon...
Understanding the Influence of Default Lighting Conditions on Occupant's Lighting Use Behavior
Previous research has shown that occupants have a tendency to keep the lighting levels of an office space if it’s close to their preferences and comfort levels. In this study, inorder to effectively reduce energy consumption of an office room without impacting occupant comfort, we are investigating the effect of default (initial) lighting settings on end-users’ behavior. Our goal is to better understand at which lighting levels the endusers are willing to keep the default lighting condition rather than adjusting it to their most preferred settings. Through this approach, we are aiming to define default lighting conditions for office spaces that are set according to the user’s preferred levels while being energy efficient. To achieve this goal, the participants are randomly assigned to one of the five different virtual rooms (in immersive virtual environments) with different default lighting levels. They are given an option of keeping the lighting levels or adjusting it to their most preferred setting in order to perform a set of office related activities (e.g., reading, object identification, etc.). The main objectives of this study are: (1) to examine the effect of end-user personality and environmental values on their tendency to keep the default lighting settings; (2) to investigate the thresholds that end-users are willing to accept the default conditions in an office setting; and (3) to measure the effect of different default lighting conditions and occupants’ behavior from the energy consumption perspective.
Results:
As part of our preliminary results, we found that people are significantly less likely to change the initial condition if they are put in a room with all blinds open compared to if all the blinds are closed and all the artificial lights are on. Additionally, we found that when participants were placed in a condition with all the artificial lights on with no natural light, 68% of them changed the default condition by opening the blinds and 86% of the participants lowered the artificial lighting levels (turned or more light bulbs off). Out of the participants that were placed in a room with only natural light (no artificial lights), 100% of participants kept at least one blind open, 38% turned one light bulb on, 50% turned two light bulbs on, and 12% turned all three light bulbs on. As part of our initial analysis, the participants with a neurotic personality are significantly less likely to keep the default condition when all the lights are turned on and the blinds are closed; but they are not less likely to change the default condition if the blinds are open. As part of our future work, we are planning to increase our sample of participants and measure the effect of default lighting conditions on the electricity consumption in office buildings.
Publications: Coming soon...
Measuring Occupants' Lighting Preferences:
In order to accurately measure the effect of design features on end-users’ behavior, preference, and performance, it is important to be able to control the environment and only manipulate the design feature of interest. Previous research has identified lighting as one of the most important design features that directly affects the end-users’ mood, performance, and comfort as well the total energy consumption in a building. In order to control all other design features while manipulating only the lighting feature, we have used immersive virtual environments to collect actual end-user lighting-related data. In this study, we aim to create end-user lighting profiles by placing participants in an immersive virtual environment and asking them to adjust the lighting levels of a room to their most desired setting in order to perform a set of office related activities. Specifically, the objectives of this study are: (1) to examine the effect of end-user personality and environmental values on their lighting-setting preferences; (2) to measure the impact of end-users’ lighting preferences on the electricity consumption of the room; and (3) to develop lighting preference profiles of end-users that could be inputted in simulation software for more accurate energy simulations.
Results:
Our preliminary analysis showed that when participants were given the freedom of choosing their most preferred lighting setting, over 90% of them preferred to have some natural light available in the room. 15% preferred to have no artificial lights available (only natural light), 38% preferred to only have one light bulb on along with a number of blinds open (either one, two, or three blinds open) to allow natural light in, 32% preferred to have two light bulbs on along with a number of blinds open, and 15% preferred to have all three light bulbs along with a number of blinds open. As a part of future analyses, we will be examining the relationship between participants’ personalities and environmental values and their lighting preferences. To better understand the effect of end-users’ preferences on the lighting-related electricity consumption, we will also analyze the participants’ lighting preferences through energy simulation.
Publications:
Heydarian A, Pantazis E, Carneiro J P, Gerber D, Becerik-Gerber B (2015). Towards Understanding End-user Lighting Preferences in Office Spaces by Using Immersive Virtual Environments. ASCE International Workshop of Computing in Civil Engineering. Austin, TX.
Heydarian A, Pantazis E, Gerber D, Becerik-Gerber B. (2015). Use of Immersive Virtual Environments to Understand Human Building Interactions and Improve Building Design. Human Computer Interaction International 2015. Los Angeles, CA.
Understanding the Impact of Green Branding on Occupants' Energy Relevant Behavior:
The objective of this study is to investigate the impact of branding a building as LEED (Leadership in Energy and Environmental Design) certified on the building occupants’ environmental and energy consumption related behaviors. This study aims to answer the following questions: (1) how does introducing a building, as a green building influence a building occupant’s environmental and energy consumption behavior in an office space?; (2) how does introducing a building as a green building influence the use of some of the amenities that could be found in green buildings, such as a personal remote control for opening/closing blinds or turning on/off lights? and (3) how do occupants' environmental attitudes affect the influence of green
while the diverse team found 26 top ranked solutions with one of the aggregation methods. The uniform team found at most 8 top ranked solutions. Hence, we are able to facilitate a better design decision-making in the early stage of design by providing the designer with a larger solution pool of high-quality solutions concerning energy and cost efficient designs. Given this larger pool, there is a higher likelihood that the designer will be able to find the best model according to her subjective evaluation.
Publications:
L. S. Marcolino, B. Kolev, S. Price, S. P. Veetil, D. Gerber, J. Musil, M. Tambe. Aggregating Opinions to Design Energy-Efficient Buildings. In the 8th Multidisciplinary Workshop on Advances in Preference Handling (M-PREF 2014), Québec, Canada, July 2014.
Gerber D., Shiordia R, Veetil S, Mahesh A (2014), Design Agency: Prototyping Multi-Agent System Simulation for Design Search and Exploration, Symposium on Simulation for Architecture and Urban Design, April 12-16, 2014, Tampa, FL.
branding on their environmental and energy consumption related behavior? A 3D model of an office space was created in an immersive virtual environment. In order to develop a green environment, the modeled office included green features, such as personal controls for blinds and lighting fixtures, adequate daylight, view, and trash can for recyclables. The following features were integrated in the room’s design: a controllable blind (personal control); controllable lighting fixtures (personal control); a specific trashcan for recyclables, and a regular trashcan. The study included two groups: group 1 (control group): no information was provided about the room being green or having any greenfeatures and group 2 (branding group): information was provided about “green buildings” and LEED certification (its goals and objectives were introduced to the participants) and the room was introduced as being part of a green LEED certified building. So far, we recruited 84 participants (42 participants in branding group and 42 participants in control group).
Results:
Our preliminary analysis showed that when participants were given the freedom of choosing their most preferred lighting setting, over 90% of them preferred to have some natural light available in the room. 15% preferred to have no artificial lights available (only natural light), 38% preferred to only have one light bulb on along with a number of blinds open (either one, two, or three blinds open) to allow natural light in, 32% preferred to have two light bulbs on along with a number of blinds open, and 15% preferred to have all three light bulbs along with a number of blinds open. As a part of future analyses, we will be examining the relationship between participants’ personalities and environmental values and their lighting preferences. To better understand the effect of end-users’ preferences on the lighting-related electricity consumption, we will also analyze the participants’ lighting preferences through energy simulation.
Publications: Coming soon...
Publications:
L. S. Marcolino, D. Gerber, B. Kolev, S. Price, E. Pantazis, Y. Tian, M. Tambe (2015). Agents vote for the environment: Designing energy-efficient architecture. AAAI Workshop on Computational Sustainability. Austin, TX.
Gerber D, Pantazis E, Marcolino L S. (2015). Design Agency: Prototyping Multi-Agent Systems in Architecture. Computer- Aided Architectural Design Futures -- New Technologies and the Future of the Built Environment. Sao Paulo, Brazil.
Gerber D, Pantazis E, Marcolino L, Heydarian A (2015). Multi Agent Systems for Design Simulation Framework: Experiments with Virtual Physical Social Feedback for Architecture. 6th annual Symposium on Simulation for Architecture and Urban Design (SimAUD). Washington D.C.
Shiordia-Lopez R, Gerber D. (2014). Context-Aware Multi Agent Systems: Negotiating Intensive Fields. 2014 Association of Computer Aided Design of Architecture (ACADIA). Los Angeles, CA.
Pantazis E, Gerber D (2014). Material Swarm Articulations–New View Reciprocal Frame Canopy. 32nd eCAADe. New Castle, UK.
Acknowledgment and Disclaimer: This material is based upon work supported by the National Science Foundation under Grant No. 1231001. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.