Optimal building envelope design based on simulated ...

G Model

ARTICLE IN PRESS

ENB-6145; No. of Pages 12

Energy and Buildings xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild

Optimal building envelope design based on simulated performance: History, current status and new potentials Yu Huang, Jian-lei Niu ∗ Department of Building Service Engineering, Faculty of Construction and Environment, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China

a r t i c l e

i n f o

Article history: Received 26 June 2015 Received in revised form 28 August 2015 Accepted 13 September 2015 Available online xxx Keywords: Design optimization Building performance simulation Building envelope Optimization algorithms Optimization objectives

a b s t r a c t Green building design is presently among the hottest research topics in the world. Maintaining a comfortable indoor environment with minimum energy consumption is a challenging task that attracts the attention of experts around the world. With the recent advances in building performance simulation tools, it is now possible to predict and assess building performance at the design stage. Simulation-based optimization of building design is a potential application that connects building performance simulation with optimization algorithms. In this paper, numerous studies on the optimization of building envelope design were assembled and reviewed. Popular optimization algorithms were compared and discussed. Targeted objectives were collected and summarized. Based on the statistical results, the limitations in this research area were identified, and some potential breakthroughs were suggested. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Since the “energy crisis” in the 1970s, saving energy has become a common-sense objective for people around the world. Buildings, industries and transportation systems are the major factors in energy consumption. Buildings are already responsible for more than 30% of the total energy consumption of human kind, and this figure is expected to grow [1]. Compared with the other factors, saving energy in the building sector is the simplest and most efficient [2]. Based on the above considerations, it is clear that energyefficient buildings, sustainable building design and retrofitting are necessary strategies for the future of human society. In addition, buildings are among the most important elements of everyday life. Normally, a person will spend over 70% of his lifetime inside buildings. A comfortable indoor environment can not only improve the occupants’ working efficiency, but also preserve their health. To achieve a comfortable indoor environment in an energy-efficient building is a goal that draws the attention of professionals in architecture and in civil, mechanical and electrical engineering. With the increase of computer calculation capacity, a large number of tools for the simulation of building performance have emerged since the later years of the last century. With their

∗ Corresponding author. E-mail addresses: [email protected] (Y. Huang), [email protected] (J.-l. Niu).

user-friendly interfaces and sophisticated calculation engines, these simulation tools can easily display the thermal, visual and acoustic performance of a building. The reliability of these simulation tools has been tested and proved by various studies [3–8]. With the assistance of such tools, researchers can investigate the effects of different design parameters on a building’s performance. They can determine the sensitivity of building performance to various parameters, thereby developing a reference for actual building design activities. This simulation-based building design process has become a common practice in the construction industry. However, the number of parameters that can affect a building’s performance is rather huge, and in many cases different parameters exert conflicting influences. To achieve an optimal design solution through building performance simulation tools often requires running a large number of simulation cases. This process can be expensive and time-consuming. Conducting a systematic and effective optimization process for building design solutions is becoming a hot topic for researchers in the building performance simulation area. There have been several previous literature review papers that have surveyed the available studies on the optimization of building design. Stevanovic´ made a summary of previous studies on the optimization of passive solar design in buildings. He focused on statistical analysis of the existing studies, trying to identify the most popular building performance simulation tools, the most popular optimization objectives and the most popular search engines used [9]. Evins reviewed 74 studies, seeking to provide a brief

http://dx.doi.org/10.1016/j.enbuild.2015.09.025 0378-7788/© 2015 Elsevier B.V. All rights reserved.

Please cite this article in press as: Y. Huang, J.-l. Niu, Optimal building envelope design based on simulated performance: History, current status and new potentials, Energy Buildings (2015), http://dx.doi.org/10.1016/j.enbuild.2015.09.025

G Model ENB-6145; No. of Pages 12 2

ARTICLE IN PRESS Y. Huang, J.-l. Niu / Energy and Buildings xxx (2015) xxx–xxx

introduction to computer-assisted optimization methods for sustainable building design, including overviews on research concerning the optimization of building envelopes, HVAC systems and the renewable energy supplies [10]. Nguyen et al. composed a summary with an introduction on the details of the optimization process used during building performance analysis. These authors discussed the main activities performed during the major phases of the optimization process and the difficulties or challenges for different optimization problems [11]. Machairas et al. and Negendahl studied the commonly used algorithms and building performance models, and explored how these models can be integrated for building design optimization [12,13]. Carlucci et al. focused on a specific optimization objective, namely visual comfort. They discussed the descriptions and means of assessing visual comfort and presented the conventional process for the optimization of visual comfort [14]. Attia et al. interviewed 28 optimization experts and made a summary of their opinions, which provided a general picture of the progress in research on building performance optimization, including the algorithms, simulation tools and expected future developments [15]. Clearly, all of these previous reviews have summarized the progress of research on building performance optimization problems in general. Actually however, there are two main objectives in building design, namely HVAC system design and building envelope design. Compared to HVAC system design, the design of building envelopes involves a much larger number of parameters. The relationships between different design parameters and the effects of those parameters on the performance of the building envelope are more complicated. Besides, the different directions of heat flow which determine the heating load and cooling load is also an important issue that should be considered for building envelope energy performance. Also, the daylight that travels directly through building envelope should also be considered carefully, for it not only affect the radiation heat gain, but also serves as natural light source which would significantly affect the lighting energy consumption. In addition, there are more evaluation systems available for analyzing the performance of the building envelope. This review paper focuses specifically on simulation-based optimization in building envelope design. The paper is divided into five parts. Part 1 gives a brief introduction to the paper’s topic. Part 2 discusses the major algorithms used for building envelope optimization. Part 3 introduces the main objectives of simulation-based optimization for building envelope design. Part 4 presents the most popular optimization software tools that were surveyed in previous studies. Part 5 gives a summary of the paper and discusses possible future work in this research area.

2. Algorithms applied in simulation-based building envelope optimization 2.1. History and early development of the methodology for simulation-based building envelope design optimization Before all the following discussion, one important factor should be pointed out clearly. The design of an energy-effective building is restricted by many external environment conditions, such as the geography location, the local location and the climate condition. Besides, many internal factors would also affect the design result, such as the purpose of the building, the operating schedule of the building as well as the occupant behavior inside the building. An optimal building design can only be achieved when all these boundary conditions were defined, thus every optimization case is restricted by these boundary conditions. An optimal result is only significant within the corresponding boundary conditions. Once the boundary conditions change, the result will also change.

As early as 1983, Gero et al. discussed a simple multi-criteria model for the optimization of a building’s energy performance. The method they applied was quite similar to the multi-criteria Pareto optimization [16]. In 1990, Bouchlaghem and Letherman reported an attempt to optimize a building envelope’s thermal performance in maintaining a comfortable indoor temperature. These researchers applied a hybrid simplex and non-random complex algorithm, with the indoor environment as their optimization objective [17]. In 1992, Sullivan et al. introduced the concept of optimizing the building envelope based on simulations. These authors applied a regression analysis as their optimization algorithm. With a series of DOE-2-based simulations, they generated a database for the energy performance of various building envelopes and lighting systems. Then they defined two variables, namely the solar aperture (which is a function of the shading coefficient and the window-to-wall ratio) and the effective daylighting aperture (which is a function of visible transmittance and the window-towall ratio). With this energy performance database, they conducted a regression analysis of the relation between electrical energy consumption and the two pre-defined variables [18]. Although these various approaches were rather preliminary and had many limitations, they were still pioneering efforts in the search for optimal solutions in the area of building design with the assistance of numerical simulations, and thus they cast light on the issues involved. Al-Homoud reported his work on the optimal thermal design for office buildings in several cities in the U.S. and Saudi Arabia. He considered 14 design variables in the optimization process. First, he defined an initial value and the upper and lower boundaries for each variable. With a selected value difference, he conducted over 700 simulation runs and listed the thermal performance of every possible combination of variables [19]. In 2001, Depecker et al. reported a very simple study on the optimal building shapes for various climates. They defined the shape coefficient, which was the surface area of the building envelope divided by the building’s volume. Then they utilized the shape coefficient to describe the shape of the buildings. From their results, they concluded that only in extremely cold climates would a building’s shape affect its energy performance [20]. Ghisi and Tinker presented a study on the optimal design of window-to-wall ratios in single office rooms. They made the minimization of total building energy consumption their primary objective and conducted 17,600 DOE-2-based simulation cases. From all of these simulation cases, they were able to choose the solutions with the best energy performance. They then summarized several design principles as references for designing building envelopes in England and Brazil [21]. During the studies of this period, researchers started to realize that although an optimization process could achieve a reduction in energy consumption, the approach used was far too timeconsuming. Their studies showed the importance of applying a solution searching engine in the process. During the 2000s, the development of mathematical and algorithmic methodologies raised the possibility of solving optimal building envelope problems more quickly and accurately. Among all of the methodologies proposed, the direct search and the stochastic population-based search (evolutionary algorithm) were the most popular. 2.2. Direct search The direct search methodology relies completely on the value of the objective function. The basic principle of direct search is a focus on searching around the current solution point. During a direct search, a current point is first defined with the value of the objective function. A series of points are then searched, and their objective function values are recorded and compared. If an objective function value that is closer to the optimization target is


G Model ENB-6145; No. of Pages 12


achieved, the corresponding point is then defined as the latest current point. The process is repeated until an optimal point is found [22]. Direct search methodologies can be generally divided into two types, namely the gradient-deterministic and the gradient-free methods. 2.2.1. Gradient-deterministic search The gradient-deterministic search can be considered the most straightforward optimization methodology. In a gradientdeterministic problem, there usually exists an objective function that can be analyzed with Taylor’s series expansion. In this situation, the optimal solution can be easily obtained by going in the direction that has a reducing gradient. Since the early 2000s, a series of optimization studies have appeared that utilized the gradientdeterministic search in dealing with various design variables. In 2003, Marsh presented a methodology for optimizing the geometric design of a shading shape. Taking the shading shape as his main design variable, he applied a ray-tracing technology to display the shadow area required, and a cut-off scheme to form the shading shape. He sought to develop a function to connect the shading shape with the solar heat gain [23]. In 2004, Al-Homoud presented a brief summary of the optimization problem in architecture, and described an optimization approach based on the direct search technique. He claimed that an effective result search method would significantly improve the speed of the optimization process. Later, with the assistance of his optimization approach, he conducted a series of studies on the optimization of designs for mosques. These buildings had special occupation schedules, with five peak values each day [24,25]. Wang et al. made an optimization on the parameters of façade design. They considered both the U-value of the wall material and the window-to-wall ratio, trying to find an indoor environment with optimal thermal comfort [26]. Adamski applied the ratio of the area of the southern part of building S1 to the area of building S as the design variables, and made the payback period the optimization objective in his research. He developed a mathematical model to describe the relationship between the building’s shape and the life cycle cost. With the assistance of computer-based simulation, he was able to select the optimal case directly [27]. Ucar and Balo introduced a simple optimization case for selecting the thickness of insulation materials. They considered the payback period as their optimization objective, and deduced the functional relationship between heat loss and the thickness of insulation materials. With that function identified, they were able to make a quick and simple selection of the optimal solution [28]. Later, Ucar and Balo reported a similar case study that involved the application of their previously proposed method [29]. A similar methodology could be inferred in the study by Lollini et al. [30]. Bambrook et al. reported a case study on the optimization of the envelope for an individual residential house in Sydney. They conducted 210 simulation cases to test the performance of popular designs for the walls, windows and shading devices, trying to reach a minimum life cycle cost [31]. Albatici and Passerini presented a simplified approach to search for the best solution for a building shape that would minimize the heating requirements for a residential building in Italy. They defined a new index, named the south exposure coefficient, and did a regression analysis on this coefficient and the shape coefficient. They claimed that both the shape coefficient and the south exposure coefficient were almost linear in relation to the heating energy consumption, and that these two indices could guide the building design regardless of the building’s volume [32]. Jiang et al. described a study to optimize a building internal envelope’s specific heat for the purpose of phase change material application. They constructed a simplified room temperature model and did an analytical optimization of room temperature, trying to build a function for connecting the building internal envelope’s specific heat and the indoor space temperature [33]. Later, Cheng et al. reported a

3

similar study on the optimization of a building external envelope’s transient performance [34]. From the above-described studies, it is clear that applications of the gradient-deterministic search methodology have continued until very recently. The extensive use of this approach has indicated that compared to more complex algorithms, the performance of the gradient-deterministic search methodology was considered satisfactory. In 2014, Asadi et al. proposed a multi-linear regression method for estimating the total energy consumption of a building. They conducted over 10,000 simulation cases and selected 17 building design variables for the multi-linear regression. A coefficient of determination (R2) of around .94–0.95 was achieved, which indicated that 94–95% of the building’s energy consumption could be determined by those 17 variables. This research showed that regression analysis based on the gradient-deterministic search methodology was still an acceptable approach in building design for the optimization of energy conservation [35]. It should also be noted that the majority of existing gradient-deterministic search studies have been focused on the minimization of building energy consumption. Only one study was about optimizing thermal comfort [27]), and this study indicated that the performance of the gradient-deterministic search methodology was inadequate for solving other optimization problems. 2.2.2. Gradient-free search A gradient-free search methodology does not require any information about the gradient of the objective function. As mentioned above, this method constantly replaces the current point with different searching rules, trying to approach the optimal solution. The basic principle of different search methods was to find a point when no further increase or decrease was observed. The differences among these methods were the determination of the step size. According to the various methods used in searching for neighboring points, the gradient-free search methodology can be further divided into the pattern search method, the simplex method and the adaptive search directions set method. All of the present gradient-free search methodologies, such as the Hooke–Jeeves algorithm, Tabu search and the orthogonal method, have been modifications of these three basic methods [36]. In 1990, Bouchlaghem and Letherman tried to apply the simplex method for the optimization of indoor thermal comfort [17]. Hasan et al. applied a hybrid GPS Hooke–Jeeves/PSO algorithm in their optimization of a Finnish house for minimum life-cycle cost [37]. Futrell et al. reported a study on the optimization of a classroom’s thermal and visual performance. In that study, the authors also selected the hybrid GPS Hooke–Jeeves/PSO algorithm to obtain the optimal design solution for the window’s size, location and optical properties. Both energy consumption and visual comfort were chosen as the objectives in this study. Futrell et al. discovered that for the south orientation, the thermal and the visual performance did not significantly conflict, but for the north orientation these values had the greatest conflict [38]. Stazi et al. conducted a life cycle assessment on the optimization design of solar wall systems. They proposed a “factorial plan technique” for selecting the optimal design. These researchers determined n parameters as the design variables and assigned these variables a 2-level value definition. Then they conducted 2n simulation runs to search for the optimal result. Stazi et al. claimed that their method was fast, simple and intuitive. However, it could only be applied on simple systems, and the results were not entirely accurate [39]. Gong et al. presented their research on applications of the orthogonal method for the optimal passive design of residential building envelopes in 25 Chinese cities. They selected 7 control parameters and defined the parameters into 4 levels. In addition, they considered 15 possible interactions between parameters. With the help of a L32 (231 ) matrix, they were able to select the optimal design solution for each




4

city. Gong et al. claimed that unlike other algorithms, an orthogonal method did not require a deep knowledge of computer programming, and thus was suitable for architects [40]. Ruiza et al. reported a case study on a Spanish residential building optimization design based on a Tabu search algorithm. They considered the building’s energy efficiency and the life cycle cost as their optimization objectives [41]. 2.3. Stochastic population-based search An evolutionary algorithm is a stochastic population-based search that was inspired by studies of biological evolution. The core of an evolutionary algorithm is the selection of high-performance individuals and the reproduction of new individuals, based on existing performance data. A typical evolutionary algorithm process includes selection, crossover and mutation. First, a series of points are generated randomly as the initial population. An evaluation is then conducted to assess the fitness of each individual. The individuals that best fit the optimal solution are chosen as the parents. The parents are applied for the reproduction of children, which form the next generation through crossover and mutation. Then the new population goes through a new round of selection. This process can be repeated until the termination of the algorithm [42]. Compared with the direct search methodology, the evolutionary algorithm has a higher calculation speed, greater accuracy and a stronger adaptability. Based on their differing implementation details, evolutionary algorithms have been classified into genetic, neuroevolution, particle swarm optimization and other types of algorithms. 2.3.1. Genetic algorithm The genetic algorithm (GA) is no doubt the most popular method applied in simulation-based design for building envelope optimization. Of all the studies surveyed in this paper, over 60% were conducted via the GA. In the GA, many individuals are considered simultaneously, so that the possibility of ending up at a local minimum is reduced [43]. The GA and its modifications are considered to be the best choices for solving building design optimization problems. As early as 2002, Coley and Schukat attempted to introduce the GA for building design. They first built a very simple thermal model in which only five variables were considered. Then they applied the GA to search for the solutions with minimum annual energy consumption. Their results were satisfactory, and a large number of optimal solutions were identified [44]. In 2003, Wang et al. presented a case study in which the GA was applied to minimize the life cycle exergy. In 2005, Wang et al. applied a GA for the optimization of a rectangular-shaped building with a fixed floor area. They considered the life cycle exergy as their objective, and tried to find the optimal solution for the building’s orientation and its building materials [45,46]. Later, Wang et al. summarized the previous optimization studies, and claimed that although all of these studies applied the GA, the methodologies they used lacked versatility. The optimization approach would fail when applied to another design case. Wang et al. suggested the development of an object-oriented framework, so that the GA method could be easily adopted with numerical simulation within a much more user-friendly interface. This approach would significantly improve the efficiency of the optimization process [47]. With their developed approach, Wang et al. conducted a systematic study to consider façade design variables, including shape, structure, material type and shading device [48]. Wright and Mourshed designed an interesting study to test the stochastic behavior and reliability of the GA in optimizing the window-to-wall ratio in a building envelope. They divided the building façade into small rectangular cells in which each cell could

be equipped with either solid wall material or glazing. They ran the GA-based optimization several times, and found that although the distribution of glazing cells was different for each optimal solution, the number of glazing cells stayed constant [49]. Tuhus-Dubrow and Krarti conducted an optimization on the shape of a U.S. residential building envelope. They chose the minimization of whole-building energy consumption as their objective. Their results showed that although there were some differences in energy consumption among different building shapes, these deviations were within 0.5%. These authors claimed that other variables such as orientation or construction materials could affect the building’s energy performance more significantly, and thus should receive more attention [50]. Sahu et al. conducted a GA-based design optimization on the shape, orientation and materials of the building envelope. They selected minimized energy consumption as their objective and did a detailed validation with the TRSYS simulation tool. They discovered that the climate had an effect on the accuracy of the optimization. If the space load situation was complex, with a widely variable combination of heating, cooling and humidity conditions, the degree of error would be larger [51]. Ioannou and Itard conducted a Monte Carlo sensitivity analysis against various design variables in terms of their effects on building energy consumption. They claimed that under heating conditions, the window thermal properties were the most critical parameters among all of the variables, because heat loss through the windows took the majority of the total heating load. A similar conclusion could be applied in other climate situations. The sensitivity of a specific design variable depended on how it affected the optimization objective [52]. In addition to the optimization of energy performance for building envelopes, the GA was applied in optimization problems involving visual comfort assessment. Torres and Sakamoto conducted an optimization of daylighting performance. They considered 21 variables in the design of window and shading systems, in an effort to minimize visual discomfort. They proposed a modification to the existing daylight glare probability index (DGP) and applied a dynamic daylighting simulation software called Radiance to calculate the values of DGP from several viewpoints inside the building. Considering the modified DGP as their optimization objective, Torres and Sakamoto selected the GA to search for the optimal solution. They discovered that the GA was especially effective for optimization of daylighting performance, because it obtained a fast convergence. They also claimed that hybrid fitness may be helpful for the accuracy of the optimization [53]. Later, Kämpf et al. reported a similar study that involved the coupling of Radiance and a GA search engine for the optimization of building shape with consistent building volume. They announced a further implementation of an urban scale optimization by using the same methodology [54]. Gagne and Andersen proposed a simulation approach for the optimization of façade design in terms of daylighting objectives (illumination level and glare). In their approach, a simple data model of the buildings was first defined, based on the design case from the designers. Then a series of new 3D building models were automatically generated during the optimizations. Micro-GAs were applied for both single-objective and multi-objective optimizations. During their research Gagne and Andersen discovered a notable limitation: the micro-GA-based optimization required much more computing time and a relatively large population size [55]. Rakha and Nassar developed a GA-based searching method for the optimization of geometric forms of ceiling areas. They claimed that although their approach was time-consuming, it could reach the optimal solution precisely and was thus useful for the optimization of daylighting [56]. Yi and Kim also reported several case studies in which the GA indicated the optimal designs of sunlight exposure time on tall residential buildings [57].




5

Table 1 Different modification approaches applied for the improvement of the genetic algorithm’s performance. Names of researchers

Modification approach

Major finding

Reference

Znouda et al.

Applying elitism selection instead of traditional wheel selection. Introducing an immigration procedure into the mutation process. Applying an omni-optimizer, a simulated binary crossover operator and a polynomial mutation in crossover and mutation operations. Application of hybrid covariance matrix adaptation evolutionary strategy and HDE algorithms.

The speed and the accuracy of the genetic algorithm search engines were largely improved.

[62]

The stopping criterion definition was very important for the reduction of computing time.

[63]

The covariance matrix was used for decorrelation of design variables. Although the calculation of the covariance matrix was time-consuming (compared with the computing time used for annual building simulation), it was acceptable. The proposed approach could save around 80% of the computing time with a negligible change in the degree of error. The degree of error could be reduced by more than 70% when design space exploration was utilized.

[64,65]

With the consideration of mutual information, the data mining with genetic algorithm optimization could save 40% of the computing time and improve the accuracy by 25%. In 1/3 of the optimization runs, the accuracy was improved by at least 5%, but the computing time did not change significantly. A small relative error (of 1.7% for energy consumption and 2.1% for indoor thermal comfort hours) was reported on the prediction.

[68]

Palonen et al.

Kämpf and Robinson Ramallo-González and Coley

Eisenhower et al. Stavrakakis et al. Geyer and Schlüter

Bucking et al.

Applying multiple algorithms for building envelope optimization. Applying a meta-model approach to avoid repeating simulation cases. Introducing design space exploration into the traditional response surface method to further improve the method’s accuracy. Considering mutual information (dependency between variables) during the evolution of the design cases.

Junghans and Darde

Combining a genetic algorithm and a simulated annealing algorithm.

Yu et al.

Applying the genetic algorithm into the back propagation neural network.

In terms of applying GA-based building envelope optimization to the issue of thermal comfort, few relevant studies were found. In the majority of cases, thermal comfort was considered a subsidiary factor of energy consumption. Hamdy et al. carried out an optimization study aiming to integrate both thermal comfort and energy performance as dual optimization objectives. They adopted the GA optimization tool of Matlab for an IDA-ICE 4.0 building performance simulation. Their results indicated that an optimal solution for both thermal comfort and energy saving would require 10 kWh/(m2a) more than an optimal solution for energy saving alone [58]. Stavrakakis et al. also reported a study on the optimal window-opening design for enabling thermal comfort, based on meta-model construction [59]. Yu et al. conducted a multiobjective optimization, trying to improve energy performance and indoor thermal comfort at the same time. During their optimization process, they applied the GA into the back propagation neural network to speed up the simulation and keep the result more accurate. They discovered that under the same level of energy consumption, the indoor thermal comfort status did not change significantly. If a specific energy consumption value was fixed, the difference in terms of thermal comfort hours among the different cases was insignificant [60]. 2.3.2. Improvement of the genetic algorithm (GA) With the ongoing development of computer technologies and other mechanisms, many professionals started to notice the limitations of the GA, as its searching speed was still not acceptable. The search results were also highly affected by the definitions of selection, crossover and mutation. Great efforts have been taken to improve the performance of the GA in building envelope optimization design. For the reduction of computing time, in general there are two methods. The first is to apply simplified models instead of complex ones. The second method is to reduce the size of the population and the number of generations. For improving the accuracy of results, a common approach is to conduct a dominating relationship-based selection before generation. In other words, this approach involves increasing the probability of survival for the

[66,59,67]

[69]

[60]

better individuals [61]. Table 1 summarizes the different approaches that have been applied for the improvement of the GA’s performance. 2.3.3. Harmony search algorithm The harmony search algorithm is another population-based optimization method. This method was inspired by improvements in the techniques that musicians use to find the best harmony [70]. The harmony search algorithm has been considered a special case among evolution strategies [71]. Fesanghary et al. presented a multi-optimization model based on a harmony search algorithm. They selected life cycle cost and CO2 emission as their optimization objectives. They applied their approach in the optimization of a residential house in the U.S., and claimed that although the method required a relatively long computing time, its results were stable and trustworthy [72]. 2.3.4. Evolutionary artificial neural networks Artificial neural networks are a series of learning models that were inspired by studies of biological neural networks. These networks can be described as systems of neurons that connect and send messages to each other, so that the network can learn from present observations and improve its performance during tasks. The core of an evolutionary artificial neural network application is the training of the neural network. Magnier and Haghighat proposed an optimization methodology in which an artificial neural network and a GA were coupled. The artificial neural network was first applied to mimic the behavior of the building simulation model. Then, the output of the artificial neural network was used as the population. The key factor in this methodology was the training of the artificial neural network. As long as the artificial neural network could output accurate building performance, the optimization result could be received quickly and reliably [73]. Zemella et al. introduced evolutionary artificial neural networks that combined evolutionary algorithms with artificial neural networks. They divided the total population into two parts, with 80% used for training a three-layer neural network, and




6

the other 20% of the population used for root predictive error testing. During the training of the neural network, the evolutionary algorithm was applied for adjustment of the structure parameters. Zemella et al. claimed that their proposed approach was more accurate and less time-consuming, due to the application of the artificial neural network [74]. Later, Gossard et al. reported similar results from tests that involved cooperation between the GA and an artificial neural network [75]. 2.4. Summary From the above-described studies it can be concluded that among the direct search methods for building envelope optimization, the gradient-deterministic direct search methodology was more popular than the gradient-free direct search methodology, even though the gradient-free direct search method was more advanced. The reason for this preference lies in the limitations of direct search. The gradient-deterministic direct search methodology was only applicable for relatively simple optimization problems. As for the gradient-free direct search methodology, it was found to be time-consuming. In addition, the accuracy of direct search was not satisfactory. When encountering simple optimization problems, researchers have tended to use the gradient-deterministic direct search. When encountering complex optimization problems, researchers have tended to use algorithms (such as evolutionary algorithms), which take less time and perform more precisely. Among the evolutionary algorithms, the GA and its modifications were the most widely applied choices, due to their vast applicability, high accuracy and speed of operation. It should also be noted that among all of the studies reviewed, the majority concerned the optimization of energy-related performance, and relatively few studies were found on optimal design for visual or thermal comfort. Also, most studies were single-objective optimizations. The limited number of multi-objective optimization reports all concerned thermal comfort and energy efficiency. 3. Optimization objectives 3.1. Classification of objectives The building envelope is the most important element in a building. Building envelopes separate the building’s indoor environment from the outside world, supplying the occupants with a stable and comfortable living space. As mentioned above, the target of sustainable building envelope design is to maintain a comfortable indoor environment with minimum energy consumption. Thus, naturally, the indices that describe the performance of the building envelope are the main focused objectives in building envelope design optimization problems. These indices include the following: (1) Energy performance index, including space cooling load, space heating load, HVAC system energy consumption and total building energy consumption. (2) Life cycle cost index, including life cycle CO2 emission, life cycle primary energy consumption and life cycle financial investment. (3) Thermal comfort index, including mean PMV level, mean PPD level, thermally comfortable hours, space temperature and operative temperature. (4) Visual comfort index, including illumination level, illumination uniformity, daylight factor, daylight autonomy, useful daylight illuminance and DGP. We conducted a statistical analysis to show the attention given to each of these possible objectives. The result is presented in Fig. 1.

Fig. 1. Statistical result for optimization cases focused on different objectives.

It should be noted that the life cycle cost index has a direct connection with the energy performance index. With some simple additional information, it is easy to transfer energy consumption into life cycle cost [76]. Thus in Fig. 1, studies focusing on life cycle cost optimization are counted under the category of studies on energy consumption optimization. From Fig. 1, it is clear that concern for energy consumption drew the most attention. Among the over 70 optimization cases surveyed in this paper, over 80% aimed at minimizing the energy consumption or life cycle cost. There were comparatively few optimization cases that focused on comfort issues. Up to the present, the factors that affect the occupant’s thermal comfort have been deeply investigated, and the indices describing thermal comfort status have been widely accepted. However, most studies on thermal comfort have been focused on the design and control of HVAC systems. Not many researchers have done detailed analyses on the effects of building envelopes on indoor thermal comfort. With respect to visual comfort, academics still lack accurate and effective measures for assessing the visual environment. The existing visual environment optimization cases were mainly based on a simple assumption that sufficient daylight and acceptable distribution of illumination could produce visual comfort. For the future, the improvement and adoption of a system to assess the occupant’s comfort is a possible direction for simulation-based building envelope optimization design. In addition, the effect of occupant behavior cannot be ignored. Until now, occupant behavior has been introduced in building simulations only in terms of fixed schedules. However, occupants are not just components of the building system that accept the controlled environment passively. Occupants take initiative to adjust their surrounding environment by modifying the HVAC system and the shading devices. The activities of the occupants therefore affect energy consumption, indoor comfort status and even the occupant’s future activities. How to consider occupant behavior more precisely in building simulations is a challenging topic for academics. 3.2. Multi-objective optimization As the GA simultaneously calculates a set of points, it is able to reach multiple Pareto optimal solutions within one calculation run, which makes it a perfect choice for multi-objective optimization problems. Caldas and Norford proposed a computing method for design optimization in the placing and sizing of window openings for office buildings. They used DOE-2 to simulate the annual




air-conditioning and lighting energy consumption, with an objective of minimizing such consumption. GAs were applied to search for design solutions, and the researchers reached several conclusions that were widely accepted by other professionals. The results indicated that the climate, orientation and type of the building can all significantly affect the results of the optimization. The analyses also showed that in many cases, similar minimum annual energy consumption could be achieved with several different configurations, which illustrated that a multi-objective optimization was possible [77]. Manzan and Pinto presented a case study for optimizing the positions and sizes of the external shading equipment. They prepared a simplified office model and conducted the energy and daylighting simulations to form the database. They claimed that with the optimization process, the differences in energy consumption between the best and the worst solutions could be as large as 17% [78]. Hamdy et al. constructed a multi-objective optimization method to guide the design of a residential building in Finland. In their method the GA was utilized, and minimizing the primary energy consumption and life cycle costs were considered as the objectives. Based on their method, Hamdy et al. proposed a series of reference data for residential house designs in Finland [79]. Later, Ferrara et al. reported a similar study on the optimal design of a residential house in France [80]. Karmellos et al. suggested that a multi-objective optimization should consist of two parts: an optimization process and a decision making process. They provided a decision maker software tool to assist in the multi-objective optimization process [81]. From the existing reports, it can be easily observed that most of the so-called multi-objective optimization studies in the building envelope design area have considered energy consumption and life cycle cost as their two objectives. However, life cycle cost and annual energy consumption actually have a direct relationship. With some extra information, it is not difficult to calculate life cycle cost (be it financial cost, primary energy cost or CO2 emission cost) from the annual energy consumption data. Actually, these studies cannot be treated as “multi-objective optimizations.” There have been very few optimization studies that combined the objectives of comfort and energy consumption. 3.3. Summary A summary of the key factors and major findings of the above literatures was displayed in Table 2. From the statistics and analyses of the targeted optimization objectives in the collected studies, several obvious conclusions can be drawn. Of all the previous studies on simulation-based building envelope optimization, over 90% have focused on the optimal solution for a single objective. Of all the single-objective optimization cases, over 80% were targeted at the minimization of energy consumption. In the few multi-objective optimization studies, energy consumption was always included as one of the objectives. At present, energy efficiency is still the top concern for scientists and engineers during the design stage of buildings, and energy consumption is the primary issue that is considered and analyzed. In addition, the present limitations in defining the status of occupant comfort and the lack of precise descriptions of occupant behavior also constrain the development of comfort-related quantitative evaluation systems, which may be a potential direction for future research. 4. Common optimization tools As the development of computer science, many software tools focused specifically for the optimization process were designed nowadays. These tools offer user-friendly interface and ports with

7

which the programming process of the optimization studies can be largely simplified. 4.1. Matlab Matlab is a world-famous multi-paradigm numerical computing tool that allows algorithm development, data visualization, numerical calculation and interaction with programs written in other computer languages. Matlab has a toolbox specifically designed for optimization. The Matlab Optimization Toolbox includes a large range of choices among algorithms. The toolbox’s ability to cooperate with other programs makes it a perfect numerical environment for optimization problems based on third-party simulation programs. Before the appearance of software developed specifically for optimization, Matlab was the first choice of researchers who aimed at simulation-assisted building design optimization. Even now, Matlab is still a popular choice, because if researchers use Matlab in the optimization process, they can use all of the other Matlab functions that provide significant improvements in data presentation and analysis. Lu et al. conducted a series of optimization processes on the passive solar and renewable energy design of a reference building envelope in Hong Kong. They applied the GA for a singleobjective optimization process and the Non-dominated Sorting Genetic Algorithm (NSGA-II) for a multi-objective optimization process in the same Matlab environment. With the assistance of Matlab software, they were able to utilize both algorithms easily, regardless of the ten-fold differences in the numbers of generations and in computing-time between the two algorithms [82]. McKinstray et al. also applied the NSGA-II in Matlab Optimization Toolbox for his research on the optimal BIPV installation for a single-story building [83], and Shea et al. applied ant colony optimization from the Matlab Optimization Toolbox in their research [84]. 4.2. GenOpt GenOpt is a generic program that was developed by the famous simulation research group at the Lawrence Berkeley National Laboratory. GenOpt can be used with any building simulation tools that have text-based input and output. The users can add self-defined optimization algorithms into GenOpt’s library, which makes GenOpt a practical and flexible optimization tool. However, GenOpt cannot carry out multi-objective optimization. The postprocessing ability is also limited [85,86]. Holst reported an optimization case study involving cooperation between the thermal performance simulation tool EnergyPlus and the decision making software GenOpt. He considered the annual primary energy consumption as the optimization objective, while trying to find the optimal solution for 14 design variables in building envelope design. He also addressed the thermal comfort status in the optimal solution, and found an improvement in thermal comfort [87]. Similarly, Hasan et al. reported an optimization case that involved cooperation between DOE-2 and GenOpt [88]. 4.3. modeFRONTIER modeFRONTIER is a platform developed by the ESTECO corporation. The advantages of this platform mainly lie in its workflow interface, variable choices of algorithms and its ease in combining with other simulation tools [89]. Shi integrated modeFRONTIER with the popular building performance simulation tool EnergyPlus to conduct an optimization on the installation of insulation material in an office building envelope [90].


G Model

ARTICLE IN PRESS


Y. Huang, J.-l. Niu / Energy and Buildings xxx (2015) xxx–xxx

8 Table 2 Key factors and major findings of reviewed literatures. Author and year

Optimization objective

Optimized parameters

Tools and algorithms

Key results

Al-Homoud in 2004 and 2009 [24,25]

Energy consumption

Direct Search

Wang et al. in 2007 [26]

Indoor thermal comfort Payback period

U-value, absorptance, emittance, glass area percentage U-value of the wall, window-to-wall ratio The ratio of the area of the southern part of building S1 to the area of building S Thickness of insulation materials

An effective result search method would significantly improve the speed of the optimization process. The optimal U-value for wall material is affected by orientations. A function was achieved which describe the relationship of S1/ S and building operation cost F.

Adamski in 2007 [27]

Direct Search Direct Search

Ucar and Balo in 2009 and 2010 [28,29], Lollini et al. in 2006 [30] Bambrook et al. in 2011[31].

Payback period

Albatici and Passerini in 2011 [32].

Heating requirements

Jiang et al. in 2012 [33], Cheng et al. in 2014 [34]. Asadi et al. in 2014 [35].

Room temperature

Building internal envelope’s specific heat

Direct Search

Energy consumption

17 popular building design parameters

Direct Search

Bouchlaghem and Letherman in 1990 [17]. Hasan et al. in 2008 [37].

Indoor thermal comfort

Thermal properties, thickness, glazing, shading, etc. Thickness of insulation materials

Simplex method

Futrell et al. in 2014 [38].

Energy consumption and visual comfort

The window’s size, location and optical properties

Stazi et al. in 2012 [39].

Life cycle cost of the solar wall system Minimum thermal load

Material, thickness of the wall, type of the window.

Factorial plan technique

Thickness of insulation materials, window type, area, orientation and shading Thickness of insulation materials, window type and area Window type, wall type and window area in wall and roof Building’s orientation and material types

Orthogonal method

Gong et al. in 2012 [40].

Ruiza et al. in 2014 [41]. Coley and Schukat in 2002 [44]. Wang et al. in 2005 [45,46].

Life cycle cost

Life cycle cost

Energy consumption and life cycle cost Annual energy consumption Life cycle exergy

Thickness of insulation materials, ventilation rate, thermal mass, window type, area, orientation and shading. The south exposure coefficient

Direct Search

A functional relationship between heat loss and the thickness of insulation materials was achieved.

Direct Search

It is possible to cost effectively reduce the space heating and cooling energy requirement of a new house in Sydney by up to 94% compared with the legislated BASIX requirements.

Direct Search

Both the shape coefficient and the south exposure coefficient were almost linear in relation to the heating energy consumption. A function for connecting the building internal envelope’s specific heat and the indoor space temperature was built. Regression analysis based on the gradient-deterministic search methodology was still an acceptable approach in building design for the optimization of energy conservation.

Hybrid GPS Hooke–Jeeves/PSO algorithm Hybrid GPS Hooke–Jeeves/PSO algorithm

For the south orientation, thermal and the visual performance did not significantly conflict, but for the north orientation the values had the greatest conflict. Their method was fast, simple and intuitive. However, it could only be applied on simple systems, and the results were not entirely accurate. An orthogonal method did not require a deep knowledge of computer programming, and thus was suitable for architects

Tabu search

Genetic algorithm

There existed a large number of optimal solutions.

Genetic algorithm

The Pareto front is important to help the designers in decision making during multi-objective problems. They developed an object-oriented framework, so that the GA method could be easily adopted with numerical simulation within a much more user-friendly interface. Though the result from the GA showed a stochastic behavior, the reliability was satisfactory. Although there were some differences in energy consumption among different building shapes, these deviations were within 0.5%. The climate had an effect on the accuracy of the optimization. If the space load situation was complex, with a widely variable combination of heating, cooling and humidity conditions, the degree of error would be larger. Under heating conditions, the window thermal properties were the most critical parameters among all of the variables. GA was especially effective for optimization of daylighting performance. Hybrid fitness may be helpful for the accuracy of the optimization.

Genetic algorithm

Wang et al. in 2005 and 2006[47,48].

Existing design guideline should be revised with lower U-value suggestion.

Wright and Mourshed in 2009 [49]. Tuhus-Dubrow and Krarti in 2010 [50].

Energy consumption Energy consumption

Window-to-wall ratio

Genetic algorithm

Envelope shape

Genetic algorithm

Sahu et al. in 2012 [51].

Energy consumption

The shape, orientation and materials of the building envelope

Genetic algorithm

Ioannou and Itard in 2015 [52].

Energy consumption

Genetic algorithm

Torres and Sakamoto in 2007 [53].

Modified daylight glare probability index

U value for window, wall, roof and floor, ventilation, occupant, thermostat Size, position, transmittance, reflectivity of window and shading systems

Genetic algorithm


G Model

ARTICLE IN PRESS


Y. Huang, J.-l. Niu / Energy and Buildings xxx (2015) xxx–xxx

9

Table 2 (Continued) Author and year

Optimization objective

Optimized parameters

Tools and algorithms

Key results

Kämpf et al. in 2010 [54].

Energy absorbed by corresponding surface Illumination level and glare index

Building shape with consistent building volume

Genetic algorithm

Urban scale optimization by using GA was possible.

Size, location, transmittance of the window, window-to-wall ratio Geometric forms of ceiling areas Distribution of buildings.

Genetic algorithm

The micro-GA-based optimization required much more computing time and a relatively large population size.

Genetic algorithm

Their proposed method was more precise.

Gagne and Andersen in 2011 [55].

Rakha and Nassar in 2011 [56]. Yi and Kim in 2015 [57]. Hamdy et al. in 2011 [58]. Stavrakakis et al. in 2012 [59].

Yu et al. in 2015 [60].

Znouda et al. in 2007 [62].

Illumination level Sunlight exposure time Energy consumption and thermal comfort. Thermal comfort

Energy consumption and thermal comfort. Energy consumption Life cycle cost.

Size and U-value of window, size of shading system Position and size of window-opening

Area, orientation, heat transfer coefficient of wall and window Building shape, U-value of the wall, shading coefficient of window Insulation thickness, U-value of window Insulation thickness, U-value of window, window-to-wall ratio, ventilation, size of shading systems Insulation thickness, orientation, U-value of window, window-to-wall ratio Insulation thickness, orientation, U-value of window, window-to-wall ratio U-value of wall, U-value and transmittance of window, U-value of shading system Wall materials and glazing type

Palonen et al. in 2009 [63]. Kämpf and Robinson in 2009 [64], Ramallo-González and Coley in 2014 [65]. Eisenhower et al. in 2012 [66], Geyer and Schlüter in 2014 [67].

Energy consumption and thermal comfort.

Bucking et al. in 2013 [68].

Energy consumption

Junghans and Darde in 2015 [69].

Life cycle cost.

Fesanghary et al. in 2012 [72].

Life cycle cost and CO2 emission.

Magnier and Haghighat in 2010 [73].

Energy consumption and thermal comfort.

Window-to-wall ratio, wall thickness

Zemella et al. in 2011 [74].

Energy consumption and annual carbon emission. Energy consumption and thermal comfort.

Popular building envelope design parameters

Gossard et al. in 2013 [75].

Energy consumption

Thermophysical properties of wall and roof

Caldas and Norford in 2002 [77].

Energy consumption

Shape and size of window

Manzan and Pinto in 2009 [78].

Energy consumption and visual comfort Energy consumption and life cycle cost. Energy consumption and life cycle cost.

Positions and sizes of the external shading equipment Insulation thickness, type of glazing and shading systems. Popular building envelope design parameters

Hamdy et al. in 2010 [79], Ferrara et al. in 2014 [80]. Karmellos et al. in 2015 [81].

Genetic algorithm Genetic algorithm

Genetic algorithm based on meta-model construction Genetic algorithm

An optimal solution for both thermal comfort and energy saving would require 10 kWh/(m2a) more than an optimal solution for energy saving alone

See Table 1

Under the same level of energy consumption, the indoor thermal comfort status did not change significantly. See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

See Table 1

Harmony search algorithm

Although the method required a relatively long computing time, its results were stable and trustworthy. The optimization result could be achieved more quickly and reliably.

Genetic algorithm coupled with artificial neural network Evolutionary artificial neural networks Genetic algorithm coupled with artificial neural network Genetic algorithm

Their proposed approach was more accurate and less time-consuming.

In many cases, similar minimum annual energy consumption could be achieved with several different configurations.

Genetic algorithm

Genetic algorithm

A series of reference data for residential house designs in Finland and Italy was proposed. A multi-objective optimization should consist of two parts: an optimization process and a decision making process. A decision making software tool was developed.




4.4. ParaGen Turrin et al. described a GA-based optimization tool named ParaGen for the application of both single- and multi-objective optimizations in the area of architecture design. They presented several projects involving ParaGen to show the practicality of this tool [91].

Based on the above conclusions, some potential directions for future work are proposed.

Caldas made a detailed introduction on the capacity of GENE ARCH. With the assistance of the GA, GENE ARCH could cooperate with the building energy simulation software DOE-2 to solve optimization problems in architectural design for projects ranging from an individual house to an urban scale. However, Caldas pointed out that when it came to complex 3D design problems, GENE ARCH’s performance was still not satisfactory, because it required very precise information on the building model [93].

(1) The comprehensive design of building envelopes requires the assessment of energy performance, thermal comfort performance and visual comfort performance. To achieve the goal of sustainability in buildings, it is of great importance to conduct multi-objective building envelope optimizations that involve all three indices. As there are many user-friendly multiobjective optimization tools, efforts can be made to improve multi-objective optimization toward the sustainable design of building envelopes, with consideration for both comfort and energy efficiency. (2) Insufficient attention is paid to the effect of building envelopes on indoor thermal and visual comfort. The descriptions and evaluation systems for estimating occupant comfort still have some limitations. In the future, professionals can improve the design of optimal building envelopes by applying more accurate and practical functions related to comfort. (3) The importance of occupant behavior has not been sufficiently accounted for. The effects of occupant behavior on a building’s energy performance and indoor environment are often ignored. In the future, a more precise and complete occupant behavior model should be added to simulations of building performance. Occupant behavior should not be considered as simply a matter of fixed schedules, but as a factor involving human initiative that influences the simulation process.

4.7. MOBO

Acknowledgement

In the year 2013, Palonen et al. reported another very powerful optimization tool MOBO. This software tool was developed under a funding project of the Academy of Finland. Via a well-designed graphical user interface, MOBO was able to couple with many building simulation programs such as IDA-ICE and TRNSYS, etc. With an extendable library, MOBO was able to handle single- and multiobjective optimization problems with popular algorithms such as NSGA-II, Hooke–Jeeves, Hybrid Algorithm as well as Random Search [94].

The work described in this paper was financially supported by the Research Grants Council of Hong Kong (Project no. 522613).

4.5. MultiOpt Chantrelle et al. developed a multi-criteria optimization tool named MultiOpt. The software was built in a TRNSYS-based simulation environment, with GA referred as the search engine. These researchers presented several case studies to prove that MultiOpt was able to manage the optimization of energy consumption, life cycle cost and thermal comfort [92]. 4.6. GENE ARCH

5. Conclusion and future work This paper has focused on the simulation-based optimization of building envelope design. Studies on this subject from the past 30 years were collected, summarized and discussed. The history and development of various methods was explained. Popular algorithms were presented and compared. Statistics concerning the achievement of objectives were compiled and analyzed. Some interesting conclusions were reached, as follows: (1) Genetic algorithms and their modifications are the most popular optimization algorithms used among the surveyed building envelope optimization studies. The applicability, speed and accuracy of these algorithms were considered in detail. (2) For relatively simple optimization problems, professionals tend to use gradient-deterministic search systems, but for complex problems, GAs have been their first choice. (3) In the area of simulation-based building envelope design, single-objective optimization is still dominant. (4) Energy consumption remains the top concern for researchers and designers. (5) Although there have been many software tools developed specially for building optimization design, Matlab is still the most popular tool for optimal solution searches.

References [1] L. Perez-Lombard, J. Ortiz, C. Pout, A review on buildings energy consumption information, Energy Build. 40 (2008) 394–398. [2] Y. Jiang, Current building energy consumption in China and effective energy efficiency measures, J. HV AC 35 (2005) 30–40. [3] Y.P. Zhou, J.Y. Wu, R.Z. Wang, S. Shiochi, Y.M. Li, Simulation and experimental validation of the variable-refrigerant-volume (VRV) air-conditioning system in EnergyPlus, Energy Build. 40 (2008) 1041–1047. [4] P.C. Tabares-Velasco, C. Christensen, M. Bianchi, Verification and validation of EnergyPlus phase change material model for opaque wall assemblies, Build. Environ. 54 (2012) 186–196. [5] R.H. Henninger, M.J. Witte, D.B. Crawley, Analytical and comparative testing of EnergyPlus using IEA HVAC BESTEST E100-E200 test suite, Energy Build. 36 (2004) 855–863. [6] G. Yun, K.S. Kim, An empirical validation of lighting energy consumption using the integrated simulation method, Energy Build. 57 (2013) 144–154. [7] J. Alstan Jakubiec, C.F. Reinhart, A method for predicting city-wide electricity gains from photovoltaic panels based on LiDAR and GIS data combined with hourly Daysim simulations, Solar Energy 93 (2013) 127–143. [8] X. Zhou, T. Hong, D. Yan, Comparison of HVAC system modeling in EnergyPlus, DeST and DOE-2.1E, Build. Simul. 7 (2014) 21–33. ´ Optimization of passive solar design strategies: a review, [9] S. Stevanovic, Renew. Sustain. Energy Rev. 25 (2013) 177–196. [10] R. Evins, A review of computational optimization methods applied to sustainable building design, Renew. Sustain. Energy Rev. 22 (2013) 230–245. [11] A.-T. Nguyen, S. Reiter, P. Rigo, A review on simulation-based optimization methods applied to building performance analysis, Appl. Energy 113 (2014) 1043–1058. [12] V. Machairas, A. Tsangrassoulis, K. Axarli, Algorithms for optimization of building design: a review, Renew. Sustain. Energy Rev. 31 (2014) 101–112. [13] K. Negendahl, Building performance simulation in the early design stage: an introduction to integrated dynamic models, Autom. Constr. 54 (2015) 39–53. [14] S. Carlucci, F. Causone, F. De Rosa, L. Pagliano, A review of indices for assessing visual comfort with a view to their use in optimization processes to support building integrated design, Renew. Sustain. Energy Rev. 47 (2015) 1016–1033. [15] S. Attia, M. Hamdy, W. O’Brien, S. Carlucci, Assessing gaps and needs for integrating building performance optimization tools in net zero energy buildings design, Energy Build. 60 (2013) 110–124. [16] J.S. Gero, N. D’Cruz, A.D. Radford, Energy in context: a multicriteria model for building design, Build. Environ. 18 (1983) 99–107.




[17] N.M. Bouchlaghem, K.M. Letherman, Numerical optimization applied to the thermal design of buildings, Build. Environ. 25 (1990) 117–124. [18] R. Sullivan, E.S. Lee, S. Selkowitz, A method of optimizing solar control and daylighting performance in commercial office buildings, in: Proceedings of the ASHRAE/DOE/BTECC Conference, Clearwater Beach, FL, U.S., 1992. [19] S. Mohammad, Al-Homoud, Optimum thermal design of office buildings, Int. J. Energy Res. 21 (1997) 941–957. [20] P. Depecker, C. Menezo, J. Virgone, S. Lepers, Design of buildings shape and energetic consumption, Build. Environ. 36 (2001) 627–635. [21] E. Ghisi, J. Tinker, Optimization energy consumption in offices as a function of window area and room size, in: Proceedings of the building simulation, Rio de Janeiro, Brazil, 2001. [22] R. Hooke, T.A. Jeeves, “Direct Search” solution of numerical and statistical problems, J. ACM 8 (1961) 212–229. [23] A. Marsh, Computer-optimised shading design, in: Proceedings of the building simulation, Eindhoven, Netherlands, 2003. [24] M.S. Al-Homoud, A systematic approach for the thermal design optimization of building envelopes, J. Build. Phys. 29 (2005) 95–119. [25] M.S. Al-Homoud, Envelope thermal design optimization of buildings with intermittent occupancy, J. Build. Phys. 33 (2009) 65–82. [26] L. Wang, N.H. Wong, S. Li, Facade design optimization for naturally ventilated residential buildings in Singapore, Energy Build. 39 (2007) 954–961. [27] M. Adamski, Optimization of the form of a building on an oval base, Build. Environ. 42 (2007) 1632–1643. [28] A. Ucar, F. Balo, Effect of fuel type on the optimum thickness of selected insulation materials for the four different climatic regions of Turkey, Appl. Energy 86 (2009) 730–736. [29] A. Ucar, F. Balo, Determination of the energy savings and the optimum insulation thickness in the four different insulated exterior walls, Renew. Energy 35 (2010) 88–94. [30] Lollini, Barozzi, Fasano, Meroni, Zinzi, Optimisation of opaque components of the building envelope: energy, economic and environmental issues, Build. Environ. 41 (2006) 1001–1013. [31] S.M. Bambrook, A.B. Sproul, D. Jacob, Design optimisation for a low energy home in Sydney, Energy Build. 43 (2011) 1702–1711. [32] R. Albatici, F. Passerini, Bioclimatic design of buildings considering heating requirements in Italian climatic conditions. A simplified approach, Build. Environ. 46 (2011) 1624–1631. [33] F. Jiang, X. Wang, Y. Zhang, Analytical optimization of specific heat of building internal envelope, Energy Convers. Manage. 63 (2012) 239–244. [34] R. Cheng, X. Wang, Y. Zhang, Analytical optimization of the transient thermal performance of building walls by using thermal impedance based on thermal-electric analogy, Energy Build. 80 (2014) 598–612. [35] S. Asadi, S.S. Amiri, M. Mottahedi, On the development of multi-linear regression analysis to assess energy consumption in the early stages of building design, Energy Build. 85 (2014) 246–255. [36] F. Glover, Future paths for integer programming and links to artificial intelligence, Comput. Oper. Res. 13 (1986) 533–549. [37] A. Hasan, M. Vuolle, K. Siren, Minimisation of life cycle cost of a detached house using combined simulation and optimization, Build. Environ. 43 (2008) 2022–2034. [38] B.J. Futrell, E.C. Ozelkan, D. Brentrup, Bi-objective optimization of building enclosure design for thermal and lighting performance, Build. Environ. 92 (2015) 591–602. [39] F. Stazi, A. Mastrucci, P. Munafò, Life cycle assessment approach for the optimization of sustainable building envelopes: an application on solar wall systems, Build. Environ. 58 (2012) 278–288. [40] X. Gong, Y. Akashi, D. Sumiyoshi, Optimization of passive design measures for residential buildings in different Chinese areas, Build. Environ. 58 (2012) 46–57. [41] P.A. Ruiza, J.G. Martín, J.M. Salmerón Lissén, F.J.S. de la Flor, An integrated optimisation method for residential building design: a case study in Spain, Energy Build. 80 (2014) 158–168. [42] T. Back, Evolutionary Algorithms in Theory and Practice: Evolution Strategies, Evolutionary Programming, Genetic Algorithms, Oxford University Press, Oxford, UK, 1996. [43] V. Congradac, F. Kulic, HVAC system optimisation with CO2 concentration control using genetic algorithms, Energy Build. 41 (2009) 571–577. [44] D.A. Coley, S. Schukat, Low-energy design: combining computer-based optimisation and human judgement, Build. Environ. 37 (2002) 1241–1247. [45] W. Wang, H. Rivard, R.G. Zmeureanu, Optimizing building design with respect to life-cycle environmental impacts, in: Proceedings of the building simulation, Eindhoven, Netherlands, 2003. [46] W. Wang, R. Zmeureanu, H. Rivard, Applying multi-objective genetic algorithms in green building design optimization, Build. Environ. 40 (2005) 1512–1525. [47] W. Wang, H. Rivard, R. Zmeureanu, An object-oriented framework for simulation-based green building design optimization with genetic algorithms, Adv. Eng. Inform. 19 (2005) 5–23. [48] W. Wang, H. Rivard, R. Zmeureanu, Floor shape optimization for green building design, Adv. Eng. Inform. 20 (2006) 367–378. [49] J. Wright, M. Mourshed, Geometric optimization of fenestration, in: Proceedings of the building simulation, Glasgow, Scotland, 2009. [50] D. Tuhus-Dubrow, M. Krarti, Genetic-algorithm based approach to optimize building envelope design for residential buildings, Build. Environ. 45 (2010) 1574–1581.

11

[51] M. Sahu, B. Bhattacharjee, S.C. Kaushik, Thermal design of air-conditioned building for tropical climate using admittance method and genetic algorithm, Energy Build. 53 (2012) 1–6. [52] A. Ioannou, L.C.M. Itard, Energy performance and comfort in residential buildings: sensitivity for building parameters and occupancy, Energy Build. 92 (2015) 216–233. [53] S.L. Torres, Y. Sakamoto, Facade design optimization for daylight with a simple genetic algorithm, in: Proceedings of the building simulation, Beijing, China, 2007. [54] J.H. Kämpf, M. Montavon, J. Bunyesc, R. Bolliger, D. Robinson, Optimisation of buildings’ solar irradiation availability, Solar Energy 84 (2010) 596–603. [55] J. Gagne, M. Andersen, A generative facade design method based on daylighting performance goals, J. Build. Perform. Simul. 5 (2011) 141–154. [56] T. Rakha, K. Nassar, Genetic algorithms for ceiling form optimization in response to daylight levels, Renew. Energy 36 (2011) 2348–2356. [57] Y.K. Yi, H. Kim, Agent-based geometry optimization with genetic algorithm (GA) for tall apartment’s solar right, Solar Energy 113 (2015) 236–250. [58] M. Hamdy, A. Hasan, K. Siren, Impact of adaptive thermal comfort criteria on building energy use and cooling equipment size using a multi-objective optimization scheme, Energy Build. 43 (2011) 2055–2067. [59] G.M. Stavrakakis, P.L. Zervas, H. Sarimveis, N.C. Markatos, Optimization of window-openings design for thermal comfort in naturally ventilated buildings, Appl. Math. Model. 36 (2012) 193–211. [60] W. Yu, B. Li, H. Jia, M. Zhang, D. Wang, Application of multi-objective genetic algorithm to optimize energy efficiency and thermal comfort in building design, Energy Build. 88 (2015) 135–143. [61] K. Deb, A. Pratap, S. Agarwal, T. Meyarivan, A fast and elitist multiobjective genetic algorithm: NSGA-II, IEEE Trans. Evol. Comput. 6 (2002) 182–197. [62] E. Znouda, N. Ghrab-Morcos, A. Hadj-Alouane, Optimization of Mediterranean building design using genetic algorithms, Energy Build. 39 (2007) 148–153. [63] M. Palonen, A. Hasan, K. Siren, A genetic algorithm for optimization of building envelope and Hvac system parameters, in: Proceedings of the building simulation, Glasgow, Scotland, 2009. [64] J.H. Kämpf, D. Robinson, Optimisation of urban energy demand using an evolutionary, algorithm, in: Proceedings of the building simulation, Glasgow, Scotland, 2009. [65] A.P. Ramallo-González, D.A. Coley, Using self-adaptive optimisation methods to perform sequential optimisation for low-energy building design, Energy Build. 81 (2014) 18–29. [66] B. Eisenhower, Z. O’Neill, S. Narayanan, V.A. Fonoberov, I. Mezic, A methodology for meta-model based optimization in building energy models, Energy Build. 47 (2012) 292–301. [67] P. Geyer, A. Schlüter, Automated metamodel generation for design space exploration and decision-making – a novel method supporting performance-oriented building design and retrofitting, Appl. Energy 119 (2014) 537–556. [68] S. Bucking, R. Zmeureanu, A. Athienitis, An information driven hybrid evolutionary algorithm for optimal design of a net zero energy house, Solar Energy 96 (2013) 128–139. [69] L. Junghans, N. Darde, Hybrid single objective genetic algorithm coupled with the simulated annealing optimization method for building optimization, Energy Build. 86 (2015) 651–662. [70] K.S. Leea, Z.W. Geem, A new meta-heuristic algorithm for continuous engineering optimization: harmony search theory and practice, Comput. Methods Appl. Mech. Eng. 194 (2005) 3902–3933. [71] D. Weyland, A rigorous analysis of the harmony search algorithm: How the research community can be misled by a “novel” methodology, Int. J. Appl. Metaheuristic Comput. 1 (2010) 50–60. [72] M. Fesanghary, S. Asadi, Z.W. Geem, Design of low-emission and energy-efficient residential buildings using a multi-objective optimization algorithm, Build. Environ. 49 (2012) 245–250. [73] L. Magnier, F. Haghighat, Multiobjective optimization of building design using TRNSYS simulations, genetic algorithm, and artificial neural network, Build. Environ. 45 (2010) 739–746. [74] G. Zemella, D. De March, M. Borrotti, I. Poli, Optimised design of energy efficient building facades via evolutionary neural networks, Energy Build. 43 (2011) 3297–3302. [75] D. Gossard, B. Lartigue, F. Thellier, Multi-objective optimization of a building envelope for thermal performance using genetic algorithms and artificial neural network, Energy Build. 67 (2013) 253–260. [76] Y. Huang, J.-l. Niu, T.-m. Chung, Energy and carbon emission payback analysis for energy-efficient retrofitting in buildings—Overhang shading option, Energy Build. 44 (2012) 94–103. [77] L.G. Caldas, L.K. Norford, A design optimization tool based on a genetic algorithm, Autom. Constr. 11 (2002) 173–184. [78] M. Manzan, F. Pinto, Genetic optimization of external shading devices, in: Proceedings of the building simulation, Glasgow, Scotland, 2009. [79] M. Hamdy, A. Hasan, K. Siren, A multi-stage optimization method for cost-optimal and nearly-zero-energy building solutions in line with the EPBD-recast 2010, Energy Build. 56 (2013) 189–203. [80] M. Ferrara, E. Fabrizio, J. Virgone, M. Filippi, A simulation-based optimization method for cost-optimal analysis of nearly zero energy buildings, Energy Build. 84 (2014) 442–457. [81] M. Karmellos, A. Kiprakis, G. Mavrotas, A multi-objective approach for optimal prioritization of energy efficiency measures in buildings: model, software and case studies, Appl. Energy 139 (2015) 131–150.




[82] Y. Lu, S. Wang, Y. Zhao, C. Yan, Renewable energy system optimization of low/zero energy buildings using single-objective and multi-objective optimization methods, Energy Build. 89 (2015) 61–75. [83] R. McKinstray, J.B.P. Lim, T.T. Tanyimboh, D.T. Phan, W. Sha, A.E.I. Brownlee, Topographical optimisation of single-storey non-domestic steel framed buildings using photovoltaic panels for net-zero carbon impact, Build. Environ. 86 (2015) 120–131. [84] K. Shea, A. Sedgwick, G. Antonuntto, Multicriteria optimization of paneled building envelopes using ant colony optimization, Intell. Comput. Eng. Architect. (2006) 627–636. [85] M. Wetter, GenOpt – a generic optimization program, in: Proceedings of the building simulation, Rio de Janeiro, Brazil, 2001. [86] F. Flager, B. Welle, P. Bansal, G. Soremekun, J. Haymaker, Multidisciplinary process integration and design optimization of a classroom building, J. Inf. Technol. Constr. 14 (2009) 595–612. [87] J.N. Holst, Using whole building simulation models and optimizing procedures to optimize building envelope design with respect to energy consumption and indoor environment, in: Proceedings of the building simulation, Eindhoven, Netherlands, 2003.

[88] A. Hasan, M. Vuolle, K. Siren, Minimisation of life cycle cost of a detached house using combined simulation and optimisation, Build. Environ. 43 (2008) 2022–2034. [89] C. Baglivo, P.M. Congedo, D. D’Agostino, I. Zac, Cost-optimal analysis and technical comparison between standard and high efficient mono-residential buildings in a warm climate, Energy 83 (2015) 560–575. [90] X. Shi, Design optimization of insulation usage and space conditioning load using energy simulation and genetic algorithm, Energy 36 (2011) 1659–1667. [91] M. Turrin, P. von Buelow, R. Stouffs, Design explorations of performance driven geometry in architectural design using parametric modeling and genetic algorithms, Adv. Eng. Inform. 25 (2011) 656–675. [92] F.P. Chantrelle, H. Lahmidi, W. Keilholz, M. El Mankibi, P. Michel, Development of a multicriteria tool for optimizing the renovation of buildings, Appl. Energy 88 (2011) 1386–1394. [93] L. Caldas, Generation of energy-efficient architecture solutions applying GENE ARCH: an evolution-based generative design system, Adv. Eng. Inform. 22 (2008) 59–70. [94] M. Palpnen, M. Hamdy, A. Hasan, MOBO a new software for multi-objective building performance optimization, in: Proceedings of the building simulation, Chambéry, France, 2013.


Optimal building envelope design based on simulated ...

Optimal building envelope design based on simulated ...

Suggest Documents

Optimal design of residential building envelope ...

Optimal design of residential building envelope ...

Building Classification Based On Simulated Annual Results

Building Envelope Design Strategies for Single ...

Optimal Design Criteria Based on Tolerance Regions

Building Envelope Survey

Building Envelope - Karpinski Engineering

Green Building Envelope

PERFORMANCE-BASED ENVELOPE DESIGN FOR RESIDENTIAL ...

Multi-Criterion Optimal Design of Building ... - onlinepresent.org

Probabilistic Performance-Based Optimal Design

The impact of sustainable building envelope design ...

Design of the Building Envelope: A Novel Multi

2012204 10 GN CSD HS Building Envelope- Design Development ...

Hygrothermal Simulation: A Tool for Building Envelope Design Analysis

Transformable building envelope design in architectural ...www.researchgate.net › publication › fulltext › Transform

Optimization of thermal comfort in building through envelope design

Hygrothermal Simulation: A Tool for Building Envelope Design Analysis

Variability of optimal solutions for building components based on ...

Generative and evolutionary techniques for building envelope design

Building envelope design as a passive cooling technique

Optimal Design on Robustness of Scale-Free Networks Based on

Uniform design simulated annealing for optimal access point ...

Building Envelope - CSIR Research Space