Elsevier

Energy

Volume 36, Issue 3, March 2011, Pages 1659-1667
Energy

Design optimization of insulation usage and space conditioning load using energy simulation and genetic algorithm

https://doi.org/10.1016/j.energy.2010.12.064Get rights and content

Abstract

Architectural design is a process to find the best solution to satisfy various design criteria. To achieve sustainable and green design, performance simulations are often used to verify these criteria and modify the design. The conventional approach of manual trial-and-error is too time-consuming to be practical. Introducing optimization technique can greatly improve the design efficiency and help designers find the optimal design. In this paper, modeFRONTIER was used as the design optimization environment to find the best insulation strategy to minimize the space conditioning load of an office building located in Nanjing, China while keeping the insulation usage at minimum. EnergyPlus was integrated into the optimization tool by writing a DOS batch file to automate the work flow. The search engine was the genetic algorithm and it proved to be able to generate a well-defined Pareto frontier in a reasonable number of runs. Based on the Pareto frontier, the designer can specify his preferences and select the final design. The case study shows that an energy simulation program can be effectively integrated into a design optimization environment to find the optimal design. The technique presented has a broad application in architectural design, especially when the design considerations are multi-objective.

Research highlights

► Used modeFRONTIER as a design optimization environment to find the best insulation strategy to minimize the space conditioning load of an office building while keeping the insulation usage at minimum. ► Developed a technique to integrate EnergyPlus into the optimization tool through writing DOS batch files. ► Applied Multi-Objective Genetic Algorithm to obtain a well-defined Pareto frontier in a reasonable number of runs to find the optimal insulation strategy. ► Established a technique that can be applied to other building energy-related optimization problems, especially when EnergyPlus is used as the energy simulation tool.

Introduction

Architectural design, in essence, is a process to search for an optimal solution that satisfies a variety of objectives. These objectives include, but not limited to, (1) creating a space or a combination of spaces that perform the intended functions of the building, (2) meeting all applicable codes and standards, (3) producing an aesthetically pleasing facility, (4) minimizing the consumption of energy, water, materials, and other resources, (5) minimizing pollution and other detrimental effects to the environment, (6) maximizing the performance of buildings such as comfortable thermal environment, adequate natural lighting, plenty of ventilation, etc. Each of these categories of objectives includes many criteria that may or may not be quantifiable. In this sense, architectural design is multi-objective and can be quite complex and challenging.

During the past several decades, the sustainable and green design has been embraced by many stake holders in the AEC (Architectural, Engineering and Civil) industry. Architects, engineers, developers, builders, and government agencies are all interested in designing and building green. Green building standards have been issued and implemented in many countries. For instance, MOC (Ministry of Construction, now Ministry of Housing and Urban-Rural Development) of China issued its first green building standard in 2006 [1]. USGBC (United States Green Building Council) has been working on the well-known LEED (Leadership in Energy and Environmental Design) system for well over a decade [2]. Applying these green building standards requires that architects need to consider more design objectives in addition to those regulated by the mandatory codes and standards. Many of these objectives are performance based and therefore, a performance simulation is needed to verify whether or not the objectives are met.

Chinese AEC firms started using building performance simulation programs around 2000. Nowadays, widely used simulation programs include energy simulation programs, CFD software tools, solar and lighting analysis programs, etc. The application of these programs has significantly improved the quality of design and fundamentally changed the conventional design process. MOC of China issued a document in 2003 to divide the architectural design process into three steps, namely, conceptual design, detailed design, and shop drawing design, i.e. design of construction drawings, generally done by the contractor and approved by the designer (Fig. 1) [3]. Adding performance simulation changes this design process as shown in Fig. 2. Note that the performance simulation can be used in all three stages to improve the quality of design and that the arrows added are two-way, indicating that the performance simulation and each design step should communicate back and forth. The idea is to use the results of performance simulation to assist the architect in modifying and optimizing the design. For instance, Palonen et al. studied how to minimize the life-cycle cost of a single-family detached house by combining simulation and optimization [4]. Manzan et al. used the energy code ESP-r and lighting program RADIANCE to optimize the design of shading devices for an office building [5].

In practice, the process shown in Fig. 2 can rarely be completed in one iteration. Multiple iterations are needed to adequately explore the design space and find the optimal design. However, even in leading AEC firms, this process is not iterated adequately. For example, a survey performed at Arup found that the average number of iterations is 2.7, significantly less than what is needed to find the optimal design [6]. The number of iterations to obtain the optimal design is dependent on the specific project. When the building is complex and the optimization is multi-objective, the number of iterations needed will be large. Several reasons explain the lack of iterations in practice.

  • The current application of performance simulation is more of a “computer-aided-check” than a “computer-aided-design”. The architect still needs to generate a conceptual design first using the conventional design approach in order to send the model into the performance simulation program for analysis. Therefore, the real function that the simulation serves is to check the design rather than generate a new design. The architect still has to rely on the inefficient “trial-and-error” method to modify the design. This gives the architect an impression that the performance simulation is a bonus, not a necessity.

  • Conducting a performance simulation is time-consuming and labor-intensive.

  • Many architectural design projects have a tight schedule, which makes the iterative process not feasible.

  • The architect is not familiar with the performance simulation tools and therefore, needs somebody else to conduct the simulation and interpret the results for him. The extra communication slows down the design process and is not always smooth.

Among all of the reasons above, the first one is probably the most indicative one. It raises an important question, i.e., how can we develop a technique that is able to combine the performance simulation with the design generation so that it can automate the work flow and avoid the inefficient “trial-and-error” method? The ultimate goal is to provide architects with an effective technique to adequately explore the design space and find the optimal design. In May 2008, the School of Architecture at SEU (Southeast University) held a consulting meeting. Six professionals (a chief architect of a 300-people AEC firm, a chief mechanical engineer of a 200-people AEC firm, a university professor specializing in renewable energy, a senior engineer of an architectural research institute, a CEO of a developer company, and a government official from the Bureau of Construction, Jiangsu Province) were invited to brainstorm solutions for the previously mentioned problem. All of them agreed that the first reason is the greatest obstacle to incorporating the performance simulation into the daily architectural design practice.

Based on the above analysis, it is clear that architectural design firms need an effective technique to take full advantage of the performance simulation programs. Compared with the conventional function of checking the design, this technique should be able to use the performance simulation tools to generate new designs and then automatically and systematically search for the optimal one. This paper proposes a technique of using modeFRONTIER, an optimization software tool, to integrate the building performance simulation program (EnergyPlus in this paper) and achieve these goals. The case study is a one-story office building located in Nanjing, China. The design objectives are reducing the insulation usage and minimizing the space conditioning load at the same time. Genetic algorithm was used as the engine to drive the optimization. In summary, this paper introduces a new optimization software tool into architectural design and presents a technique of integrating EnergyPlus into the optimization tool by writing DOS batch files. Furthermore, the paper presents, through the case study, how to obtain the true Pareto optimal design and apply multi-criteria decision-making tools to determine the final design. In the end, a general procedure of performing optimization in architectural design and the key elements involved are proposed.

The studied building has three functional and thermal zones: Zone 1 is a conference/training room, and Zones 2 and 3 are two office spaces. Fig. 3 shows a 3D model of the building. The roof of Zone 2 is taken out to show the inside of the zone. The building has six walls, one facing north, one facing east, two facing west, and two facing south. Based on the thermal design code of China [7], Nanjing is in the so-called Hot-Summer–Cold-Winter zone and therefore, insulation is required. The typical practice is to determine the required overall conductance of the wall in accordance with the code and then select the type and thickness of insulation to achieve, combined with other components of the wall, the required overall conductance. Generally the insulation is uniformly distributed on all walls, i.e., the insulation on each wall has the same thickness. However, since the walls facing the four orientations are under different solar radiation, the effectiveness of applying insulation on each wall is not the same. In other words, given a total amount of insulation, there exists an optimal distribution that is able to achieve the best result of keeping the space conditioning load of the building at minimum. Hence, the design problem is how to find the optimal distribution of the thermal insulation. From another perspective, the problem can be equally stated as how to find the best balance between the insulation usage and the space conditioning load. In essence, the architect is trying to achieve two objectives simultaneously: (1) keeping the usage of thermal insulation as low as possible, and (2) keeping the space conditioning load at minimum.

Section snippets

Energy simulation

It is obvious that an energy simulation needs to be performed to solve the design optimization problem described above. EnergyPlus was used to simulate the space conditioning load of the building. EnergyPlus is an energy analysis and thermal load simulation program [8]. It has been widely used, like its predecessor programs, BLAST and DOE-2, by architects, engineers, and researchers. More information about EnergyPlus can be found in Refs. [9], [10] and its typical applications on mechanical

Results

Three cases were run to produce data for analysis, namely 5 × 5 (5 designs of experiment and 5 genetic generations), 10 × 5 (10 designs of experiment and 5 genetic generations), and 10 × 10 (10 designs of experiment and 10 genetic generations). In each case, the numbers of insulation strategies and the energy simulations runs were 25, 50, and 100, respectively. Note that there were some invalid runs in each case and they will be discussed later.

Table 2 summarizes the results of the insulation

Pareto frontier

The data shown in Table 2 gives an overview of what the optimization work flow produces. However, the real objective was not finding the minimum space conditioning load or the minimum total thickness of insulation alone. Rather, it was to find the most reasonable balance between these two objectives, i.e., the optimal design solution. This is where the “Pareto frontier” can help the designer make the decision. A rigorous mathematical definition of the Pareto frontier is beyond the scope of this

Conclusions and future work

Architectural design, in essence, is a process to find the best solution to balance a variety of design considerations. Therefore, the designer needs to make multi-objective decisions. It is challenging when these design objectives are complex and even contradictory. To achieve sustainability, many design objectives considered are related to physical performances of the building and thus, they need to be simulated using computer programs. The Chinese AEC industry started adopting these

Acknowledgement

This paper is supported by the National Natural Science Foundation of China (51008058) and the Ph.D. Programs Foundation of Ministry of Education of China (20100092120017).

Cited by (70)

  • Exploring data-driven building energy-efficient design of envelopes based on their quantified impacts

    2021, Journal of Building Engineering
    Citation Excerpt :

    Previously, parametric or even optimization techniques have been performed to investigate the impacts of building envelopes on energy consumption [19–21]. However, most of these experiments were conducted only on a single building [19]. In other words, these studies may not be instructive to other buildings.

View all citing articles on Scopus
View full text