Conceptual Design Algorithm Configuration Using Generative Design Techniques

1. Introduction

2. Trend of Overseas Research on Generative Design in the Construction Field

In this study, we have used the Web of Science as a search engine to investigate the overseas research trend, and analyzed and classified 35 studies related to generative design as shown in Table 1. below. Each study is analyzed based on the categories of creative design, detailed design, performance evaluation, and service & platform, and detailed analysis was performed[8].

Table 1.

Generative design research analysis

3. Research on Building Layout Design Algorithm Configuration

In this study, we have used Autodesk’s Revit and Dynamo programs to generate alternatives using the generative design technique in the planning and early design stage of buildings. Dynamo is a visual programming program and is usually used along with Revit. Unlike common programming languages, Dynamo has the characteristic of using visual elements so that users who are not familiar with programming languages and can write algorithms that computers can understand.

In this study, therefore, we have used this program to build a methodology for defining problems in the generative design technique, based on which various layout design alternatives were generated. Finally, we chose a representative design based on the evaluation criteria among numerous design alternatives.

3.1. Defining Attribute Classification and Methodology for Problem Definition

The design layout algorithm for the concept stage was implemented by dividing the task into four steps as shown in Step 2 in Fig. 1. The first step is the site selection. The most commonly used dimension is 10 × 10 m, which is a basic criterion. Second, 441 arbitrary transparent points (21 X 21) are placed in the form of grids in the site as shown in Fig. 2. Third, an algorithm was created to select four points randomly from the 441 arbitrary points. Finally, a square of 500–3,000 mm is created at each of the four points, and they are combined to create a layout alternative.

Fig. 1.

Research methodology

Fig. 2.

Methodology based on grid division

Furthermore, with respect to the designation of building lines in Article 46 of the Building Act, we have added an algorithm to connect two points retracted by a certain distance from the intersection point of the road boundary line, which is the border of the corner of the site. This is to apply the most basic part of the legal conditions, and the possibility of various conditions that will be added in the future were examined.

3.2. Generation of Layout Design Alternatives

Fig. 3. shows the output of random result values for concept design alternatives based on the design automation algorithm. The number of results can be hundreds to thousands, as configured in the algorithm. Some result values, among countless alternatives, show design characteristics similar to those generated when actual designers and architects perform conceptual designs on a typical site.

Fig. 3.

Basic results through random output generative design

Additionally, an algorithm was added to calculate the building coverage ratio. Based on this, alternatives with low and high building coverage ratios are arranged as shown in Fig. 4. The building coverage ratio refers to the area occupied by the land relative to the site area. By adding this value to the algorithm, a difference in design according to the building density can be identified. In other words, it can be seen that the alternatives with low building coverage ratios have relatively monotonous shapes. In contrast, the alternatives with high building coverage ratios have diverse shapes, such as “ㄱ,” “ㄴ,” and “ㄷ”.

Fig. 4.

Design output according to the difference in building coverage ratio (left: low, right: high)

Next, an algorithm was added to find the horizontal and vertical lengths of the buildings, and then the sum in each direction was computed to represent them as a ratio. This is used as a criterion for distinguishing buildings that are long in the north-south direction as opposed to building that are long in the east-west direction. Fig. 5. shows the results with high and low aspect ratios according to the algorithm. It can be seen that the long buildings in the east-west direction have more area facing the south, which has the most solar radiation, and those in the north-south direction are favorable for the formation of wind corridor.

Fig. 5.

Design output according to the difference in aspect ratio of the building (left: low, right: high)

Finally, the design results of the buildings were extracted according to the difference in the number of masses as shown in Fig. 6. below. A rectangle corresponding to each point was set as a single mass. When the areas were shared with each other, the number of masses were recognized as one. In this study, an alternative with many masses is represented in the form of four buildings without overlapping rectangles. As a result, the number of masses increases, and the direction of buildings is relatively insignificant in terms of environmental performance, and relatively diverse shapes are produced in the results.

Fig. 6.

Design results according to differences in the number of buildings (left: low, right: high)

3.3. Selecting Evaluation Criteria and Representative Designs

The generative design method can generate countless alternatives randomly, but it can also produce alternatives with optimized results of the desired items based on various evaluation criteria and control conditions. The six items shown in Fig. 7 below can be used as evaluation criteria: building area, building coverage ratio, the number of buildings, aspect ratio, and total perimeter length. Based on these criteria, alternatives with a low building coverage ratio (high utilization rate of the surrounding land), a relatively high aspect ratio, and large number of masses (diversity) were extracted as shown in Fig. 8.

Fig. 7.

Derivation of results based on the setting of evaluation criteria

Fig. 8.

Comprehensive design output (appropriate for building coverage ratio and high aspect ratio with a large number of south-facing locations)

4. Conclusion

In this study, we reviewed previous studies by classifying generative design techniques that are used and studied in a variety of ways, and then implemented a basic generative design technique that can be used in the conceptual design stage to generate various alternatives. The results presented in this study are summarized as follows.

(1) Previous studies on generative design can be classified into four categories: creative concept design, detailed design, performance evaluation, and service and platform.

(2) According to previous studies, AI-based design automation methods can revolutionize the paradigm of the existing construction and design fields and dramatically improve construction productivity. In particular, when 35 items from a previous study were compared, it was found that generative design techniques were the most advanced in the field of creative conceptual design.

(3) The implementation of generative design technique in the conceptual design stage was conducted based on three steps: (1) defining attribute classification and methodology for problem definition, (2) generation of layout design alternatives, and (3) selecting evaluation criteria and representative designs.

(4) In terms of methodology, we used a method of dividing the zone with grids and generating and combining arbitrary rectangles in the zone. Numerous alternatives were generated through the basic methodology, and three criteria (building coverage ratio, building aspect ratio, and number of masses) were chosen to evaluate the results.

(5) The minimum and maximum values were selected in detail for the three criteria, based on which layout designs with optimal alternatives were generated.

Using previous studies and generative design techniques, we constructed a relatively simple methodology and generated numerous design alternatives. Then, we extracted optimal alternatives from many alternatives based on the evaluation methodology.

In the future, numerous alternatives can be generated at the concept and detailed design stage by modifying the algorithms and applying deep learning-based machine learning methods. Furthermore, generative design techniques can be applied and used systematically through platforms using various evaluation methods.

The purpose of this study was to examine the applicability of the basic concept of generative design in the building layout process. The algorithm based on the procedure was implemented using a visual programming method. As a result, we have confirmed that the algorithm can perform various designs that can be created in the actual layout design process. Furthermore, by applying the evaluation methodology, we confirmed the feasibility of separately extracting alternatives according to the criteria of building coverage ratio, aspect ratio, and the number of masses. However, the limitation of this study is that relatively simple algorithms were used. For the ease of applying algorithms, we selected a simplified site of 10 × 10 m and considered rectangles of 500–6,000 mm, which were as small as architectural spaces. Some elements that are essential in the actual layout design process, such as the building function, path planning, and layout of functions inside the site, as well as external criteria of the site, were not considered. However, they can be included in follow-up studies. It is also necessary to analyze various laws, regulations, and design criteria, which should be considered in the construction process, and add them to the generative design algorithm. Furthermore, additional research is required to add functionalities to consider and set the relationships between various relational variables. One of the problems of this study is that it is configured to draw only parallel lines when creating rectangles on the site. Thus, shapes of diagonal lines cannot be produced for the site. It is expected that more optimized alternatives can be generated according to the designer’s intention by diversifying and refining the items of the evaluation criteria.

Acknowledgments

This study was supported by the Research Operation Cost Support (Main Project) Program of the Korea Institute of Civil Engineering and Building Technology, Ministry of Science and ICT (Project Number 20230117-001, Smart Construction Technology Research for Driving Future Construction Industry and Creating New Market).

References

• 강남우, 제너레이티브 디자인과 인공지능 기반 설계 자동화, 한국 :기계저널, 제59권 제8호, 2019.8, pp.24-28.
  N.W. Gang, Generative design and AI-based design automation, Korea: Journal of the KSME 59(8), 2019.8, pp.24-28.
• K.S. Lee, K.J. Han, J.W. Lee, Feasibility study on parametric optimization of daylighting in building shading design, Sustainability, 8(12), 2016.11. [https://doi.org/10.3390/su8121220]
• 추승연, 지능형 공간계획 및 계획설계 제안 기술, 한국 : 대한건축학회, 제65권 제9호, 2021.08, pp.33-36.
  S.Y Choo, Intelligent space planning and planning design proposal technology, Korea : The Architectural Institute of Korea, Vol. 65, No. 9, 2021.08, pp.33-36.
• 김근재 외 3인, 건축 기획업무 효율성 향상을 위한 공간 인접성 그래프 추출 알고리즘 개발, 한국 : 대한건축학회논문집, 제38권 제1호, 2022, pp.67-74.
  G.J. Kim et al., Development of spatial adjacency graph extraction algorithm for improvingpre-design efficiency in architectural design process, Korea : The Architectural Institute of Korea, Vol.38 No.1, 2022, pp.67-74.
• 국토교통부, 스마트 건설 활성화 방안 S-Construction 2030, 2022.
  Ministry of Land, Smart construction revitalization plan S-Construction 2030, 2022.
• Silver D. et al., Mastering the game of Go with deep neural networks and tree search, nature, 529(7587), 2016, pp.484-489. [https://doi.org/10.1038/nature16961]
• Castro Pena, M. Luz et al., Artificial intelligence applied to conceptual design. a review of its use in architecture, Automation in construction, 124, 2021, p.103550. [https://doi.org/10.1016/j.autcon.2021.103550]
• 경북대학교 산학협력단, 인공지능 기반 건축설계 자동화 기술개발, 국토교통부, 2020, pp.153-154.
  KyungPook National University, AI-based architectural design automation technology development, Ministry of Land, 2020, pp.153-154.
• J.T. Jin, J.W. Jeong, Optimization of a free-form building shape to minimize external thermal load using genetic algorithm, Energy and buildings, 85, 2014, pp.473-482. [https://doi.org/10.1016/j.enbuild.2014.09.080]
• Dincer A.E., Cagdas G., Tong H., A computational model for mass housing design as a decision-support tool, Procedia environmental sciences, 22, 2014, pp.270–279. [https://doi.org/10.1016/j.proenv.2014.11.026]
• Khalili Araghi, S., Stouffs R., Exploring cellular automata for high density residential building form generation, Automation in construction, 49(A), 2015, pp.152-162. [https://doi.org/10.1016/j.autcon.2014.10.007]
• Manrique J.D. et al., Automated generation of shop drawings in residential construction, Automation in construction, 55, 2015, pp.15-24. [https://doi.org/10.1016/j.autcon.2015.03.004]
• Hua Hao, Irregular architectural layout synthesis with graphical inputs, Automation in construction, 72(3), 2016, pp.388-396. [https://doi.org/10.1016/j.autcon.2016.09.009]
• Zhang L., Wang Y., Shape optimization of free-form buildings based on solar radiation gain and space efficiency using a multi-objective genetic algorithm in the severe cold zones of China, Elsevier Science B.V., Amsterdam, 132, 2016, pp.38-50. [https://doi.org/10.1016/j.solener.2016.02.053]
• Dino I.G., An evolutionary approach for 3D architectural space layout design exploration, Automation in construction, 69, 2016, pp.131-150. [https://doi.org/10.1016/j.autcon.2016.05.020]
• Song H., Ghaboussi J., T.H. Kwon, Architectural design of apartment buildings using the Implicit Redundant Representation Genetic Algorithm, Automation in construction, 72(2), 2016, pp.166-173. [https://doi.org/10.1016/j.autcon.2016.09.001]
• Herr C.M., Ford R.C., Cellular automata in architectural design: From generic systems to specific design tools, Automation in construction, 72(1), 2016, pp.39-45. [https://doi.org/10.1016/j.autcon.2016.07.005]
• Gerber David J., Pantazis Evangelos, Wang Alan, A multi-agent approach for performance based architecture: Design exploring geometry, user, and environmental agencies in façades, Automation in construction, 76, 2017, pp.45-58. [https://doi.org/10.1016/j.autcon.2017.01.001]
• Guo Z., Li, B., Evolutionary approach for spatial architecture layout design enhanced by an agent-based topology finding system, Frontiers of architectural research, 6(1), 2017, pp.53-62. [https://doi.org/10.1016/j.foar.2016.11.003]
• Sharafi Pezhman et al., Automated spatial design of multi-story modular buildings using a unified matrix method, Automation in construction, 82, 2017, pp.31-42. [https://doi.org/10.1016/j.autcon.2017.06.025]
• Zhang Jiansong., El-Gohary, Nora M., Integrating semantic NLP and logic reasoning into a unified system for fully-automated code checking, Automation in construction, 73, 2017, pp.45-57. [https://doi.org/10.1016/j.autcon.2016.08.027]
• Chatzikonstantinou, Ioannis., Sariyildiz, I. Sevil., Addressing design preferences via auto-associative connectionist models: Application in sustainable architectural Façade design, Automation in construction, 83, 2017, pp.108-120. [https://doi.org/10.1016/j.autcon.2017.08.007]
• Liu Hexu et al., BIM-based automated design and planning for boarding of light-frame residential buildings, Automation in construction, 89, 2018, pp.235-249. [https://doi.org/10.1016/j.autcon.2018.02.001]
• As Imdat, Pal Siddharth, Basu Prithwish, Artificial intelligence in architecture: Generating conceptual design via deep learning, International Journal of Architectural Computing, 16(4), 2018, pp.306-327. [https://doi.org/10.1177/1478077118800982]
• Tafraout S. et al., Automatic structural design of RC wall-slab buildings using a genetic algorithm with application in BIM environment, Automation in construction, 106, 2019, p.102901. [https://doi.org/10.1016/j.autcon.2019.102901]
• Karan Ebrahim, Asadi Somayeh, Intelligent designer: A computational approach to automating design of windows in buildings, Automation in construction, 102, 2019, pp.160-169. [https://doi.org/10.1016/j.autcon.2019.02.019]
• Fang Yuan, S. Cho, Design optimization of building geometry and fenestration for daylighting and energy performance, Solar energy, 191, 2019, pp.7-18. [https://doi.org/10.1016/j.solener.2019.08.039]
• Agirbas Asli, Facade form-finding with swarm intelligence, Automation in construction, 99, 2019, pp.140-151. [https://doi.org/10.1016/j.autcon.2018.12.003]
• Cubukcuoglu Cemre et al., OPTIMUS: Self-adaptive differential evolution with ensemble of mutation strategies for grasshopper algorithmic modeling, Algorithms, 12(7), 2019, p.141. [https://doi.org/10.3390/a12070141]
• Baghdadi Abtin, Heristchian Mahmoud, Kloft Harald, Design of prefabricated wall-floor building systems using meta-heuristic optimization algorithms, Automation in construction, 114, 2020, p.103156. [https://doi.org/10.1016/j.autcon.2020.103156]
• Alomani Abdullah, El-Rayes Khaled, Automated generation of optimal thematic architectural layouts using image processing, Automation in construction, 117, 2020, p.103255. [https://doi.org/10.1016/j.autcon.2020.103255]
• Ligler, Heather, Reconfiguring atrium hotels: Generating hybrid designs with visual computations in Shape Machine, Automation in construction, 132, 2021, p.103923. [https://doi.org/10.1016/j.autcon.2021.103923]
• Xu Chengran et al., Optimal brick layout of masonry walls based on intelligent evolutionary algorithm and building information modeling, Automation in construction, 129, 2021, p.103824. [https://doi.org/10.1016/j.autcon.2021.103824]
• Zhang et al., Generative design and performance optimization of residential buildings based on parametric algorithm, Energy and buildings, 244, 2021, p.111033. [https://doi.org/10.1016/j.enbuild.2021.111033]
• Liu Jiepeng et al., Automatic and optimal rebar layout in reinforced concrete structure by decomposed optimization algorithms, Automation in construction, 126, 2021, p.103655. [https://doi.org/10.1016/j.autcon.2021.103655]
• Guo Dongming, Onstein Erling, Rosa Angela Daniela La, A semantic approach for automated rule compliance checking in construction industry, IEEE access : practical research, open solutions, 9, 2012, pp.129648-129660. [https://doi.org/10.1109/ACCESS.2021.3108226]
• Soliman-Junior Joao et al., Automated compliance checking in healthcare building design, Automation in construction, 129, 2021, p.103822. [https://doi.org/10.1016/j.autcon.2021.103822]
• Ghannad Pedram, Y.C. Lee, Automated modular housing design using a module configuration algorithm and a coupled generative adversarial network (CoGAN), Automation in construction, 139, 2022, p.104234. [https://doi.org/10.1016/j.autcon.2022.104234]
• Zhou et al., Automated structural design of shear wall structures based on modified genetic algorithm and prior knowledge, Automation in construction, 139, 2022, p.104318. [https://doi.org/10.1016/j.autcon.2022.104318]
• Xu Chengran et al., Automated steel reinforcement detailing in reinforced concrete frames using evolutionary optimization and artificial potential field, Automation in construction, 138, 2022, p.104224. [https://doi.org/10.1016/j.autcon.2022.104224]
• W.J. Sung, Y.H. Jeong, Site planning automation of apartment complex through grid-based calculation in grasshopper, Automation in construction, 138, 2022, p.104216. [https://doi.org/10.1016/j.autcon.2022.104216]
• Otero Roi et al., Semi-automatic roof modelling from indoor laser-acquired data, Automation in construction, 136, 2022, p.104130. [https://doi.org/10.1016/j.autcon.2022.104130]
• Zaraza Julian et al., Generative design to reduce embodied GHG emissions of high-rise buildings, Automation in construction, 139, 2022, p.104274. [https://doi.org/10.1016/j.autcon.2022.104274]