ArticlePDF Available

Workflow for applying optimization-based design exploration to early-stage architectural design - Case study based on EvoMass

Authors:

Abstract and Figures

The role of optimization-based design exploration in early-stage architectural design has been increasingly recognized and valued. It has been widely considered an effective approach to achieving performance-informed and performance-driven design. Nevertheless, there is little research into how such design exploration can be adapted to various early-stage architectural design tasks. With this motivation, this paper revolves around a computer-aided design workflow for early-stage building massing design optimization and exploration while presenting three workshop case studies to demonstrate how the workflow can be intertwined with the design process. The design workflow is based on EvoMass, an integrated building massing design generation and optimization tool in Rhino-Grasshopper. The case study illustrates task-specific applications of the design workflow for synthesizing building design, finding design precedents, and understanding the interrelationship between formal attributes and building performance. The paper concludes by discussing the relevant efficacy of the design workflow for architectural design.
Content may be subject to copyright.
For Peer Review
1Workflow for Applying Optimization-based Design Exploration to Early-
2stage Architectural Design – Case Study Based on EvoMass
3
4Abstract
5The role of optimization-based design exploration in early-stage architectural design has been
6increasingly recognized and valued. It is considered an effective approach to achieving
7performance-informed and performance-driven design. Nevertheless, there is little research
8into how such design exploration can be adapted to various early-stage architectural design
9tasks. With this motivation, this paper revolves around a computer-aided design workflow for
10 early-stage building massing design optimization and exploration while presenting three
11 workshop case studies to demonstrate how the workflow can be intertwined with the design
12 process. The design workflow is based on EvoMass, an integrated building massing design
13 generation and optimization tool in Rhino-Grasshopper. The case study illustrates task-specific
14 applications of the design workflow for synthesizing building design, finding design precedents,
15 and understanding the interrelationship between formal attributes and building performance.
16 The paper concludes by discussing the relevant efficacy of the design workflow for
17 architectural design.
18
19 Keyword
20 Performance-based design; design workflow; design exploration; early design stage; parametric
21 design; optimization
22
23 1 Introduction
24 In the past two decades, a growing amount of research has focused on performance-based
25 design optimization and its application to building design. By integrating parametric modeling,
26 building performance simulation, and computational optimization, performance-based design
27 optimization has been proven to be an effective and efficient approach to assisting designers in
28 coping with complex design tasks related to building performance.
29
30 More recently, researchers and designers have shown interest in the utility of performance-
31 based design optimization in early-stage architectural design 1,2. For early-stage architectural
32 design, the focus of performance-based design optimization is shifting from producing high-
33 performing solutions to exploring the design space to extract information and overcome data-
Page 1 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
34 poor situations 3,4. Such applications can also be referred to as optimization-based design
35 exploration. It is argued to be an approach that enables designers to gain insight into the
36 performance implications in architectural design and make performance-informed decisions
37 and evidence-based design 58.
38
39 Over the past few years, several design tools have been developed to support optimization-
40 based design exploration, such as Design Explorer9,10, Performance Map 11, and Octopus 12.
41 Moreover, the paper's author also developed a design tool called EvoMass13 to support
42 designers to conduct agile building massing design exploration at the outset of design. Despite
43 the availability of these design tools, it is still not easy for designers to fully exploit the potential
44 of optimization-based design exploration in their design. This is partly due to the lack of
45 applicable design workflows integrating optimization-based design exploration into designers’
46 ideation, abstraction, and exploration processes.
47
48 In response, the study of this paper focuses on a computer-aided design workflow for enabling
49 optimization-based design exploration based on the use of EvoMass and presents three task-
50 specific applications of the design workflow. The applications are collected from three design
51 examples completed in a five-day international computational design workshop. The workshop
52 is intended to instruct designers to apply optimization-based design exploration to architectural
53 massing design. Along with the workshop, the participant astutely tailored and adapted the
54 design workflow to their design processes. With this opportunity, this study outlines the
55 application of the design workflow to offer precedents for future applications, tool development,
56 and research on optimization-based design exploration.
57
58 To place this study into context, the paper first provides an overview of optimization-based
59 design exploration and the design workflow adopted in relevant studies. The main body of the
60 paper describes EvoMass and its corresponding design workflow as well as the details of the
61 workshop while demonstrating how the design workflow was adapted to the three case-study
62 design examples. Finally, the paper concludes by discussing the relevant efficacy of these
63 workflows in early-stage architectural design and pointing out future research directions.
64
65 1.1 Optimization-based design exploration and design workflow
66 Applying computational design optimization to performance-based building design, also
67 known as performance-based design optimization14 or performance-driven design
68 optimization15, is not new in research. Over the past two decades, numerous studies have shown
69 that computational design optimization can solve various building design problems associated
Page 2 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
70 with building performance, such as energy efficiency 16, daylighting 17, and material usage 16.
71 Due to its root in numerical optimization, performance-based design optimization is also widely
72 considered a problem-solving design approach, making it more readily applicable to detailed
73 or late-stage design tasks. This is because these types of design tasks are typically well defined,
74 and common optimization workflows based on algorithms, such as canonical genetic
75 algorithms, can provide designers with a specific solution with excellent performance16.
76
77 More recently, the application of performance-based design optimization to early-stage
78 architectural design has been explored. In contrast to late-stage design, architects often face
79 design problems that are highly “ill-defined” and “wicked” at the early design stage18. Therefore,
80 the need for problem-solving is replaced by problem framing, and as a result, the focus of design
81 optimization is also shifted from purely maximizing the performance to design exploration and
82 information extraction1921. With shifting focus, several studies have been undertaken to explore
83 and investigate the applications of optimization-based design exploration in practice and
84 education, including building massing 4,22, building layout 23,24, and façades/envelopes 25,26.
85 These studies demonstrate that the information extracted from optimization allows designers to
86 compare the advantages and disadvantages of different design alternatives or understand the
87 trade-offs characterizing the design problem.
88
89 In terms of the design workflow, most existing studies and applications follow the conventional
90 design optimization workflow, where optimization occurs after designers’ ideation and
91 conceptual exploration 20. This is because, for most existing design tools, design concepts have
92 to be conceived and determined before parametric modeling can be undertaken and the design
93 space constructed 27. Therefore, such workflows can be characterized as post-optimization
94 design explorations. As the design concept and the corresponding design space are
95 predetermined, the scope for the optimization search and the subsequent post-optimization
96 design exploration is confined 28 and “locked-in”29, which may, in turn, strengthen the design
97 fixation and limit its utility to explore unknown design solutions.
98
99 For early-stage design, the overall design process often involves several parallel and iterative
100 explorations toward different alternative directions before narrowing down to a few design
101 options for further development 3,13,3032. However, there are few attempts utilizing optimization
102 to such parallel design exploration workflows. One exception is a design decision support tool
103 called UrbanSOLve, which can facilitate iterative design workflows for small-scale urban
104 (neighborhood) design 33.
105
Page 3 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
106 Regarding architectural design, there is still a lack of applicable tools and design workflows to
107 support this type of parallel and iterative design exploration workflow for probing different
108 alternative design directions. In response, we developed EvoMass primarily for enabling an
109 iterative and parallel exploration for building massing design, and on this basis, a hypothetical
110 optimization-based design exploration workflow has been developed and proposed in our
111 previous studies 7,13. In this paper, we further test the design workflow in a computational design
112 workshop, thereby investigating how designers can tailor the workflow according to their
113 specific design intent and goals as well as the utility of the design workflow in their design
114 process.
115
116 2 Method
117 This paper and the presented workshop examples focus on applying optimization-based design
118 exploration to building massing design. This is because, serving as the interface and envelope
119 between interior spaces and the exterior urban environment, building massing is often taken as
120 the starting point for architectural design 34. In addition, building massing forms can greatly
121 impact building performance, such as daylighting and energy efficiency, and urban
122 environments, such as outdoor thermal/wind comfort and local microclimate 35. Due to these
123 factors, the design of building massings is critical to both architecture and building performance,
124 and it has a profound effect on the subsequent design of building façades and interior spatial
125 configurations.
126
127 To provide a general context for the following analysis of the presented case study, this section
128 provides a brief overview of EvoMass and describes the basic design workflow. After that, the
129 pertinent details of the workshop are presented.
130
131 2.1 EvoMass
132 EvoMass is developed to support rapid building massing prototyping and coding- and
133 programming-free design optimization for early-stage architectural design with a special focus
134 on understanding the interrelationship between building form and performance 13. The
135 development of EvoMass is based on several studies on building massing design generation
136 and evolutionary optimization7,13,36,37. Since previous studies have elaborated on the technical
137 details of EvoMass, this paper only describes the pertinent concept of the embedded algorithms.
138
139 One major feature distinguishing EvoMass from other relevant tools is that designers can
140 customize EvoMass to generate building massing designs for various design conditions through
141 several user-defined parameters, freeing the designer from onerous parametric modeling. In
Page 4 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
142 addition, the embedded optimization algorithm allows the design to be optimized to meet the
143 performance objective. The result of the optimization provides a variety of high-performing
144 solutions that are diverse and differentiated in design, which is intended to help designers
145 identify the pertinent architectural implications related to building performance. Two primary
146 parts of EvoMass help to accomplish this goal.
147
148 The first part contains two independent components for building massing design generation.
149 The two generative components encode subtractive and additive form generation principles 7,37
150 and generate building massing design by accumulating several mass units (the additive
151 component) or creating several voids within a predefined initial volume (the subtractive
152 component). This strategy allows the design generated by these two components to exhibit a
153 wide variety of topological configurations for building massing forms.
154
155 Designers can customize the component with a few user-defined parameters, such as the
156 number of additive mass units or subtractive voids and the spatial boundary (for the additive
157 component) or the initial volume (for the subtractive component). With different input user-
158 defined parameters, the two components can be applied to generate building massings for
159 different design tasks, such as high-rise towers and middle-rise strip-shaped buildings, as shown
160 in Figure 1.
161
162 Figure 1 Random sampling design generated by the two components in EvoMass
163 The second part of EvoMass is a component encapsulating a hybrid evolutionary algorithm,
164 called the steady-state island evolutionary algorithm (SSIEA). SSIEA is developed to achieve
165 an explorative design optimization process, which splits the design population into several
166 subpopulations, each of which can focus the optimization search on a different region in the
167 design space 36,38. Accordingly, in contrast to canonical evolutionary algorithms typically
Page 5 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
168 producing results with a family of similar design variants, the optimization based on SSIEA
169 can offer optimized solutions with high design diversity and differentiation, which intensifies
170 the design information feedback.
171
172 In terms of utilization, the two parts of EvoMass can be easily connected to other simulation
173 tools, such as Ladybug and ClimateStudio (Figure 2), on the Rhino-Grasshopper platform.
174 After entering the user-defined parameters in EvoMass, the user can readily run the
175 optimization. The use of EvoMass significantly reduced the technical demand and flattened the
176 learning curve for mastering computational design optimization. Thus, the use of EvoMass can
177 facilitate the designer to concentrate on applying optimization-based design exploration in their
178 design without spending too much time acquiring technical expertise.
179
180 Figure 2 A design optimization framework based on EvoMass and DIVA (source: 13)
181
182 2.2 Previous Examples
183 EvoMass has been applied to a few studio design tasks in the School of Architecture and Urban
184 Planning, Nanjing University (Figure 3). The examples in Figure 3 illustrate how EvoMass can
185 handle different types of building designs, including high-rise office buildings and middle-rise
186 educational buildings. In these design tasks, the student conducted various design explorations
187 to gather design inspiration or for site/massing studies, thus using the information to complete
188 the final design. Using EvoMass, they can integrate building performance into their design
189 process while eliminating tedious and error-prone manual design exploration.
Page 6 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
190
191 Figure 3 Examples of previous applications of EvoMass based on studio design tasks
192 However, as the student involved in these examples was inexperienced with computational
193 design, the teaching of these design studio tasks preliminarily focused on technical aspects of
194 computational design optimization and building performance. At the same time, due to a rigid
195 teaching plan and target, the application of EvoMass was unable to be freely explored in these
196 studio design tasks. Thus, these examples are insufficient to fully demonstrate the potential of
197 EvoMass in architectural design for design exploration and ideation.
198
199 2.3 Design Workflow
200 In the previous work of EvoMass, we developed a design workflow for conducting
201 optimization-based design exploration (Figure 4). This design workflow assumes that designers
Page 7 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
202 will conduct multiple parallel design optimizations to gather design information relevant to the
203 design problem or is of interest to them. At the same time, feedback from the optimization can
204 encourage designers to iteratively reframe their design problem and thereby promote
205 exploration toward other design directions. The parallel and iterative design exploration
206 workflow is made possible by EvoMass’ ability to generate a wide variety of building massing
207 design alternatives without extra parametric modeling. Accordingly, designers can swiftly set
208 up another optimization framework and recalibrate the optimization on different performance-
209 related design spaces by modifying the user-defined parameters of EvoMass or the simulated
210 performance.
211
212 Figure 4 Optimization-based design exploration workflow using EvoMass
213 The design workflow also implies that the use of EvoMass is not intended to directly provide
214 solutions to designers. In contrast, optimization-based design exploration based on EvoMass
215 aims to provide designers with information to overcome data-poor situations and design
216 fixation while helping them narrow down the scope of the design space for more detailed and
217 in-depth exploration. This also agrees with Bradner’s observation that “the computed optimum
218 was often used as the starting point for design exploration, not the end product3. Hence, this
219 also highlights a drawback in the previous application in studio design tasks, in which the
220 student tends to get stuck by the design in the optimization result provided by EvoMass.
221
222 2.4 Workshop
223 The workshop, with the title of Agile Performance-based Building Massing Design
224 Optimization and Exploration, was held from 27 June to 1 July 2021 and as a part of the 2021
225 Inclusive Futures workshop series. Aiming at optimization-based design exploration, the
226 workshop is centered on how designers can employ optimization-based design exploration and
227 experience the difference in the design process when using optimization-based design
228 exploration. As a result, teaching technical know-how is not the main goal of this workshop,
Page 8 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
229 and preliminary computational design skills and the experience of using Rhino-Grasshopper
230 are required for participants to attend this workshop.
231
232 The workshop provides an opportunity to examine how optimization-based design exploration
233 and EvoMass can be applied to various design situations beyond previous studio design tasks.
234 Therefore, there were no predefined design tasks for the workshop, and the participants were
235 encouraged to conduct a design based on their routine design jobs while considering factors
236 such as geography, climate, culture, and design intent instead of focusing only on building
237 performance. In this way, the participant could freely explore how optimization-based design
238 exploration and EvoMass could be adapted to their design.
239
240 As highlighted in the workshop title, “agileness” is a critical aspect for optimization-based
241 design exploration because time-consuming and technically demanding issues are generally
242 accepted as two barriers for computational design optimization to be applied to rush early-stage
243 architectural design 39. Thus, the use of EvoMass in this workshop addresses the technical
244 challenge and onerous parametric modeling and provides designers with a handy tool for their
245 design jobs.
246
247 The workshop was taught online over five consecutive days, each day consisting of a three-
248 hour teaching session. The first two days provided technical tutorials covering the operation of
249 EvoMass and introduced the basic design workflow, as illustrated in Figure 4, which provides
250 a scaffold to facilitate them to conduct optimization-based design exploration. In the third and
251 fourth days, the participants began to apply EvoMass to their design tasks and customized the
252 design workflow according to their specific design goals. The workshop concluded with a final
253 presentation and review on the fifth day, where three experts from the National University of
254 Singapore, Princeton University, and Nanjing University were invited as the review panelists.
255
256 Apart from EvoMass, Ladybug was used for the building performance simulation. Owing to
257 the time constraint of the five-day workshop, only solar irradiation and sunlight hours were
258 considered the building performance metrics in this workshop because the simulation speed for
259 both of these factors is relatively fast, typically less than 10 seconds each. These two metrics
260 have been adopted in our previous studies, providing a technical and methodological reference
261 for the participant 37,40.
262
263 The iteration number was also kept small to further shorten the optimization process, typically
264 from 800 to 1500. With these simulation and optimization setups, each optimization run can be
265 completed in 2 to 5 hours. Although the smaller number of iterations might be insufficient for
Page 9 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
266 the optimization to reach full convergence, the general design tendency was typically
267 identifiable. Moreover, the participant was also advised to run longer optimization processes
268 with more iterations during the nighttime. During the workshop, the participants ran the
269 optimization after the teaching session and then presented and discussed the optimization
270 results in the next session (Figure 5).
271
272 Figure 5 Screenshots of the workshop teaching session (participants showing their optimization results)
273
274 We acknowledged that the two metrics have a certain correlation; thus, some optimization
275 results showed similar design patterns despite using the two different performance metrics.
276 Furthermore, the shortened optimization search process may also impede the optimization
277 result. Hence, it should be stressed that other performance metrics such as daylighting and
278 wind-related factors as well as longer and more accurate optimization search need to be
279 incorporated into their design when they apply the workflow for future design tasks with fewer
280 time constraints.
281
282 3 Case study
283 After the five-day teaching and design exercise, 10 out of 26 participants completed four
284 designs in this workshop, and three representative designs are presented as the case study here.
285 The one that is not selected is because the participant only ran one optimization and did not
286 show overt intent related to design exploration or information extraction. In contrast, the three
Page 10 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
287 examples, based on Iran, Egypt, and China, explore the application of the optimization-based
288 design exploration workflow to early-stage architectural design for distinct design intent. Listed
289 in Table 1, these examples cover a wide variety of design contexts, but due to the time
290 constraints of the workshop, the degree of completion of the designs varies.
291 Table 1 Summary of the design parameters of the case study
Location
Functionality
Design objective
Keyword
Participant
Example
1
Bushehr,
Iran
Vernacular
buildings
Explore high-performing
building typologies that
are suitable for the
existing traditional urban
context
Design synthesis
one practitioner
and three
postgraduate
students
Example
2
Al-
Dakhliya,
Egypt
Government
office
building
Explore high-performing
solutions as a design
reference for conceptual
development
Finding
precedents
One postgraduate
student
Example
3
Shanghai,
China
University
library
building
Explore design solutions
considering both building
performance and design
constraints
Meta-optimization
Three postgraduate
students and one
undergraduate
student
292
293 3.1 Example 1
294 The first example is based on the city of Bushehr, located in southern Iran. The city lies on a
295 vast plain along the coastal region of the Persian Gulf. As a trade center of Iran in the past
296 centuries, the city has a mixture of traditional vernacular courtyard buildings and modern
297 architecture. The building site is situated near the water and within the traditional city context
298 of Bushehr. As the warmest region of Iran, Bushehr's average daily high temperature is 30
299 degrees Celcius, and in the warmest season, the temperature can reach 45 degrees. Thus, the
300 participants focused their design on preventing the building from overheating.
301
302 Since there were four participants, it allowed for a more explorative design process. Based on
303 the two design generative components in EvoMass and two performance metrics simulated by
304 Ladybug, each participant carried out optimizations focusing on a particular combination of the
305 generative model and the performance metric (Figure 6). As such, different design directions
306 can be explored. The collaborative design exercise also amplifies the design information
307 feedback from the optimization.
Page 11 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
308
309 Figure 6 The optimization result for design exploration (redraw based on participants’ work)
310 In this example, a key finding revealed by the optimization result is the difference in the
311 architectural implications related to building performance when using the two generative
312 components. When using the subtractive generative component, the optimal design solution
313 closely resembles the traditional building typology with inner courtyards and narrow alleys,
314 which leads to mutual shading effects. In contrast, when using the additive generative
315 component, the optimal design solution displays overhanging and cantilevered blocks, creating
316 self-shading features and stepped volumes, exhibiting features in line with solar envelopes 41.
317 Both optimization results reveal design strategies that can reduce the solar heat absorbed by the
318 building surface, preventing the building from overheating.
319
320 The optimization results also led the participant to rethink the traditional use of vertical
321 elements to create shading effects. They attributed this to the limitation imposed by traditional
322 structural and constructional techniques since large spans, or overhanging structures were
323 difficult and costly to construct until modern times. Thus, with modern structural techniques,
324 more efficient passive cooling design strategies such as self-shading and the solar envelope can
325 also be applied to building design under such climatic conditions.
326
327 Finally, based on the optimization result, the participants synthesized the extracted information
328 into an integrated design solution incorporating the building features and passive cooling
329 strategies revealed by the optimization (Figure 7). As shown in the final design, the participant
330 primarily utilized the vertical component to enhance self-shading effects. Meanwhile, they also
331 inserted small courtyards and alleys into the building volume to further enhance mutual shading.
332 The courtyard and alley compensate for the inferior daylighting due to the deep floor plan
333 caused by the large overhanging and cantilevered blocks. Furthermore, the extracted
Page 12 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
334 information also facilitated the participant to synthesize building performance into other design
335 aspects, such as window types and space organization.
336
337 Figure 7 The final design (redraw based on participants’ work)
338
339 3.2 Example 2
340 The second example is a redesign of a government office building in Al-Dakhilya Governorate,
341 Egypt. The participant argued that the existing design of the building does not make the most
342 benefit from the local bioclimatic resource. This region has a typical arid climate, and in the
343 hottest month, the temperature reaches 43 degrees Celcius. Therefore, the participant specified
344 that the design objective was to increase the sunlight hours on the east façade surface using
345 natural daylight with lower solar irradiation. It is partly because sunlight cannot be used as a
346 source of natural daylighting because of the intense solar irradiation at noon and in the afternoon.
347 In addition, another objective was to maximize the solar irradiation on the roof surface for
348 energy generation using PV panels.
349
350 Aiming at this objective, the participant conducted two design optimizations based on the two
351 generative components. However, since the participant had to do all the design work and
352 optimization, this limited the scope of design exploration using optimization and made this
353 participant unable to complete a design at the end of the workshop. Nonetheless, instead of
354 synthesizing the design information into an integrated design as the participant in Example 1
355 did, this participant utilized optimization-based design exploration to understand the
356 architectural implications of building performance and then found design precedents for
357 inspiration (Figure 8).
358
Page 13 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
359
360 Figure 8 Design condition and the optimization result (redraw based on participants’ work)
361 Based on using two generative models, the optimization results also display distinct features in
362 response to the optimization objectives. For the result based on the additive component, the
363 major volume of the building rests on the west side of the building plot to escape from the
364 context shading caused by the building on its east side, which ensures that the east façade
365 surfaces of the building are permitted to receive the maximum amount of sunlight. At the same
366 time, the building massing design in the optimization result also features a stepped volume that
367 maximizes the solar irradiation that the roof-mounted PV panels can capture.
368
369 In contrast, when using the subtractive component, the building massing features a long atrium
370 extending across the central axis of the building. This strategy increases the area of the east-
371 facing façade for capturing sunlight in the morning. In addition, as the subtractive component
372 typically produces building massing with a large roof surface that allows for more PV panels
373 to be installed, the optimization, indeed, attempts to balance the trade-off between minimizing
374 the reduction on the roof surface and creating more openings on the roof to allow more sunlight
375 to reach the inner part of the building.
376
377 This participant further abstracted the optimization result and found design precedents for
378 inspiration based on the extracted information. According to the architectural implication
379 revealed by the optimization result, the participant found several real-world building designs
380 that incorporated the same strategies appearing in the optimized design solutions (Figure 9). In
381 design practice, precedents serve as valuable sources of inspiration for aesthetic ideas and
Page 14 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
382 formal styles 42. Thus, these design precedents can provide designers with a starting point for
383 conceptual development and help them narrow down the design space for further exploration.
384 Moreover, designers can also connect the performance implications revealed by the
385 optimization to these precedents, allowing them to understand these design precedents' formal
386 features from a performance perspective. Last, if the participant was given the time to run more
387 iterations of optimization-based design explorations, more design precedents could be found
388 and be superimposed one upon the other to serve as traces of a creative stream of
389 consciousness43.
390
391 Figure 9 Design precedents found based on the information abstracted from the optimization result (redraw based
392 on participants’ work)
393
394 3.3 Example 3
395 The last example is a library building design in a constrained building plot, where several trees
396 on the building plot need to be preserved. Thus, the participants in this example framed their
397 design problem to maximize sunlight hours and focused on satisfying design constraints
398 imposed by the building plot. Therefore, they explored the application of soft constraints to
399 intervene in the optimization process to find a building design that does not “invade” the
400 boundary of the trees. Technically, they used a penalty function to punish the design with a
401 building volume overlapping on the tree boundaries by lowering the fitness to decrease the
402 likelihood of the design surviving in the subsequent evolutionary process.
403
404 As there were four participants involved in this project, they could explore more of the potential
405 of the two different generative components to the design task. They used additive and
406 subtractive components to generate building massing design variants and conducted
407 optimization. Unlike the first example focusing on extracting design information from the
408 optimization result, the participants in this example were more interested in interacting and
409 “playing” with EvoMass. Hence, beyond simply using the two generative components, they
410 modified the user-defined design parameters in the generative components, including the
411 number of column-grid units and the spacing of the column grid, and repeated the optimization
412 using different user-defined parameters. By modifying the design parameter, the participants
413 define different design spaces that embody a different set of building massing design typologies
Page 15 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
414 for the optimization.
415
416 For the optimization based on the additive component, with a larger number of column-grid
417 units and a smaller column-grid spacing, the generated design can have more flexible
418 topological configurations (Figure 10-left). In contrast, with fewer column-grid units and a
419 larger column-grid spacing, the generated design tends to be relatively bulky and massive
420 (Figure 10-right). For the optimization based on the subtractive components, it was found that
421 the number of subtractive voids significantly affects the efficiency of the optimization to
422 identify valid design solutions. Using fewer voids makes the design generation over-
423 constrained, resulting in a more challenging optimization process to find legitimate solutions
424 (Figure 11). Moreover, these results also helped the participant identify different passive
425 energy-saving design strategies such as self-shading, stepped roofing, and scattered small
426 courtyards.
427
428 Figure 10 The optimization results based on different design parameters using the additive component (redraw
429 based on participants’ work)
Page 16 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
430
431 Figure 11 The optimization results based on different design parameters using the subtractive component (redraw
432 based on participants’ work)
433
434 With different combinations of design parameters explored, the purpose of conducting
435 optimization goes beyond merely optimizing the building performance but rather to determine
436 which design space, as well as the building typology underpinning that design space, is more
437 suitable to the given design task. This process can be regarded as a “meta-optimization”, where
438 the goal of the design workflow transforms from finding the optimal solution within one single
439 design space to finding the design space that can produce more balanced design solutions.
440 However, due to the time constraints, their exploration of the design parameter was still rather
441 limited, and if there were more iterations of design exploration, a more comprehensive
442 comparison of design spaces could be made.
443
444 4 Discussion
445 The presented examples illustrate how EvoMass and the corresponding optimization-based
446 design exploration workflow can be applied to early-stage architectural design for design
447 ideation and synthesis. While the workshop output undoubtedly contains many flaws because
Page 17 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
448 of the five-day time constraints, it still demonstrates that optimization-based design exploration
449 can potentially be integrated into the design process and open new avenues for design
450 development in the age of sustainable and digital design. As shown by the three examples, the
451 participant generally followed the basic design workflow in Figure 4, while three task-specific
452 variations of the design workflow can be identified in response to their distinct design intention.
453
454 In the first example, the designer synthesized the architectural implications associated with
455 building performance into an integrated design solution. The design workflow begins with a
456 parallel exploration process of building design toward different performance objectives.
457 Afterward, the optimization provides information to the designer, and the designer extrapolates
458 the design implications and conceives design solutions, in nova, to challenge the “computed
459 optimal”. Additionally, the extracted information can encourage the designer to consider the
460 impact of building performance on other design aspects, including orientation, façade design,
461 and spatial configuration.
462
463 In the second example, the designer utilized the design workflow to find precedents for
464 abstraction and inspiration. Therefore, the execution of the workflow is meant to narrow down
465 the scope of design exploration and offer a “starting point” for design exploration, as Bradner
466 pointed out 3. The associated design precedents also inspire the designer with aesthetic ideas
467 and formal styles. As such, the designer can develop a more detailed and task-specific
468 parametric model and conduct the corresponding optimization to search for the solution that
469 can balance the requirements of building performance, functionality, and design intent.
470 Moreover, finding precedents can also help the designer leap out of the optimization result.
471
472 In the third example, the design workflow progresses with the exploration of different design
473 spaces representing different sets of building design typologies into optimization. By
474 conducting the design exploration, the designer can connect the pursuit of building performance
475 improvement with building massing design, which also promotes the investigation of how
476 building massing with different formal attributes can reversely affect building performance and
477 the design optimization result. Accordingly, the workflow no longer simply aims to optimize
478 the design toward specific building performance but instead serves to find the desired design
479 space that can produce satisficing solutions.
480
481 In summary, compared to the previous studio design tasks, the three examples show more
482 engaging applications of EvoMass to optimization-based design exploration in dynamic
483 architectural design scenarios. In addition, in the previous application to design studio tasks,
484 we spent three to four weeks instructing the student to conduct the optimization-based design
Page 18 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
485 exploration, in which the instructor (the author) had a noticeable effect on the student's design
486 process. In contrast, in the presented workshop design, the five-day length leaves the design
487 process mostly conducted by the participant depending on their own design intuition without
488 having too much instructor intervention. Therefore, it allows the application of EvoMass and
489 the optimization-based design exploration workflow to be freely adjusted to their design.
490
491 In terms of the overall pattern of the design process, one panelist summarizes that by running
492 the optimization, the designer is not seeking a design solution but a design space. Therefore,
493 this tendency could potentially serve as a guiding principle for future applications and research
494 of optimization-based design exploration to early-stage architectural design. In addition, it also
495 implies that the application of EvoMass in building massing design exploration is not the end
496 of the design process, but rather a starting point. The information revealed by the optimization
497 is conducive to more informed and detailed design exploration.
498
499 The three presented examples also specify the application of the optimization-based design
500 exploration workflow to early-stage architectural design using EvoMass. Although we
501 proposed a basic design workflow as shown in Figure 4, it remains relatively abstract. It is not
502 evident for many users to fully comprehend the idea of optimization-based design exploration
503 and how the design workflow can be connected to their design tasks. As a result, it is common
504 to see many applications of EvoMass focusing on seeking specific solutions rather than
505 exploring the design space as expected. In this regard, the presented application of the design
506 workflow can be utilized explicitly by designers who intend to use optimization-based design
507 exploration with EvoMass. Meanwhile, it should be stressed that the three task-specific design
508 workflows are not mutually exclusive. Designers can synergize these workflows to further
509 enhance the information feedback from different design aspects when using optimization-based
510 design exploration.
511
512 Last but not least, one panelist also commented that it is impressive to see the participant
513 conducting such sizeable design optimization and exploration in such a short time. This
514 underscores the potential utility of optimization-based design exploration using EvoMass for
515 practical design scenarios. In the rush early-stage architectural design task, the time is typically
516 insufficient for designers to conduct such design exploration because they need to encode
517 multiple design schemes and build up the parametric models from scratch. EvoMass, by
518 contrast, allows the designer to swiftly explore the design from various aspects without being
519 burdened by parametric modeling, thereby facilitating the designer to concentrate on the design
520 exploration process.
521
Page 19 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
522 4.1 Limitations
523 Due to the five-day constraint, the exploration conducted by the participants was still limited.
524 However, it can be extrapolated that the exploration can be amplified if the designer has enough
525 time. In addition, the workshop also reveals several defects that need to be considered in the
526 future application and teaching of optimization-based design exploration using EvoMass. First,
527 as pointed out by one panelist, the participant in the first example did not evaluate the
528 performance of the synthesized design against the optimal designs produced by the optimization.
529 Therefore, if the time is available, it is recommended for designers to conduct post-evaluations,
530 which can also be an opportunity for further improvement and detailed design exploration.
531
532 Second, during the workshop, we noticed that the participant was less critical of the
533 optimization result even when it was infeasible from the design perspective. In many cases, the
534 infeasible result is due to inappropriate stimulation setups or optimization objectives defined
535 by the designer. Thus, as one panelist suggested, designers should be more reflective when
536 analyzing optimization results. Designers should rethink their optimization setups when the
537 optimization result is infeasible rather than simply taking it for granted.
538
539 Third, when teaching the operation of EvoMass in the workshop, we found that the participant
540 could be overwhelmed by the technical issues associated with design generation and
541 optimization, which leads them to confuse the two procedures. As a result, we noticed that only
542 the third example shows that the participant was consciously aware that the design generation
543 setup could greatly impact the optimization result, while the other two groups conspicuously
544 omitted this aspect. Therefore, we consider that it is more advisable to separate the teaching of
545 these two procedures and to place more emphasis on design generation in future teaching.
546
547 4.2 Future Research
548 This study provides a starting point to investigate how optimization-based design exploration
549 applies to architectural design. Thus, based on the three task-specific design workflows, a
550 systematic and detailed investigation of each workflow will be carried out for further
551 structuralizing and formalizing the design process. In addition, examining these workflows to
552 practical design tasks for real-world projects is another critical aspect to be considered.
553
554 In addition, we acknowledge that the presented example and application are all based on
555 EvoMass. Hence, it might not be able to be generalized to designs using other tools. Even so,
556 this study can also serve as a groundwork for future tool development concerning other design
557 aspects, such as spatial configuration and building façades. For instance, we are conducting
558 another track of research related to the application of optimization-based design exploration to
Page 20 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
559 building façade design. That research focuses on how different façade schemes can affect
560 building performance and the building massing form 44, which can also provide design
561 information and abstract design concepts for performance-based building design.
562
563 5 Conclusion
564 To conclude, this study aims to further advance the research and application of optimization-
565 based design exploration to early-stage architectural design with a special focus on the design
566 workflow. The study presents three case-study applications based on EvoMass that integrate
567 the optimization-based design exploration workflow into early-stage architectural massing
568 design. The presented applications demonstrate that the design workflow can provide design
569 information from different design aspects, and the information is conducive for designers to
570 instill building performance into their design process. In addition, the presented case studies
571 also provide precedents for future teaching, research, tool development, and applications of
572 optimization-based design exploration to early-stage architectural design.
573
574
575 References
576 1. Touloupaki E, Theodosiou T. Performance simulation integrated in parametric 3D
577 modeling as a method for early stage design optimization - A review. Energies; 10. Epub
578 ahead of print 2017. DOI: 10.3390/en10050637.
579 2. Cichocka J, Browne WN, Ramirez ER. OPTIMIZATION IN THE ARCHITECTURAL
580 PRACTICE An International Survey. In: Proceedings of the 22nd CAADRIA
581 Conference, pp. 387–397.
582 3. Bradner E, Iorio F, Davis M. Parameters tell the design story: Ideation and abstraction
583 in design optimization. In: 2014 Proceedings of the Symposium on Simulation for
584 Architecture & Urban Design. Tampa, Florida: Society for Computer Simulation
585 International, 2014, pp. 172–197.
586 4. Chen KW, Choo TS, Norford L. Enabling Algorithm-Assisted Architectural Design
587 Exploration for Computational Design Novices. Comput Des Appl 2019; 16: 269–288.
588 5. Wortmann T. Architectural Design Optimization-Results from a User Survey. In:
589 Architecture Across Boundaries 2019. Suzhou, China, 2019.
590 6. Wortmann T, Schroepfer T. From optimization to performance-informed design. In:
591 Rockcastle S, Rakha T, Davila CC, et al. (eds) 2019 Proceedings of the Symposium on
592 Simulation for Architecture & Urban Design. Atlanta, GA, USA, 2019, pp. 261–268.
593 7. Wang L, Janssen P, Chen KW, et al. Subtractive Building Massing for Performance-
594 Based Architectural Design Exploration : A Case Study of Daylighting Optimization.
595 Sustain 2019; 11: 6965.
596 8. Chen KW, Janssen P, Schlueter A. Multi-objective optimisation of building form,
597 envelope and cooling system for improved building energy performance. Autom Constr
Page 21 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
598 2018; 94: 449–457.
599 9. Brown NC, Mueller CT. Design variable analysis and generation for performance-based
600 parametric modeling in architecture. Int J Archit Comput 2019; 17: 36–52.
601 10. Brown NC, Jusiega V, Mueller CT. Implementing data-driven parametric building
602 design with a flexible toolbox. Autom Constr 2020; 118: 103252.
603 11. Wortmann T. Surveying design spaces with performance maps: A multivariate
604 visualization method for parametric design and architectural design optimization. Int J
605 Archit Comput 2017; 15: 38–53.
606 12. Vierlinger R. Multi Objective Design Interface. University of Applied Arts Vienna,
607 2013. Epub ahead of print 22 April 2013. DOI: 10.13140/RG.2.1.3401.0324.
608 13. Wang L, Chen KW, Janssen P, et al. Enabling optimisation-based exploration for
609 building massing design: A coding-free evolutionary building massing design toolkit in
610 rhino-grasshopper. In: RE: Anthropocene, Design in the Age of Humans - Proceedings
611 of the 25th International Conference on Computer-Aided Architectural Design
612 Research in Asia, CAADRIA 2020. 2020, pp. 255–264.
613 14. Oxman R. Performance-Based Design: Current Practices and Research Issues. Int J
614 Archit Comput 2008; 6: 1–17.
615 15. Shi X. Performance-based and performance-driven architectural design and
616 optimization. Front Archit Civ Eng China 2010; 4: 512–518.
617 16. Shi X. Design optimization of insulation usage and space conditioning load using energy
618 simulation and genetic algorithm. Energy 2011; 36: 1659–1667.
619 17. González J, Fiorito F. Daylight Design of Office Buildings: Optimisation of External
620 Solar Shadings by Using Combined Simulation Methods. Buildings 2015; 5: 560–580.
621 18. JANSSEN PH. THE ROLE OF PRECONCEPTIONS IN DESIGN. In: GERO J (ed)
622 Design Computing and Cognition ’06. Dordrecht: Springer Netherlands, pp. 365–383.
623 19. Janssen P. A design method and computational architecture for generating and evolving
624 building designs. The Hong Kong Polytechnic University, 2005.
625 20. Wortmann T, Tunçer B. Differentiating parametric design: Digital workflows in
626 contemporary architecture and construction. Des Stud 2017; 52: 173–197.
627 21. Turrin M, Von Buelow P, Stouffs R. Design explorations of performance driven
628 geometry in architectural design using parametric modeling and genetic algorithms. Adv
629 Eng Informatics 2011; 25: 656–675.
630 22. Zhang Y, Tindale A, Garcia A, et al. How to integrate optimization into building design
631 practice: lessons learnt from a design optimization competition. In: Proceedings of
632 BS2013: 13th Conference of International Building Performance Simulation
633 Association. 2013, pp. 1860–1868.
634 23. Petrov M, Walker J, Optimization AD, et al. Optioneering Methods for Optimization -
635 Methods of exploring primary and secondary performance criteria in urban. In:
636 Anthropologic: Architecture and Fabrication in the cognitive age - Proceedings of the
637 38th eCAADe Conference. 2020, pp. 29–36.
638 24. Natanian J, Wortmann T. Simplified evaluation metrics for generative energy-driven
639 urban design: A morphological study of residential blocks in Tel Aviv. Energy Build
640 2021; 240: 110916.
641 25. Pasternak A, Kwieciński K. High-rise Building Optimization: A Design Studio
Page 22 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
642 Curriculum. In: eCAADe 33. Vienna, Austria, 2015, pp. 305–314.
643 26. Negendahl K, Nielsen TR. Building energy optimization in the early design stages: A
644 simplified method. Energy Build 2015; 105: 88–99.
645 27. Victor Okhoya, Bernal M, Chen C, et al. CONSTRUCTING DESIGN SPACES: Case
646 Studies in Parametric Building Performance Analysis at Perkins and Will. Perkins+Will
647 Res J 2018; 10.02: 77–98.
648 28. Kilian A. Design Innovation through Constraint Modeling. Int J Archit Comput 2006;
649 4: 87–105.
650 29. Harding JE, Shepherd P. Meta-Parametric Design. Des Stud 2017; 52: 73–95.
651 30. Harding J, Joyce S, Shepherd P, et al. Thinking Topologically at Early Stage Parametric
652 Design. In: Advances in Architectural Geometry 2012. Vienna: Springer Vienna, pp.
653 67–76.
654 31. Cristie V, Joyce SC. Versioning for parametric design exploration process. Autom
655 Constr 2021; 129: 103802.
656 32. Woodbury RF, Burrow AL. Whither design space? AIE EDAM Artif Intell Eng Des Anal
657 Manuf 2006; 20: 63–82.
658 33. Nault E, Waibel C, Carmeliet J, et al. Development and test application of the
659 UrbanSOLve decision-support prototype for early-stage neighborhood design. Build
660 Environ 2018; 137: 58–72.
661 34. Akin O, Moustapha H. Strategic use of representation in architectural massing. Des Stud
662 2004; 25: 31–50.
663 35. DeKay M, Brown GZ. Sun, wind, and light: Architectural design strategies. John Wiley
664 & Sons, 2013.
665 36. Wang L, Janssen P, Ji G. SSIEA: a hybrid evolutionary algorithm for supporting
666 conceptual architectural design. Artif Intell Eng Des Anal Manuf 2020; 34: 458–476.
667 37. Wang L, Chen KW, Janssen P, et al. Algorithmic generation of architectural massing
668 models for building design optimisation: Parametric modelling using subtractive and
669 additive form generation principles. In: RE: Anthropocene, Design in the Age of
670 Humans - Proceedings of the 25th International Conference on Computer-Aided
671 Architectural Design Research in Asia, CAADRIA 2020. 2020, pp. 385–394.
672 38. Wang L, Janssen P, Ji G. Diversity and efficiency. In: Intelligent and Informed -
673 Proceedings of the 24th International Conference on Computer-Aided Architectural
674 Design Research in Asia, CAADRIA 2019. 2019, pp. 593–602.
675 39. Attia S, Hamdy M, O’Brien W, et al. Assessing gaps and needs for integrating building
676 performance optimization tools in net zero energy buildings design. Energy Build 2013;
677 60: 110–124.
678 40. Chen Y, Lu Y, Gu T, et al. From Separation to Incorporation - A Full-Circle Application
679 of Computational Approaches to Performance-Based Architectural Design. In:
680 Proceedings of the 2021 DigitalFUTURES. Springer Singapore, pp. 189–198.
681 41. Knowles RL. The solar envelope: its meaning for energy and buildings. Energy Build
682 2003; 35: 15–25.
683 42. Eckert C, Stacey M. Sources of inspiration: A language of design. Des Stud 2000; 21:
684 523–538.
685 43. Simitch A, Warke V. The language of architecture: 26 principles every architect should
Page 23 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
686 know. Rockport Pub, 2014.
687 44. Zhang H, Wang L, Guohua JI. The synergy of building massing and facade: An evo-
688 devo approach for performance-based design optimization combining facade design
689 with building massing. In: A. Globa, J. van Ameijde, A. Fingrut, N. Kim TTS Lo (ed)
690 Projections - Proceedings of the 26th International Conference of the Association for
691 Computer-Aided Architectural Design Research in Asia, CAADRIA 2021. The Chinese
692 University of Hong Kong and Online, Hong Kong, 2021, pp. 451–460.
693
Page 24 of 24
https://mc.manuscriptcentral.com/ijoac
International Journal of Architectural Computing
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
... EvoMass has been applied to the teaching of several design studios and workshops. 40,41 In these applications, users were instructed to conduct performance-based design optimization for building massing design, and during the optimization process, they also established their own design approaches to integrating the optimization into their design development activities. By observing their utilization, we extracted two optimization-aided design approaches that are typically adopted by the users and conducted research further formalizing the approaches. ...
Article
Full-text available
The role of using performance-based design optimization in early-stage architectural design to prevent poorly designed buildings has been increasingly recognized by researchers and designers. Recently, a large amount of research has been made focusing on technical advancement, which, however, also reflects the limited research on how this technique can be applied to the design process and how it can aid designers when confronting ill-defined design problems. In this regard, this paper centers on the design approaches assisted by using performance-based design optimization. The paper proposes two optimization-aided design approaches that were identified in the previous applications of a design tool, called EvoMass, and showcases these approaches in case-study designs. The case studies demonstrate how the use of performance-based design optimization can facilitate designers’ reflection and exploration. With the demonstration of the design approaches, we discuss the utility of performance-based design optimization in assisting architects in the early design stages.
... The design optimization process generates and evaluates a large number of design variants and identifies highperforming solutions for the design problem. When design optimization is applied in practice, an iterative process consisting of multiple evolutionary runs is typically required (Chen et al., 2022;Janssen et al., 2022;Likai Wang, 2022). In each iteration, the designer tweaks the process by adjusting the design generation and evaluation, so as to improve the quality of design variants being evolved. ...
Conference Paper
Full-text available
This paper presents a framework for rapid design optimization, which is aimed to support the iterative design optimization process. The framework consists of Rhino-Grasshopper and an evaluation server. In order to speed up the optimization process, three strategies are implemented in the framework, including parallel execution, early abortion, and multi-resolution simulations. To examine the efficacy of the developed framework, a case-study design optimization is conducted, and different combinations of the strategies are tested and compared. The case study investigates the impact of the adopted strategies on the optimization process in terms of search efficiency and effectiveness, and the result of the case study also demonstrates that optimization can be significantly improved by the use of the adopted strategies.
... EvoMass has been applied to several studio design tasks associated with performance-based architectural design (L. Wang, 2022). In these applications, EvoMass demonstrates its capability of assisting designers in early-stage building massing design exploration and information extraction. ...
Conference Paper
Full-text available
Over the past decade, computational design optimization has been increasingly applied to the studies related to early-stage architectural design exploration and information extraction, in which various optimization approaches have been adopted such as single-objective optimization (SOO) and multi-objective optimization (MOO). Despite this, the impact of optimization approaches on information extraction has been relatively under-researched. In this regard, this paper presents a comparison study of different optimization approaches, with a special focus on search efficiency and information extraction. In the comparison study, we investigate the impact of different optimization approaches on information extraction and design cognition and finally, discuss the utility of these approaches to architectural design ideation and synthesis.
... Previous studies demonstrated that the optimization of EvoMass can provide designers with multiple high-performing solutions that would be challenging for human designers to conceive. This also highlights the utility of EvoMass in stimulating architects' design exploration and helping them overcome design fixation (Chen et al., 2022;Wang, 2021). In this study, the combination of EvoMass for building massing and façade designs further increases the complexity of the design optimization task. ...
Article
Full-text available
In performance-based architectural design optimization, the design of building massings and façades is commonly separated, which weakens the effectiveness in performance improvement. In response, this study proposes a hybrid massing-façade integrated design generation and optimization workflow to integrate the two elements in an evolutionary design process. Compared with the existing approaches, the proposed workflow emphasizes the diversity of building design generation, with which various combinations of building massing forms and façade patterns can be systematically explored. Two case studies and a corresponding comparison study are presented to demonstrate the efficacy of the proposed workflow. Results show that the optimization can produce designs coupling the potential of building massings and façades in performance improvement. In addition, the optimization can provide information that supports early-stage architectural design exploration. Such information also enables the architect to understand the performance implications associated with the synergy of building massing and façade design.
Chapter
This paper investigates team psychological safety (N = 34 teams) in a synchronous online engineering design class spanning 4 weeks. While work in this field has suggested that psychological safety in virtual teams can facilitate knowledge-sharing, trust among teams, and overall performance, there have been limited investigations of the longitudinal trajectory of psychological safety, when the construct stabilizes in a virtual environment, and what factors impact the building of psychological safety in virtual teams.
Chapter
Benchmark designs can play a critical role in giving performative feedback to the design exploration process. This paper explores the idea of using computational design optimization as a way of generating such site-specific benchmark designs. A case study of the proposed benchmarking approach is presented focusing on public residential precinct design in Singapore. A key feature of this approach is a parametric typology that can generate designs that vary significantly in their overall configuration while at the same time remaining feasible and valid. In addition, an evolutionary optimization system is then used to evolve a set of high-performing site-specific benchmark designs. The case study successfully demonstrates how these benchmarks can be used to give performative feedback to the design exploration process.
Article
Design support tools are increasingly used to address complex issues. However, existing tools are limited in the interpretability of their solutions, the participation of architects, and the developability of the framework. They might provide architects with “non-guide assistance” throughout the performance design process. Venis, a designer-centric support tool that provides interactive decision-making and automatic optimization, was proposed to address these concerns. The former mechanism provides architects with reasonable, interpretable suggestions for schemes, where the intentions of architects are fully integrated into the deepening design. The latter mechanism inspires architects in the absence of clear concepts by generating numerous alternatives and assisting in scheme selection through comprehensive metrics. Venis also incorporates real-time performance evaluation, full-process visualization, a user-friendly interface, and generalizable design scenarios for architects. Feedback from professionals has proven its validity. The results show that Venis can assist architects in fully exploring and improving performance efficiently and reasonably.
Article
This paper describes a method to tailor computational design algorithms to evaluate human-centric outputs in architectural projects by presenting a case study that evaluates teamwork affordance for primary care clinics. We argue that computational design assessment techniques fall short of evaluating those human experiences that stem from multiple interactions between individuals and the surrounding environment. This research suggests that future computational design algorithms could benefit from incorporating scenario-based methods developed in the field of evidence-based design, in which studies are concerned about improving qualitative goals. Through this case study, we describe the process of prototyping a computational design algorithm based on the Functional Scenario Analysis approach, applying the algorithm to evaluate case studies, visualizing findings, and extracting design strategies based on the analysis results. This method offers a new vision of how computational design can benefit informed decision-making processes and generate design alternatives aligned with a project’s design goals.
Article
Full-text available
In performance-based architectural design optimization, the design of building massings and façades is commonly separated, which weakens the effectiveness in performance improvement. In response, this study proposes a hybrid massing-façade integrated design generation and optimization workflow to integrate the two elements in an evolutionary design process. Compared with the existing approaches, the proposed workflow emphasizes the diversity of building design generation, with which various combinations of building massing forms and façade patterns can be systematically explored. Two case studies and a corresponding comparison study are presented to demonstrate the efficacy of the proposed workflow. Results show that the optimization can produce designs coupling the potential of building massings and façades in performance improvement. In addition, the optimization can provide information that supports early-stage architectural design exploration. Such information also enables the architect to understand the performance implications associated with the synergy of building massing and façade design.
Chapter
Full-text available
In performance-based architectural design, most existing techniques and design approaches to assisting designers are primarily for a single design problem such as building massing, spatial layouts, or facade design. However, architectural design is a synthesis process that considers multiple design problems. Thus, for achieving an overall improvement in building performance, it is critical to incorporate computational techniques and methods into all key design problems. In this regard, this paper presents a full-circle application of different computational design approaches and tools to exploit the potential of building performance in driving architectural design towards more novel and sustainable buildings as well as to explore new research design paradigms for performance-based architectural design in real-world design scenarios. This paper takes a commercial complex building design as an example to demonstrate how building performance can be incorporated into different building design problems and reflect on the limitations of existing tools in supporting the architectural design.
Conference Paper
Full-text available
The complex interactive relationship between wind flow and building design poses a great challenge in architectural design. Recent research has been conducted to combine Computational Fluid Dynamics (CFD) and computational design optimization to solve the problem. However, due to the time-consuming simulation process and the assertion-oriented computational optimization application, such CFD-based design optimization frameworks are not easy to integrate with architects' early-stage design exploration. To address these issues, this paper proposes an agile wind-driven building massing design optimization framework incorporating EvoMass and GH_Wind in the Rhino-Grasshopper environment. EvoMass is an integrated evolutionary building massing design tool, and GH_Wind is a simulation tool embedded with a Fast Fluid Dynamics (FFD) solver. Combining these two tools allows for fast wind-driven design optimization, thereby enabling architects to apply it to early-stage design exploration. To demonstrate its efficacy, a case study is presented to illustrate how the proposed design optimization framework can provide architects with useful design information and, thereby, facilitate more performance-informed design for early-stage architectural design.
Conference Paper
Full-text available
One of the problems lies in performance-based architectural design optimization is the separation of building massing design and facade design. The separation of design processes significantly weakens the synergizing of building massing and facades for more progressive performance improvement. In order to overcome this weakness, this paper presents a performance-based design optimization workflow combining facade with building massing design using an Evo-Devo approach. This workflow enables architects to make a rapid design exploration of different façade design schemes incorporating building massing design optimization. For demonstration, a case study is presented to show how this approach can facilitate early-stage architectural design exploration, and how the combination of the two factors can outperform the results produced by separated design processes.
Conference Paper
Full-text available
Architectural design optimization (ADO) is the notion of satisfying performance criteria within a constraint parameter space to find the best solution with the least trade-offs. This process tends to produce a vast amount of designs of varying quality and requires design space exploration (DSE) to review the myriad of designs so that an optimal candidate for design development can be chosen. This paper will explore different applied workflows for using optimization in an urban design competition, which seek to overcome the limitations of using optimization and DSE with a time constraint.
Conference Paper
Full-text available
Using performance-based optimization to explore unknown design solutions space has become widely acknowledged and considered an efficient approach to designing high-performing buildings. However, the lack of design diversity in the design space defined by the parametric model often confines the search of the optimization process to a family of similar design variants. In order to overcome this weakness, this paper presents two parametric massing generation algorithms based on the additive and subtractive form generation principles (food4rhino.com/node/2974). By abstracting the rule of these two principles, the algorithms can generate diverse building massing design alternatives. This allows the algorithms to be used in performance-based optimization for exploring a wide range of design alternatives guided by various performance objectives. Two case studies of passive solar energy optimization are presented to demonstrate the efficacy of the algorithm in helping architects achieve an explorative performance-based optimization process.
Conference Paper
Full-text available
This paper presents an evolutionary design toolkit for performance-based building massing design optimisation (food4rhino.com/node/2974). The toolkit is aimed to assist architects in exploring a wide range of building massing design alternatives guided by various performance objectives, thereby encouraging architects to incorporate evolutionary design optimisation for enriching design ideation at the outset of the design process. The toolkit is implemented in the Rhino-Grasshopper environment and includes components of a diversity-guided evolutionary algorithm and two pre-defined parametric models capable of generating a wide range of massing designs. The evolutionary algorithm can yield diverse design variants from the optimisation process and present more informative results with higher design differentiation. The pre-defined parametric models require minimal customisation from the architects. By using the toolkit, architects can readily explore high-performing building design with performance-based design optimisation with ease, and the coding-free optimisation workflow also streamlines the design process.
Article
Full-text available
Significant research has been undertaken focusing on the application of evolutionary algorithms for design exploration at conceptual design stages. However, standard evolutionary algorithms are typically not well-suited to supporting such optimization-based design exploration due to the lack of design diversity in the optimization result and the poor search efficiency in discovering high-performing design solutions. In order to address the two weaknesses, this paper proposes a hybrid evolutionary algorithm, called steady-stage island evolutionary algorithm (SSIEA). The implementation of SSIEA integrates an island model approach and a steady-state replacement strategy with an evolutionary algorithm. The combination aims to produce optimization results with rich design diversity while achieving significant fitness progress in a reasonable amount of time. Moreover, the use of the island model approach allows for an implicit clustering of the design population during the optimization process, which helps architects explore different alternative design directions. The performance of SSIEA is compared against other optimization algorithms using two case studies. The result shows that, in contrast to the other algorithms, SSIEA is capable of achieving a good compromise between design diversity and search efficiency. The case studies also demonstrate how SSIEA can support conceptual design exploration. For architects, the optimization results with diverse and high-performing solutions stimulate richer reflection and ideation, rendering SSIEA a helpful tool for conceptual design exploration.
Article
Many design versions are produced during design exploration. However, these versions are typically maintained manually often using an individual's self-determined logic, and thus hamper design workflow when a specific file cannot be located, similar ideas and mistakes revisited, or even loss of a potentially good design direction. These issues, in contrast, have been effectively addressed in software engineering by using a versioning system. In this work, we propose that integrating a versioning system into parametric design exploration workflow provides following benefits: (1) version control, (2) cross-iteration versions comparison, (3) feedback platform, and (4) design project analytics. A computational design workshop facilitated with our customised cloud-based ver-sioning tool revealed student's positive responses of versioning as a personal documentation and group collab-oration tool.
Article
For almost two decades, the Zero Energy Buildings (ZEB) standard has epitomized a commitment to the high energy performance of buildings. Nevertheless, the applicability of ZEB in hot climates is currently limited and furthermore, in light of the current limitations of traditional building energy modeling methods, new methods are necessary to effectively evaluate the energy balance potential of larger districts. To help bridge this gap, this paper introduces solar-based (both sun-hours and solar irradiation) and geometry-based prediction metrics to use in optimization studies to evaluate the impact of urban morphology on the energy balance of residential blocks in hot urban contexts. These prediction metrics are derived from the simulated energy performance of 1,944 parametric variations of residential blocks in Tel Aviv, which is then followed by a regression analysis in which these metrics recorded high correlation with energy demand, energy supply and the balance between them. To test the applicability of these metrics for optimization, the RBFMOpt method is employed in a multi-objective optimization study of the energy supply and demand of a nine-block residential district in Tel Aviv. Detailed energy simulations are performed for the best non-dominated results from the solar and geometric optimization studies and compared to the non-dominated results from a full energy optimization run. The results indicate that these metrics - the solar and geometric area-weighted exposure and shading indices - can serve as effective energy performance indicators to inform early stage morphological decisions making. This workflow promotes urban energy optimization towards more harmonized energy supply and demand driven approaches.