Download as pdf or txt
Download as pdf or txt
You are on page 1of 176

ALGORITHM-AIDED BUILDING INFORMATION MODELING

Connecting Algorithm-Aided Design and Object-Oriented Design

Harri Humppi
Tampere University of Technology
School of Architecture
Master’s Thesis 2015
ABSTRACT
TAMPERE UNIVERSITY OF TECHNOLOGY Currently, there are only a few studies that investigate AAB and
Master’s Degree Programme in Architecture most of them only focus on certain aspects related to AAB. The study
HUMPPI, HARRI: Algorithm-Aided Building Information Modeling: argues that valid research results can only be achieved by forming a com-
Connecting Algorithm-Aided Design and Object-Oriented Design prehensive image of the subjects that are related to AAB.
Master of Science Thesis, 164 pages This study has three main approaches. Firstly, the study is closely
November 2015 linked to design research of CAD because it looks into AAB from various
Major: Architecture perspectives of CAD. Secondly, the study is related to software research
Examiner: Professor Ilmari Lahdelma because it investigates many design approaches that should be taken into
Supervisor: Toni Österlund account, when AAB tools will be developed. Thirdly, the study uses prac-
tical design as learning method, so the study is related to researching
Keywords: BIM (Building Information Modeling), AAB (Algo- approach called ‘Research by Design’.
rithm-Aided Building Information Modeling), parametric modeling, The study is divided into three main parts that proceed from prac-
algorithm, parameter. tice to theory. Part A focuses on design solutions that AAB tools can gen-
erate. Part B examines modeling, simulation and optimization methods
New design methods have induced an undergoing transition from analog and processes that are related to AAB. Part C investigates the theory of
to digital design methods. This transition has started only some decades digital design that is seen as a starting point for the further development
ago and the study affirms that there are still many significant reforms of AAB.
to come. New innovations can only arise when stabilized and standard The main result of the study is that Algorithm-Aided Building
processes are rethought and challenged by new ways of working. On this Information Modeling (AAB) can connect Algorithm-Aided Design
basis, this study examines a new approach to Computer-Aided Design (AAD) to Object-Oriented Design (OOD). After all, the result is rather
(CAD). In this study, this new approach is called Algorithm-Aided Build- obvious but this connection opens many new possibilities to CAD. The
ing Information Modeling (AAB). connection will enable many design processes that haven’t been possi-
The starting point of the study is the awareness that AAB tools ble before.
somehow bridge Algorithm-Aided Design (AAD) and Building Infor-
mation Modeling (BIM), so this study started by investigating this rela-
tion. The study examines if AAB could reshape old digital design meth-
ods by offering a novel approach to CAD.

i
TIIVISTELMÄ
TAMPEREEN TEKNILLINEN YLIOPISTO Tällä hetkellä on vain muutamia tutkimuksia, jotka kohdistuvat
Arkkitehtuurin koulutusohjelma algortimiavusteiseen tietomallintamiseen ja suurin osa niistä keskittyy
HUMPPI, HARRI: Algoritmiavusteinen tietomallintaminen: Algoritmi- vain tarkoin rajattuihin algortimiavusteisen tietomallintamisen osa-alu-
avusteisen ja olio-pohjaisen suunnittelun yhdistäminen eisiin. Selvitys väittää että jatkotutkimusten kannalta on tärkeää muo-
Diplomityö, 164 sivua dostaa kokonaisvaltainen kuva aiheista, jotka ovat kytköksissä algorit-
Marraskuu 2015 miavusteiseen tietomallintamiseen.
Pääaine: Arkkitehtuuri Tutkielmalla on kolme tutkimuksellista päälähestymistapaa. Ensin-
Tarkastaja: Professori Ilmari Lahdelma näkin tutkielma on läheisesti kytköksissä tietokoneavusteisen suunnitte-
Ohjaaja: Toni Österlund lun tutkimukseen, sillä tutkielma lähestyy algoritmiavusteista tietomal-
lintamista monista näkökulmista. Toisaalta tutkielma on kytköksissä
Avainsanat: Tietomallintaminen (BIM), Algoritmiavusteinen tietomal- ohjelmistojen tutkimukseen, sillä tutkielma perehtyy moniin näkökul-
lintaminen (AAB), parametrinen mallintaminen, algoritmi, parametri. miin, jotka tulisi ottaa huomioon kun algortimiavusteisia työkaluja kehi-
tetään. Tutkielma hyödyntää myös käytännönläheistä suunnittelua oppi-
Uudet suunnittelumenetelmät ovat aikaansaaneet murroksen analogi- mismenetelmänä, joten tutkielma on verrattavissa ‘Research by Design’
sista digitaalisiin suunnittelumenetelmiin. Tämä murros on alkanut vain tutkimustapaan.
muutamia kymmeniä vuosia sitten ja tutkielma vahvistaa että digitaali- Tutkielma on jaettu kolmeen pääosaan, jotka etenevät käytännöstä
siin suunnittelumenetelmiin on vielä tulossa monia merkittäviä uudis- teoriaan. Osa A keskittyy suunnitteluratkaisuihin, joita voidaan tuottaa
tuksia. Uudet innovaatiot voivat nousta vain kyseenalaistamalla tällä algoritmiavusteisen suunnittelun kautta. Osa B tarkastelee mallinnus-,
hetkellä vakiintuneet prosessit. Tällä perusteella, tutkielma tutkii yhtä simulointi- ja optimointimenetelmiä ja prosesseja algoritmiavusteiseen
uutta lähestymistapaa tietokone-avusteiseen suunnitteluun (CAD, Com- tietomallintamisen kannalta. Osa C tutkii digitaalisen suunnittelun teo-
puter-Aided Design). Tässä selvityksessä, tätä uutta lähestymistapaa kut- riaa, joka nähdään tutkielmassa algoritmiavusteisen tietomallintamisen
sutaan nimellä algoritmiavusteinen tietomallintaminen (AAB, Algori- päälähtökohtana.
thm-Aided Building Information Modeling). Tutkielman tärkein tutkimustulos on tieto siitä että algoritmiavus-
Tutkielman lähtökohtana on että algoritmiavusteiset tietomallin- teinen tietomallintaminen luo uuden yhteyden olio-pohjaisen suunnitte-
tamistyökalut voivat yhdistää algoritmiavusteisen suunnittelun (AAD, lun (OOD, Object-Oriented Design) ja algoritmiavusteisen suunnittelun
Algorithm-Aided Design) ja tietomallintamisen (BIM, Building Infor- välille. Kaiken kaikkiaan tämä tutkimustulos on melko ilmiselvä, mutta
mation Modeling), joten tutkielma alkoi perehtymällä tähän suhtee- se avaa monia uusia mahdollisuuksia tietokoneavusteiseen suunnitte-
seen. Tutkielma selvittää voisiko algoritmiavusteinen tietomallintami- luun. Löydetty yhteys mahdollistaa monia suunnitteluprosesseja, jotka
nen uudistaa vanhoja digitaalisen suunnittelun menetelmiä tarjoamalla eivät ennen ole olleet mahdollisia.
uudenlaisen lähestymistavan tietokoneavusteiseen suunnitteluun.
ii
FOREWORD
The main motivation of this study is simply the need to find out how (2013, pp. 186–195). Thanks to them, the right studying approach was
things are and how they should be in the field of CAD. However, this study found. These concepts are explained as a part of this study.
has been a great challenge for me. During the study, my mood has varied The main objective of the study is not only to concentrate on the
from hopefulness to hopelessness. This is mostly due to the challenging subject but also to gain a comprehensive understanding of digital design.
objective that aims to connection of two digital design approaches. These Thus, this study investigates many themes that have been unknown to
design approaches partially overlap but on the other hand they have the author. Known themes have been the basis for the study, but the main
developed from distinctive premises and towards distinctive objectives. focus has been in the investigation of ambiguous themes. For me, the
All in all, the study shows that the border areas of distinctive fields studying process itself has been more important than the result of the
are probably the best platforms for emerging developments. It is reas- study.
suring to note that when the development has steered towards a certain I would like to point out the approach that this study offers into
objective the development will sooner or later happen. When the devel- the theme and design in general. This study argues that by understand-
opment has happened there is rarely need for turning back. Hopefully, ing the approach the theme can be understood more comprehensively,
some day these two approaches of CAD can interact more fluently. which is essential if we want to use and develop it in the future.
The process of this study has led to think many aspects of the sub- The study emphasizes that comprehensive knowledge of terms has
ject. At the beginning of the study the main intention was to find out how led the study to right direction. Misunderstanding of some terms has
new digital design tools work and what they are able to produce. Testing challenged the study significantly, so the struggle for finding and under-
has also been mapping the problems and advantages of the tools used. standing terms has taken much effort. Many terms have been unclear and
There is often a need to examine the limitations of tools that they could even unknown, so a comprehensive glossary is attached to the study.
be used creatively. On this basis, the main turning point in the study was
to note that this point of view was not fully successful. Instead of study-
ing what tools can do, the study should aim to examine why these tools
should be able to do something, and if these new tools correspond to this
need.
The keys for understanding this reversed attitude and design in
general, have been the concepts of ‘the design space’ and ‘the solution
space’. These terms and their relations to design processes and design
tools have been examined by Benjamin (2012, pp. 14–25) and by Scheurer

iii
THE MAIN DIAGRAM
HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?
Figure 1  THE MAIN DIAGRAM
The Figure presents a colored preview of the main diagram in Figure 9.
The red terms are the ones from which the study has begun. iv
MAIN TERMS
Here are listed the main terms that are used or mentioned in the study. geometry. Also known as: Generative BIM, Parametric BIM. Like BIM,
Other terms are listed at the end of the study. The study suggests to read AAB refers also to the result of an AAB process: Algorithm-Aided Build-
the main terms and checking the others if needed. ing Information Model.

Remarks: AAD
*Proposal by the author - the term is not commonly and unambiguously Algorithm-Aided Design. An unsettled term referring to algorithm-aided
adopted. design which doesn’t utilize objects. Used geometry consists only of
generic geometry which doesn’t contain semantic data. The term AAD
appears at least in the following publications: Österlund (2013a; Öster-
MAIN TERMS lund, 2013b), Tanska and Österlund (2014), http://www.adaptiveurban-
lighting.fi/; Tedeschi (2014). Also known as: Algorithmic Design, Para-
CAD metric Design (PD) and Generative Design.
Computer-Aided design. Term refers generally to all computer-aided
design. As a result of historical development, the term is sometimes used AAM*
to refer only to computer-aided drawing methods in 2D. Also known as: Algorithm-Aided Modeling. Refers to modeling methods that are defined
Digital Design. through scripting. Also known as: Parametric Modeling.

CAAD AAA*
Computer-Aided Architectural Design. Term referring only to the archi- Algorithm-Aided Analysis. Refers to analysis methods that are defined
tectural part of CAD. Compared to CAD software CAAD software con- through scripting.
tain features that support architectural design. For example, Archicad,
Autodesk Architecture and Revit can be considered to be CAAD soft- AAO*
ware. Algorithm-Aided Optimization. Refers to optimization methods that are
defined through scripting.
AAB*
Algorithm-Aided Building Information Modeling. An introduced term AAS*
referring to algorithm-aided building design, which utilizes BIM pro- Algorithm-Aided Simulation. Refers to simulation methods that are
cesses and objects. Objects contain parameters that are integrated to the defined through scripting.

v
BIM OOA
Building Information Modeling. A process of computer-aided design Object-Oriented Analysis. A design method that is based on analysis in
which utilizes objects for modeling. Thus, BIM is Object-Oriented Mod- object-oriented design tools. “The main difference between object-ori-
eling (OOM) method. Objects are controlled with parameter data. ented analysis and other forms of analysis is that by the object-oriented
approach we organize requirements around objects, which integrate both
Geometry behaviors (processes) and states (data) modeled after real world objects
A geometrical entity that doesn’t contain data, except the geometry itself. that the system interacts with. In other or traditional analysis method-
ologies, the two aspects: processes and data are considered separately.”
GB* (https://en.wikipedia.org/)
Geometry-based. Refers to issues that are geometry-oriented. In this
study such issues are geometry-oriented modeling and simulation. Cf. OOD
OO. “Object-Oriented Design is the process of planning a system of inter-
acting objects for the purpose of solving a software problem. It is one
GBD* approach to software design.” (https://en.wikipedia.org/)
Geometry-Based Design. A computer aided design approach that uses
geometry in modeling. OOM
Object-Oriented Modeling. A design method that is based on modeling
Object in object-oriented design tool.
“In the class-based object-oriented programming paradigm, “object”
refers to a particular instance of a class where the object can be a com- OOS
bination of variables, functions, and data structures” (https://en.wikipe- Object-Oriented Simulation. A design method that is based on simula-
dia.org/). An object contains data (geometry, parameters, relationships) tion in object-oriented design tool.
that can be modified parametrically. Also known as: Element.
OOP
OO Object-Oriented Programming. “Object-oriented programming (OOP)
Object-oriented. Refers to issues that are object-oriented. In this study is a programming paradigm based on the concept of “objects”, which are
such issues are Object-Oriented Design (OOD), Object-Oriented Pro- data structures that contain data, in the form of fields, often known as
gramming (OOP), Object-Oriented Modeling (OOM) and Object-Ori- attributes; and code, in the form of procedures, often known as meth-
ented Analysis (OOA) and Object-Oriented Simulation (OOS). Cf. GB. ods.” (https://en.wikipedia.org/)

vi
TABLE OF CONTENTS
ABSTRACT
TIIVISTELMÄ
FOREWORD
THE MAIN DIAGRAM
MAIN TERMS
TABLE OF CONTENTS

1. INTRODUCTION TO THE THEME����������������������������������������������������������������������� 1


1.1. THREE ERAS OF CAD����������������������������������������������������������������������������������� 3
1.1.1. The 2d Drafting era����������������������������������������������������������������������������������� 5
1.1.2. The Bim era����������������������������������������������������������������������������������������������� 5
1.1.3. The Design Computing era���������������������������������������������������������������������� 9

2. INTRODUCTION TO THE STUDY���������������������������������������������������������������������� 13


2.1. TERMS OF THE DESIGN ��������������������������������������������������������������������������� 15
2.1.1. Purpose of the study������������������������������������������������������������������������������� 17
2.1.2. Clarification of the main terms�������������������������������������������������������������� 17
2.1.3. Other terms �������������������������������������������������������������������������������������������� 19
2.2. METHODS OF THE STUDY ���������������������������������������������������������������������� 21
2.2.1. Studying process������������������������������������������������������������������������������������� 21
2.2.2. Studying tools����������������������������������������������������������������������������������������� 21
2.2.3. Sources of inspiration����������������������������������������������������������������������������� 23

3. GIGA-MAPPING����������������������������������������������������������������������������������������������������� 27
3.1. CONCEPTS OF OOD AND AAD��������������������������������������������������������������� 29

vii
A
4. THE TEST PROJECT����������������������������������������������������������������������������������������������� 33
PRACTICE 4.1. THE MODELING WORKFLOW����������������������������������������������������������������� 35
4.2. THE DIGITAL DESIGN PROCESS�������������������������������������������������������������� 39
4.3. SUMMARY���������������������������������������������������������������������������������������������������� 39

5. CURRENT PROJECTS�������������������������������������������������������������������������������������������� 43
What?

5.1. FONDATION LOUIS VUITTON���������������������������������������������������������������� 45


5.2. CITY-LIFE RESIDENCES����������������������������������������������������������������������������� 47
5.3. LOUISIANA STATE MUSEUM AND SPORTS HALL OF FAME�������������� 49

B
6. INTEROPERABILITY��������������������������������������������������������������������������������������������� 53
6.1. PRODUCT DATA MODEL EXCHANGE FORMATS�������������������������������� 55
6.2. TYPES OF DATA EXCHANGE ������������������������������������������������������������������� 57
6.2.1. Direct Data Exchange����������������������������������������������������������������������������� 57
6.2.2. Indirect Data Exchange�������������������������������������������������������������������������� 59
6.2.3. Shared Mapping Process ������������������������������������������������������������������������ 59
PRACTICE

7. PARAMETRIC MODELING����������������������������������������������������������������������������������� 61
THEORY /

7.1. BIM PROCESSES������������������������������������������������������������������������������������������ 63


7.1.1. Advantages and drawbacks of BIM�������������������������������������������������������� 65
7.2. AAM PROCESSES���������������������������������������������������������������������������������������� 67
How?

7.2.1. Advantages and drawbacks of AAM������������������������������������������������������ 69


7.3. AAB PROCESSES������������������������������������������������������������������������������������������ 69
7.3.1. Creation of new objects�������������������������������������������������������������������������� 71 viii
7.3.2. Modification of existing objects������������������������������������������������������������� 71
7.3.3. Advantages and drawbacks of AAB������������������������������������������������������� 73
7.4. AAB TOOLS�������������������������������������������������������������������������������������������������� 75
7.4.1. Rhino-Grasshopper-Archicad Connection������������������������������������������� 76
7.4.2. Grevit������������������������������������������������������������������������������������������������������ 77
7.4.3. Lyrebird��������������������������������������������������������������������������������������������������� 78
7.4.4. VisualARQ���������������������������������������������������������������������������������������������� 79
7.4.5. Elefront��������������������������������������������������������������������������������������������������� 80
7.4.6. Geometry Gym��������������������������������������������������������������������������������������� 81
7.4.7. Dynamo for Revit����������������������������������������������������������������������������������� 82

8. SIMULATION AND OPTIMIZATION������������������������������������������������������������������ 83


8.1. SIMULATION����������������������������������������������������������������������������������������������� 85
8.2. OPTIMIZATION������������������������������������������������������������������������������������������ 87
8.2.1. Heuristic methods���������������������������������������������������������������������������������� 87
8.2.2. Metaheuristic methods��������������������������������������������������������������������������� 89
8.3. SIMULATION PROCESSES IN OOD���������������������������������������������������������� 89
8.4. SIMULATION AND OPTIMIZATION PROCESSES IN AAD������������������ 91
8.5. AAB AND SIMULATION AND OPTIMIZATION PROCESSES�������������� 93
8.6. TESTING SIMULATION PROCESSES������������������������������������������������������� 95
8.6.1. Structural analysis���������������������������������������������������������������������������������� 95
8.6.2. 4D BIM and 5D BIM������������������������������������������������������������������������������ 95
8.6.3. Clash detections�������������������������������������������������������������������������������������� 95
8.7. TESTING OPTIMIZATION PROCESSES��������������������������������������������������� 96
8.7.1. Optimization: form finding ������������������������������������������������������������������� 96
8.7.2. Structural optimization�������������������������������������������������������������������������� 96

ix
C
9. THEORY OF DIGITAL DESIGN�������������������������������������������������������������������������� 109
9.1. COLLABORATION - ARCHITECT AND ENGINEER��������������������������� 113
9.2. THE DESIGN SPACE AND THE SOLUTION SPACE������������������������������ 115
9.3. DESIGN EXPLORATION AND EXPLOITATION����������������������������������� 117
THEORY
9.4. CONTROL OR NON-CONTROL�������������������������������������������������������������� 117
9.5. FROM STATIC TO DYNAMIC DESIGN PROCESS��������������������������������� 119
Why?
9.6. PRE- AND POST-RATIONALIZATION��������������������������������������������������� 121
9.7. STANDARDIZATION OR CUSTOMIZATION��������������������������������������� 123
9.8. AAB IN HYBRID PROCESS���������������������������������������������������������������������� 125

10. SUMMARY AND CONCLUSIONS�������������������������������������������������������������������� 127


10.1. SUMMARY������������������������������������������������������������������������������������������������ 129
10.1.1. C. Why should AAB be developed? �������������������������������������������������� 129
10.1.2. B. How to use design methods of AAB?�������������������������������������������� 131
10.1.3. A. What can AAB produce in practice?��������������������������������������������� 132
10.2. CONCLUSIONS���������������������������������������������������������������������������������������� 133
10.2.1. Combining OOD and AAD��������������������������������������������������������������� 133
10.2.2. Two levels of object standards������������������������������������������������������������ 133
10.2.3. Suggestions for further studies����������������������������������������������������������� 133
10.3. FINAL COMMENT���������������������������������������������������������������������������������� 135

GLOSSARY������������������������������������������������������������������������������������������������������������������ 137
TERMS
TOOLS

SOURCES�������������������������������������������������������������������������������������������������������������������� 153
REFERENCES
FIGURES x
1. INTRODUCTION TO THE THEME

1
2
This study focuses on two distinct building design approaches called
Building Information Modeling (BIM) and Algorithm-Aided Design
(AAD). Currently, BIM and AAD are main approaches of digital design.
BIM is introduced first because the development of BIM has proceeded
wider than in AAD. The study investigates the combination of these two
design approaches. These two building design approaches have been
practically separate until recent years. There has risen an interest to com-
bine them which could lead to more fluent building design processes.
AAD and BIM don’t take full advantage of each other, so the interopera-
bility could be strengthened by improving the integration between these
two methods (Boyekens, 2013, p. 1). Algorithm-Aided building Informa-
tion Modeling (AAB) is the selected term to describe the combination of
BIM and AAD. The main target of the study is to examine AAB because
it could enable processes that have been before impossible.

1.1. THREE ERAS OF CAD


In order to understand digital design approaches, the historical back-
ground of them, presented in Figure 2, needs to be understood. The fig-
ure shows that the history of practical Computer-Aided Design (CAD)
can be split into three eras: 2D Drafting era, Building Information era
and Design Computation era (Aish, 2013, pp. 40–41). The diagram also Figure 2  THREE ERAS OF CAD - FLOWS OF INFORMATION
shows that BIM, AAD and AAB are subsets of Computer-Aided Design
(CAD). The figure introduces the main data-flows from design to con- The diagram presents data-flows from design to construction. The lower
struction. In reality, these eras and data-flows are overlapping. gray arrow shows the ordinary data-flow in AAM that is used in AAD.
The red arrow shows the flow in AAB. The diagram is adapted from the
one made by Aish (2013, p. 40), but some changes are made so that the
diagram fits the purposes of the study.

3 CHAPTER 1: INTRODUCTION TO THE THEME


Era Generative Data Generated Data Derived Data Reality
THE 2D DRAFTING ERA
2D Drawings (multiple,
independent, uncoordinated)

THE BIM ERA


Performance simulation Building

Inherently via drawing


2D Drawings (Consistent; as interpretation and craft
BIM (”manually” modeled)
Building Information Modeling derived from the model) construction
(BIM)

Analytical composite models

THE COMPUTATIONAL DESIGN ERA

Script Object model


Algorithm-aided Building Information
Modeling Performance simulation Performance optimization
(AAB)
Performance simulation Performance optimization

Review only
Script Geometrical model

Fabrication Data Building


Algorithm-Aided Modeling (geometric normalisation)
(AAM) Direct manufacturing
and consturction

CHAPTER 1: INTRODUCTION TO THE THEME 4


1.1.1. The 2d Drafting era
The 2D CAD era started in the 1980s, when the first computers were
adopted to present building designs as multiple 2D drawings. The only
essential change was that computers partially replaced traditional paper
drawings. However, the CAD era failed to utilize the creative potential
of computers. (Aish, 2013, pp. 40–41)

1.1.2. The Bim era


The development of BIM started already in the 1980s (Aish, 2013, p. 43;
Garber, 2014, p. 14) and nowadays BIM is a commonly used method for
controlling building design processes. BIM provides the ability to digi-
tally coordinate often complex building processes from design to actual
construction (Garber, 2014, p. 14). Nowadays, BIM has reached a stable
position in the field of building design. It has also overcome many lim-
itations of 2D CAD drawing.
According to The National Building Information Model Standard
Project Committee (https://www.nationalbimstandard.org/) the defini-
tion of BIM is following: “Building Information Modeling (BIM) is a dig- Figure 3  BUILDING INFORMATION MODELING (BIM)
ital representation of physical and functional characteristics of a facility.
A BIM is a shared knowledge resource for information about a facility The figure presents an example project, which is modeled with a BIM
forming a reliable basis for decisions during its life-cycle; defined as exist- software. The result can be called as a Building Information Model
ing from earliest conception to demolition.” (BIM). The project consists of objects which can be exported to Industry
Foundation Classes (IFC). As we can see, the project consists mainly of
According to Eastman et al. (2008, p. 17) the main benefits of BIM wall and floor objects.
compared to CAD in the design stage are: Beyond the project was a desire to design a project that takes full
– Earlier and More Accurate Visualizations of a Design advantage of BIM tools. The project showed that the model is challenging
– Automatic Low - Level Corrections When Changes Are Made to Design to update because BIM software are not capable to generate fast itera-
tions.

5 CHAPTER 1: INTRODUCTION TO THE THEME


CHAPTER 1: INTRODUCTION TO THE THEME 6
– Generate Accurate and Consistent 2D Drawings at Any Stage of the
Design
– Earlier Collaboration of Multiple Design Disciplines
– Easily Check against the Design Intent
– Extract Cost Estimates during the Design Stage
– Improve Energy Efficiency and Sustainability

As mentioned in the definition, BIM processes utilize the idea of


shared information model, which unites all the disciplines related to the
design process. BIM focuses on parametric building components that
are used to design the 3D object model. An example model is presented
in Figure 3.
Representational 2D-drawings can be extracted from the model.
The model consists of parametric objects, but it also can also contain
geometry that doesn’t utilize parameters. The behavior of objects is
defined by all exposed parameters. Objects are scripted inside of the
host software. That’s why BIM software limit the amount of modeling
with user-defined algorithms. (Boeykens, 2012, p. 453.) Figure 4  MATURITY DIAGRAM OF BIM
In order to understand BIM more comprehensively The Maturity The Maturity Diagram of BIM has four levels. Level 0 is based on general
Diagram is presented in Figure 4. The diagram is commonly used to 2D CAD processes. The diagram shows that BIM can be divided into
describe the development and current adoption of BIM. The diagram was Data and Process managements which contain common BIM standards
originally prepared by Mervyn Richards and Mark Bew in 2008 (Sinclair, and processes. Level 1 utilizes both 2D and 3D information in projects.
2012, p. 3). The diagram has four levels (level 0, 1, 2 and 3). Level 1 also acknowledges the need for data management processes. On
The diagram shows that BIM is based on standardization of design Level 2 all disciplines use their own 3D models. The collaboration is
and construction processes. The diagram also acknowledges that both based on information exchange between disciplines by using common
Data and Process management are the main sectors of BIM (Sinclair, file formats such as IFC. On Level 3 all disciplines work with the same
2012, p. 3). The Maturity Model doesn’t suggest to have any relationship shared project model. The risk for conflicting information disappears.
(RIBA, 2012, p. 3.) Modified from (RIBA, 2012, p. 3).

7 CHAPTER 1: INTRODUCTION TO THE THEME


LEVEL 3

Asset life cycle management


LEVEL 2 Data management
iBIM
LEVEL 1
BIMs
Standards for
interoperability:

BSIM
BrlM
2D 3D
AIM

FIM
SIM
LEVEL 0 IFC, IFD, IDM

CPIC ISO BIM


Avanti Process management
CAD BS 1192:2007
User guides CPIC, Avanti, BSI
Source: Bew and Richards 2008

CHAPTER 1: INTRODUCTION TO THE THEME 8


with AAD because these two fields are seen as separate and alternative
ways for building design.
BIM software have some advantages compared to traditional 2D
CAD software. Firstly, Building Information Models (BIMs) are struc-
tured with semantic information. Semantic information means data,
which is embedded into the property values of building objects. In BIM
the semantic structure is predefined into components. Secondly, BIM
software are logical to control because they are structured on multiple
levels. BIM software can produce all of the needed documentation for
different stages of a project. (Boeykens, 2012, pp. 453–454.)
On the other hand, BIM software are criticized because they use
predefined tools to generate buildings (Peters and Peters, 2013, p. 9). In
addition, BIM utilizes parametric functionality mostly on building object
level. Thus, the model behaves like an assembly of independent objects.
(Boeykens, 2012, p. 454.) Thirdly, BIM software don’t utilize the full com-
putational and generative power of computers. Thus, utilization of algo-
rithms can expand the field of BIM.
Figure 5  ALGORITHM-AIDED DESIGN (AAD)
1.1.3. The Design Computing era
The design computing era introduced the distinction between genera- The exploded view of a school project represents an example project,
tive description of the building and the generated model (Aish, 2013, pp. which utilizes methods and processes of AAD. The geometry doesn’t
43–45). Scripts and graphs are used to generate buildings. Iterative exe- contain any information except the geometry itself, so the geometry and
cutions of the script enables the exploration of many design alternatives. other data are exchanged separately with other design tools.
The study begun by dividing The Design Computing Era into AAD and Beyond the project was a desire to design a project that consists
AAB, depending on the algorithmic modeling method. The study will only of geometry produced through algorithm-aided design. The project
eventually show why this starting point was not fully sufficient. showed that large scale algorithm-aided projects are challenging to
In Algorithm-Aided Modeling Design (AAD) algorithms are used control, if the designer loses control of the script. In AAD, all objects are
to model the geometry. In Algorithm-Aided Building Information Mod- built “from scratch”, which in large projects leads to complex scripts.

9 CHAPTER 1: INTRODUCTION TO THE THEME


CHAPTER 1: INTRODUCTION TO THE THEME 10
eling (AAB) algorithms are used to model objects. At the moment AAD
is slowly consolidating its position in building design. Instead, the devel-
opment of AAB, on the other hand, has only recently started.
Can be stated that AAD has been an avant-garde movement which
has started to stabilize it’s position as a building design method. The main
focus in AAD is the use of algorithms to generate the geometry. AAD
focuses on modeling the geometry which doesn’t contain any informa-
tion except the geometry itself. An example model is presented in Fig-
ure 5. The flow of information moves inside of the script but the result-
ing geometry doesn’t contain any parameter information.
At the beginning of the study, AAB is seen as a combination of
AAD and BIM. The study supposes that this combination could offer new
possibilities for building design because it produces objects as a result of
modeling process.
All in all, the development of AAB is closely linked to the devel- Figure 6  ALGORITHM-AIDED BUILDING INFORMATION
opment of digital tools and especially to the tools that allow integration MODELING (AAB)
of BIM and AAD. Around AAD has developed a lively user community, The figure presents an example project, which utilizes methods and
which shares information, develops new tools and helps to find solutions processes of AAB. The final model contains only objects. Thus, the infor-
to problems that designers confront. AAB plugins for Grasshopper and mation can be further utilized in design processes. Objects are based on
Dynamo represent the latest development of AAB. An example model IFC (Industry Foundation Classes), so building objects can be exported
is presented Figure 6. to other software.
Currently the mainstream of AAD doesn’t utilize objects, which Beyond the process was a desire to design a project that consists
would allow more interactive design processes. There is lot of literature only of objects that are produced through algorithm-aided design. The
on AAD and BIM separately, but literature of the combination of these project shows that the development of these kind of tools has just begun.
subjects barely exists. In contrast, some internationally renowned archi- The figure is divided into three parts. The left one shows the structural
tects and building designers have done some building projects, which wire-frame model that is used to control and host objects. The part in the
take advantage of AAB. That’s why the study argues that additional study middle shows object types and the right part shows a visualization of the
of AAB is needed. final geometry.

11 CHAPTER 1: INTRODUCTION TO THE THEME


CHAPTER 1: INTRODUCTION TO THE THEME 12
2. INTRODUCTION TO THE STUDY

13
14
This study investigates CAD generally as well as its relation to AAB and ‘design solution’. The main objective of the section is to clarify these terms
CAD. In this sense the study is design research. The focus of the study because the study investigates them in the context of CAD. Regarding
is not only in design but also in the design methods and in design tools. AAB, the meaning of these terms has not settled, so this study investi-
All in all, BIM and AAD are relatively well known. Instead, the informa- gates their relation to AAB. In this sense, the study is a concept analysis.
tion on AAB is scattered or it doesn’t exist at all. Only some books and As we will see, the relation of these concepts is not always unambiguous.
articles pay attention to AAB and the main priority is often in BIM or In order to understand the study and these design terms more com-
AAD. Theoretical literature of AAB is still missing, so there is a real need prehensively, Figure 7 presents them and their relations conceptually. In
to investigate AAB more comprehensively. the figure the main design terms and their relations are clarified.
This study argues that especially during the development of new To relate these terms to CAD, each of them is described shortly.
tools we should be aware of why we are actually developing them. Thus, Designer in the context of building design is usually the designer, an
the study of AAB shouldn’t be only an exploration of current situation, architect or an engineer or other member of the design team. Of course,
which may naturally focus on the investigation of current tools of AAB. in the collaborative design the design team designs together.
The study suggests that if we aim to develop design processes related to Design contains all aspects that are related to the design event. In
AAB, the theoretical context of AAB should be investigated comprehen- practice, different design approaches can be combined in the same design
sively. process. For example, AAD can be combined with traditional design
The paragraph by Sevaldson (2005, p. 32) has the same approach methods like drawing. Design is seen to be conceptually dependent on
that the study calls for: “Processes become what we make them become. external and internal influences. For example laws and regulations are
They are not there as stable phenomena to be observed and explored seen as external influences that affect to the design. In the design pro-
by means of one or other technique. They are themselves emergent cess the main external influences are the building industry and assembly
and adaptive, evolutionary, reflexive and a knowledge-building activ- because they affect particularly into digital design. Other less important
ity. Design research in this sense is about initiating change rather than aspects are not taken into account.
observing and understanding what is there, or solving preset problems.” Design methods (or methodologies) refer to actions that happen in
the design process. More specifically, they describe the interaction that
happens between the designer and the tool. In the study, design methods
2.1. TERMS OF THE DESIGN refer to methods that are used in CAD. The most essential design meth-
In order to start the study, this section reminds the reader of some basic ods in the study are modeling, analysis, simulation and optimization.
design terms that are strictly related to the design process. These terms All in all, the design process is a series of methodological steps that the
are ‘designer’, ‘design’, ‘design process’, ‘design method’, ‘design tool’ and designer takes in order to find a sufficient solution. The term ‘method’

15 CHAPTER 2: INTRODUCTION TO THE RESEARCH


ITERATIVE

HUMAN MIND
DESIGN PROCESS

COMPUTER
OTHER DESIGN
APPROACHES

EXTERNAL
INFLUENCES

DESIGN
TECNIQUES

DESIGN DESIGN DESIGN


DESIGNER DESIGN
METHODS TOOLS SOLUTION

DESIGN AND MODELING PROCESS

PRACTICE
THEORY

A. B. C.
WHY?, HOW?, WHAT? WHY?, HOW?, WHAT? WHY?, HOW?, WHAT?
DIRECTION OF THE PRESENTATION
DIRECTION OF THE INVESTIGATION

Figure 7  THE MAIN DIAGRAM OF DESIGN CHAPTER 2: INTRODUCTION TO THE RESEARCH 16


or methodologies basically refers to the same thing as the term ‘design On this basis, it can be stated that the study conceptually seeks answers
strategy’ (eg. Sevaldson, 2005, p. 31). The study prefers to use the term to three questions: why, how and what. These questions are also pre-
‘design method’ if needed. sented in Figure 7. These research question are related to three main
In the study design techniques are seen to be ways of using design parts of the study.
methods. For example, commonly used paneling is technique that is used
in modeling. There are a wide range of different design techniques, so 2.1.2. Clarification of the main terms
they are not presented in the diagrams the study introduces. However, The preceding clarification of the main design concepts is required for
it’s essential to understand that design methods can be used many ways the abbreviations BIM, AAD and AAB. In reality there doesn’t exist set-
by using different design techniques. tled terms, which would describe AAD and AAB (See: Glossary). Those
Design tools are used to generate the design solution. In the field terms are used because they describe well the character of these two
of CAD, tools are usually software, add-ons, plugins or self made scripts. design approaches. In addition, they can be derived from the term CAD
In order to emphasize the relation between the designer and the solu- because all of them contain the word ‘aided’, which refers to the aiding
tion, the circle of design solution is detached from the center in Figure role of computers and algorithms in design processes. All in all, design
7. and research would benefit from stable terms.
Lastly, the design process is the time-sensitive event that leads the BIM has manifold meaning because it refers to the design solu-
designer to the final solution. All of the terms mentioned are needed to tion (Building Information Model) and to the modeling method (Build-
understand the design process. Often the design process is not linear, so ing Information Modeling). However, BIM is mainly used to refer to the
iterations are needed to find the final solution. process (e.g. Eastman et al., 2008, p. vii) because it primarily describes
the activity in the design process.
2.1.1. Purpose of the study AAD (Algorithm-Aided Design) refers literally to design that uti-
As mentioned the study investigates AAB comprehensively form differ- lizes algorithm-aided design methods and tools to search for the design
ent points of view. The main research questions that support this objec- solution. In AAD the focus shifts from the design solution to the design
tive are: process, so maybe that is why the design solution in algorithm-aided
design doesn’t have settled term like BIM (Building Information model).
A. What can AAB produce in practice? The design solution of an algorithm-aided design process is called with
B. How to use design methods of AAB? various generally used terms because the design solution itself is not algo-
C. Why should AAB be developed? rithmic. For example, ‘model’, ‘geometry’ and ‘building’ can refer to the
design solution.

17 CHAPTER 2: INTRODUCTION TO THE RESEARCH


Figure 8  THE RELATION OF BIM AND AAD

Various conceptual options were considered to clarify the relation


between BIM, AAD and AAB. The first (1.) option failed because it pres- BIM AAB AAD
ents AAB as an independent field of modeling. AAB is not independent,
it is more like a combination of BIM and AAD.
The second option (2.) failed because BIM refers to modeling and
AAD refers to design, so these concepts were not equal. The study consid-
ered BIM as a “design method” until the concept of OOD was found. 1.
The study ended up using the third option (3.) because it describes
the main themes well. AAB is connecting OOD and AAD. OOD and
AAD both refer to design so they are equal.
BIM AAB AAD
MODELING
BIM (Building Information Modeling)
AAM (Algorithm-Aided Modeling)
AAB (Algorithm-Aided Building Information Modeling) 2.

DESIGN
OOD (Object-Oriented Design)
AAD (Algorithm-Aided Design)

OOD AAB AAD

3.

CHAPTER 2: INTRODUCTION TO THE RESEARCH 18


As BIM, AAB (Algorithm-Aided Building Information Model- The term ‘parametric modeling’ is often used separately to refer
ing) refers to the modeling process. Just like BIM, AAB can also be used either to BIM or AAD. This study argues that the term should be used to
to refer to the model (Algorithm-Aided Building Information Model). describe all modeling methods that use parameters to control the geom-
However, in the study AAB is mainly used to refer to the modeling pro- etry. Figure 32 shows that many modeling methods are based on model-
cess. This study suggests that AAB cannot yet to be related to design, ing with parameters.
because it is more like connecting two design approaches than an inde- Figure 10 shows the difference of an object and geometry. In this
pendent approach to design. Perhaps in the future if AAB stabilizes its study the term ‘geometry’ refers to an item that doesn’t contain any
position as a design approach, then can AAB be considered as an inde- data other than the geometry itself. Instead, the term ‘object’ is used to
pendent concept of design. describe an item that contains data. Also the term ‘data’ is often mixed
The study considered many conceptual alternatives for perceiving with information. The study prefers to use term ‘data’ because it refers
the relation between BIM, AAD and AAB. These alternatives are pre- to knowledge that is presented or coded in some form suitable for better
sented in Figure 8. The main result of the study is that AAB should be usage or processing with computer.
understood as a connection between OOD (Object-Oriented Design) Terms ‘parameter’ and ‘attribute’ are used interchangeably to
and AAD. The Main Diagram in Figure 9 clarifies the situation where describe the data that is used to control a geometry or an object. The
AAB is located between OOD and AAD. The main diagram forms the study doesn’t find an exact definition that would explain the difference
basis for this study by dividing the study into three parts: A, B, and C. of these terms. Actually, there is no settled definition that would make a
difference between terms ‘parameter’ and ‘attribute’ (Weygant, 2011, p.
2.1.3. Other terms 25). This study prefers to use the term ‘parameter’.
Some other terms related to CAD are also confusing because they are There are also some other terms that can be compared with the
used arbitrarily. This section presents some terms that are considered to term ‘parameter’. Terms ‘property’, ‘feature’ and ‘character’ are used to
be important in the study. All of these terms are also mentioned in the describe the nature of geometry. This study prefers to use only the term
Glossary. ‘property’ because it is settled in the context of object-oriented program-
The meaning of ‘Design Computing’ and ‘Computational Design’ ming (OOP).
is basically the same. They refer to design that utilizes the computational Another relatively unknown term is ‘topology’. Koleravic (2013, p.
power of a computer. ‘Digital design’ refers to all design that takes place 57) defines the term: “According to its mathematical definition, topology
in digital environment. AAD and AAB are subsets of Digital Design as is a study of intrinsic, qualitative properties of geometric forms that are
Figure 2 shows. Harding (2015, p. 7) have similar explanation of the rela- not normally affected by changes in size or shape – that is, which remain
tion of AAD and its relationship to Digital Design. invariant through continuous one to one transformations or elastic defor-

19 CHAPTER 2: INTRODUCTION TO THE RESEARCH


HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?

Figure 9  THE MAIN DIAGRAM CHAPTER 2: INTRODUCTION TO THE RESEARCH 20


mations, such as stretching or twisting.” Koleravic (2013, p. 57) also con- 2.2.1. Studying process
tinues: “The structuring of dependencies - the topology of the paramet- As mentioned, the study is a combination of theoretical and practical
ric model - determines how it transforms as the parameters are varied.” investigation. The study proceeds from practical studying of AAB to the-
Shortly, topology refers to structural connections that define relations of oretical studying of digital design. The diagram and related concepts are
parametric model (Eastman et al. 2008, p. 46). presented in Figure 9. The theory focuses on the border areas of the dia-
There are many terms that refer to CAD tools. The study uses the gram and the practical study is focused on the center of the diagram.
term ‘tool’ to commonly describe all CAD software, add-ons and plugins. Studying has been split into three phases corresponding to the
The term ‘software’ refers to independent CAD tools, the term ‘add-on’ research questions. These phases are presented in Figure 11. The pre-
refers to integrated tools and the term ‘plugin’ refers specifically to tools sentation of the study has the same direction than the investigation had.
that can be installed to Graphical Algorithm Editors (GAEs). In other words, the presentation flows from practice to theory. This study
relies that the order helps the reader understand the subject.

2.2. METHODS OF THE STUDY 2.2.2. Studying tools


As mentioned the study is design research. The study partly utilizes the The study trusts that an adventurous, even playful attitude can lead to
‘research by design’ method for the investigation because the study exam- finding valid results. The novel studying subject needs to be confronted
ines the relationship of theory and design. They are not opposite points experimentally. As a consequence of the digital character of the subject,
of view because both have not only differences but also similarities and many computational tools are utilized.
connections (Wang and Groat, 2013, p. 21). Thus, the border between The study utilizes a wide range of tools because not only the tools
theory and study is blurred and both can be used to enhance the under- of AAD but also of OOD are needed to investigate the theme. However,
standing of the theme. the main focus is still in AAB tools, so this section presents the main dig-
The study is seen partially to be literature review because it gath- ital tools that enable AAB.
ers and analyses multiple sources of information. As mentioned before, AAB tools work in Graphical Algorithm Editors (GAEs) which
the study is seen to be sort of formal ‘concept analysis’ because it investi- are used to define visual scripts. The main GAEs are presented in Fig-
gates the relationship between concepts. However, the study doesn’t focus ures 12 and 13. These GAEs are called Grasshopper (for Rhinoceros) and
only on the literature but also presents various software, reference- and Dynamo (for Revit). Currently, they seem to be the most potential tools
test projects. for controlling AAB. These and other tools that are used in the study are
listed at the end of the study in the section called Tools.

21 CHAPTER 2: INTRODUCTION TO THE RESEARCH


OBJECT GEOMETRY
-geometry -geometry
-parameters
-relations

Figure 10  GEOMETRY AND OBJECT

PRACTICE
-Learning the main AAB tools -Extended learning of main AAB tools
-Mapping current projects related to AAB -Learning data-exchanges from GAE
to simulation and optimization tools

A. WHAT? B. HOW? C. WHY? STUDYING PROCESS,


PRESENTATION OF THE STUDY

-Mapping and reading theory -Mapping and reading theory -Extended theoretical studying
related to digital design of simulations and optimizations of digital design research
-Mapping and reading current -Mapping and reading theory -Extended theoretical studying
researhes of AAB of data-exchanges of data-exchanges

THEORY
Figure 11  EMPHASIS OF THEORY AND PRACTICE CHAPTER 2: INTRODUCTION TO THE RESEARCH 22
2.2.3. Sources of inspiration Regarding the study, the most interesting detail in the project is the
External sources of inspiration are always used in design. Similarly exter- experimental plan for freeform brick constructions. Aalto attempted to
nal sources of inspiration are used in the study. The following example develop a new form of standard brick, which would allow making capri-
isn’t directly related to the theme but it shows that study can use many cious and curved wall constructions. Aalto (1953, pp. 11–15) describes
sources of inspiration. Probably only a few designers have thought that the main aims of the experiment: “An attempt to develop a type of a stan-
Alvar Aalto would have any relation to AAB, but in this study Alvar Aalto, dard brick or standard object so that it becomes possible to make walls in
and especially his interest to freeform structures has been a resource that a capricious curved form without having to change the standard pieces,
has been a motivating resource. in other words, a sort of further development of the now practically for-
Aalto believed that play has a decisive importance when we build gotten bricks but adapted for other, more up to date purposes.”
communities for people. He also stated that experimental work and play The freeform brick construction was never carried out (Alvar Aalto
should be united. (Aalto, 1953, pp. 11–15.) The study trusts that playful Museum). However, this test shows that Aalto was also interested to per-
approach can lead us to finding things that would not be found other- ceive the relationship between standardization and customization. This
wise. relationship is one of the main challenges that this study faces. The quote
Aalto noted that when constructional structures, shapes and knowl- by Alvar Aalto (Source unknown) has exactly the same approach that
edge are mixed with play, we are on the right path. Technology and econ- characterizes AAB: “Flowers of the apple trees are standardized but they
omy should always be linked to charm which makes life richer. (Aalto, are all different. Thus should we learn to build”.
1953, pp. 11–15.) Similarly the study explores technologies which could All in all, The Experimental House shows that Aalto was also inter-
allow playfulness to arise. ested in controlling new forms. Alvar Aalto’s Experimental House is just
Experimental House in Muuratsalo by Alvar Aalto is the main proj- one of the built and unbuilt projects, which show the interest Aalto had
ect that has inspired the study. The plan for the house is presented in Fig- in freeform structures. Ideologically the study continues Aalto’s free-form
ure 14. The house symbolizes and represents the key approach that is an wall project, which was never finished.
atmosphere, where the designer has playful approach and he’s not afraid
of failing. The house project itself has predefined categories for experi-
mentation and an aim to develop project during the process. Alvar Aalto
describes the building group as a protected architect’s atelier and exper-
imental center, where even those ideas can be tested, which are not yet
mature to be tested out. (Aalto, 1953, pp. 11–15.)

23 CHAPTER 2: INTRODUCTION TO THE RESEARCH


Graphical algorithm editor (GAE) Interface of CAD software Graphical algorithm editor (GAE) Interface of BIM software

Figure 12  GRASSHOPPER FOR RHINOCEROS Figure 13  DYNAMO FOR REVIT

CHAPTER 2: INTRODUCTION TO THE RESEARCH 24


25 CHAPTER 2: INTRODUCTION TO THE RESEARCH
Figure 14  EXPERIMENTAL HOUSE
The plan is for Experimental House by Alvar Aalto (1898-1976). The
free-form brick construction is marked with white circle. (photo of a post-
card. Copyright: Alvar Aalto Museum)

CHAPTER 2: INTRODUCTION TO THE RESEARCH 26


3. GIGA-MAPPING

27
28
3.1. CONCEPTS OF OOD AND AAD

OBJECT-ORIENTED OBJECT-ORIENTED OBJECT-ORIENTED


DESIGN (OOD) DESIGN METHODS DESIGN TOOLS
OOD/OOM/BIM PROCESSES

BUILDING INFORMATION MODELING


Here the study introduces a GIGA-map that was developed (BIM)
along the studying process. The map is the same as the main
diagram in Figure 9, but the map gathers more terms that are OBJECT-ORIENTED SIMULATION (OOS)
SIMULATION METHODS:
related to OOD and AAD. AAB tools are located in the center SIMULATION

INDIRECT AND DIRECT


4D BIM - Scheduling
SOFTWARE

DATA EXCHANGES
of the map. The map shows that AAB can connect not only e.g. Phasing simulation
the tools of OOD and AAD but also corresponding design 5D BIM - Estimation
e.g. Cost simulation
methods and design approaches. 6D BIM - Sustainability
ANALYSIS
e.g. Energy simulation SOFTWARE
7D BIM - Facility management
e.g. Life cycle simulation

OBJECT-ORIENTED ANALYSIS (OOA)


ANALYSIS METHODS
CONSTRUCTION e.g. Structural analysis

-STANDARDIZED CONSTRUCTION
-NO NEED TO “RE-DESIGN”
FABRICATION

MASS-PRODUCTION
-STANDARDIZED MANUFACTURING
-NO NEED TO “RE-DESIGN” CONSTRUCTION

29
CONSTRUCTION

INPUT
DESIGN “BY HAND” “DESIGNING
CONSTRUCTION”

PARAMETERS
AND OBJECTS

ALGORITHM AIDED ALGORITHM AIDED ALGORITHM AIDED


DESIGN TOOLS DESIGN METHODS DESIGN (AAD)
AAD/AAM PROCESS
AAB CAN COMBINE PROCESSES
PARAMETERS AND OBJECTS SIMULATION
PARAMETERS AND GEOMETRY
PLUGINS ALGORITHM AIDED
ANALYSIS PLUGINS MODELING (AAM)
e.g. Karamba
PARAMETRIC MODELING TOOLS SIMULATION PLUGINS ALGORITHM AIDED
e.g. Kangaroo
SIMULATION (AAS)
AAB TOOLS
DEFINED BY DESIGNER

DEFINED BY DESIGNER
INDIRECT AND DIRECT

e.g. Geometry Gym IFC

ANALYSIS METHODS
DATA EXCHANGES

OPTIMIZATION
PROPERTY SETS

PROPERTY SETS
ALGORITHM AIDED e.g. Structural analysis
MODELING TOOLS PLUGINS AND SCRIPTS
e.g. Grasshopper,
Dynamo
OPTIMIZATION PLUGINS (Heuristic) ALGORITHM AIDED
e.g. “Script: Based on rules of thumb” ANALYSIS(AAA)
OPTIMIZATION PLUGINS (Metaheuristic) SIMULATION METHODS
e.g. Evolutionary Algorithm (EA) e.g. Physics simulation

PARAMETRIC BIM
ALGORITHM AIDED
e.g. ArchiCAD, Revit, OPTIMIZATION (AAO)
ViauslARQ, Tekla OPTIMIZATION
“ALGORITHM AIDED” BUILDING INFORMATION MODEL (AAB)
METHODS (Heuristic)
“ALGORITHMIC” GEOMETRY MODEL
e.g. Form-finding MASS-
OPTIMIZATION
GAE (Graphical algorithm editor) METHODS (Metaheuristic) CUSTOMIZATION
e.g. Evolutionary optimization -CNC-MACHINES
“DESIGNING FABRICATION”
OUTPUT

BIM

30
Part A focuses on issues that are
related to the design solution. The
main approach of Part A emphasizes
practice over theory.
PRACTICE
What?

31
HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?

WHITE = STUDY SUBJECT 32


4. THE TEST PROJECT

33
34
A test project was used to learn AAB tools. This knowledge has helped to tial to comprehending the practice of AAB. The concept requires little
understand and learn the practice of AAB. To this end, instead of investi- computational power, so the time needed for modifications during the
gating various architectural aspects, the test project focuses on the inves- modeling process is minimal. In addition, the project is easier to con-
tigation of aspects that are related to the practice of AAB. trol than a project that is complex. The project is seen as a digital proto-
The test project was utilized especially in the early stages of the type that can be compared to physical prototypes produced in the AEC
study. In the early stages of the study the main objective was more practi- industry. The finished digital project is presented in Figure 16.
cally oriented. The main study objective was to learn the tools and under-
stand their limitations. There were intentions to produce various test
projects that could serve the objectives of the study. However, during the 4.1. THE MODELING WORKFLOW
study the main intention has steered more into theory because it seemed The modeling process was controlled mainly in GAE (Graphical Algo-
to have more things to learn. The study argues that only one test proj- rithm Editor). The modeling process was seen as a part of the design
ect can be sufficient to demonstrate a significant portion of the various process. The test project was modeled in Grasshopper (for Rhinoceros)
practical aspects of AAB. because the latter offered comprehensive range of plugins that could be
It was a challenge to find a concept for the test project that would used in the design process. Geometry Gym plugins were the main tools
fit to all desired study objectives. The main challenge was in finding a that were used in AAB to generate objects and export them to external
concept that would be able to show the main aspects of AAB and pro- analysis and simulation software.
duce questions to investigate in the theory part of the study. The test proj- The project was modeled in eight stages that are roughly presented
ect needed to have a conceptual starting point and the study considered in Figure 17. In the figure the amount of details grows gradually from
many alternative concepts that would fit this purpose. left to right. Form finding in the project was handled not only manually
The selected concept consists of interior wall that is located between but also algorithmically. Manual form finding was controlled by modify-
two zones. The concept is presented in Figure 15. The first zone is seen ing two Main Curves in Rhinoceros. They defined the borders of shapes.
as a public lobby that needs to have a novel character. The second zone These manual modifications were updated automatically to the graph and
contains secondary serving spaces, so their character is calm and func- then into to the model preview in Rhinoceros.
tional. The interior wall is confronted by imaginary forces from the first The main idea of algorithm-aided form finding was to apply hori-
zone. zontal forces into the structure. This enabled controlling the shapes with
The study argues that the concept exceeds only the minimal require- only one parameter, which was the force that was applied to the struc-
ments to investigating AAB in practice. This approach has some main ture. Algorithm-aided form finding was made with the aid of a Grass-
advantages. Above all, the study can focus only on things that are essen- hopper plugin called Kangaroo.

35 CHAPTER 4: THE TEST PROJECT


Zone 1 Zone 2

Figure 15  THE CONCEPT

Figure 16  THE FINISHED PROJECT CHAPTER 4: THE TEST PROJECT 36


During the modeling process it was observed that the project can C. Vertical curves
be modeled by using various approaches. This means that BIM, AAD – Dividing lines into (4) segments
and AAB could be combined in the same project. BIM objects could – Applying Kangaroo force (-0.5) into division points
be imported from a BIM software and combined into the model. How- – Defining interpolated curves through moved points
ever, iterative modeling in a BIM software would lead to extra work that
would not necessarily help to achieve the main aims of the study. The D. Vertical polylines
final model consists of objects, except the Wooden Cover. It consists only – Dividing curves into (3) segments
of geometry that can be used in the construction process. Both geom- – Defining tangent lines on the points
etry and objects were controlled parametrically in Grasshopper. These – Joining lines as polylines
include, for example, proportions and structural sizes of the model. – Defining beams with Geometry Gym IFC plugin
In the following list the main stages of the modeling process are
presented. In reality, the process is not as linear because visual scripting E. Support beams
allows the modification of the script from any location. – Projecting middle points of polylines on the back wall and floor
– Defining support lines between middle points and projected points
The modeling process of the wall: – Defining beam with Geometry Gym IFC plugin

A. Back Wall F. Steel details


– Manually modeling a host polyline of the back wall – Defining three types of frames for steel details.
– Defining the wall by Geometry Gym IFC plugin – Defining Tekla objects with Geometry Gym to Tekla plugin

B. Main curves, back line, planes and lines G. Wood plates


– Manually modeling main curve 1 and main curve 2 – Defining solid mass for wooden cover
– Manually modeling one straight back line – Defining border curves of wooden cover
– Dividing back line by (10) perpendicular planes – Defining plates of wooden cover
– Intersecting main curves with planes
– Creating (10) straight lines between intersection points H. Upper beams
– Defining lines between the upper main curve 1 and back polyline
– Defining beams with Geometry Gym IFC plugin

37 CHAPTER 4: THE TEST PROJECT


H

The Main Curve 1


Back wall Wooden cover

Polyline
Line Curve

Back line A Support


Line

G
F
E
D Floor
The Main Curve 2 C
B

Figure 17  PHASES OF MODELING CHAPTER 4: THE TEST PROJECT 38


4.2. THE DIGITAL DESIGN PROCESS The project shows that digital design tools and processes are chal-
The modeling process, as usual, is seen to be a part of the digital design lenging to learn only by reading theory. Instead, they should be learned
process. The modeling process produces a model that can be used for practically through trial and error.
various digital design tasks. Figure 18 presents the design process con-
ceptually from design to fabrication. The design process has two main
data flows because the project combines design methods from both AAD
and AAB. Figure 19 presents the visual script that was used to control the
design process. In the figure data flows from left to right.
The main intention of the study was not only to learn AAB tools
but also to investigate interoperability and processes from GAE to analy-
sis and simulation software. Thus, various simulation and analysis meth-
ods were conceptually integrated into the design process. A large part of
the time was spent on the learning of tools that enabled these processes.
The aim of these methods was to analyze the performance of the
project. For example, the evaluation of structural performance was made
in Grasshopper with plugin called Karamba. The analysis is presented
in Figure 20. Data exchange tests were also made to external simulation
tools. Resulting models are presented in Figures 21–23.

4.3. SUMMARY
The project showed that AAB tools still have many limitations. So to use
AAB tools effectively, the designer needs to be able to understand these
limitations.
Structural detailing turned out to be one of the largest challenges in
the modeling process. Objects didn’t fit always into design objectives. In
addition, used tools didn’t enable to model details by using ‘reverse engi-
neering’ methods. They mean that some parts are taken out of objects.

39 CHAPTER 4: THE TEST PROJECT


STRUCTURAL ANALYSIS
Geometry Gym Karamba to Robot exporter
-Robot

STEEL JOINTS Fabrication plans Manufacturing


-Tekla -Drawings, dwg -Robot

AAB Geometry Gym to Tekla exporter

THE MODELING PROCESS STRUCTURAL ANALYSIS MEP


Rhinoceros and Grasshopper Karamba CLASH DETECTION
-Solibri
STEELFRAME (AAB) 4D BIM
-Grasshopper: Geometry Gym exporter
Geometry Gym, Kangaroo to Navisworks BIM Visualization
-Archicad elements COMBINATION MODEL -Plan, Section, 3D
WOOD SKIN (AAD) -Archicad/Rhino
-Grasshopper

AAM
WOOD SKIN (AAD) AAD to Archicad exporter Fabrication plans
-Details -Drawings, dwg, Manufacturing
gcode, numbering -Laser cutting

Figure 18  THE DESIGN PROCESS CHAPTER 4: THE TEST PROJECT 40


Beams Beams
Karamba Objects (IFC)
Objects

Beams Beams 4D BIM

OBJECTS
Geometry Gym IFC Geometry Gym Geometry Gym IFC
Objects Direct exchange of objects to Tekla Time scheduling parameters to Navisworks
Form finding Curves
Geometry Geometry
Back wall
Objects (IFC)

Vertical steel details


Geometry

OBJECTS
Front lines for Steel details
wooden cover Horizontal steel details Geometry Gym IFC,
Geometry Geometry Direct exchange of objects to Tekla

GEOMETRY
Wooden cover
Geometry

41 CHAPTER 4: THE TEST PROJECT Figure 19  GRASSHOPPER DEFINITION


Figure 20  STRUCTURAL SKETCHING ANALYSIS WITH KAR- Figure 22  4D BIM (TIME SCHEDULING) IN NAVISWORKS
AMBA MANAGE

Figure 21  STRUCTURAL STEEL FRAME DETAILING IN TEKLA Figure 23  CLASH DETECTION IN TEKLA BIMSIGHT

CHAPTER 4: THE TEST PROJECT 42


5. CURRENT PROJECTS

43
44
5.1. FONDATION LOUIS VUITTON

The Fondation Louis Vuitton is a museum and a cultural center that is other information was defined into a 3D model. 2D drawings were sup-
located in the park called Bois de Boulogne. The park is located in 16th plemented with 3D files to support the quality control of the fabrication.
arrondissement of Paris, France. The building was designed by Frank The project was challenging because for example the structural core of
Gehry (www.foga.com) and it was opened in October 2014. The primary the museum is covered with ca. 16000 ceramic tiles. All of these tiles are
design features of the building are the structural solid core referring to unique. (TEKLA.)
‘ice bergs’ and transparent covers referring to “sails”. The structural core One of the most interesting aspects of the project is that pre-ratio-
is made of steel frameworks covered with concrete panels. The covering nalization of the glass cover is managed by using developable surfaces.
“sails” are made of glass and supported by steel and wood beams. These These are ‘single-curved’ surfaces that can be folded onto the plane with-
three structural layers are shown in Figure 24. out stretching or tearing. Developable surfaces can be fabricated of pli-
The project was chosen as a reference project because the design able materials that are produced as plates. For example metals are often
process of the building has taken full advantage of current BIM technol- used in developable surfaces. The covering surface is made out of devel-
ogies. Free-form geometries are challenging to control in a BIM environ- opable strips. Many other projects by Frank Gehry are also good exam-
ment, but the project shows that there can be found advanced ways to ples of the use of developable surfaces and the use of them in digital
combine BIM and AAD. The role of AAB has in this project relied mostly design. (Pottmann et al., 2014, p. 6.)
on the combination of commercial software and custom made automa- The project is indeed one of the most interesting BIM projects that
tion tools. The project has also received many international BIM awards can be compared with AAB. All objects that were defined in self-made
including the BIM Excellence Award given by the American Institute of scripts can be regarded as AAB. The amount of these objects remained
Architects (AIA) (TEKLA). unknown for the author. The project shows that even challenging struc-
The project required worldwide collaboration between multiple tures can be designed with objects if the design team shares a common
parties. Gehry Technologies (www.gehrytechnologies.com) provided a will to do it.
3D project delivery system called Digital Project (www.digitalproject3d.
com) that was used for the 3D modeling, data exchange and BIM col-
laboration. The model server was developed to control model versions,
concurrent distribution, and tracking. (TEKLA.)
The design process included many parties and multiple design soft-
ware, so advanced interfacing between software was a necessity to con-
trol the design process. Only 2D drawings that were produced in the
design process were made for the fabrication of parts and assemblies. All

45 CHAPTER 5: REFERENCE PROJECTS


Figure 24  FONDATION LOUIS VUITTON CHAPTER 5 REFERENCE PROJECTS 46
5.2. CITY-LIFE RESIDENCES

City-Life Residences by Daniel Libeskind (libeskind.com) are located In reality, the design process of the residences has also been chal-
in Milan, Italy. The residences are part of the City-Life master plan. The lenging from a structural point of view. The most challenging part of
area comprises residential buildings, three skyscrapers, a museum and structural design has been the design of the penthouses because their
a park (Franchi et al., 2012, p. 1). The residential project by Libeskind structural dimensions are long. Furthermore, as a consequence of archi-
consists of two construction phases. The first five-buildings are already tectural requirements, balconies needed to be cantilevered up to 3,5
finished and the second three-buildings are scheduled to be completed meters. A vibration analysis was performed for the balconies to be sure
in 2017 (Liebeskind). The main architectural character of the project is that they would work properly. (Franchi et al., 2012, p. 4.)
dynamic structure that twists asymmetrically in the vertical direction. All in all, the project shows the potential that AAB could be able
Some of the most challenging structures of the project are the balconies, to offer. Current tools of AAB could have been used for the design of the
the penthouses and the covered street in the courtyard. The project is project and especially balconies. However, there would still be many chal-
presented in Figure 25. lenges related to the details of the structure. That’s why the most fluent
The project was chosen as a reference because it represents mod- way is still to design balconies by using the common modeling methods
ern architectural building design. The study didn’t find out how the archi- of BIM and AAD.
tects controlled the modeling process. Instead, the study suggests that the
design process of the project could have taken advantage of AAB.
The study suggests that the design process of similar project could
benefit from AAB. Especially, the balconies could more easily be con-
trolled by using AAB. By using BIM, on the other hand, the control of
the balconies would be frustrating. All of the balconies are unique, so
iterative updates in BIM software would always need manual, time con-
suming and labor demanding modifications.
For example, by using AAB the modeling process could take advan-
tage of the ‘wall by face’ technique. Building masses could be modeled
in CAD software and linked to GAE. Balconies could then be placed on
top of masses, so each update to the masses would automatically also
change the wall structure. Thus, masses would be host geometry for the
walls. Thus, the modeling process would also enable fast iterations with
performance tools.

47 CHAPTER 5: REFERENCE PROJECTS


Figure 25  CITY-LIFE RESIDENCES CHAPTER 5 REFERENCE PROJECTS 48
5.3. LOUISIANA STATE MUSEUM AND SPORTS HALL OF FAME

Louisiana State Museum and Sports Hall of Fame is located in Natchi- Advanced Cast Stone also hired Design Method (www.methodde-
toches, Louisiana, USA. The museum was designed by Trahan Archi- sign.com) to design the steel structures supporting the cast stone panels.
tects (www.trahanarchitects.com) and it was opened at the end of June, Design Method also provided precise anchor points for the cast panels.
2013 (Stasiuk, 2013). The entrance of the museum is free-form circula- (Stasiuk, 2013.)
tion and atrium space. The steel structures and cast panels of the entrance The design of automation algorithms was challenging task because
are shown in Figure 26. of the free-form panel structure. The design process included various
The project was chosen as a reference because it combines various design methods and tools. For example, the structural analysis was made
design methods related to BIM, AAD and AAB. Additionally, multiple in two steps. Analysis of structural behavior was made with Karamba. The
design tools were used to control the project (Stasiuk, 2013). The project main structural analysis was made in Robot. The geometry was exported
contained many parties, so collaboration between these has been essen- to Robot by using the Geometry Gym plugin. (Stasiuk, 2013.)
tial. The project has obviously been one of the main reference projects
The architect office originally designed the free-form structure of to inspire this study. The project shows well the current capabilities of
cast stone panels by using surfaces in geometry-based animation soft- AAB tools. The project also shows that designers engaging with processes
ware. However, a model based fabrication and construction coordina- in digital design need to be able to handle multiple design approaches.
tion process was required to realize the project. Thus, the project was Digital design processes are not parallel, so they need to be managed cre-
designed with 3D object models. The model enabled many trades to par- atively.
ticipate into the design process. (Case, 2013, pp. 215–216.)
Advanced Cast Stone (www.advancedarchitecturalstone.com)
managed the fabrication of over 1000 unique cast stone panels. They
hired CASE (www.case–inc.com) to provide all fabrication modeling ser-
vices. CASE also managed the entire BIM development of the project.
Remote coordination meetings were arranged to identify and resolve
clashes. (Case, 2013, pp. 215–216.)
CASE created fabrication models of the architects’ original design
surfaces. These models were used to coordinate the whole process. They
were also used to finalize the support structure, resolve panel conflicts
and fabricate the final modules for the cast stone panels with robots.
(Case, 2013, pp. 215–216).

49 CHAPTER 5: REFERENCE PROJECTS


Figure 26  THE ENTRANCE OF LOUISIANA STATE MUSEUM
AND SPORTS HALL OF FAME CHAPTER 5 REFERENCE PROJECTS 50
Part B focuses on issues that are
related to the design methods and
design tools in AAB. The main
approach of Part B is between theory
and practice.
PRACTICE
THEORY /

How?

51
HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?

WHITE = STUDY SUBJECT 52


6. INTEROPERABILITY

53
54
A single computer software cannot support all of the tasks that are needed al., 2008, p. 65.) If these formats would be used to contain all geomet-
in the design of a project, so interoperability is needed to enable flu- rical information, including geometry, relations and parameters - com-
ent and automated design processes. Interoperability refers to ability to plex projects would became too complex to handle (Eastman et al., 2008,
exchange data between software. Thus, it eliminates the need to repli- p. 70).
cate data. Interoperability can facilitate fluent processes and automation. Data models were developed to support product and object model
(Eastman et al., 2008, p. 66.) exchanges within different industries. The development of data mod-
Custom digital processes are often related to exploration of mul- els started in the 1980s. Data models are developed by the International
tiple design alternatives. An optimal solution usually cannot be found Organization for Standardization (ISO). These developments are mainly
in a parallel process, so iterative design processes are needed to find the based on the EXPRESS data modeling language. (Eastman et al., 2008, p.
optimal design alternative. In AAB data exchanging is needed mainly 65.)
because of two reasons. Firstly, modeling data may be produced and con- The main building product data models are Industry Foundation
trolled in CAD environment that doesn’t support BIM, so objects need Classes (IFC) and CIMsteel Integration Standard Version 2 (CIS/2).
to be exchanged with BIM software. This is the case especially with AAB IFC is intended for building planning, design, construction and man-
tools that are integrated into Grasshopper. However, the data exchange of agement. CIS/2 is intended for structural steel engineering and fabrica-
objects is not needed, if an AAB tool is integrated into the BIM software. tion. Both building product models are based on EXPRESS and they can
This is the case with Dynamo. Secondly, as in BIM processes, the object contain information that consists of geometry, relations and parameters.
data produced in AAB can be exchanged with simulation tools. They are EXPRESS allows software to define an object in various ways, so national
used to evaluate the performance of the model. By allowing designers to efforts aim to standardize the data that is required for specific processes.
effectively link an AAB tool to simulation tools, multiple design alterna- (Eastman et al., 2008, p. 65.)
tives can be explored. Main data exchanges related to AAB are presented The currently most used IFC schema, IFC2x3 schema was released
in Figure 27. in 2006. The newest schema, IFC4, was released in March 2013. Improve-
ments in IFC schema help to decrease earlier limitations. (Buildingsmart)
IFC4 offers some new features, which essentially facilitate AAB processes.
6.1. PRODUCT DATA MODEL EXCHANGE Some of the new features are related to the creation of geometry.
These enhancements are essential because many IFC based data exchange
FORMATS problems are related to the translation of geometry. Firstly, IFC4 expands
Interoperability has traditionally relied on file-based exchange formats
to support NURBS (Non-Uniform Rational B-Spline) curves and sur-
that are designed to contain only the geometrical information. Popular
faces (Buildingsmart), which is one of its main advantages. NURBS can
file formats such as DXF and IGES belong to this category. (Eastman et

55 CHAPTER 6: INTEROPERABILITY
DEFINED BY DESIGNER

DEFINED BY DESIGNER
INDIRECT AND DIRECT

INDIRECT AND DIRECT

AAB TOOLS
DATA EXCHANGES

DATA EXCHANGES

PROPERTY SETS

PROPERTY SETS
OOD SIMULATION
TOOLS
BIM
TOOLS
AAM
TOOLS
SIMULATION
PLUGINS AAD

GAE (Graphical algorithm editor)

Figure 27  DATA EXCHANGES


CHAPTER 6: INTEROPERABILITY 56
be used for complex structures that are common in AAB. For example, mentioned data exchange types belong to the indirect category. This study
Rhinoceros and Grasshopper are based on NURBS geometry. focuses on the first and third types because the AAB tools of this study
Secondly, IFC4 also expands to support the incorporation of para- mainly support these types of data exchange.
metrics and constraints. This offers new opportunities for advanced use
of IFC, especially for BIM content developers. (Mirtschin, 2014,p. 7.) 6.2.1. Direct Data Exchange
IFC4 also expands to support tapering in extrusions, arbitrary sweeps, In the direct data exchanges, objects are exchanged directly between GAE
non-planar surfaces and surface bounds (Buildingsmart). The other new and an external BIM or analysis and simulations tool. Direct translators
features of IFC4 are listed in many sources (e.g. Buildingsmart; Mirtschin, need to be called from one or both software. Direct data exchange pro-
2014). cesses rely on middle-ware software interfacing capabilities (Eastman
et al., 2008, p. 67). Direct data exchange processes are presented in Fig-
ure 28.
6.2. TYPES OF DATA EXCHANGE Direct data exchange procedures have some challenges. The file
The incompatibility exists on all levels of data exchange: at the data model translation needs to be made between each software, so this approach will
level, the data schema level, and the data file level. To meet these incom- lead to the need of many translators. That’s why all file translators need
patibilities there exist two alternative approaches for data exchange: direct to be developed independently. In addition, data formats update occa-
data exchange and indirect data exchange. (Janssen et al., 2015, p. 516.) sionally, so the translators need to be maintained. (Janssen et al., 2015,
pp. 516–517.) The challenges of direct data exchange may lead to data
Data exchanges between two software are typically carried out by using translators, which are not update and thus reliable. Application Program-
the following data exchange types (Eastman et al., 2008, pp. 67–68): ming Interfaces (APIs) of software also have constraints, which may lead
1. Direct, proprietary links between specific BIM tools to poor capability for the translators.
2. Proprietary file exchange formats, primarily dealing with Software companies prefer to develop direct data exchanges because
geometry then they can provide better support and it keeps customers from using
3. Public product data model exchange formats their competitors’ software (Eastman et al., 2008, p. 67).
4. XML-based exchange formats Both Grasshopper and Dynamo have some plugins, which enable
direct data exchange processes between their GAEs and external sim-
Data exchange types can be further categorized to direct file exchanges ulation software. There are also some plugins that enable direct data
and indirect data exchanges (Janssen et al., 2015, p. 516). The first type exchanges from Grasshopper to Revit. For example Grevit and Lyre-
(1.) obviously belongs to the direct data exchange category. The other bird use a direct data exchange procedure. Additionally, Geometry Gym

57 CHAPTER 6: INTEROPERABILITY
TOOL TOOL
TOOL A TOOL A
A A

TOOL TOOL TOOL TOOL


TOOL F TOOL B TOOL F TOOL B TO
F B F B

SHARED
SHARED DATA
DATA SCHEMA
SCHEMA

TOOL TOOL TOOL TOOL


TOOL E TOOL C TOOL E TOOL C TO
E C E C

TOOL TOOL
TOOL D TOOL D
D D

TOOL A TOOL A TOOL A SCHEMA TOOL A


TOOL A TOOL A TOOL A SCHEMA TOOL A TO

File translator
File translator
DATA DATA DATA DATA DATA
DATA FILE DATA FILE DATA DATA
FILE DATA FILE D
FILE FILE FILE FILE
FILE F
FILE

Figure 28  DIRECT DATA EXCHANGE Figure 29  INDIRECT DATA EXCHANGE
The upper diagram shows that direct data exchange is possible only The upper diagram shows that shared data schema can link many tools.
between two tools. The lower diagram presents the data exchange process The lower diagram presents data exchange between two tools. Both
between two tools. Both figures are adapted from: (Janssen et al., 2015, p. figures are adapted from: (Janssen et al., 2015, p. 517).
517).

CHAPTER 6: INTEROPERABILITY 58
plugins allow direct data exchange between Grasshopper and external 6.2.3. Shared Mapping Process
simulation software. Due to the limitations and challenges related to direct and indirect data
exchange, Janssen et al. (2015, p. 519) propose a graphical mapping
6.2.2. Indirect Data Exchange interface as a new way to exchange model data. Their approach allows
In indirect data exchange a shared data schema is used to link GAE and the users to define their own data mappers, if direct or indirect pro-
external BIM and simulation tools. Indirect data exchange is based on cesses don’t exist. In addition, a graphical mapping interface can com-
an exchange format that is human readable (Eastman et al., 2008, p. 68). bine direct and indirect translation processes. These shared mapping pro-
Levels of indirect data exchange vary depending of the amount of seman- cesses between tools are presented in Figure 30. In addition, Figure 31
tic information linked to the object (Janssen et al., 2015, p. 516). Indi- shows a user defined mapping process. The figure shows that mapping
rect data exchange processes are presented in Figure 29. between two data files involves three stages: parsing, mapping and seri-
On the low level the amount of information is limited by the geom- alizing (Janssen et al., 2015, p. 8).
etry of the object. DXF, IGES, and SAT are examples of low-level file The proposed mapping platform can be implemented as a web
exchange formats. Low-level data exchanges cannot exchange any param- based application. Users can define their own mappings and custom
eter data, so they don’t fulfill the requirements of AAB. On the high-level mappings can be shared collaboratively. The platform aims to be flexible
the object can contain much information related to the reality. Particu- and user-friendly. (Janssen et al., 2015, pp. 517–519.)
larly during the last decade, a STEP-based IFC standard has been used The platform needs to be able to translate varying input data to
for high-level data exchange. (Janssen et al., 2015, p. 516.) required output data. The simplest type of mapping is called ‘declar-
Both exporting and importing software need to support the data ative equivalency mapping’, which is used for semantically equivalent
translation format. In addition, the format itself limits the possibilities data translations. More complex mappings called ‘procedural query map-
of data exchange. pings’, are handled with the aid of graphical programming. (Janssen et
Grasshopper and Dynamo have some plugins, which enable data al., 2015, pp. 517–519.)
exchange between their GAE platform and external software. Geome- Graphical mapping processes have potential that can be used to
try Gym is an example of plugin that enables high-level data exchanges connect GAE to external BIM and simulation software. At the moment all
between Grasshopper and external software. At the moment Geometry AAB plugins of Dynamo and Grasshopper use independent approaches
Gym offers the most comprehensive data exchange between Grasshop- for modeling and exchanging data. An adaptive graphical mapping pro-
per and external object-oriented software. cess could unify these processes and allow various OpenBim processes
to appear.

59 CHAPTER 6: INTEROPERABILITY
TOOL
A
TOOL
A

TOOL TOOL
F B
TOOL TOOL
F B
SHARED TOOL A MAPPER TOOL B
MAPPING
SHARED TOOL A MAPPER TOOL B
PROCESS SOURCE INPUT USER DEFINED OUTPUT TARGET
MAPPING DATA DATA MAPPING PROCEDURE DATA DATA
PROCESS SOURCE INPUT
SCHEMA USER DEFINED
SCHEMA OUTPUT TARGET
SCHEMA SCHEMA
DATA DATA MAPPING PROCEDURE DATA DATA
TOOL TOOL SCHEMA SCHEMA SCHEMA SCHEMA
E C SOURCE TARGET
TOOL TOOL DATA MODEL INPUT EXCECUTION OF OUTPUT DATA MODEL
E C SOURCE DATA USER DEFINED TARGET
DATA
DATA MODEL INPUT EXCECUTION
SET OUTPUT
OF MAPPING PROCEDURE DATA
SETMODEL
DATA USER DEFINED DATA
TOOL
SET MAPPING PROCEDURE SET
D
TOOL SOURCE TARGET
PARSER SERIALIZER
D DATA FILE MAPPING DATA FILE
SOURCE TARGET
PARSER DATA SERIALIZER
MODEL
DATA FILE MAPPING DATA FILE
DATA MODEL

TOOL A MAPPER TOOL A


TOOL A MAPPER TOOL A

DATA DATA
FILE FILE
DATA DATA
FILE FILE

Figure 30  SHARED MAPPING PROCESS Figure 31  MAPPING PROCESS


The upper diagram shows that one shared mapping processes can link The diagram shows the mapping process between two data files. The
many tools. The lower figure shows the mapping process between two figure is adapted from: (Janssen et al. 2015, p. 518).
tools. Both figures are adapted from: (Janssen et al., 2015, p. 517).

CHAPTER 6: INTEROPERABILITY 60
7. PARAMETRIC MODELING

61
62
This chapter discusses the modeling methods and tools of BIM, AAD and 7.1. BIM PROCESSES
AAB. These modeling methods and tools have many similarities but also As can be seen in Figure 32, BIM is based on object-oriented paramet-
many differences. The main aim of this chapter is to give an overview of ric modeling. The modeling process is controlled manually. This section
each modeling method by presenting their main modeling processes. investigates parametric modeling in BIM by introducing the main mod-
The main modeling methods related to CAD are presented in Fig- eling process, which is initially is presented in Figure 33 and then dis-
ure 32. The figure shows the main relations between these modeling cussed further. This section proceeds along the modeling processes from
methods. The study suggests that the main differences of these mod- input to actions and output.
eling methods can be observed by dividing the methods to parametric In general, object-oriented parametric modeling refers to model-
and non-parametric methods, algorithmic and non-algorithmic model- ing that controls objects through parameters and rules. Object-oriented
ing methods and to object-oriented and geometry-based methods. On parametric modeling was first developed in the 1980s. (Eastman et al.,
this basis, and as the figure shows, the commonly used term ‘parametric 2008, p .25.)
modeling’ can refer to many modeling methods. In general, the term is In BIM software parameters are hierarchically arranged to affect
often misleadingly used to refer to specific type of modeling. objects on many levels. Depending on the type of the parameter, para-
The figure shows that there are some fundamental similarities and metric values can be fixed, user defined or determined by the their rela-
differences between modeling in OOD and AAD. Both are based on para- tion to other another object. Objects contain a set of rules and relations
metric modeling. OOD is based on modeling with objects while AAD is that defines the behavior of the object. (Eastman et al., 2008, p. 29.)
based on modeling with geometry. Algorithms are integrated into BIM Modeling happens manually in BIM software by choosing desired
tools but in AAD tools they can be relatively freely edited by the designer. objects and placing them into the modeling space that can be previewed
AAM (Algorithm-Aided Modeling) enables the designer to define geom- via various 3D and 2D viewports. The main direction of modeling can
etry in GAE. The only modeling method that combines parametric, algo- be seen to be top-down and bottom-up. In reality, modeling processes
rithmic and object oriented modeling simultaneously is AAB. use both of these directions. Modeling processes in BIM are often start
Parametric modeling methods are further discussed in the follow- from the large scale and then go towards smaller details. That’s why BIM
ing sections. Their modeling processes are described shortly by divid- processes are often intended to be top-down modeling processes. On the
ing modeling process into three steps called input, actions, and output. other hand, BIM objects seems to fit the bottom-up approach because
This study argues that a comparison of the modeling processes can show the designer builds the model out of objects by determining parameters.
the main advantages and drawbacks these modeling methods and tools During the modeling process more and more data is embedded into the
have. On this basis, modeling methods called BIM and AAM are com- BIM objects (Boeykens, 2012, p. 455).
pared. Then the study introduces modeling process in AAB.

63 CHAPTER 7: MODELING
BIM (Building Information Modeling)/ AAM (Algorithm-Aided Modeling)
OOM (Object-Oriented Modeling) GEOMETRY-ORIENTED PARAMETRIC/ALGORITHMIC MODELING
-PARAMETRIC
OBJECT-ORIENTED PARAMETRIC MODELING -ALGORITHMIC
-PARAMETRIC -(NO OBJECTS)
-(NON-ALGORITHMIC) GEOMETRY-ORIENTED PARAMETRIC MODELING
-OBJECTS -PARAMETRIC
-(NON-ALGORITHMIC)
-(NO OBJECTS)

e.g. Dynamo, Grasshopper + Geometry Gym, Grevit, Lyrebird

e.g. 3DS MAX

e.g. Grasshopper, Dynamo

OOD MODELING
METHODS
MODELING
TOOLS AAD
GEOMETRY-ORIENTED MODELING
-(NON-PARAMETRIC)
-(NON-ALGORITHMIC)
-(NO OBJECTS) e.g. Rhinoceros, SketchUP

AAB (Algorithm-Aided
e.g. ArchiCAD, Revit, Building Information Modeling)
VisualARQ, Tekla
OBJECT-ORIENTED PARAMETRIC/ALGORITHMIC MODELING
-PARAMETRIC
-ALGORITHMIC
-OBJECTS

OBJECT-ORIENTED ALGORITHMIC MODELING


-(NON-PARAMETRIC)
-ALGORITHMIC e.g. BIM script and Lena
-OBJECTS

Figure 32  MODELING METHODS OF CAD CHAPTER 7: MODELING 64


Software contain pre-defined objects that are integrated into the challenging to model creatively without following predefined procedures.
system. However, pre-defined objects complete only the basic types of Objects are scripted inside the software, so they limit the freedom of
construction. User-defined parametric objects, on the other hand, can modeling with user-defined algorithms (Aish, 2013, p. 43; Boeykens,
overcome the limitations of predefined objects. (Eastman et al., 2008, pp. 2012, p. 453).
35–36.) Objects in BIM software are independent, which makes it challeng-
The internal structure of an object is defined within a directed ing to built relationships between them (Denis, 2014, p. 13). BIM objects
graph. In BIM software the graph is not usually visible for direct editing. are driven parametrically but the data embedded to the objects doesn’t
The internal structure consists of object families, operations and links. usually affect to the surrounding objects. Only some built in functions
(Eastman et al., 2008, p. 44.) In BIM software the graph is controlled like automatic ‘wall-joints’ take advantage of relationships between BIM
through panels and windows that contain the structural relations of the objects.
model. Compared to traditional 2D drawing procedures, iterations in BIM
Topological structures are often built automatically into the soft- are fast. BIM software are good at manual modifications with indepen-
ware. For example, the designer can place a wall into the model and the dent objects. On the other hand, comparison between BIM and AAM
tool will automatically associate the wall to its context. Connections in shows that current BIM software’ lack of relationships between objects
mathematics are called topology. Connections carry three kinds of data: cannot lead to as effective iterations as in AAM. In BIM, construction
what can be connected, what the connection consists and how the con- drawings are usually drawn many times during the design to construc-
nection is composed in response to various contexts. (Eastman et al., tion process.
2008, of pp. 45–46.) For example, hosts of objects are defined in topo- The modeling tools of BIM lack add-ons to analyze, visualize and
logical structures. Hosted objects are located in the coordinate system of generally utilize the data the model contains. In addition, representations
the host object, so they move as a unit (Eastman et al., 2008, p. 30). Win- are often limited to general 2D drawings and 3D views. (Boeykens, 2012,
dows for example are always hosted by walls. p. 454.)
An output of the BIM process consists of objects that contain the Iterations in BIM are challenging. There have been some efforts to
geometry, relations and parameters. The output model can be called as make this better. For example, Revit offers an option to lock the measure-
BIM or object model. ments and borders of objects. Thus, these measurements are permanent
and locked objects are permanently related to each other.
7.1.1. Advantages and drawbacks of BIM
BIM is often criticized because it steers the designer too much and the
final products of modeling are too obvious. In BIM software, it may be

65 CHAPTER 7: MODELING
PERFORMANCE EVALUATION SOFTWARE
LEGEND Object-oriented
External software

DEFINED IN FIGURE e.g. Robot

EXTERNAL SOURCE
PARAMETRIC PERFORMANCE EVALUATION ADD-ON
BIM SOFTWARE Object-oriented
AAB Integrated to BIM

SIMULATION (NOT SHOWN) e.g. 360 Structural Analysis for Autodesk Revit

OPTIMIZATION (NOT SHOWN)


P MANUAL MODELING OBJECT MODEL EXTERNAL
Object-oriented MODELING
O OBJECTS Manual / built in relations SOFTWARE
O Parameters
P PARAMETERS

G GEOMETRY
e.g. Tekla
DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE
MAPPINGS Object based
External software
MAIN DATA FLOW e.g. Navisworks
COMMON DATA FLOW

SECONDARY DATA FLOW INPUT ACTIONS OUTPUT

PERFORMANCE EVALUATION SOFTWARE


Geometry based
External software

e.g. Robot
Figure 33  BIM PROCESSES CHAPTER 7: MODELING 66
PARAMETRIC CAD
SOFTWARE PERFORMANCE EVALUATION
ADD-ON
Geometry based
7.2. AAM PROCESSES Visual scripting means that the process is controlled by editing
As can be seen in Figure 32, AAD is based on geometry-based paramet- graphical algorithms in GAE. Visual programming allows the designer
ric modeling called AAM. The modeling process is controlled by a script to take advantage scripting, without being a programmer (Denis, 2014,
that uses algorithms to perform tasks. This section investigates paramet- p. 6). One of the main advantages of visual scripting is that it allows the
ric modeling in AAD by introducing the main modeling process that is designer to perceive the algorithmic process or solution in a diagram and
presented in Figure 34 and further discussed in this section. The section get real time feedback (Denis, 2014, p. 9).
proceeds along modeling processes from input to actions and output. Utilization of visual scripting can be seen as a turning point for
Internally defined input data of a script usually consist of numeric, the development of AAD. However, the most comprehensive advantage
textual and geometric data that is defined in CAD software or inside of of scripting can be achieved; if textual and visual scripting methods are
a GAE. The range of input data can be extended by feeding data from combined in the same project (Aish, 2013, p. 47). This study concen-
external data sources. (Boeykens, 2012, pp. 453–454.) trates mainly on visual scripting because it is better suited for architec-
Modeling processes in AAD combine both top-down and bot- tural design purposes.
tom-up directions. Sometimes the main geometrical forms are designed Usually an interface of a GAE is relatively simple because its main
before the utilization of scripts. In such cases, AAD modeling processes purpose is to control flows of the data. In GAE visual scripting is based on
use top-down design approach. On the other hand, scripting processes connecting “nodes” containing functions. Parameters are input data for
are usually based on bottom-up approaches. Scripts are usually started these nodes. Nodes execute the built-in algorithm and then give another
from small entities and continued to larger assemblies. (Boeykens, 2012, parameter as an output. The script is built by combining required nodes.
p. 456.). The focus in AAD shifts from geometry to defining topological
There are two main approaches to control scripting: textual and structures. Algorithm-aided design utilizes topological structures to
visual scripting. According to Aish (2013, p. 47), text based scripting define and control the geometry. (Koleravic, 2013, p. 55.) In other words,
can be called ‘imperative scripting’ and graphical scripting can be called this means that designer builds the project by determining relationships
‘associative scripting’. These terms are not widely known, so this study and dependencies inside of the script. Thus, modification of one param-
prefers to use the more descriptive terms ‘textual and visual scripting’. eter or algorithmic part of the script can affect the whole model. All con-
Textual scripting is the conventional way of scripting – scripts are nections and relationships together form the topology of the project.
written into rows of text. Textual scripting is characterized by for loops To control the project the designer needs to understand the topo-
(for iteration) and if statements (for conditionals) (Aish, 2013, p. 47), logical structure of the project. The project’s topological constraints
which are used to expand the possibilities of visual scripting. determine the possible manipulations of the project. The project’s topo-
logical structure determines how the project changes when parameters

67 CHAPTER 7: MODELING
SECONDARY DATA FLOW INPUT ACTIONS OUTPUT

PERFORMANCE EVALUATION SOFTWARE


Geometry based
External software

e.g. Robot

PARAMETRIC CAD
SOFTWARE PERFORMANCE EVALUATION
ADD-ON
Geometry based
Integrated to CAD software
GAE

PERFORMANCE EVALUATION PLUGIN


Geometry based
Integrated to GAE
e.g. Scan-and-Solve
e.g. Millipede

P P GRAPH P GEOMETRIC MODEL EXTERNAL


MODELING
GEOMETRY GEOMETRY GEOMETRY SOFTWARE
Geometry-oriented
FORM FINDING

e.g. Autocad
Eg. Kangaroo OPTIMIZATION
G Meta-Heuristic methods
FITNESS FUNCTION

FITNESS VALUE (S)


e.g. Galapagos
GENOTYPES

DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE


Object based
External software

e.g. Excel

INPUT ACTIONS OUTPUT

Figure 34  AAM PROCESSES CHAPTER 7: MODELING 68


are changed. Too complex topology may lead to surprising results. Then define built in parameters. Thus, object-oriented software could “under-
the designer moves from predictable to unpredictable results. (Kolarevic, stand” the model as a combination of objects and not just as a combina-
2013, p. 56.) tion of geometries. (Zarzucki, 2012, p. 1; Denis, 2014, p. 8.)
The script produces an output of the modeling process. An out- An automation process of AAD aims that construction drawings
put of the AAD modeling process contains the geometry and other para- are produced automatically. However, AAD tools are only used only for
metric data that is not attached to the geometry. Parameters can contain conceptual design and not for producing representational drawings.
numerical and textual data. (Denis, 2014, p. 10.) The process is then continued in BIM software to
Visual scripting enables the designer to control geometry in many produce documentation for construction and representation. For exam-
stages of the building design process at the same time. For example, geo- ple plans, sections and elevations of the building are usually produced
metrical parts of a 3D model can be arranged on layout sheets, so the in a BIM software. Required data exchange process from GAE to BIM
designer can prepare the construction stage in the early stage of the proj- software breaks the link between the geometry and the script (Boeykens,
ect. 2012, p. 454). The only way to update the transferred model is to make
the needed changes in GAE and replace the old model in the BIM soft-
7.2.1. Advantages and drawbacks of AAM ware with a new one. The BIM software doesn’t “understand” the geom-
Some of the main advantages of BIM software are missing in AAD tools. etry, so it cannot recognize potential problems (Denis, 2014, p. 9). All in
These include structural modeling, embedded data and links to construc- all, modeling processes in GAEs produce separate geometry and para-
tion documentation (Boeykens, 2012, p. 454). metric data files, so the modeling processes are challenging to continue
AAD tools can deal well complex freeform geometries because they in external tools.
are based on integrated scripting and visual programming (Boeykens,
2012, p. 454). Geometry consists only of geometrical parts such as points,
lines, curves and surfaces, which cannot contain complex semantic data, 7.3. AAB PROCESSES
unlike objects in object-oriented design software. Geometrical models As we see in Figure 32, AAB is also based on object-oriented parametric
are purely visual compositions of geometry, so they don’t contain the modeling. A modeling process is controlled via a script that uses algo-
database dimension and material-based knowledge associated with dig- rithms to perform tasks. This section too proceeds along the modeling
ital models (Zarzucki, 2012, p. 1). processes from input to actions and output.
On this basis, visual scripting would be more powerful if param- This study agues that there exist two kinds of AAB processes. More
eters could be linked to the objects of the model. Then visual scripting obvious is the process that creates new objects. The other modeling pro-
would not only bu used to generate geometry and relations but also to cess modifies existing objects. These processes can be combined in the

69 CHAPTER 7: MODELING
Definition of objects SIMULATION SOFTWARE Main
Object-oriented (or geometry-oriented) advantages
External software of AAB
e.g. Geometry Gym -
Karamba to Robot exporter e.g. Robot

PARAMETRIC CAD
GAE SIMULATION PLUGINS Heuristic methods SIMULATION ADD-ON
SOFTWARE
Object-oriented Object-oriented
Integrated to GAE (or geometry-oriented)
Integrated to CAD software

e.g. Karamba for Grasshopper “Rules of thumb”


e.g. 360 Structural
“Collected” Analysis for Autodesk Revit
property set
EXTERNAL
P P MAIN GRAPH Definition of objects OBJECT MODEL MODELING
SOFTWARE
GEOMETRY GEOMETRY Object-oriented
Heuristics

e.g. Tekla
e.g. Form finding in
G Kangaroo Meta-Heuristic methods

FITNESS FUNCTION

FITNESS VALUE (S)


OBJECTS
e.g. Galapagos
GENOTYPES

Modification of parameters
O

P P

DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE


Object-oriented
External software

e.g. Navisworks

INPUT ACTIONS OUTPUT

Figure 35  AAB PROCESSES CHAPTER 7: MODELING 70


same script. This study, however, doesn’t investigate the combination of is integrated into the interface. In the following sections the study pres-
these processes because they would get too complex. Instead, this study ents the main tools that can offer this functionality.
focuses mainly on the AAB process that creates new objects. Both pro- The process produces an object model, so the modeling process can
cesses are presented in the following sections and in Figure 35. be continued in object-oriented modeling software. However, the link to
AAB doesn’t exclude other computer-aided modeling methods, the script breaks if the software is external.
so the results of different modeling methods can be combined in the
same model. The same script can be use to produce both geometries 7.3.2. Modification of existing objects
and objects. Manually produced geometry and objects can also be com- This section focuses on processes that aim at the modification of exist-
bined into same model. However, manually modeled objects and geom- ing objects. As can be seen in Figure 35 this kind of a process can utilize
etry also needs to be updated manually, so manually controlled itera- objects and parameters as input. In this section the focus is in the pro-
tions are much slower than iterations through scripts. cess that uses geometry and parameters as an input.
Objects and parameters are the inputs of the process. Thus, the
7.3.1. Creation of new objects definition needs to use nodes that can import object-oriented data into
This section focuses on processes that aim at the creation of new objects. GAE. The data of objects is structured, so nodes need to be able to han-
As can be seen in Figure 35, this kind of process can utilize geometry and dle these structures.
parameters as an input. In this section the focus is on the process that The script is used to modify parameters of imported objects. The
use geometry and parameters as an input. scripts can be used iteratively to modify parameters or it can be used to
The script in GAE is divided into two parts. The first part defines automate tasks that are needed in modeling. For example, some tasks,
the host geometry for objects, so the definition needs only nodes that can such as ordered numbering of spaces, are not integrated into Revit, so
utilize geometry and parameters. The topological structure of the object Dynamo can be used to do these tasks through visual scripting.
model is defined in this definition, so relations between objects can be There are only a few tools that allow a fluent modification of objects
modified here. through visual scripting. In this study the main tools that can modify
The second part of the definition defines the objects themselves. In object data are the Geometry Gym plugin for Grasshopper and Dynamo
this definition the hosting geometry and needed parameters are linked for Revit. This functionality in other tools doesn’t exist at all or it is rel-
into one or more nodes that define the geometry. atively limited.
All in all, there are only a few advanced tools that can be used to
define objects algorithmically. There are some plugins for Grasshopper
that offer functionality to define objects. In Dynamo this functionality

71 CHAPTER 7: MODELING
Modeling method Key Concept Implementation of Function / Output
AAM (Algorithm Aided Modeling) - Geometry: objects are just - Geometry and relation between - Possibility to export the 3D of
geometry but the desiger them the design into CAD software
interprets them as walls, - No “components” with - Not a presentation tool. Needs
insulation or components parameters inside them another software to produce
-Real-time feedback - Easy to change the shape of the plans/sections
design
BIM (Building Information - Data - Definition of components, - All technical drawings are within
Modeling) - Objects or families with different materials, schedules the BIM database
characteristics are understood by - Difficult to change the geometry - Plans/sections are produced
the software as components with
physical properties

Figure 36  COMPARISON CHART BETWEEN AAD AND BIM


The diagram is adapted from the one made by Denis (2014, p. 14) but
some changes are made so that (better) terms fit to the purposes of the
study. CHAPTER 7: MODELING 72
7.3.3. Advantages and drawbacks of AAB manual modifications are in danger of disappearing. All in all, the pos-
One of the main advantages of AAB is that scripts can be used to con- sibilities to combining manual modeling and visual scripting should be
trol large amounts of objects simultaneously. One change to the script further developed.
can affect a large amount of objects. A topological structure of the object
model is defined in a script. Thus, the script can be used to control the
overall shape of the project. This makes modeling faster, more open for
modeling iterations.
Creation of precise joints is one of the main problems in AAB
processes. This is because objects are based on object-standards. Thus,
objects will always contain limitations. The built in script of an object
defines the generality of the object. Standard objects are usually based
on Cartesian coordination. Thus, the problem is that these objects don’t
behave precisely when they are used in the other coordinate systems that
are commonly used in scripts. The only way for precise AAB is to use
Cartesian coordinates in the script.
BIM software contain pre-defined and user defined objects. Dynamo
shows that, AAB can take advantage of user defined objects by using them
in GAE. Thus, object modeling is not only limited to the objects that a
BIM software offers. Then the BIM software can “understand” what the-
ses objects are. However, the challenge is that these user-defined objects
may be difficult to export to external software for further design.
GAE that is integrated to the BIM software should allow visual
scripting and manual modeling to happen simultaneously for same
objects. Scripts could control the overall shape of the model while man-
ual modifications could be made in BIM software. Manual modifications
should be attached to the objects even if the script updates. The chal-
lenge is to find the right ways to treat manual modifications if the script
modifies the model. BIM software should at least warn the designer if

73 CHAPTER 7: MODELING
CHAPTER 7: MODELING 74
7.4. AAB TOOLS The section focuses on modeling processes, so resulting objects are not
Following sections investigate AAB tools. The Chapter 6 showed that shown. Objects are based on standards so, these tools produce objects
existing data exchanges can be divided into direct and indirect data that have similar geometry. Instead, the data that is embedded into the
exchanges. AAB processes are based on data exchanges if these tools are objects varies depending of the AAB tool. Following examples of model-
not integrated into a system that can manage objects. ing processes present the main parameters that are used to define objects.
If GAE is integrated into CAD, the management of objects can hap-
pen in the CAD, if the requires add-on is installed into the software. For
example, Grasshopper (GAE) is integrated into Rhinoceros (CAD), so
there are some add-ons that enable the management of objects in Rhi-
noceros. Instead, if GAE is integrated into BIM software the manage-
ment of objects can happen directly in the BIM software. For example,
Dynamo (GAE) is integrated into Revit (BIM), so objects can be man-
aged in Revit. On this basis AAB tools are divided into four classes.:

AAB TOOLS BASED ON DATA EXCHANGE


Direct data exchange:
– Rhino-Grasshopper-Archicad Connection
– Grevit
– Lyrebird
Indirect data exchange:
– Geometry Gym (IFC)

INTEGRATED AAB TOOLS


Integrated to CAD:
– Visualarq
– Elefront
Integrated to BIM:
– Dynamo for Revit

75 CHAPTER 7: MODELING
7.4.1. Rhino-Grasshopper-Archicad Connection The tool has various nodes that are divided into the following five groups.
These groups are called: Parameter components, design components,
Grasshopper‐ArchiCAD Live Connection tool can link Grasshopper to document components, reshape components and input nodes. (Graphi-
Archicad. The tool is used to define objects in Grasshopper and exchange soft, 2015, p. 8.) An example modeling process is presented in Figure 37.
them with ArchiCAD. (Graphisoft, 2015, p. 4.) The key features and ben- The example process of the figure focuses on Design Components that
efits of the tools are (Graphisoft, 2015, p. 4): produce 3D geometry.

– Translating simple geometry created in Rhino into BIM construction


objects
– Bi‐directional connection – no need to export/import files for data
exchange
– Dynamic (live) connection while editing
– Direct and simultaneous graphical feedback from both Rhinoceros
and Archicad

Send command Input data Object data Output data

Automatic / manual send Trigger Object prameters

Base control geometry


-basecurves Design Component
-origin points
Export/import
Main parameter to Archicad ArchiCAD
-parameter 1 -Selection of objects
-parameter 2 -Defining parameter values

Figure 37  DEFINING OBJECTS WITH THEGRASSHOPPER‐AR-


CHICAD LIVE CONNECTION TOOL CHAPTER 7: MODELING 76
7.4.2. Grevit Figure 38 shows that all objects are defined in object node. The
Grevit is a Grasshopper plugin that can link Grasshopper to Revit and defined objects are connected to a ‘Send’ component, which controls all
Autocad. Grevit can define and send native Revit objects into Revit. sending operations. From Revit objects can be exported to other design
Grevit has specific nodes for each object type and one main node tools.
for sending native object data. Typical parameters are built into object
nodes but designer can also define specific parameters in Grasshopper.
Sent Grevit objects can be updated by sending object data into Revit.
Grevit has rough “single surface” preview for objects in Rhinoceros.
Modeling tests showed that Grevit still has some limitations.
Regardless, Grevit is still a potential modeling-aid at least for the early
phases of BIM design.
Send command Input data Object Send node Output data

Send button Element information

Sending settings Errors

Levels
Send
-Contains the selected object types
and parameters from Revit -Export/import
Host geometry
-baselines Revit/Autocad
-points

Typical parameters Revit (Autocad) object


-built in BIM software

Controlling parameters
values in Revit/Autocad
-parameter 1
-parameter 2
...

77 CHAPTER 7: MODELING Figure 38  DEFINING OBJECTS WITH GREVIT


7.4.3. Lyrebird objects cannot be updated anymore. One restrictive lack of functionality
Lyrebird is a Grasshopper plugin developed by the design technology is that there is no preview to for the generated geometry in Rhinoceros.
studio LMNts. Lyrebird is another example of linking Grasshopper to All in all, Lyrebird is probably the simplest public plugin that exists
Revit. The first release of Lyrebird can create and control all Revit fami- between Grasshopper and Revit so far. Modeling tests showed that Lyre-
lies from Grasshopper. bird still has many limitations. Anyway the core idea to have only one
Lyrebird contains only one node that is used to model all Revit node for all kind of Revit families is simple and in that way powerful.
objects as can be seen in Figure 39. Lyrebird utilizes families which are
open in the current Revit project. Lyrebird also has also some basic con-
cepts to control geometry. Objects in particular set can be selected. If
needed their connection to Lyrebird can be erased from objects and then

Send command Input data Object Output data

Send Element information

Base control geometry Errors


-baselines
-origin points
-adaptive component points
Export/import
Object
-Contains selected object type Revit
Orientation and parameters from Revit -Selection of particular set of objects
-Erasing Lyrebird data from objects
Level

Parameters’ values to Revit


-parameter 1
-parameter 2 BIM software
...

Figure 39  DEFINING OBJECTS WITH LYREBIRD CHAPTER 7: MODELING 78


7.4.4. VisualARQ
VisualARQ provides a modeling add-on for Rhinoceros and plugin for
Grasshopper. Both tools use a shared object library, so both tools can be
used collaboratively. Modeled objects can be previewed in Rhinoceros.
Objects can be used as host geometry for new objects.
VisualARQ for Rhinoceros contains an IFC exporter, which allows
sending all VisualARQ objects to other BIM programs. VisualARQ also
contains many nodes that can deconstruct parameters that object nodes
contain.
In Figure 40 VisualARQ has two main nodes for each physical
object category. The ‘Object options’ node and ‘Object’ node both for the
creation of the object.

Input data Object options Object Manual BIM Combination model

Referencing BIM geometry Manually modelled


VisualARQ objects
Combined VisualARQ
model in Rhinoceros
Base control geometry Baking
-baselines
-points

Export
Object
IFC
Element parameters Object options
Import

BIM software

79 CHAPTER 7: MODELING Figure 40  DEFINING OBJECTS WITH VISUALARQ


7.4.5. Elefront The created objects contain data which Rhinoceros can read. Ele-
Elefront is a Grasshopper plugin; which enables designer to create “para- front is an example of a tool that allows the designer to define their own
metric” objects in Rhinoceros. Elefront differs from the other presented objects but which offer only limited capabilities. Elefront doesn’t offer an
Grasshopper plugins because it allows the designer to define “paramet- object library, so the created object cannot be called “real” objects.
ric objects” of out of basic geometry in Rhinoceros.
As can be seen in Figure 41 Elefront allows the designer to cre-
ate smart objects by defining parameter names and their values for each
object. This group of objects can then be selected, and the selection can
be filtered and modified for further work.

Input geometry Keys and key values Parameters Baking Rhino geometry

Geometry

Keys “Smart geometry”


-key 1
-key 2 Bake
“Parameter names” Rhino
-key 3
...
Parameters
Key values
-value 1
-value 2
“Parameter values” -value 3
...

Input geometry Export

Rhino Reference objects Filter objects


Export Optional

Iterative loop

Figure 41  DEFINING OBJECTS WITH ELEFRONT CHAPTER 7: MODELING 80


7.4.6. Geometry Gym Generated IFC models can for example contain data on building
Geometry Gym, developed by Jon Mirtschin, is a Grasshopper plugin services, schedules and timetables (4D and 5D BIM). Thus the generated
which focuses on open BIM processes. Geometry Gym utilizes neutral IFC models can be utilized comprehensively in various IFC supported
OpenBIM formats such as IFC2x3, IFC2x4, CIS/2 and Steel Detailing tools.
Neutral Format (SDNF). All in all, Geometry Gym is a really potential plugin because it is
Objects can be previewed in Rhinoceros. Geometries of objects based on OpenBIM processes and formats. For a newcomer the struc-
can also be baked into Rhinoceros as for further work. Geometry Gym ture of IFC is complicated to work with but the plugin allows various
allows many importing and exporting functionalities from Rhinoceros processes, when the designer gets familiar with it.
and Grasshopper. In Grasshopper Geometry Gym enables modification As can be seen in Figure 42, IFC building object generation requires
of IFC files. that all needed IFC parameters are defined. The diagram below shows
only the main order for building an IFC object.

Send command IFC building project IFC element type IFC element Output data
standard case

Element geometry preview

IFC-building IFC-storey

IFC element standard case


Element data
IFC-project IFC element type
Send settings
Send button
IFC materials
-material 1 IFC material layer/ Export
-material 2 IFC material profile
IFC
...
Import
Base control geometry
-baselines
-points BIM software

81 CHAPTER 7: MODELING Figure 42  DEFINING OBJECTS WITH GEOMETRY GYM


7.4.7. Dynamo for Revit In Figure 43 is shown the basic object creation and modification
Dynamo is open source GAE for Autodesk Revit. The main advantage processes in Dynamo. Dynamo contains object node for main object
of Dynamo compared to Grasshopper is that Dynamo is integrated into types. Parameters of the object can be defined in GAE. The modification
BIM software. On this basis, Dynamo is one of the most promising GAEs of created or imported objects happens by setting parameters to objects.
regarding AAB.
This enables many new possibilities for object creation, object mod-
ification and data management. Dynamo can extend the capabilities of
Revit, by enabling custom design processes. Objects can be controlled
from both Revit and Dynamo.

Selected Input Input data Object Input parameters Input data


geometry

Model elements Element type

Object Specific parameter in Revit Set parameter


Base control geometry -parameter 1
-baselines -parameter 2
-points ...

Typical parameters
-built in

Element type

Figure 43  DEFINING OBJECTS WITH DYNAMO CHAPTER 7: MODELING 82


8. SIMULATION AND OPTIMIZATION

83
84
This chapter investigates performance evaluation methods and their rela- information, so the performance evaluation methods are used to predict
tion to AAB. Figure 44 presents relations of the main simulation and opti- answers to these criteria.
mization methods that are discussed in the study. The figure shows also Currently the simulation tools are mainly used to analyze the proj-
analysis methods that are seen to be part of simulation methods, because ects. Utilization of the simulation processes need specialized expertize.
simulation methods produce analysis information. The use of the simulation methods varies in each project and people who
As figure shows, object-oriented simulation methods and tools are use them are not always the same. Thus, the usefulness of the current sim-
related to the field of OOD. Script based simulation methods are part of ulation tools is limited. Interpretation of the results is made by experts,
AAD. There are also geometry-based simulation methods, but they are usually in the late phases of the project (Boyekens, 2011, p. 456).
not discussed further in the study. In this study all used optimization The development of the simulation methods has focused on sev-
tools are controlled through GAE, so they are seen to be part of AAD. eral fields. These include decision support tools that offer user-friendly
This study explores how the simulation and optimization methods interfaces and visualization techniques. In addition, the development of
and tools can be integrated into AAB processes. The main intention of performance based optimization of the models offers possibility to take
the chapter is to study, how AAB could make simulation and optimiza- full advantage of computational power. (Malkawi, 2005, p. 90.)
tion processes more effective. The role of the simulation tools should be rethought, so that they
Aim of the simulation processes is to find optimal result for a design would support dynamic design processes. In other words, the simulation
problem. Optimal result can be found by using iterative process. It con- tools should not only be able to make analyses but also to unite the design
sist of evaluation of a current design proposal, a proposition for alterna- process. The interoperability between the model and simulation meth-
tives, solving problems in collaboration and determination of the basis ods should exist during the whole design process. Comprehensively inte-
for next the loop (Mirtschin, 2011, p. 2). grated simulation needs a framework that can be used whenever needed.
The development of these process-driven simulations is still in its early
stages. (Malkawi, 2005, pp. 89–91.) An integration of AAB and simula-
8.1. SIMULATION tions can facilitate process driven simulations. However, this integration
There is a need to predict the behavior of the building during its life cycle. would need much more research.
That’s why there are many performance criteria that need to be evalu- Modeling tools have shifted from procedural to object-oriented
ated during the design process. These include cost estimation, accessi- approach and simulation tools have followed the shift. Object-oriented
bility, energy, lightning, structure, durability, acoustics, transport, plan- scripting has allowed performance simulations to be more flexible and
ning and many others. These criteria are usually based on quantitative expendable. Shift to object-oriented approach has enabled collaborative
exchange of simulations. It has also enabled simulation tools to develop so

85 CHAPTER 8: SIMULATION AND OPTIMIZATION


OOS (Object-Oriented Simulation) AAS (Algorithm-Aided Simulation)
OOA (Object-Oriented Analysis) AAA (Algorithm-Aided Analysis)
OBJECT-ORIENTED PARAMETRIC SIMULATION AAA (Algorithm-Aided Optimization)
-PARAMETRIC
-(NON-ALGORITHMIC) GEOMETRY ORIENTED PARAMETRIC/ALGORITHMIC SIMULATION AND OPTIMIZATION
-OBJECTS
-PARAMETRIC
-ALGORITHMIC

e.g. Karamba

SIMULATION SIMULATION
TOOLS METHODS
OOD SIMULATION
METHODS
SIMULATION
TOOLS
MODELING
TOOLS AAB OPTIMIZATION OPTIMIZATION
TOOLS METHODS
AAD

e.g. Scan-and-Solve for Rhinoceros

e.g. Galapagos, Octopus

SIMULATION SIMULATION
METHODS TOOLS
e.g. Robot
“GOD”
GEOMETRY-ORIENTED SIMULATION
-(NON-PARAMETRIC)
-(NON-ALGORITHMIC)
-(NO OBJECTS)
Figure 44  OBSERVED SIMULATION AND OPTIMIZA-
TION METHODS CHAPTER 8: SIMULATION AND OPTIMIZATION 86
that they support modeling to analysis processes by utilizing object-ori- second challenge is to find quantitative measurement that can be used
ented data exchange. Practically this means utilization of IFC in simula- in the optimization process. (Scheurer, 2013, p. 193.)
tion software. (Malkawi, 2005, p. 88.) The optimization methods can be divided into two main catego-
The simulation tools can be used to support performance-driven ries. These are called heuristic and meta-heuristic algorithms. They don’t
design with the aid of optimization and partial automation. Thus, we can give direct answer to a design problem. Instead of controlling the result
take advantage of the human computer interaction. Addition, frame- itself, these algorithms will lead us to control the result through process.
works and standards are developed to facilitate integration. (Malkawi, Both methods utilize external analyzes in their processes. The simula-
2005, p. 87.) On this basis, the optimization methods are investigated in tion methods can provide analysis data for these processes. That’s why
the next section. Later examples show how these methods can be utilized the integration of the simulation and the optimization methods should
in practice. be developed.

8.2.1. Heuristic methods


8.2. OPTIMIZATION Heuristic methods are often used to optimise a single design objective.
Generative computational methods have led to a shift from making the Heuristic methods are good for solving problems that can be clearly
form to finding the form (Kolarevic, 2005, p.195). This means that com- stated and for searching design solutions that are foreknown. Heuris-
puter can be utilized to produce and find design solutions in constraints tic methods usually include some kind of rules of thumb or experience
that we define. In AAD design solution can be found manually by alter- of designer that is embedded into the optimization algorithm. Thus, the
ing geometry and parameters of the script. nature of heuristic methods is predictive and they don’t guarantee com-
Generative optimization methods are used in design, if the solu- pletely optimal result. An optimization process can be less predictive if
tions space is too large to be explored manually. That’s why computer is there exists a feedback of the process. For example analysis methods can
used to explore optimal design on behalf of the designer. (Mueller and provide feedback if they are integrated into the design process. (Hard-
Smith, 2013, p. 147.) The concept of the ‘solution space’ is further dis- ing, 2015, p. 20, 35.)
cussed in the Chapter 9. Optimization means of finding the best possi- Heuristic methods guide how the problem is solved. Instead, meta-
ble solution for the design problem. Optimization is used to deal with heuristic algorithms distance themselves from the question how the prob-
vast amount of solutions in the solution space. lem is solved. They are used for problems that cannot be clearly identified.
The first challenge in the optimization processes is to find unbiased Unlike in heuristic algorithms, knowledge of designer is not embedded
definition of the qualities that the design solution needs to contain. The into the metaheuristic algorithms. Thus, the metaheuristic algorithms

87 CHAPTER 8: SIMULATION AND OPTIMIZATION


PERFORMANCE EVALUATION SOFTWARE
LEGEND Object-oriented
External software

DEFINED IN FIGURE e.g. Robot

EXTERNAL SOURCE
PARAMETRIC PERFORMANCE EVALUATION ADD-ON
BIM SOFTWARE Object-oriented
AAB Integrated to BIM

SIMULATION e.g. 360 Structural Analysis for Autodesk Revit

OPTIMIZATION
P MANUAL MODELING OBJECT MODEL EXTERNAL
Object-oriented MODELING
O OBJECTS Manual / built in relations SOFTWARE
O Parameters
P PARAMETERS

G GEOMETRY
e.g. Tekla
DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE
MAPPINGS Object based
External software
MAIN DATA FLOW e.g. Navisworks
COMMON DATA FLOW

SECONDARY DATA FLOW INPUT ACTIONS OUTPUT

PERFORMANCE EVALUATION SOFTWARE


Geometry based
External software

e.g. Robot
Figure 45  MODELING AND SIMULATION PROCESSES IN OOD CHAPTER 8: SIMULATION AND OPTIMIZATION 88
PARAMETRIC CAD
SOFTWARE PERFORMANCE EVALUATION
ADD-ON
Geometry based
are able to adapt to the changing objectives and constraints of the design. rent state of the performance. The metaheuristic methods take shortcuts
(Harding, 2015, pp. 35–37.) to find design solutions because the search of the whole design space
The heuristic methods can be divided into three types depending would waste computing time. Some of the most well known metaheuris-
how they are related to external analyses. The first category don’t utilize tic algorithms are Evolutionary Algorithms (EAs) and Simulated Anneal-
the analyses at all, so they provide the result without information outside. ing (SA). The choice for an appropriate metaheuristic algorithm is a trade
These algorithms are typically based on problems that require numeri- between the computing time and the nature of the solution space. For
cal data which cannot be produced analytically. All data embedded into example, Grasshopper plugin called Galapagos is based on Evolutionary
the algorithm is provided by the designer of the algorithm. Structural Algorithms (EAs). (Harding, 2015, pp. 35–51.)
form-finding is maybe the most well known method that belongs to this Some metaheuristic algorithms can be used to optimize many
group. (Harding, 2015, pp. 20–26.) This type of heuristic algorithm is design objectives simultaneously. These kinds of methods are multi-ob-
used in Figures 53 and 54. jective optimization processes. For example, Grasshopper plugin called
The second type of the heuristic methods uses the external analy- Octopus can manage multiple design objectives. (Harding, 2015, p. 48.)
sis but the information from the analysis stays constant. Thus, these algo- Octopus uses evolutionary algorithms called SPEA-2 and HypE (http://
rithms are not based only on the knowledge of the designer but also on www.food4rhino.com/project/octopus?ufh).
some type of external data. (Harding, 2015, pp. 20–26.) This type of heu-
ristic algorithm is used in Figure 55.
The third type of the heuristic methods uses the external analysis 8.3. SIMULATION PROCESSES IN OOD
but the provided information from the analysis is updated at each itera- Figure 45 shows the main modeling and simulation processes related
tion. The heuristic algorithm generates changes to the model, so the data to OOD. BIM enables designers to produce objects that can be used in
from the analysis changes. In other words, there exists a feedback loop in object-oriented simulation tools. BIM is object-oriented design method,
the process. (Harding, 2015, pp. 20–21.) This type of heuristic algorithm so parameters are used to define objects. The parameters define proper-
is used in Figure 56. ties of the objects.
Object data and property sets are needed so that the information
8.2.2. Metaheuristic methods can be exported to other software. Usually external object-oriented soft-
Metaheuristic algorithms are based on iterative search processes. An aim ware doesn’t use individual parameters, so they are organized into prop-
of the process is to find a design solution that has maximal ‘fitness’. Fit- erty sets (Eastman et al., 2008, pp. 46–47).
ness refers to the performance of the design solution. A process needs The property sets are often integrated into BIM software. However,
to contain some sort of analysis at each iteration to understand the cur- user-defined property sets are needed to support a wider range of tools

89 CHAPTER 8: SIMULATION AND OPTIMIZATION


SECONDARY DATA FLOW INPUT ACTIONS OUTPUT

PERFORMANCE EVALUATION SOFTWARE


Geometry based
External software

e.g. Robot

PARAMETRIC CAD
SOFTWARE PERFORMANCE EVALUATION
ADD-ON
Geometry based
Integrated to CAD software
GAE

PERFORMANCE EVALUATION PLUGIN


Geometry based
Integrated to GAE
e.g. Scan-and-Solve
e.g. Millipede

P P GRAPH P GEOMETRIC MODEL EXTERNAL


MODELING
GEOMETRY GEOMETRY GEOMETRY SOFTWARE
Geometry-oriented
FORM FINDING

e.g. Autocad
Eg. Kangaroo OPTIMIZATION
G Meta-Heuristic methods
FITNESS FUNCTION

FITNESS VALUE (S)


e.g. Galapagos
GENOTYPES

DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE


Object based
External software

e.g. Excel

INPUT ACTIONS OUTPUT

Figure 46  MODELING AND SIMULATION PROCESSES IN AAD CHAPTER 8: SIMULATION AND OPTIMIZATION 90
if pre-defined property sets don’t support the required data exchanges. options is a challenge for the whole design team. AAD enables many
Many of the current BIM software define default parameters for objects. simulation and optimization processes, which can be used to evaluate
However, external object-oriented simulation software need specific architectural design options. These new processes help designers to make
property sets, so designers or the software needs to define all required design decisions.
parameters to objects. (Eastman et al., 2008, p. 47.) There are multiple simulation plugins that can be installed into
Property sets are challenging to control because each pair of the GAEs. They enable designer to generate simulations automatically. Thus,
tools needs a specific property set. There are at least three ways to define design options and performance evaluations can be explored iteratively.
properties (Eastman et al., 2008, pp. 46–47.): Also external simulation software can also be used to evaluate the
performance. AAM processes generate geometry and parameters that
– By predefining them in the object libraries so they are added to the are separate, so the main challenge regarding data exchanges is that the
design model when an object instance is created. geometry and parameter data need to be exchanged separately if an exter-
– By the user adding them as - needed for an application from a stored nal software needs them. GAEs are integrated into CAD software that
library of property sets. offer also some simulation functionalities. They commonly behave like
– By the properties being assigned automatically, as they are exported to external simulation software because data exchanges are needed to exe-
an analysis or simulation application. cute these simulations.
There are also some optimization plugins that can be used in GAEs.
The first way is good for standard data exchange processes. The second The design objective is not integrated into metaheuristic optimization
is relatively time consuming because it requires computational power. algorithms. That’s why even one plugin can be really multi–functional.
The third way is light but it requires the development of comprehen- The metaheuristic algorithms need analysis data that can be generated
sive material tag that can be used for all data exchanges. Tag is used to in fitness function or in simulation plugins.
attach a property set to each object. (Eastman et al., 2008, pp. 46–47.) Heuristic algorithm contains design objectives, so single heuris-
tic algorithm fits only to specific optimization task. That’s why heuristic
algorithms are often integrated into plugins that are meant only to spe-
8.4. SIMULATION AND OPTIMIZATION cific design tasks. Many heuristic algorithms need analysis data that can
PROCESSES IN AAD be generated with simulation plugins.
Figure 46 shows the main modeling, simulation and optimization pro-
cesses in AAD. AAM enables designer to rapidly produce and explore
multitude of design options. Effective evaluation of different design

91 CHAPTER 8: SIMULATION AND OPTIMIZATION


Definition of objects SIMULATION SOFTWARE Main
Object-oriented (or geometry-oriented) advantages
External software of AAB
e.g. Geometry Gym -
Karamba to Robot exporter e.g. Robot

PARAMETRIC CAD
GAE SIMULATION PLUGINS Heuristic methods SIMULATION ADD-ON
SOFTWARE
Object-oriented Object-oriented
Integrated to GAE (or geometry-oriented)
Integrated to CAD software

e.g. Karamba for Grasshopper “Rules of thumb”


e.g. 360 Structural
“Collected” Analysis for Autodesk Revit
property set
EXTERNAL
P P MAIN GRAPH Definition of objects OBJECT MODEL MODELING
SOFTWARE
GEOMETRY GEOMETRY Object-oriented
Heuristics

e.g. Tekla
e.g. Form finding in
G Kangaroo Meta-Heuristic methods

FITNESS FUNCTION

FITNESS VALUE (S)


OBJECTS
e.g. Galapagos
GENOTYPES

Modification of parameters
O

P P

DESIGN AND CONSTRUCTION MANAGEMENT SOFTWARE


Object-oriented
External software

e.g. Navisworks

INPUT ACTIONS OUTPUT


Figure 47  AAB CONNECTS MODELING, SIMULATION
AND OPTIMIZATION PROCESSES CHAPTER 8: SIMULATION AND OPTIMIZATION 92
8.5. AAB AND SIMULATION AND in GAE. Thus, iterative processes become fast and exchange processes
can be even automatically managed.
OPTIMIZATION PROCESSES
Figure 47 shows that AAB can connect the main modeling, simulation
and optimization processes of OOD and AAD. AAB is managed in GAE
so, geometry-based simulation and optimization plugins can be used
to define the host geometry for objects. However, the real advantage of
AAB is that objects can be used in external object-oriented simulation
software.
AAB enables double-iterative simulation processes. This means
that the simulation process can be managed in two stages. The process
consists of two loops, which are utilized sequentially. This means that pre-
liminary simulations are done in GAE. These simulations are fast so the
process can contain multiple iterations. Designer can manage these sim-
ulations independently. When the sufficient precision level is achieved,
the process can be continued in external software. AAB generates objects,
so secondary simulations can be handled in external object-oriented soft-
ware. Simulation plugins need parameters so designer needs to define
them in the graph. The study suggests that also this organized set of
parameters can be called as property set.
There are also some plugins that are used not only for simulations
but also for modeling the geometry. Karamba is an example of such a
plugin. In Karamba colored geometry is used to visualize the behavior
of the model.
Optimization algorithms can be used to optimise parameters that
are input into object definition. However, the real advantage of object
definition is also in this case representational.
All in all, AAB can make external simulation processes faster
because all parameters related to simulation and objects can be defined

93 CHAPTER 8: SIMULATION AND OPTIMIZATION


CHAPTER 8: SIMULATION AND OPTIMIZATION 94
8.6. TESTING SIMULATION PROCESSES 8.6.2. 4D BIM and 5D BIM
Multiple test projects were used to study simulation processes in GAE. 4D (time) and 5D (cost) BIM are used for project management and con-
The main aim was to investigate currently possible simulation pro- struction. 4D BIM refers to the integration of object model with con-
cesses which in a way or another are related to AAB. Test processes are struction schedule in order to visualize the sequence of construction.
presented in following figures: 4D BIM can be used to review construction phases generally, sort out
construction order, and examine critical paths and logistical issues. (Sin-
– Structural analysis (Figures 48 and 49) clair, 2012, p. 9.)
– 4D BIM (Figure 50) Geometry based models are challenging to use in 4D and 5D BIM
– Clash detection (Figures 51 and 52) software. That’s why in 5D BIM also the cost data is integrated into object
models. Cost information and quantity schedules help to estimate the
8.6.1. Structural analysis costs of the design. (Sinclair, 2012, p. 9.)
Structural analysis tools allow designers to evaluate structural perfor- One of the main advantages is that GAE can be used not only for
mance of the model. Data exchanges to structural analysis software has modeling objects but also for controlling time and cost aspects of the
commonly been controlled by authoring spreadsheets. Geometrical data objects. Integration of GAEs with 4D and 5d BIM tools needs further
exchange has been done through CAD formats such as DWG and DXF. development. Time and cost aspects should be taken into account already
These formats contain only geometrical information, so analysis parame- in the early stages of the design.
ters such as constraints, materials and loads need to be defined manually.
Manual work leads to extra coordination and maintenance. (Mirtschin, 8.6.3. Clash detections
2011, p. 5.) Clash detection is analysis method, which is used to locate clashes
AAB processes offer new ways to evaluate alternative design between objects. Multiple disciplines provide object models, so com-
options. Analysis parameters for structural analysis can be defined and bination models are used to locate clashes. Often possible clashes are
controlled in GAE. (Mirtschin, 2011, p. 5.) Pre-defined parameters are located visually in CAD or BIM program but . AAB generates objects so
able to automate structural analysis processes. Thus, iterative processes they can be exported to clash detection software. Clash detection can be
are possible. Generally geometrical data contains points, support points, made with other object models.
loads and profile lines or curves. Parameter data contains directional
forces and reactions.

95 CHAPTER 8: SIMULATION AND OPTIMIZATION


8.7. TESTING OPTIMIZATION PROCESSES 8.7.2. Structural optimization
Multiple test projects were also used to study the optimization process- Structural optimization is needed when the structure needs to be defined
es in GAE. The main aim was to investigate currently available optimi- by many constraints or the design process aims to take into account many
zation tools that in a way or another are related to AAB. Test processes design objectives. Structural optimization aims to minimize or maximize
are presented in following figures: of objectives. “Structural optimization is an inverse process in which
parameters are implicitly/indirectly optimized to find the geometry of a
– Form finding (Figures 53 and 54) structure such that an objective function criterion is minimized.” (Adri-
– Structural optimization aenssens et al. 2014, p. 3.)
– Size (Figure 55) Structural optimization has traditionally three main classes,
– Topology (Figure 56) depending on the variable. These classes are called: shape optimization,
– Shape (Figure 57) topology optimization and sizing optimization. The shape optimization
aims to optimize the shape, without modifying the topology. The topol-
8.7.1. Optimization: form finding ogy optimization aims to optimize the structure, whose shape is not
Form finding method is used to find optimal form for intended structure. defined yet. Connectivity of the nodes in the structure is defined. Also
Form finding can be divided into ‘classical’ and ‘modern’ form finding. the existence or absence of objects is defined. The sizing optimization
In classical form finding the form is defined by the relationship between aims to optimize structural cross-sections without affecting the shape
form and forces. Modern form finding has broader meaning. It means or topology. (Adriaenssens et al. 2014, p. 4.)
the process of finding appropriate architectural and structural shape. Examples explores each optimization class individually. Grasshop-
(Coenders and Bosia, 2006, s. 275.) per plugin called Karamba is used for all examples because it can be used
This chapter focuses on classical form finding. There ‘form-active’ for all optimization classes.
systems are used to find ‘form-passive’ shell structures. The definition of
the classical form finding is (Adriaenssens et al. 2014, p. 2): “Form find-
ing is forward process in which parameters are explicitly/directly con-
trolled to find an “optimal” geometry of a structure which is in static equi-
librium with a design loading.” All form finding examples are made in
Grasshopper because the study didn’t find any form finding tools pub-
lished for Dynamo.

CHAPTER 8: SIMULATION AND OPTIMIZATION 96


Figure 48  STRUCTURAL ANALYSIS 1
1.Grasshopper: Definition of the host geometry.
The example presents an object-oriented data exchange process from
Karamba to Robot. The example shows that preliminary object-oriented
analysis plugin can be connected to external analysis software.
Preliminary analyses can be done iteratively in Karamba because
the analysis updates automatically. Geometry Gym plugin for Karamba
is used to export Karamba object data to Robot for further structural
analysis. The most relevant parameters in the process are point loads,
supports, beam cross-sections and beam materials.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper: 2. Karamba: Parameters and definition of objects.
Definition of the host geometry. 2. Karamba: Parameters and definition
of objects. 3. Karamba: Structural analysis. 4. Geometry Gym export to
Robot. 5. Robot: Structural analysis model.

5. Robot: Structural analysis model.

3. 4.
1. 2.

97 CHAPTER 8: SIMULATION AND OPTIMIZATION


Figure 49  STRUCTURAL ANALYSIS 2
0. Rhinoceros: Input curves for the graph.
The example presents an object-oriented data exchange process from
Dynamo to Robot. The example shows that preliminary object-oriented
analysis plugin can be connected to external analysis software.
Dynamo Plug-in for Autodesk Robot Structural Analysis is a free
plugin for Dynamo. Plugin enables to model parametric and structural
frame models. It allows designers to control Autodesk Robot Structural
Analysis software from Dynamo. (https://beta.autodesk.com/)
In the example, input curves are imported from Rhinoceros to
Dynamo. The import is managed by Rhynamo, that is open-source
plugin for Dynamo. Dynamo shows a preview of parameters that Robot 3. Dynamo: Definition of the host geometry.
Structural Analysis (RSA) calculation generates. Thus, designer can
analyze the structure already in Dynamo. The most relevant parameters
in the process are point loads, supports, beam cross-sections and beam
materials.

The process: 0. Rhinoceros: Input curves for the graph. 2. Rhynamo:
Curve import from Rhinoceros. 3. Dynamo: Definition of the host geom-
etry. 4. Dynamo to Robot plugin: Parameters and definition of objects. 5.
Robot Structural Analysis (RSA). 6. Robot: Structural analysis model. 7.
Dynamo: Previewing analysis results. 6. Robot: Structural analysis model.

2. 4. 7.
3.
5.

CHAPTER 8: SIMULATION AND OPTIMIZATION 98


Figure 50  4D BIM SIMULATION
1. Grasshopper: Definition of the host geometry.
The example presents an object-oriented data exchange process from
Grasshopper to Navisworks. The example shows that GAE can manage
iterative 4D BIM simulations in external object-oriented software.
In the example, Geometry Gym plugin is used to define IFC
objects. Objects are scheduled in the graph, so they have the correct
building order. Geometry Gym plugin is used to input sequence data
to objects. Object data is sent to Autodesk Navisworks, which is used
to visualize construction sequences. Iterative data exchange process is
handled manually. The most relevant parameters in the process are the
start time and the duration of the sequence. 2. Geometry Gym: Parameters and IFC definition of objects.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper:
Definition of the host geometry. 2. Geometry Gym: Parameters and
IFC definition of objects. 3. Object scheduling. 4. Geometry Gym: Time
sequence parameters. 5. Navisworks: Sequence animation.

5. Navisworks: Sequence animation.

2.
4.
3.
1.

99 CHAPTER 8: SIMULATION AND OPTIMIZATION


Figure 51  CLASH DETECTION 1
1. Grasshopper: Definition of the host geometry.
The example presents an object-oriented data exchange process from
Grasshopper to Navisworks. The example shows that external anal-
ysis software can execute clash detection processes with objects that are
defined in GAE.
In the example, Geometry Gym plugin is used to define IFC
objects. The graph defines IFC walls and beams. Objects are exported to
Tekla BIMsight that is used to analyze clashes between walls and beams.
Iterative data exchange process is handled manually.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper: 2. Geometry Gym: Parameters and IFC definition of objects.
Definition of the host geometry. 2. Geometry Gym: Parameters and IFC
definition of objects. 3. Clash detection in TeklaBIMsight.

3. Clash detection in TeklaBIMsight.

2.
1.

CHAPTER 8: SIMULATION AND OPTIMIZATION 100


Figure 52  CLASH DETECTION 2
1. Rhynamo: curve geometry from Rhinoceros to Dynamo.
The example presents an object-oriented data exchange process from
Grasshopper to Navisworks. The example shows that external anal-
ysis software can execute clash detection analyzes with objects that are
defined in GAE. In addition, clash data can be exported back to GAE
and clashing elements can be visualized in BIM software. Clashes are
located instantly but corrections of objects are handled manually.
DynaWorks is Dynamo package, which interacts with Navisworks
(http://bim42.com/tag/dynamo/). Rhynamo is used to export host geom-
etry from Rhinoceros to Dynamo. Beam and wall objects are defined in
Dynamo. Objects are exported to Navisworks, which is used for clash 4. Navisworks: Clash detection.
detections. Dynaworks can export clash data back to Dynamo. Object
IDs are used to select clashing objects. Clashing information is added
to “Comment” parameter of objects. Selection filter is used to highlight
clashing walls.

The process: 0. Rhinoceros: Input curves for the graph. 1. Rhynamo:


Curve import from Rhinoceros. 2. Dynamo: Definition of the host geom-
etry. 3. Dynamo: Parameters and definition of objects. 4. Navisworks:
Clash detection. 5. DynaWorks: Analysis data import to Dynamo. 6. 6. Revit: Selecting and highlighting clashing objects.
Revit: Selecting and highlighting clashing objects.

1. 2. 3. 5.

101 CHAPTER 8: SIMULATION AND OPTIMIZATION


Figure 53  FORM FINDING 1
0. Modeling curves in Rhinoceros.
The first type (See Section 8.2.1.) the heuristic algorithms is used in the
example, so analysis is not needed to optimise the structure. The example
shows that the heuristic algorithms can be used for form finding and defi-
nition of objects is often done afterwards.
In the example, Kangaroo is used for form finding of the host
geometry. The most relevant parameter in the process is the horizontal
force. Geometry Gym plugin is used to define IFC beam objects. The most
relevant parameters in the object definition are cross-sections and mate-
rials.
1. Grasshopper geometry definition.
The process: 0. Modeling curves in Rhinoceros. 1. Grasshopper geom-
etry definition. 2. Kangaroo: Parameters and solver. 2. Geometry Gym:
Parameters and IFC definition of objects.

2. Geometry Gym: Parameters and IFC definition of objects.

1. 2. 3.

CHAPTER 8: SIMULATION AND OPTIMIZATION 102


Figure 54  FORM FINDING 2
0. Rhinoceros: Input curves for the graph.
The first type (See Section 8.2.1.) of the heuristic algorithms is used in the
example, so analysis is not needed to optimise the structure. The example
shows that the object-oriented simulations can be combined with the
heuristic optimization algorithms.
In the example Karamba is used for the form-finding process.
Beam objects are defined with Karamba. Vertical force is applied on the
beam framework. The form-finding process searches the most optimal
structure. At the end of the process the optimized structure is analyzed
with Karamba.
1. Grasshopper: Definition of the host geometry.
The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper:
Definition of the host geometry. 2. Karamba: Parameters and definition
of objects. 3. Karamba: Deformation analysis, 4. Karamba: definition of
objects. 5. Karamba: Structural analysis.

5. Karamba: Structural analysis.

2. 5.
1.
3. 4.

103 CHAPTER 8: SIMULATION AND OPTIMIZATION


Figure 55  SIZING OPTIMIZATION
1. Grasshopper: Definition of the host geometry.
The second type (See Section 8.2.1.) of the heuristic algorithms is used in
the example, so analysis is done only once in the optimization process.
The example shows that the object-oriented simulations can be combined
with the heuristic optimization algorithms.
Karamba is used to define and analyze beam framework. Anal-
ysis generates normal forces and bending moments. These parameters
are input to the heuristic algorithm. Short heuristic algorithm, written in
C# generates new diameter and thickness values for beams. The heuristic
algorithm is written by Clemens Preisinger (2011).
3. Karamba: Structural analysis.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper:


Definition of the host geometry. 2. Karamba: Parameters and definition
of objects. 3. Karamba: Structural analysis. 4. Heuristic algorithm. 5.
Karamba: Definition of objects. 6. Karamba: Structural analysis.

6. Karamba: Structural analysis.

1. 3. 6.
2. 5.

4.

CHAPTER 8: SIMULATION AND OPTIMIZATION 104


Figure 56  TOPOLOGY OPTIMIZATION
1. Grasshopper: Definition of the host geometry.
The third type (See Section 8.2.1.) of the heuristic algorithms is used in
the example, so analysis is updated at each iteration. The example shows
that the object-oriented simulations can be combined with the heuristic
optimization algorithms.
Topology optimization (TO) methods generally fall into two cate-
gories: homogenization and evolutionary methods (Bulman et al., 2001).
This example uses the latter. More specifically, the example uses Discrete
Topology Optimization method. This method often deals with truss-like
structures, where the existence of objects is optimized (Richardson, 2014,
p.178). 3. Karamba: Preview of structural analysis.
In the example, Karamba is used to execute structural topology
optimization. Karamba generates objects and analysis that are used
as a part of the optimization process. Shape of the structure remains
untouched. The objective of the process is to place structural objects
along the route of the forces. The optimization of the structure is made
by evolutionary structural optimization algorithm called BESO (Bidirec-
tional Evolutionary Structural Optimization for beam structures). The
study didn’t examine, which parameters are input to BESO.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper: 5. Karamba: Structural analysis.
Definition of the host geometry. 2. Karamba: Parameters and definition
of objects. 3. Karamba: Preview of structural analysis. 4. Karamba: struc-
tural optimization (BESO) 5. Karamba: Structural analysis.

3. 2. 3. 5.
1.

105 CHAPTER 8: SIMULATION AND OPTIMIZATION


Figure 57  SHAPE OPTIMIZATION
1. Grasshopper: Definition of the host geometry.
The example uses the metaheuristic algorithms for optimization (See
Section 8.2.2.), so analysis and fitness value are updated at each iter-
ation. The example shows that the object-oriented simulations can be
combined with the metaheuristic optimization algorithms.
In the example Karamba generates object framework and struc-
tural analysis of the framework. The optimization is done with Evolu-
tionary Algorithm (EA) plugin called Galapagos. Location points for
columns are defined by ten genomes. The objective of the optimization is
to find solution for minimal displacement, so Galapagos is used to seek 4. Karamba: Structural analysis.
minimum fitness-value for displacement. Shell and beam objects are
combined into single Karamba model showing how an integrated shell
and beam structure can be analyzed. The color displays the displacement
of objects.

The process: 0. Rhinoceros: Input curves for the graph. 1. Grasshopper:
Definition of the host geometry. 2. Defining genotype constraints for
column locations. 3. Karamba: Parameters and definition of objects. 4.
Karamba: Structural analysis. 5. Galapagos evolutionary optimization.
6. Karamba: Structural analysis.
6. Karamba: Structural analysis.

2. 4./ 6. 5.
3.
1.

CHAPTER 8: SIMULATION AND OPTIMIZATION 106


Part C focuses on issues that are
related to design, design context and
design processes in AAB. The main
approach of Part C is theoretical.
THEORY
Why?

107
HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?

WHITE = STUDY SUBJECT 108


9. THEORY OF DIGITAL DESIGN

109
110
As in the architectural design processes generally - the design process One of the main aims of the study is to investigate how AAB can
in CAD is a struggle for finding an optimal design solution for design support designer from these two aspects. There is no need to choose side
intent. So how do we define an optimal solution? or another because there can be found stability between these two sides
There has currently arisen an interest towards performance and (Benjamin, 2012, p. 15). The struggle between creativity and performance
optimization in architectural design. Performance-based design is criti- is seen as a starting point for the development and the use of CAD tools
cized because an optimal design don’t necessarily equal with good design. because both aspects should be considered CAD.
The design process needs also take creativity into account. Thus, an opti- Especially AAD has potential to support the search of creativity
mal solution should be characterized by two aspects: creativity and per- that lacks in OOD. The character of OOM methods are commonly static
formance. (Benjamin, 2012, pp. 14–15.) because they don’t utilize the full power of computers. On the other hand,
A design process of a building has always been characterized by cre- OOD can support the search of performance that is somehow lacking in
ativity and performance based design decisions. Computer-aided mod- AAD. Geometry-based models are challenging to simulate in external
eling methods are used to produce representational forms that support design tools. Instead, object-oriented models can be used collaboratively
the creative exploration of the design solution. Computer-aided simula- and in many design tools that support the search of performance. In the
tions and optimizations focus on the search of performance. reality, the division is not as rough as described. However, this study sug-
In order to deal with creativity, there is a need to understand how gests that AAB has potential that could be used to find balance between
it emerges in design processes. Actually, what is creativity in the design creativity and performance. A conceptual image of the balance is pre-
processes? According to Sevaldson (2005, p. 29) there are many interpre- sented in Figure 58. In addition, Figure 59 shows that AAB can be an
tations but he notes that the emergence of the unanticipated is common essential part of design process that searches the balance. Only the devel-
feature for all clarifications. He continues: “Creation implies the arrival opment and investigation of AAB can show how well this new design
of something new, something that has not been imagined before in rela- method can adapt to the challenge.
tion to the context of what has previously existed.” Following sections deal with many concepts that are somehow
There is also a need to understand how performance emerges in related to the search of balance between creativity and performance. The
design processes. The performance of a design solution expresses how study encourages reader to think how design methods and tools of this
well quantitative features of the object appear in the reality. Terms “per- study can support the existence of creativity and performance.
formance” and “efficiency” refer almost to the same thing. In this study
the term “performance” is used primarily because it doesn’t have such a
negative connotation.

111 CHAPTER 9: THEORY OF DIGITAL DESIGN


eality

Building

erently via drawing


rpretation and craft
struction
PERFORMANCE CREATIVITY

BALANCE
Figure 58  THE BALANCE BETWEEN DESIGN INTENTS

Building
ect manufacturing
d consturction
Parameters BIM Simulations Optimization Visualization
(generators) (assembly) (evaluators) (evaluators) (evaluator)

Figure 59  AAB IN CONCEPTUAL DESIGN PROCESS


The diagram is adapted from the one made by Boyekens (2012b) but
some changes are made that the diagram fits to the purposes of the study.

CHAPTER 9: THEORY OF DIGITAL DESIGN 112


9.1. COLLABORATION - ARCHITECT AND opposite aspects. Only then the collaborative design can lead to design
solutions that are characterized by both aspects.
ENGINEER New computational tools should not be integrated to existing pro-
Digital design has enabled the emergence of freeform, non-standard
cesses. Instead, new methods of digital design should be seen as a possi-
structures. These structures require new modes of thinking and collab-
bility to rethink the practice of collaborative design. (Harding, 2014, p.
oration. Architects and engineers should work collaboratively from the
4.) Digital design is based on modeling with parametric models which
beginning of the project. Engineers should “speak the language” of archi-
can used collaboratively. These parametric models can be seen as a new
tects and be able to support architectural design intent. Architects and
platform for collaborative debate. This requires that design objectives are
engineers have different design values, so comprehensive understand-
clearly defined. Parametric models are controlled by parameter that con-
ing of the values can lead the design team to productive collaboration.
tain values, so these values can be modified collaboratively. All paramet-
(Kloft, 2005, pp. 137–138.)
ric modeling is based on the modification of parameters. However, only
So how should the relation between architects and engineers
in AAD parameters can be used to control the model comprehensively.
change? Engineering have traditionally been pressured by speed, eco-
Commonly single designer defines these parametric models.
nomics and reliability. This has led engineers to rely on standardization.
Scripts contain complex structures, so they are challenging to under-
Thus, the role of engineers has commonly been strongly prescriptive.
stand for other designers. The nature of visual scripting often leads to
Engineers should be able to distinguish performance and prescriptive
messy structures in scripts. That’s why the legibility of the visual script is
based design approaches. Architectural design solutions are becoming
key to productive collaboration. Well structured parametric models can
more complex. That’s why engineers must be able to offer new solutions
enable productive collaboration by allowing all disciplines to attend to
that are able to brake old boundaries and to adapt to new challenges.
the development of the script. (Harding, 2014, p. 77.) Figure 60 shows
(Schwitter and Keough, 2005, p. 113.) All in all, architecture and engi-
an example of double iterative process that aims to better design process
neering should not be seen as two confronting design approaches. Col-
and collaboration.
laborative design can lead to design solutions which please both archi-
The study suggests that AAB has potential to be used as a new mod-
tects and engineers.
eling method for collaborative design. AAB offers new ways for inte-
Architects are usually concerned more about qualitative features.
grating practices of architecture and engineering. These ways have been
Engineers prefer to be more concerned about quantitative features. As
impossible before the development of AAB.
performance algorithms are increasingly driving architectural design pro-
cesses, the authorship of architects comes into question (Marble, 2012, p.
26). Architects and engineers should be able to understand these often

113 CHAPTER 9: THEORY OF DIGITAL DESIGN


2. Slow iterations

1. Fast iterations
Pre-simulation as staring
GAE information for the main
simulation: analysis data
OBJECT DATA INTEGRATED PRE-SIMULATION + parameters MAIN SIMULATION

ARCHITECT ARCHITECT ENGINEER

Figure 60  DOUBLE ITERATIVE COLLABORATION

CHAPTER 9: THEORY OF DIGITAL DESIGN 114


9.2. THE DESIGN SPACE AND THE SOLUTION study suggests that AAB could help us to reconsider these relationships.
In this study the main focus has been in the digital design space.
SPACE In some articles also occurs the concept of ‘the solutions space’
In order to understand digital design processes in general (and the rela-
(e.g. Scheurer, 2013, pp. 188–191) and the ‘fitness landscape’ (e.g. Rut-
tionship between design, design methods, design tools, design solutions
ten, 2013, pp. 132–135). These two terms have almost the same mean-
and design processes), concepts of the ‘design space’ and of the ‘solution
ing as the term the ‘design space’. The solution space focuses on solu-
space’ are examined. Abstraction of the design space is presented in Fig-
tions that the process can produce. On the other hand, the solution space
ure 61 and comparable solution space is presented in Figure 63. Also
implies that there is at least one problem that is being solved in the pro-
some abstract fields of them are presented in these figures. These fields
cess. Thus, the term doesn’t only refer to the result of the design. The solu-
are discussed more specifically in this and following sections.
tion space refers to processes that is not necessarily solved in the mind
The concept of the ‘design space’ can be used as a conceptual model
of the designer. On this basis, the term refers especially to processes that
to illustrate and design processes (Westerlund, 2005, p. 1). The concept
are solved by a computer. Thus, the term is often used to describe pos-
refers to design action that happens in human mind. The concept of the
sible solutions of optimization processes. Instead, the term fitness land-
design space means a group that consists of all possible design permu-
scape is used only the context of optimization (e.g. Rutten, 2013, pp. 132–
tations (Benjamin, 2012, p. 15). Generally speaking the concept is used
135).
in the field of design study to highlight the freedom to explore different
Visualization of the design space is challenging because design con-
alternatives and choose the most appropriate option. It is worth to men-
tains also qualitative features that cannot be precisely presented in visu-
tion that the term ‘design’ has double meaning. Thus, the design space
alizations. Instead, visualization of the solution space is relatively easy if
refers not only to the result of the design but also to the design process.
the solution space consists only of design solutions produced in quan-
One definition by Sevaldson (2010, p. 20) notes that the design
titative processes. By visualizing the solutions space, we can extend our
space consist of separate “fields” around the project: “The design space
perception of digital tools and methods (Benjamin, 2012, p. 16).
is understood as the physical, digital and social space created around
Figure 62 shows that design is searching of design solution that ful-
the project”. On this basis, this study argues that the design space can be
fills qualitative and quantitative design objectives. In the design space the
understood to consist of smaller subsets. Limitations of materials form
design process is incidental and design objectives can alter during the
the “material design space”, social relations form the “social design space”
process. Instead, Figure 64 shows that search processes in the solution
and digital tools form the “digital design space”. Thus, these concepts
spaces are usually constrained and the number of possible design solu-
can be compared and further discussed. Maybe by conceptually rethink-
tions is limited. The figure shows that computational search algorithms
ing the interaction and relationships of the social -, physical- and digi-
tal design space can lead to new understanding of design processes. The

115 CHAPTER 9: THEORY OF DIGITAL DESIGN


The theoretical design space

The design space


Knowledge about
the design space

Exploration
The final design solution

The constrained design space


“Solution experts”

Quality and quantity


Exploitation

The obvious The design


process
design space s
ces
“Fashion design” n pro
i g r
des signe
The he de
of t
“Domain experts”

Knowledge about Design solutions


the problem domain

Figure 61  THE DESIGN SPACE Figure 62  DESIGN PROCESS OF DESIGNER
Partially adapted from: (Keeling, 2010) Inspired by: (Harding, 2014, p.39)

CHAPTER 9: THEORY OF DIGITAL DESIGN 116


can “take shortcuts” but they cannot change design objectives during the usually to find innovative and novel solutions for the design problem. The
search. performance of these solutions doesn’t necessarily need to be maximal.
This study suggests that terms the ‘design space’, the ‘solutions By exploring the design space, designer can map multiple solutions for
space’ and the ‘fitness landscape’ can be used in suitable contexts. This the intended design. Usually the aim of the process is not to find single
study uses primarily term the ‘design space’, because it refers to design. global maximum. Instead, designer can map multiple local maximums,
In this study, term the ‘solution space’ is used if the focus is in design that which perform well compared to global maximum (Benjamin, 2012, p.
generates solutions computationally. 15). Exploration processes are often followed by exploitation processes.
In the design processes characterized by exploitation, the interest
9.3. DESIGN EXPLORATION AND of the designer is to find the most suitable solution for the design prob-
lem. This means that designer is looking for the global maximum, which
EXPLOITATION performs as well as possible. In exploitation processes the design space
There exists both externally and internally defined constraints in design
is usually constrained by designer. Simpler the design space is, the faster
processes. Practical design processes are always constrained. External
the optimal solution can be found (Benjamin, 2012, p. 15). Quite often
constraints refer to constrains that are not in the control of designer.
exploitation in design processes means fine adjustments of the design.
External constraints constrain the design space. Internal constraints
Beyond the design space is the “theoretical” design space. It refers
are defined by designer, so designer can control them. The designer can
to all theoretically possible design solutions. External design constraints
define design constraints that are used to facilitate the search. A con-
don’t allow designer to search these design solutions. For example design
strained design space is presented in Figure 61.
tools have limitations so they cannot produce solutions that are possible
There are two distinctive approaches for searching in the design
in design theory. That’s why there is a need to develop design tools that
space. These are called exploration and exploitation. (Benjamin, 2012, p.
support more comprehensive design spaces and design searches.
15.) These terms are originally used by statisticians but they can be uti-
lized also in the context of architectural design. These approaches differ
in the way that they relate to design constraints. Figures 60 and 61 show
conceptual exploitation processes in the design space. 9.4. CONTROL OR NON-CONTROL
Design space exploration (DSE) is in the context of computation A subject of control and non-control refers to the control of the design
but this study suggests that it can be also used also to describe design process. The question is that how much control we can keep to ourselves
in general. Design space exploration refers to the procedure of explor- and how much we need to give computers? In digital design processes
ing design solutions for implementation (Benjamin, 2012, p. 15). In the the question of the control of the process is essential because these pro-
design processes characterized by exploration, the interest of designer is cesses enable designer to lose the control to computers. The study sug-

117 CHAPTER 9: THEORY OF DIGITAL DESIGN


The theoretical solution space
“Knowledge” about
the solution space

The solution space


Exploration

The constrained solution space The final solution

Exploitation The search process of


the genetic algorithm
of
ocess
rch pr orithm
The obvious Th e s e a
netic
alg
the ge
solution space

Fitness
s
ces
pro
ch The search process of
sear
The the brute-force algorithm

Predefined Solutions
problem domain

Figure 63  THE SOLUTION SPACE Figure 64  COMPUTATIONAL SEARCH PROCESS
Partially adapted from: (Keeling, 2010) Adapted from: (Harding, 2014, p.39)

CHAPTER 9: THEORY OF DIGITAL DESIGN 118


gests that loosening the control may be beneficial especially in the early trol is that designer consciously loses the control to the computer. Then
stages of the design process. the computer is seen as a tool that produces alternative design solutions
The relation between human and computer needs to be under- that can be evaluated by the designer.
stood, if we want to search creativity and performance in digital design There are several common problems that we will face if the control
processes. Performance based design objectives are quantitative so com- is given to computers and unpredictable result from computer is tried to
puters can naturally be used to solve quantitative problems. be utilized in the design. The result is not accepted because it is seen as
Instead, design objectives related to creativity are more challenging an accidental result that has emerged without the control of the designer.
because these design objectives are qualitative. They cannot be expressed Creativity is usually seen as an internal character of the designer, so the
numerically. If we want to take full advantage of computers, we need to unpredictable result underestimates the craftsmanship of the designer.
accept that computers are purely rational tools. As long as we believe that However, it is essential to understand that the designer is allowed to use
creativity is something that comes from deep inside of us we will tend the result and thus take advantage of computation. Another problem is
to resist the potential of an external technology to influence this process that designers are used to having an image of the result, so the design pro-
(Sevaldson 2005, p. 16). cess has often a clear objective. That’s why unpredictable solution from
Generative computing can support loss of control by being able to computer doesn’t have any meaning for designer who is used to having
generate multiple design solutions which may be partly unanticipated. a preliminary image of the solution. (Sevaldson, 2005, p. 49.)
Thus, the control of design has shifted from design solution to design pro-
cess. In other words, the control has shifted from controlling the design
solution to controlling the process itself. Constraints of the process con- 9.5. FROM STATIC TO DYNAMIC DESIGN
trol the generation of design solutions (Sevaldson, 2005, p. 16.)
Two negative consequences would occur if the control of design
PROCESS
The nature of design model can be static or dynamic. The character of
would be left entirely to computational solvers. Firstly, designer would
the static design process is rigid, direct and not flexible for modification.
deal less with creativity, so designer would become workhorse. Secondly,
Instead, dynamic design process is adaptable and open for variation.
the design process would lead to design solutions that are purely formal-
Thus, dynamic design processes are iterative. Fast iterations enable pos-
istic. The design solution would not have any meaning and relation to
sibility for the design team to map multiple alternative design options
the culture. (Sevaldson, 2005, p. 29.)
and then execute required performance processes. Even if the nature of
The lose of control can happen intentionally or unintentionally.
the resulting design solution is static, that doesn’t necessarily mean that
Unexperienced designer may not be able to use computers and software,
the nature of design process should be static.
so the control of the result is lost to computer. Another way of losing con-

119 CHAPTER 9: THEORY OF DIGITAL DESIGN


Design Freedom Project Knowledge

Concept Design Detailed Design Construction

Figure 65  DESIGN FREEDOM CHAPTER 9: THEORY OF DIGITAL DESIGN 120


Until current advances in computer software, sketching and cre- of the design process. As Figure 65 shows the design freedom decreases
ation of physical scale models have been the main tools to explore design after the early stages of the design process. That’s why early stages of the
options. These design tools are mainly used because they support cre- design process should be iterative.
ative thinking. However, these tools have limitations, so designers use The nature of object–oriented BIM processes is dynamic in the
CAD tools to overcome these limitations. Computer sketches are more sense that they enable low–level changes to happen easily. For example,
dynamic, so they can be dynamically modified (Sevaldson, 2005, p. 135). they enable fast modification of objects that have the same type or prop-
In addition, computer sketches can contain much more information and erty. However, these modeling processes are static compared to AAM
the representation of information can be altered. Advances in computer processes that define high–level relations of the model. AAB tools should
software have enabled designers to create geometric forms that are con- be developed, so that they would be dynamic both on low– and high–
strained only by the imagination. Especially, the development of digital levels.
design tools have enabled designer to investigate forms that have been
challenging to control before.
All in all, design processes should develop from common static and 9.6. PRE- AND POST-RATIONALIZATION
direct modeling processes to iterative design processes. These iterative New novel forms don’t necessarily perform well, so they need to be ratio-
design processes contain the evaluation of the current design proposal, nalized. Pre- and post-rationalization mean strategies that aim to ratio-
proposition of new design alternatives, resolving of conflicts and rede- nalize the designed form so that it will eventually be possible to materi-
termination of design objectives for the next iteration (Mirtschin, 2011, alize. These strategies can be implemented in various ways (Sevaldson,
p. 2). 2015, p. 221). In the study the main aim of rationalization is seen to evalu-
New digital design tools create a conceptual state where the design ate and realize the performance of the design. Terms pre- and post–ratio-
process can vary between non-rational probing and rationalization nalization are used in many references (Sevaldson, 2015; Harding, 2015).
(Sevaldson, 2005, p. 10). On this basis, the study suggests that design Post-rationalization is seen to be more common but problematic
tools should be so dynamic that not only low–level changes but also high– strategy for finding performance. Post-rationalization means that the
level changes could be performed iteratively. This means that the over- required performance of the structure is achieved after the modeling
all form of the design can be altered iteratively. Dynamic design process process. Figure 66 shows an example of design process where the archi-
enables designer to take advantage of performance evaluation tools that tectural competition model is post-rationalized.
are usually used in the late stages of the design process. In this study Performance based design objectives are seen to be less important
these tools are called as “high-resolution” tools. Instead, low-resolution within the design process and designers have belief that the design can be
tools are performance evaluation tools that can be used in the early stages rationalized afterwards. The search of performance and the use of opti-

121 CHAPTER 9: THEORY OF DIGITAL DESIGN


Regulatory Structural
Constraints Engineering

Client Winning Mechanical Design Freeze /


requirements Competition & Electrical Tender

Site Constraints Facade


Engineering

Post-Rationalization

Figure 66  POST-RATIONALIZATION AFTER COMPETITIONS


Adapted from: (Harding, 2014, p. 51) CHAPTER 9: THEORY OF DIGITAL DESIGN 122
mization methods in the late stages of design means that the nature or design and manufacturing processes. Finding a structurally optimized
design process is static. Thus, post-rationalization strategies are problem- and geometrically clearly defined form was requirement for building.
atic, if we aim to take the full advantage of computational power. Post-ra- (Kloft, 2005, pp. 137–138.) These requirements are still valid and nowa-
tionalization techniques are top-down methods, so they cannot help us days we have much more advanced tools to design these forms.
to generate new design alternatives (Rahim, 2005, p. 186). In this sense, Standardization occurs still in all stages of common design and construc-
post-rationalization methods are reaching their limits. tion processes. Standards are everywhere - there exist design standards,
Pre-rationalization means that the performance of the structure is industrial standards, computer standards, material standards and so on.
taken into account before or during the modeling process. Aspects related In this study standardization is dealt especially from the view of stan-
not only to creativity but also to performance can be evaluated dynam- dard building objects, software and file types.
ically throughout the modeling process. The search of performance and Formerly, analog design processes were based on drawing build-
the use of optimization methods can be seen as part of dynamic design ing documents by hand. These processes were time consuming and sus-
process (Rahim, 2005, p. 186). ceptible to mistakes. Architects and engineers used different standards
There are only some performance evaluation tools that can be used of representation. The language of design was ambiguous. All in all, the
during the design process. There is also lack of generative performance process from design to construction wasn’t as efficient as it is with digital
based design tools that would lead to new synergies between architec- design tools. (Scheurer, 2012, p. 111.) However, possible mistakes were
ture and engineering (Kolarevic, 2005, p. 200). There is a need to develop usually possible to handle, case by case.
tools that can manage rationalization task already in the early stages of Nowadays, the digital design and digital processes are based on
the design. computation. Computer languages are unambiguous, universal lan-
guages. Software needs to understand the data of digital model in order
to obtain the correct result of the model. In computation standards are
9.7. STANDARDIZATION OR needed that software can exchange information. Multitude of incom-
patible standard are common problem concerning digital processes.
CUSTOMIZATION (Scheurer, 2012, p. 111.)
Formal languages of architecture have occurred mostly as a result of
As a consequence of industrial revolution, industrial processes are
the technical capabilities during the corresponding era. Industrializa-
commonly based on mass-production. Logics of form-making has been
tion enabled many new methods of construction. As a consequence,
derived from industrial construction standards. “Modernism, for exam-
new forms emerged in the building design. New techniques and mate-
ple, was a language of gridded panelized construction systems derived
rials emerged simultaneously with these new forms. However, design
directly from these standards”. (Denari, 2012, p. 30.) Even BIM software
solutions were limited because undeveloped state of computer-aided

123 CHAPTER 9: THEORY OF DIGITAL DESIGN


CUSTOM

GENERALITY OF
THE STANDARD
Mass production The use of Mass customization
advantages

STANDARD

Figure 67  MASS PRODUCTION AND MASS CUSTOMIZATION Figure 68  THE DEVELOPMENT OF STANDARDIZATION

CHAPTER 9: THEORY OF DIGITAL DESIGN 124


are based on these standards. The role of performance based design is not dards used in BIM are no more enough to satisfy AAB. If the limitations
emphasized in architectural design based on standardization because the of object standards can be pushed forward, the use of AAB can lead to
behavior of standard objects and structural systems is already known. success.
The question is that is it possible to come up with global building
standard, which could be used during the whole process from design to 9.8. AAB IN HYBRID PROCESS
production. This format should be able to be used by all members of the Hybrid Process is a process that uses a wide range of design methods,
team. In order to develop standard the file should cover the whole pro- strategies and tools in the same design process (Sevaldson, 2015, 318).
cess. IFC is an ongoing attempt to develop unified data file. (Scheurer, The term is a suggestion by Sevaldson but the same approach is used com-
2012, p. 111.) monly in architectural design. In this approach the design process stra-
In mass customization the efficiency can emerge in structures and tegically ranges between different aspects of design. According to Sevald-
construction even if all building objects are singular. Customization was son (2005, p. 10): “This new and richer design process ambulates between
before challenging because there were no tools to design custom objects rationality and intuition, between research and exploration, between the
or produce them. That would have been laborious process. (Denari, 2012, casual and the heuristic, the linear and the networked.“
p. 30.) In customization “unique standard” is developed to control the This study suggest that also AAB should be seen as a part of the
building process. Scheurer notes (2012, p. 113) that digital tools are “uni- Hybrid Process that combines different design approaches into sin-
versal machines“ because the functionality of a computer is not defined gle process. AAB tools are still under heavy development, so they can-
in its hardware. not handle the whole design process independently. However, AAB is a
AAB is based on the use of standard objects that are controlled potential approach and if its tools can mature, they will probably be able
algorithmically. AAB means sort of “mass-customization of standard to serve our design processes much better. Figure 69 presents conceptu-
objects”. The responsibility of the model is still on designer but stan- ally how Hybrid Design combines different design approaches and pro-
dardization can help designer to identify if the object can be produced cesses.
or not. If limitations of algorithm meet the requirements in the model, All in all, OOD and BIM are based on the idea that they can con-
we can argue that the object can be produced. AAB can work only inside nect different design tools. This aim fits into the strategy of Hybrid Pro-
the limitations of the object standards. As Figure 68 shows the general- cess. On the other hand, AAD encourages designer to use tools creatively.
ity of standards can be increased. By connecting OOD and AAD, this study argues that AAB may be an
At the moment the object standards in software are based on basic important part of Hybrid Processes in the future.
BIM, which works only in Cartesian coordinate system. In AAB the same
objects are used in various angles and coordinated so the object stan-

125 CHAPTER 9: THEORY OF DIGITAL DESIGN


GBD GO
(GEOMETRY- MODELING,
BASED SIMULATION
DESIGN)

ESS
OC
PR
HYBRID
CRAFT DESIGN
“DESIGN” 1.
HYBRID
DESIGN
S
ES PRO
DRAWING,
ROC CESS
MODEL MAKING P
HYBRID
DESIGN OOD BIM AAB AAM AAD
(OOM) AAS
OOS AAA
OOA AAO

HYBRID DESIGN
2.

PR
DESIGN SOLUTION

OC
GBD

ESS
(GEOMETRY-
BASED
DESIGN)

MODELING
DESIGNER

Figure 69  HYBRID DESIGN


Diagram shows two conceptual examples of Hybrid design. Example 1
shows hybrid design that combines methods of OOD, GBD and tradi-
tional crafting. Example 2 shows a hybrid design that combines methods
of OOD, AAD, GBD and possibly AAB. CHAPTER 9: THEORY OF DIGITAL DESIGN 126
10. SUMMARY AND CONCLUSIONS

127
128
At the beginning of the study I had a relatively ambiguous image of the 10.1. SUMMARY
relation between BIM and AAD. The study has been a challenging jour- As the study has shown, the studying process has contained both practi-
ney for me, attempting to clarify concepts and their relationships to AAB. cal and theoretical investigations. The study proceeded from the exem-
I have had a great opportunity to map field of digital design from vari- plary design solution to design theory. The same order is used for the
ous view of points. I believe that this comprehensive approach to inves- presentation of the study.
tigation has been more beneficial than more restricted approach would In reality, a design process of a building or a software has an oppo-
have been. I see the study as a tool that I have used to learn the subject. site order. Firstly, there is a need to know why are we designing before we
Hopefully the study can help the reader to learn as it has helped me. can know what are we looking for. In other words, in order to develop
The starting point of the study was to examine the relation of BIM AAB methods and tools we need to know why are we developing them.
and AAD. However, the main diagram of Figure 70 shows that the start- Figure 70 shows direction of design and development of AAB.
ing point was not fully successful. As can be seen in the figure, theses If we know the objectives of the design, we can develop design
terms don’t have the same meaning: BIM refers to modeling and AAD methods and design tools that support our design. The quote by Scheurer
refers to design. There were found only a few articles that discussed com- (2013, p. 194) summarizes this approach of the study:“-tools don’t find
prehensively of AAB and all of them had the relationship between BIM solutions, solutions produce tools”. I think that this means that in order
and AAD as a starting point. On the basis of this study I think that BIM to develop tools, we need to find out what are we actually looking for.
and AAD are in a way over-hyped concepts because they are not the only Computational tools don’t offer solutions for design. That’s why wee need
terms that are important when AAB is examined. The field of the digital to find them ourselves.
design would benefit from unambiguous terms. At the moment, there In the following sections I shortly summarize primary points of
are no terms that would fully satisfy our needs, so this study has pro- all main parts. These sections have the mentioned order that aims at the
posed many new terms. I hope that further studies can recommend to development of AAB.
use them or suggest better ones.
All in all, this study affirms that AAB is a part of larger assembly. 10.1.1. C. Why should AAB be developed?
The main result of the study is that AAB methods and tools can connect Part C presented many theoretical aspects of digital design. The focus was
OOD and AAD. The result means that these design approaches can be to examine why do we actually design. In this study design was seen as
used collaboratively. The relation of OOD and AAD should be seen as the cooperation of the designer and the design tool. The design has con-
the starting point for further studies. Researchers and designer should text that influences the design work. These influences make each design
at least to be aware of this relation, even if the target of the work would project different. Theoretical studying of digital design can helps us to
be more restricted than in this study. challenge our current design approaches.

129 CHAPTER 10: SUMMARY AND CONCLUSIONS


HUMAN MIND COMPUTER
THEORY PRACTICE

DESIGN SOLUTION
(OBJECT / GEOMETRY)

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAD/AAM PROCESS
AAB COMBINES PROCESSES

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

C. B. A.
WHY? HOW? WHAT?

Figure 70  DIRECTION OF DESIGN AND DEVELOPMENT CHAPTER 10: SUMMARY AND CONCLUSIONS 130
We should challenge us to think if design processes that we use are design methods, there is a need to use also tools that are geometry-based.
the most effective one. This study affirmed that there are still numerous These methods allow to build all kind of models. After all, objects have
approaches that can reform digital design. The development of digital advantages, so AAB should be seen to be part a part of Hybrid Process.
design has started only a few decades ago, so the whole design approach Can also be questioned how well object-oriented modeling meth-
is still immature. The question is much about the speed of development. ods can support creative search of design solution. Geometry-based
Do we want to develop our design methods ourselves or do we want modeling methods are more flexible, because the designer doesn’t need
to wait that someone will do that for us. Maybe the development will to choose the type and behavior of a modeling item. On the other hand,
sooner or later happen. Digital design tools are relatively easy to develop object-oriented modeling methods, such as AAB, can facilitate the search
because computers don’t complain of running tools that we built into of performance already in the early stages of design. AAB can enable pro-
them. Instead, a collaboration between designers is much more challeng- cesses that iteratively search for the optimal design solution. This search
ing task because all them think differently. also allows designers to explore design options that would not be con-
We cannot address the responsibility of fluent design processes sidered without iterative processes. In this sense, I believe that AAB can
either to architect or engineers. The whole design team should be able to facilitate the creative search of the optimal design solution.
find a common vision to design. I think that if we aim to take full advan-
tage of digital design processes, the relation of architects and engineers 10.1.2. B. How to use design methods of AAB?
needs to be rethought. History has shown that this relation has always Part B investigated how AAB methods and tools can be used and how
been a challenge. The quote by Le Corbusier (1920, p. 1) summarizes AAB can facilitate other design methods such as simulations and opti-
this is: “THE Engineer’s Aesthetics and Architecture, are two things that mizations. Interoperability between tools is a prerequisite that tools can
march together and follow each other: the one being now at its full height, be used fluently in design processes.
the other in an unhappy state of retrogression” The development of AAB tools has properly begun only a few years
Part C questions if AAB methods and tools are able to serve our ago. That’s why these modeling tools have many limitations. Neverthe-
design work. After all, this study has shown that AAB is a really poten- less, they can show that AAB can also be used to control objects in real
tial modeling approach that can be used to unify our design processes. design projects.
AAB is still immature, so it cannot alone serve our design; also other Main results of Part B are emphasized here. On of the main remarks
modeling methods are needed that complex design tasks can be man- is that the part clarified the relation between different parametric model-
aged. AAB is object-oriented modeling method, which has some advan- ing methods. This comparison is important that we can realize the main
tages and drawbacks. Objects are based in standards, so they cannot be advantages and drawbacks of each modeling method. AAB is the only
used on tasks that they are not suitable. To take full advantage of digital modeling method that can control objects through scripting. This fact

131 CHAPTER 10: SUMMARY AND CONCLUSIONS


is probably the most important point that can be addressed in this con- mization methods. Optimization can also focus on the model level or
text. This means that AAB can be used to control object-oriented mod- on the object level. On the model level optimization processes can opti-
eling processes much more freely than before. mize relations of the model. On the object level optimization methods
AAB can be seen as a creator of new objects or modifier of existing can be used to find the parameters that correspond to design objectives.
objects. The study focused on modeling processes that aim on the cre- The study showed that optimization methods are used to find optimal
ation of new objects, because this approach is seen to be the main advan- parameters that are fed into script. Objects are defined at the end of the
tage of AAB. Instead, processes that redefine objects should be further script so, the main advantage of AAB in optimization processes is mainly
examined. representational.
The study showed that in AAB the designer can create and mod-
ify large amounts of objects through rules that the designer has defined. 10.1.3. A. What can AAB produce in practice?
Scripts can be used to built relations between objects on two levels. Part A examined what AAB can currently generate in practice. Digital
On the model level these relations are used to dynamically control the design processes are much more complex than traditional design pro-
whole model. Parameters can be used to modify the model iteratively. cesses that have relied on basic working methods. That’s why they need
On the object level scripts can be used to define parameters of objects. to be learned in practice. Designer needs to find the limitations of tools
The designer can build rules that define parameters of objects. Also on that they can be used creatively and effectively. In this study, the test proj-
object level parameters can be varied iteratively. All in all, there is no only ect was used for this purpose.
one way to use AAB because scripting allows the designer to determine The development of AAB would definitely benefit from realized
own rules that define the model. projects, because AAB is still relatively unknown. Currently, the use of
The study also suggests that performance evaluation methods AAB tools needs much expertise, so there are only a few projects that
should be used on two levels. Preliminary simulations should be effort- have taken advantage of AAB. This study emphasizes that the use of AAB
less and take place already in the early stages of design. More advanced necessarily don’t mean that designed projects would represent some kind
main simulations need special expertise, so they should be made in col- of style or contain specific forms. AAB is just a modeling method that
laboration with engineers. AAB can facilitate these design methods, can be used as the designer want to use it. AAB enables the designer to
because preliminary analyses can be made in GAEs. Main simulations explore solutions that wouldn’t be possible without AAB, so if needed
can be made in external object-oriented tools. Even fully automated data AAB can be used to control forms that could not be possible to handle
exchanges to simulation software can facilitate these processes. otherwise.
The study also examined optimization processes that take place in
GAEs. The study investigated if AAB could somehow support these opti-

CHAPTER 10: SUMMARY AND CONCLUSIONS 132


10.2. CONCLUSIONS In AAD the interoperability between tools is lacking because
In the previous sections I shortly summarized primary points of the parameters and geometry are exchanged separately. I think that this
study. These sections formed a basis for the development of AAB. In this assembly should be rethought. AAB can offer an approach for integrat-
section I further discuss of subjects that I believe to be essential for the ing parameters to design items.
development of AAB. At the moment AAB is based on objects standards that are used
in BIM software, so objects restrict design freedom. However, IFC offers
10.2.1. Combining OOD and AAD two levels of standards; standard-case and representational. Modeling
As mentioned, the main result of the study is that AAB can connect process of standard-case objects can be continued in external modeling
OOD and AAD. The development of AAB will continue, so how about tools. Instead, representational objects can only contain basic parame-
if the connection between OOD and AAD could be even stronger. Could ter data such as the type of the object.
OOD and AAD be even totally merged? Figure 71 shows conceptually AAB could take advantage of similar levels. All items of the model
how OOD and AAD could be brought even closer to each other. This could contain at least some type of parameter data, so they would be
would require that BIM, AAM and AAB could be used simultaneously objects. The study has shown that the use of objects have many advan-
to control same objects. Then the same objects that are defined in the tages, so in the future the creation of objects in GAE could be an ordinary
script of GAE could be manually modified in BIM software. Designer way of modeling. All in all, I believe that there are still many object-re-
could simultaneously use script based and manual modeling methods. lated reforms to come.
Dynamo shows the first significant steps that are taken into this direction.
However, much research and development is needed that these modeling 10.2.3. Suggestions for further studies
methods can be used simultaneously. Maybe this unification can eventu- There are some aspects that I feel to be very suitable for further devel-
ally lead us to find a totally new approach to digital design. opment and study.

10.2.2. Two levels of object standards The relation of AAB to construction


Modeling process, such as AAB, can define three main aspects: param- The study didn’t focus on the relation of AAB to building industry and
eters, relations and geometry. Currently, objects can combine parame- assembly, because the main intention was to get an overall image of AAB
ters and the geometry but not complex relations. Complex relations are an its relation to design. AAB produces objects that are all based object
defined in the modeling software and they are challenging to exchange standards. However, all of these objects can vary in the limitations of the
to other software. objects. To use these objects in industry there arises a need to investi-
gate processes from design to construction. These processes need further

133 CHAPTER 10: SUMMARY AND CONCLUSIONS


DESIGN SOLUTION
(OBJECT / GEOMETRY)

ASSEMBLY

DESIGN
INDUSTRY ASSEMBLY

DESIGN METHODS
INDUSTRY
DESIGN TOOLS
OOD/BIM PROCESS AAB COMBINES PROCESSES AAD/AAM PROCESS

DESIGNER OOD BIM AAB AAM AAD


(OOM) AAS
OOS AAA
OOA AAO

Figure 71  COMBINING OOD AND AAD CHAPTER 10: SUMMARY AND CONCLUSIONS 134
examinations. It is not enough if we can define objects in digital envi-
ronments because we need also to be able to produce them if we want to
implement our digital models.

AAB tools
The main developments of AAB are still in the field of tools. AAB tools
are still in the early stages of development, so there are still many chal-
lenges that need to be solved. AAB tools are integrated into GAEs, but
they can work in cooperation with BIM software. This study suggests that
we should consider if AAB tools should have more structured interfaces
like interfaces of BIM. Much of the development should be targeted to
user-friendliness of AAB tools.

10.3. FINAL COMMENT


I think that the development of AAB can be one of the most remarkable
approaches to develop digital design. Maybe some day AAB can be part
of an ordinary way of design. After all, I hope that this study can awake
more discussion of AAB. There are only a few researches of AAB, so it
has to be examined further if we want to develop it. I hope that this study
can encourage others to do further studies of AAB. It is fascinating to
think that the development of AAB has just begun. I believe that there is
an unique chance to be part of the development.

135 CHAPTER 10: SUMMARY AND CONCLUSIONS


CHAPTER 10: SUMMARY AND CONCLUSIONS 136
GLOSSARY

137
138
TERMS
Here are listed all essential terms that are related to the study. geometry. Also known as: Generative BIM, Parametric BIM. Like BIM,
AAB refers also to the result of an AAB process: Algorithm-Aided Build-
Remarks: ing Information Model.
*Proposal by the author - the term is not commonly and unambiguously
adopted. AAD
**Not mentioned in the study but essentially linked to the theme. Algorithm-Aided Design. An unsettled term referring to algorithm-aided
design which doesn’t utilize objects. Used geometry consists only of
generic geometry which doesn’t contain semantic data. The term AAD
appears at least in the following publications: Österlund (2013a; Öster-
MAIN TERMS lund, 2013b), Tanska and Österlund (2014), http://www.adaptiveurban-
lighting.fi/; Tedeschi (2014). Also known as: Algorithmic Design, Para-
CAD metric Design (PD) and Generative Design.
Computer-Aided design. Term refers generally to all computer-aided
design. As a result of historical development, the term is sometimes used AAM*
to refer only to computer-aided drawing methods in 2D. Also known as: Algorithm-Aided Modeling. Refers to modeling methods that are defined
Digital Design. through scripting. Also known as: Parametric Modeling.

CAAD AAA*
Computer-Aided Architectural Design. Term referring only to the archi- Algorithm-Aided Analysis. Refers to analysis methods that are defined
tectural part of CAD. Compared to CAD software CAAD software con- through scripting.
tain features that support architectural design. For example, Archicad,
Autodesk Architecture and Revit can be considered to be CAAD soft- AAO*
ware. Algorithm-Aided Optimization. Refers to optimization methods that are
defined through scripting.
AAB*
Algorithm-Aided Building Information Modeling. An introduced term AAS*
referring to algorithm-aided building design, which utilizes BIM pro- Algorithm-Aided Simulation. Refers to simulation methods that are
cesses and objects. Objects contain parameters that are integrated to the defined through scripting.

139
BIM OOA
Building Information Modeling. A process of computer-aided design Object-Oriented Analysis. A design method that is based on analysis in
which utilizes objects for modeling. Thus, BIM is Object-Oriented Mod- object-oriented design tools. “The main difference between object-ori-
eling (OOM) method. Objects are controlled with parameter data. ented analysis and other forms of analysis is that by the object-oriented
approach we organize requirements around objects, which integrate both
Geometry behaviors (processes) and states (data) modeled after real world objects
A geometrical entity that doesn’t contain data, except the geometry itself. that the system interacts with. In other or traditional analysis method-
ologies, the two aspects: processes and data are considered separately.”
GB* (https://en.wikipedia.org/)
Geometry-based. Refers to issues that are geometry-oriented. In this
study such issues are geometry-oriented modeling and simulation. Cf. OOD
OO. “Object-Oriented Design is the process of planning a system of inter-
acting objects for the purpose of solving a software problem. It is one
GBD* approach to software design.” (https://en.wikipedia.org/)
Geometry-Based Design. A computer aided design approach that uses
geometry in modeling. OOM
Object-Oriented Modeling. A design method that is based on modeling
Object in object-oriented design tool.
“In the class-based object-oriented programming paradigm, “object”
refers to a particular instance of a class where the object can be a com- OOS
bination of variables, functions, and data structures” (https://en.wikipe- Object-Oriented Simulation. A design method that is based on simula-
dia.org/). An object contains data (geometry, parameters, relationships) tion in object-oriented design tool.
that can be modified parametrically. Also known as: Element.
OOP
OO Object-Oriented Programming. “Object-oriented programming (OOP)
Object-oriented. Refers to issues that are object-oriented. In this study is a programming paradigm based on the concept of “objects”, which are
such issues are Object-Oriented Design (OOD), Object-Oriented Pro- data structures that contain data, in the form of fields, often known as
gramming (OOP), Object-Oriented Modeling (OOM) and Object-Ori- attributes; and code, in the form of procedures, often known as meth-
ented Analysis (OOA) and Object-Oriented Simulation (OOS). Cf. GB. ods.” (https://en.wikipedia.org/)

140
OTHER TERMS BIM tool
“A software application used to generate and manipulate building infor-
4D BIM mation models. The term can be further qualified to denote specific appli-
A model that incorporates the dimension of time used to visualize a con- cation areas. For example, ‘ BIM Design Tool ’ is often used to refer to
struction schedule (AIA, 2007, p. 53). tools used primarily for architectural design, such as Revit ® Building,
Bentley Architecture, Digital Project ™ and ArchiCAD ®.” (Eastman et
5D BIM al., 2008, p. 468.)
A model that incorporates cost data. 5D BIM used to automate quantity
takeoffs for cost estimating. (AIA, 2007, p. 53.) Building model repository**
“A building model repository is a database system whose schema is based
AEC on a published object-oriented format. It is different from existing Project
Architecture, Engineer, Construction. Usually refers to the industry or Data Management (PDM) systems and web-based project management
software of these three sub-categories. systems in that the PDM systems are file based, and carry CAD and anal-
ysis package project files. Building model repositories are object-based,
Algorithm allowing query, transfer, updating and management of individual proj-
A prescribed set of defined rules or instructions for the solution of a prob- ect objects from a potentially heterogeneous set of applications.” (East-
lem (Daintith and Wright, 2008). man et al., 2008, p. 467.) Also known as: Model server.

API CAE**
Application Programming Interface. An interface that is defined in terms Computer-Aided Engineering. Refers to computer-aided design empha-
of a set of functions and procedures. API can facilitate integration of new sizing the relevance of engineering. “Includes traditional technical com-
features into existing software. (Daintith and Wright, 2008.) puting graphics utilizing design (CAD) and computer-aided manufac-
turing (CAM).” (MOT-IT).
BIM system**
“A software system that incorporates a BIM design tool and other appli- CAM**
cations that utilize the BIM data. The system may be connected through Computer-Aided Manufacturing. Generally refers to numerical control
a local area network or the Internet.” (Eastman et al., 2008, p. 468.) (NC), with its sub-groups (CNC and DNC) and robotics. (MOT-IT)

141
CI** and providing users with a unified view of these data.
Continuous Integration. On of the central tenets of Agile software devel-
opment (Schwitter and Keough, 2012, p. 228). Data mapper
A mapper is tool or service that aids in the transformation of data ele-
Data ments from one dataspace into another dataspace. (e.g. Janssen et al.,
Subset of information that can be managed in computer. 2015). Also known as: Mapper.

Data exchange Data mapping


“Data exchange is the problem of taking data structured under a source “The mapping is defined as a one-way transformation process, where the
schema and creating an instance of a target schema that reflects the source data representation of the source tool is mapped to the data representa-
data as accurately as possible” (Fagin et al., 2005, p. 1.) tion of the target tool.” (Janssen et al., 2015, p. 516.)
“In data exchange, data structured under one schema (which we call a “In computing and data management, data mapping is the process of cre-
source schema) must be restructured and translated into an instance of a ating data element mappings between two distinct data models.” (https://
different schema (a target schema). Data exchange is used in many tasks en.wikipedia.org/) Also known as: Schema mapping.
that require data to be transferred between existing, independently cre-
ated applications” (Fagin et al., 2005, p. 90.) Data transformation**
Data exchange processes can be direct or indirect: Cf. Direct data “In metadata and data warehouse, a data transformation converts a set
exchange and indirect data exchange. Cf. Data integration. of data values from the data format of a source data system into the data
“Data exchange is similar to the related concept of data integration except format of a destination data system.” (https://en.wikipedia.org/)
that data is actually restructured (with possible loss of content) in data “Data transformation can be divided into two steps:
exchange” (https://en.wikipedia.org/) 1. data mapping maps data elements from the source data system to the
destination data system and captures any transformation that must occur
Data file 2. code generation that creates the actual transformation program”
“A data file is a computer file which stores data to be used by a computer (https://en.wikipedia.org/)
application or system” (https://en.wikipedia.org/). Cf. Data exchange.
Data model
Data integration** “An object schema suitable for representing a building. A building data
Data integration involves combining data residing in different sources model may be used to represent schemas for file exchange, for XML -

142
based web exchange, or to define a database schema for a repository. The Deterministic methods**
main examples of building data models are IFC and CIS/2.” (Eastman et “In computer science, a deterministic algorithm is an algorithm which,
al., 2008, p. 467.) given a particular input, will always produce the same output, with
the underlying machine always passing through the same sequence of
Database schema states.”(https://en.wikipedia.org/) Also known as: Deterministic meth-
Defines the structure of a database in a formal language. Focuses on ods. Cf. Heuristic and Metaheuristic algorithms.
the data. (Uschold, 2011.) Database schema is the structure of the data-
base that defines the objects in the database (https://en.wikipedia.org/). Direct data exchange
For example, IFC schema defines the structure and relationships of IFC Data exchange process that don’t use data schemas between data exchange
objects. Cf. Ontologies. Also known as: Schema. processes. Data is directly translated between tools. Also known as: Direct
“Global data schema” refers to schema data translation (e.g. Janssen et al., 2015), Direct data mapping. Cf. Indi-
rect data exchange.
Digital design
The use of computing machines as a part of the design. In other words, Drawingless**
design that takes place in digital environment. Cf. Computational Design. In CAD drawing-less refers to presentation style that don’t use 2D lay-
Also known as CAD. Cf. Computational Design. outs. Instead, presentation happens in 3D. Cf. Paperless. (Wu, 2015.)

Computational design DSE


Refers to design in which, the creation and application of computer algo- Design space exploration. Activity of exploring design options before
rithms enables to harness the full computational power of computers. the implementation.
Also known as: Design Computing, Generative Design. AAD is a subset
of Digital Design (Harding, 2014, p. 7). Cf. Digital Design. Fitness function
“A fitness function is a particular type of objective function that is used to
Schema matching** summarize, as a single figure of merit, how close a given design solution
Finding related properties in data mapping process. Also known as: is to achieving the set aims.” (https://en.wikipedia.org/) Cf. Fitness value.
Semantic matching, Data mapping.
Fitness value
Numerical value that describes the performance of the model. Also

143
known as: Fitness score (e.g. Harding, 2015, p.48). Cf. Fitness function. Hybrid design*
Term refers to design that utilizes many digital and traditional techniques
GAE and design strategies. The term is derived from term ‘Hybrid process’ that
Graphical Algorithm Editor. Platform which enables graphical script- is suggestion by Sevaldson (2015).
ing. GAE is a subset of Graphical User Interfaces (GUIs). Also known
as: Visual Programming/Scripting Interface. Hybrid process
The term is suggestion by Sevaldson: “The Hybrid process is a process
Generative in which the many digital and traditional techniques and design strate-
In digital design the term refers to the ability to originate and produce. gies melt together.” (Sevaldson, 2015, p. 318.)

Genotype IPD**
Data that is manipulated by metaheuristic algorithm. Usually refers to Integrated Project Delivery. IPD is an approach that integrates people,
genotypes of Evolutionary Algorithms. (Harding, 2015, p. 41.) Cf. Phe- systems, business structures and practices into an effective process. (AIA,
notype. 2007, intro.)

Graph IFC
Script that is defined graphically in GAE. Also known as: Directed Acy- Industry Foundation Classes. IFC is commonly used neutral data
clic Graph (DAG) (e.g. Harding, 2015, p. 42), visual script and schema. exchange format for OpenBIM processes. IFC was developed by an asso-
ciation of firms called the International Alliance of Interoperability (IAI),
Heuristic methods nowadays buildingSMART. (http://www.buildingsmart.org)
“In computer science, heuristic methods are techniques for solving prob-
lems that include some sort of rules of thumb or experience embedded Iteration
in the algorithm” (Harding, 2014, pp. 20–21.). Cf. Metaheuristic meth- “The repetition of a numerical or nonnumerical process where the results
ods and deterministic methods. from one or more stages are used to form the input to the next. Gen-
erally the recycling of the process continues until some preset bound
Host geometry is achieved, or the process result is constantly repeated.” (Daintith and
Geometry that defines the placement an object (e.g. Denis, 2014). Wright, 2008.)

144
Indirect data exchange Mesh
Data exchange process that uses data schemas as a part of the exchange “A mesh is a collection of vertices and polygons that define the shape of
process between tools. Also known as: Indirect data translation (e.g. Jans- an polyhedral object. Meshes in Rhino consist of triangles and quadri-
sen et al., 2015), indirect data mapping. Cf. Direct data exchange. laterals.” (Integrated Manual in Rhinoceros 5.)

Interoperability Metaheuristic methods


“The ability of BIM tools from multiple vendors to exchange building In digital design metaheuristic methods are techniques for solving prob-
model data and operate on that data. Interoperability is a significant lems that are not possible to solve by inferring the solution like in heu-
requirement for team collaboration.” (Eastman et al., 2008, p. 468.) ristic methods. These methods can solve the problem by using iterative
search methods such as Evolutionary Algorithms (EA) and Simulated
LOD** Annealing (SA). (Harding, 2014, pp. 36–40.) Cf. Metaheuristic methods
Level Of Detail. Levels that describe the development of BIM object from and deterministic methods.
conceptual stage to the highest level of representation. LOD 100, 200,
300, 400 and 500 are the main levels. (AIA, 2008, p. 1.) Modeling
“The act of creating a model of something for a particular purpose, such
MDO** as to describe it, understand it, or derive some properties” (Daintith and
Multidisciplinary Design Optimization. Application of numerical opti- Wright, 2008.)
mization techniques to solve design problems involving multiple disci-
plines or designs. MOO
Multi-Objective Optimization. Instead of single objective, the method
Mass production aims to optimize multiple objectives simultaneously. Also known as:
In AEC industry the term refers to the production of large amounts of Multi-Criteria Design Optimization (MCDA).
standardized building elements.
Node
Mass customization Independent part of a visual script. Often contains algorithms to execute
In AEC industry the term refers to the production of large amounts of series of tasks. Also known as: Component, Block.
varied building elements.

145
NURBS Parametric modeling
Non-Uniform Rational B-Spline. Mathematical representations of geom- Modeling method utilizing parameters to adjust the resulting geome-
etry, that can accurately describe any 2D or 3D geometry. (https://www. try or objects.
rhino3d.com/nurbs)
Parametric object
Ontologies** Object that allows parameters to dictate its attributes. Also known as:
Defines a structure of a set of concepts. Focuses on meaning. (Uschold, parametric element and parametric component.
2011.) Cf. Ontologies.
Performance evaluation
OpenBIM Refers to design methods called analysis, simulation and optimization.
“OpenBIM is a universal approach to the collaborative design, reali-
zation and operation of buildings based on open standards and work- Phenotype**
flows” (http://www.buildingsmart.org). Approach emphasizes open data Design solution that is generated and evaluated by metaheuristic algo-
exchange processes. OpenBIM enables fluid design processes in AAB. rithm. Usually refers to genotypes of Evolutionary Algorithms. (Hard-
ing, 2015, p. 41.) Cf. Genotype.
Optimization
In digital design optimization is a method that is used to find the best Post-rationalization
alternative for the design problem. Refers to rationalization that takes place after the main process (Hard-
ing, 2015, pp. 19–54; Sevaldson, 2005, pp. 221–223).
Paperless**
In CAD paperless refers to communication media that works without Pre-rationalization
paper. (Wu, 2015.) Cf. Drawingless. Refers to rationalization that takes place before (or during) the main pro-
cess (Sevaldson, 2005, pp. 221–223; Harding, 2015, pp. 55–71).
Parameter
“Information passed to a subroutine, procedure, or function” (Daintith Property set
and Wright, 2008.) Parameter can be numerical or textual data. Also Organized set of properties of an object (s). Property set can be pre-
known as: attribute, character, property, variable. defined or defined by designer. Property sets enables data exchanges
between software. (Eastman et al., 2008, pp. 46–47.)

146
Shape optimization p. 198; Sevaldson, 2008, pp. 30–35; Benjamin, 2012, pp. 14–25; Scheurer,
Structural optimization method that optimizes the geometry of the struc- 2013, pp. 186–195). Cf. The solution space.
ture by modifying variables. The method don’t affect to the topology.
(Adriaenssens et al., 2014, p. 4.) The solution space
The concept of the design space refers to limited or unlimited group
Simulation that consists of all possible design solutions. Refers to process that is
Imitation of the behavior or the aspect of some existing or intended sys- solved computationally. The concept is mentioned in many sources (e.g.
tem. Provides an abstract model of the behavior or aspect. (Daintith and Scheurer, 2013, pp. 188–191; Harding, 2015). Also known as: “fitness
Wright, 2008.) landscape” (e.g. Rutten, 2013, pp. 132–135). Cf. The design space.

Solver Top-down and bottom-up


“A solver is a generic term indicating a piece of mathematical software... Strategies of information processing. Refers to the direction of a process.
that ‘solves’ a mathematical problem.” Many plugins for GAEs are called “Top-down and bottom-up are both strategies of information process-
solvers because they solve specific problems. ing and knowledge ordering, used in a variety of fields including soft-
ware, humanistic and scientific theories (see systemics), and management
Technical standard** and organization. In practice, they can be seen as a style of thinking and
Principle or norm how the thing should be done. (https://en.wikipedia. teaching.” (https://en.wikipedia.org/)
org/)
Topology
Textual scripting Refers to structural connections that define relations of parametric model
Scripting process based on text. Known also as true scripting (e.g. Denis, (Eastman et al. 2008, p. 46; Kolarevic, 2013, pp. 55–56).
2014, p. 5) and imperative scripting (e.g. Aish, 2013, p. 47).
TO
The design space Topology optimization. Structural optimization method that optimizes
The concept of the design space refers to limited or unlimited group of the structure by defining the existence or absence of geometrical entities
that consists of all possible design solutions. Refers not only to design or objects. (Adriaenssens et al., 2014, p. 4; Harding, 2014, p. 32.)
solutions but also to the design process. (Benjamin, 2012, p. 15.) The con- Visual scripting
cept is mentioned in many sources (e.g. Leerberg, 2004; Kolarevic, 2005, Scripting process based on editing graphical algorithms. Known also

147
as visual programming. Known also as associative scripting/program-
ming (e.g. Aish, p. 47) and dataflow programming (e.g. Harding, 2014 ,
pp. 42–43).

148
TOOLS
Here are listed all tools that are used or mentioned in the study. ALGORITHM-AIDED DESIGN SOFTWARE

Remarks: Grasshopper
*Mentioned but not used as studying tool Grasshopper (Grasshopper3D) is a graphical algorithmic interface inte-
grated with 3D modeling tools of Rhinoceros. (http://www.grasshop-
per3d.com) Grasshopper is GAE platform, which utilizes algorithms.

CAD SOFTWARE Dynamo


Dynamo for Revit is a visual programming extension to Autodesk Revit.
Rhinoceros (http://dynamobim.org). Dynamo is GAE platform, which utilizes algo-
Rhinoceros (Rhino, Rhino3D) is CAD modeling software, which can rithms. Dynamo enables GAE to interact with BIM. Dynamo utilizes
create, edit, analyze, document, render, animate, and translate NURBS DesignScript, which has common notation for both visual and textual
curves, surfaces, and solids, point clouds, and polygon meshes. (https:// scripting (Aish, 47).
www.rhino3d.com)

PLUGINS FOR RHINO


BIM SOFTWARE
BIMscript & LENA*
Revit “...Technology and a solution designed to streamline and accelerate the
Revit is BIM software, which is developed by Autodesk. Revit includes process of BIM content creation” (http://info.bimobject.com/bimscript).
functionality for all disciplines related to building design (architecture,
MEP, and structure). (http://www.autodesk.com) Rhino to Archicad
Rhino to Archicad is Rhino plugin but it enables data exchanges from
Archicad Grasshopper to Archicad. (http://www.graphisoft.com/downloads/
Archicad is BIM software, which is developed by GRAPHISOFT. Archi- addons/interoperability/rhino.html)
cad is directed into architectural design. (http://www.graphisoft.com/
archicad/) Scan-and-Solve for Rhino*
Structural simulation of Rhino solids. (http://www.scan-and-solve.com)

149
PLUGINS FOR GRASSHOPPER Karamba
Karamba is interactive plugin for modeling structural frameworks.
Elefront Karamba is used to analyze 3D beam and shell structures under arbi-
Elefront is used to add attribute information to Rhinoceros geometry. trary loads. (http://www.food4rhino.com/project/karamba?ufh)
(http://www.food4rhino.com/project/elefront?ufh)
Lyrebird
Galapagos Lyrebird for Grasshopper is used to model AAB objects and send them
Galapagos is plugin that contains two generic solvers: one using genetic to Revit. (http://lmnarchitects.com/tech-studio/bim/superb-lyrebird/)
algorithm and one using a simulated annealing algorithm (Rutten, 2013,
p. 132). Millipede*
Millipede is geometry based structural analysis and optimization plugin
Geometry Gym for grasshopper. (http://www.grasshopper3d.com/group/millipede)
Geometry Gym provides OpenBIM tools for architects, engineers and
other building designers. These tools are used to model AAB objects Octopus*
and exchange them with other building design software. Geometry Gym Octopus uses evolutionary principles to parametric design and problem
utilizes OpenBIM formats and direct API interaction. (https://geome- solving. Octopus can be used for multiobjective optimization. (http://
trygym.wordpress.com) www.food4rhino.com/project/octopus?ufh)

Grevit VisualARQ
Grevit for Grasshopper is plugin for modeling AAB objects and sending VisualARQ Grasshopper Nodes allow designer to use VisualARQ objects
them to Revit or AutoCad Architecture. (https://github.com/moethu/ in Grasshopper. VisualARQ is BIM plugin working in Rhinoceros. (http://
Grevit) www.visualarq.com)

Rhino-Grasshopper-ARCHICAD connection
Kangaroo Physics Rhino-Grasshopper-ARCHICAD connection is plugin for modeling
Kangaroo is a Live Physics engine for interactive simulation, form-find- AAB objects and sending them to Archicad (Graphisoft, 2015, p. 4).
ing, optimization and constraint solving. (http://kangaroo3d.com)

150
PLUGINS FOR DYNAMO neering, and construction professionals to holistically review integrated
models and data with stakeholders to gain better control over project out-
Dynamo Plug-in for Robot Structural Analysis comes.” (http://www.autodesk.com/products/navisworks/)
Dynamo Plug-in for Robot Structural Analysis is used to control Autodesk
Robot Structural Analysis software from Dynamo. TeklaBIMsight
(https://beta.autodesk.com/) TeklaBIMsight is used to detect clashes in BIM models (http://www.tekla-
bimsight.com).

EXTERNAL PERFOMANCE EVALUATION TOOLS

Autodesk 360 Structural Analysis for Autodesk Revit*


Structural analysis software for Revit. (https://structuralanalysis.360.
autodesk.com)

Autodesk Robot Structural Analysis


Autodesk Robot Structural Analysis is used for simulation, analysis,
and code-based structural design. (http://www.autodesk.com/products/
robot-structural-analysis/overview)

Tekla Structures
Tekla Structures structural design and documentation software. (http://
www.tekla.com/products/tekla-structures)

DESIGN AND CONSTRUCTION MANAGEMENT TOOLS

Autodesk Navisworks Manage


“Navisworks® project review software products enables architecture, engi-

151
152
SOURCES

153
154
REFERENCES
AIA. (2007). Integrated Project Delivery: A Guide. AIA (The American BOEYKENS, S. (2012). Bridging building information modeling and
Institute of Architects). 62 p. Referred: 31.08.2015. http://www.aia.org/ parametric design. pp. 453–358. In: Gudnason, G., Scherer, R., (Eds.)
groups/aia/documents/pdf/aiab083423.pdf. eWork and eBusiness in Architecture, Engineering and Construction.
Taylor & Francis Group, London. Referred: 13.08.2015. https://lirias.
AIA. (2008). Organizing the Development of a Building Informa- kuleuven.be/bitstream/123456789/344398/1/boebbi.pdf.
tion Model. AIA (The American Institute of Architects). 4 p. Referred:
08.10.2015. http://www.aia.org/groups/aia/documents/pdf/aiab078868. BOEYKENS, S. (2012b). Bridging Building Information Modeling and
pdf. Parametric Design. [Lecture slides] Referred: 13.9.2015. http://www.
slideshare.net/stefanboeykens/bridging-building-information-model-
AALTO, A. (1953). Koetalo. Arkkitehti. In: Lukkarinen, P., Holma, M., ing-and-parametric-design.
(Eds.) The experimental House. Kirjapaino Kari Oy. 48 p. (Originally
published in: Arkkitehti (The Finnish architectural review). pp. 9–10.) BULMAN, S., Sienz, J., Hinton, E. (2001). Comparisons between algo-
rithms for structural topology optimization using a series of benchmark
ADRIAENSSENS, S., Block, P., Veenendaal., D, Williams, C. (2014). studies, Computers & Structures, Vol. 79(12). pp. 1203–1218.
Shells Structures for Architecture: Form finding and optimization. Rout-
ledge. 323 p. BUILDINGSMART. IFC4 – the new buildingSMART Standard. Referred:
6.10.2015. http://www.buildingsmart-tech.org/specifications/ifc-re-
AISH, R. (2013). First Build Your Tools. pp. 36–49. In: Peters, B., Peters, leases/ifc4-release/buildingSMART_IFC4_Whatisnew.pdf.
T., (Eds.) Inside Smartgeometry: Expanding The Architectural Possibil-
ities of Computational Design, John Wiley & Sons Ltd. 272 p. CASE. (2013). Mind The Gap: Stories Of Exchange. pp. 206–217. In:
Peters, B., Peters, T., (Eds.) Inside Smartgeometry: Expanding The Archi-
ALVAR AALTO MUSEUM. Muuratsalo Experimental House [Web tectural Possibilities of Computational Design, John Wiley & Sons Ltd.
page]. Referred: 8.10.2015. http://www.alvaraalto.fi/experimentalhouse.
htm. COENDERS, J., Bosia, D. (2006). Computational tools for design and
engineering of complex geometrical structures: From a theoretical and a
BENJAMIN, D. (2012). Beyond efficiency. pp. 14–25. In. Marble, S., practical point of view. pp. 271–279. In: Oosterhuis, K., Feireiss, L., (Eds.)
(Ed.) Digital Workflows in Architecture : Design – Assembly – Indus- Game Set And Match II: On Computer Games, Advanced Geometries,
try. Birkhäuser. and Digital Technologies. Episode Publishers.

155
DENARI, N. (2012). Precise Form For An Imprecise World. pp. 28–32. FLAGER, F., Adya, A., Haymaker, J. (2009). AEC Multidisciplinary
In: Marble, S., (Ed.) Digital Workflows in Architecture : Design – Assem- Design Optimization: Impact of High Performance Computing. Stanford
bly – Industry. Birkhäuser. University. Center for Integrated Facility Engineering. 10 p. Referred:
31.08.2015. http://cife.stanford.edu/sites/default/files/TR186_0.pdf
DAINTITH, J., Wright, E., (Eds.) (2008). A Dictionary of Computing
(6 ed.). 608 p. Referred: 30.09.2015. http://www.oxfordreference.com/ FRANCHI, G., the AMiS Staff. (2012). Structural Design Of Residential
view/10.1093/acref/9780199234004.001.0001/acref-9780199234004 Buildings, Architect Studio Daniel Libeskind, Citylife Area, Milano. 11
p. Referred: 2.10.2015. http://www.agenziamilanostrutture.it/download/
DENIS, F. (2014). Tool For Augmented Parametric Building Informa- STRUCTURAL_DESIGN.pdf
tion Modelling For Transformable Buildings. Masters thesis. BRUFACE
– ULB/VUB. 84 p. GARBER, R., (Ed.) (2014). Bim Design: Realising The Creative Poten-
tial Of Building Information Modeling. John Wiley & Sons Ltd. 248 p.
EASTMAN, C., Teicholz, P. Sacks, R. Liston, K. (2008). BIM Handbook:
A Guide to Building Information Modeling for Owners, Managers, GRAPHISOFT. (2015). User Guide for Grasshopper‐ARCHICAD Live
Designers, Engineers, and Contractors. John Wiley & Sons, Inc. 506 p. Connection Add‐On (BETA). Referred: 9.10.2015. http://www.graphi-
soft.com/ftp/marketing/addons/archicad-rhino-grasshopper/Grasshop-
FAGIN, R., Kolaitis, P.G., Miller, R.J., Popa, L. (2005). Data per-ARCHICAD%20Live%20Connection-User%20Guide.pdf?_ga=1.12
exchange: semantics and query answering. Theoretical Com- 8603532.109967165.1444324748.
puter Science. pp. 89–124. Referred: 23.10.2015. http://ac.els-cdn.
com/S030439750400725X/1-s2.0-S030439750400725X-main.pdf ?_ HARDING, J. (2014). Meta-Parametric Design: Developing a Computa-
tid=af0981e0-797a-11e5-8806-00000aacb361&acdnat=1445600533_ tional Approach for Early Stage Collaborative Practice. Doctor of Engi-
a0f5039cf223464a57625e576430e6ee neering. University of Bath, Department of Architecture and Civil Engi-
neering. 303 p. Referred: 21.09.2015. http://opus.bath.ac.uk/44302/1/
FENESAN, O., (2014). BIM Implementation in Parametric Building johnharding_thesis.pdf.
Modeling. VIA University College Aarhus. 47 p. Referred: 22.09.2015.
https://www.ucviden.dk/student-portal/files/27631688/Dissertation_ JANSSEN, P., Stouffs, R., Chaszar, A., Boeykens, S., Toth, B. (2015), Custom
Oriana_Fenesan_179740_30.10.2014.pdf. Digital Workflows with UserDefined Data Transformations Via Property
Graphs. pp. 511–528. In: Gero. J. S., Hanna. S., (Eds.). Design Comput-

156
ing and Cognition, Vol. 14, Springer International Publishing. Referred: LEERBERG, T. (2004). In: The Danish Center for Integrated Design. Aar-
13.08.2015. http://www.researchgate.net/publication/262932584_Cus- hus University.
tom_Digital_Workflows_with_User-defined_Data_Transformations_
via_Property_Graphs LIBESKIND. Citylife Residences [Web page]. Referred: 2.10.2015. http://
libeskind.com/work/city-life/.
KEELING, M. (2015). Exploring the Design Space [Web page]. Referred:
4.10.2015. http://www.neverletdown.net/2010/09/exploring-de- MIRTSCHIN, J. (2011). Engaging Generative BIM Workflows. 8 p.
sign-space.html. Referred 13.08.2015. http://rhino4you.com/1/upload/engaging_gener-
ative_bim_workflows_submit.pdf.
KLOFT, H. (2005). Non-Standard Structural Design For Non-Stan-
dard Architecture. pp. 135–148. In: Kolarevic, B., Malkawi, A., (Eds.) MIRTSCHIN, J. (2014). Generative BIM using IFC4. 19 p. Referred:
(2005). Performative architecture: beyond instrumentality. Spon Press. 6.10.2015. http://rhino4you.com/1/upload/rtc2014_attendee_handout.
Referred: 22.09.2015. https://digitalfabricationworkshop.files.wordpress. pdf.
com/2014/04/performative-architecture-beyond-instrumentality.pdf.
MALKAVI, A. M., (Ed.) (2005). Performance Simulation: Research And
KOLAREVIC, B. (2005). Towards The Performative In Architecture. Tools. pp. 85–96. In: Kolarevic, B., Malkawi, A., (Eds.) (2005). Performative
pp. 204–214 In: Kolarevic, B., Malkawi, A., (Eds.) (2005). Performative architecture: beyond instrumentality. Spon Press. Referred: 22.09.2015.
architecture: beyond instrumentality. Spon Press. Referred: 22.09.2015. https://digitalfabricationworkshop.files.wordpress.com/2014/04/perfor-
https://digitalfabricationworkshop.files.wordpress.com/2014/04/perfor- mative-architecture-beyond-instrumentality.pdf.
mative-architecture-beyond-instrumentality.pdf.
MOT-IT. MOT IT-Ensyklopedia. https://mot-kielikone-fi.libproxy.tut.fi/
KOLAREVIC, B. (2013). Parametric Evolution. pp. 50–59. In: Peters, B., mot/ttkk/netmot.exe?motportal=80.
Peters, T., (Eds.) Inside Smartgeometry: Expanding The Architectural
Possibilities of Computational Design, John Wiley & Sons Ltd. 272 p. MUELLER, V., Smith, M. (2013). Generative Components And Smart
Geometry. pp. 142–153. In: Peters, B., Peters, T., (Eds.) Inside Smart-
LE CORBUSIER. (1985). Towards a New Architecture. Dover Publica- geometry: Expanding The Architectural Possibilities of Computational
tions Inc. 320 p. Design, John Wiley & Sons Ltd.

157
PETERS, B., Peters, T., (Eds.) (2013). Inside Smartgeometry: Expanding SCHEURER, F. (2012). Digital Craftsmanship: From Thinking To Mod-
The Architectural Possibilities of Computational Design. John Wiley & eling To Building. pp. 110–129. In. Marble, S., (Ed.) Digital Workflows
Sons Ltd. 272 p. in Architecture : Design – Assembly – Industry. Birkhäuser.

POTTMANN, H., Eigensatz, M., Vaxman, A., Wallner, J. (2014). Archi- SCHEURER, F. (2013). Encoding Design. pp. 186–195. In: Peters, B.,
tectural Geometry. 22 p. Referred: 1.10.2015. http://www.geometrie. Peters, T., (Eds.) Inside Smartgeometry: Expanding The Architectural
tuwien.ac.at/ig/sn/2015/ag/ag.pdf. Possibilities of Computational Design, John Wiley & Sons Ltd.

PREISINGER, C. (2015). Parametric structural modeling: Karamba, User SCHWITTER, C., Keough, I., (2012). Continuous Integration. pp. 228–
Manual for Version 1.1.0. 127 p. Referred 1.8.2015. http://www.karam- 243. In. Marble, S., (Ed.) Digital Workflows in Architecture : Design –
ba3d.com/wp-content/uploads/gh/Install/Karamba_1_1_0_Manual.pdf. Assembly – Industry. Birkhäuser.

RAHIM, A. (2005). Performativity: Beyond Efficiency And Opti- SEVALDSON, B. (2005). Developing Digital Design Techniques. Inves-
mization In Architecture. pp. 85–96. In: Kolarevic, B., Malkawi, A., tigations on Creative Design Computing. Doctoral thesis. Oslo, Oslo
(Eds.) Performative architecture: Beyond Instrumentality. Spon Press. School of Architecture and Design. 357 p. Referred: 15.09.2015. https://
Referred: 22.09.2015. https://digitalfabricationworkshop.files.wordpress. www.academia.edu/195661/Developing_Digital_Design_Techniques.
com/2014/04/performative-architecture-beyond-instrumentality.pdf.
SEVALDSON, B. (2008). Rich Design Research Space. FORMakademisk,
RICHARDSON, J. N., Adriaenssens, S., Coelho, R. F., Bouillard, P. (2014). Vol. 1(1), pp. 28–44. Referred 15.09.2015. https://journals.hioa.no/index.
pp. 171–179. In: ADRIAENSSENS, S., Block, P., Veenendaal., D, Wil- php/formakademisk/article/view/119.
liams, C. Shells Structures for Architecture: Form finding and optimiza-
tion. Routledge. SEVALDSON, B. (2010). Giga-Mapping: Visualisation For Complexity
And Systems Thinking In Design. 20 p. In the proceedings of the Nor-
RUTTEN, D. (2013). Galapagos On The Logic And Limitations Of des, the Nordic Design Research Conference: Making Design Matter!,
Generic Solvers. pp. 132–135. In: Peters, B., De Kestelier, X., (Eds.) Com- Helsinki. Referred: 30.09.2015. http://www.nordes.org/opj/index.php/
putation Work: The building of Algorithmic Thought. AD (Architec- n13/article/download/104/88.
tural Design).

158
SCHWITTER, C. (2005). Engineering Complexity: Performance-based UscholdM/ontologies-and-db-schema-whats-the-difference.
Design In Use. pp. 111–122. In: Kolarevic, B., Malkawi, A., (Eds.)
(2005). Performative architecture: beyond instrumentality. Spon Press. WANG, D., Groat, L.N. (2013). Architectural Research Methods, (2nd
Referred: 22.09.2015. https://digitalfabricationworkshop.files.wordpress. Edition). John Wiley & Sons, Inc. 468 p.
com/2014/04/performative-architecture-beyond-instrumentality.pdf.
WESTERLUND, B. (2005). Design space conceptual tool - grasping the
SINCLAIR, D., (Ed.) (2012). BIM Overlay to the RIBA Outline Plan of design process. 7 p. In the proceedings of the Nordes, the Nordic Design
Work. RIBA Publishing. 20 p. Referred 1.8.2015. https://www.architec- Research Conference: In the Making, Copenhagen. Referred: 19.09.2015.
ture.com/Files/RIBAProfessionalServices/Practice/General/BIMOver- http://www.nordes.org/opj/index.php/n13/article/view/254.
laytotheRIBAOutlinePlanofWork2007.pdf.
WEYGANT, R., S. (2011). Bim Content Development: Standards, Strat-
STASIUK, D. (2013). Louisiana State Museum and Sports Hall of Fame egies and Best Practices. Wiley, 2011. 448 p.
[Web page]. Referred: 1.10.2015. http://www.grasshopper3d.com/pro-
files/blogs/louisiana-state-museum-and-sports-hall-of-fame. WU, O. (2015). Drawingless or Paperless? [Web page]. Referred:
24.09.2015. http://blogs.solidworks.com/solidworksblog/2015/04/draw-
TANSKA, T., Österlund, T. (2014). Algoritmit puurakenteissa, ingless-or-paperless.html.
Menetelmät, mahdollisuudet ja tuotanto. DigiWoodLab. Oulun Ylio-
pisto, Arkkitehtuurin tiedekunta. 176 p. ZARZYCKI, A. (2012). Parametric BIM as a generative design
tool. Referred: 4.9.2015. https://www.researchgate.net/publica-
TEDESCHI, A. (2014). AAD_Algorithms-Aided Design, Parametric tion/256497793_Parametric_BIM_as_a_generative_design_tool. 11 p.
Strategies Using Grasshopper. Edizioni Le Penseur. 496 p.
ÖSTERLUND, T., (2013a). Algorithm Aided Design Methods for Creative
TEKLA. Fondation Louis Vuitton: A dream come constructable [Web Computational Evolution. pp. 262–280. In: Herneoja, A., Hirviniemi,
page]. Referred: 2.10.2015. http://www.tekla.com/uk/references/fonda- H., Hirvonen-Kantola, S., Joutsiniemi, A., Luusua, A., Mäntysalo, R.,
tion-louis-vuitton-dream-come-constructable. Niskanen, A., Niskasaari, K., Pihlajaniemi, H., Rönkkö, E., Soikkeli, A.,
Soudunsaari, L., Suikkari, R., and Tolonen, K., (Eds.) Proceedings of the
USCHOLD, M. (2011). Ontologies and DB Schema: What’s the Dif- 3rd Symposium of Architectural Research 2011: Research & Praxis (Ark-
ference? [Web page]. Referred: 5.10.2015. http://www.slideshare.net/ kitehtuuritutkimuksen päivät 2011: tutkimus ja käytäntö). Publications

159
A58 . University of Oulu.

ÖSTERLUND, T., (2013b). Design Possibilities of Emergent Algorithms


for Adaptive Lighting System. Nordic Journal of Architectural Research,
Vol. 25(1). pp.159–184.

160
FIGURES
The copyright of all figures belongs to the author, unless otherwise men-
tioned in the caption.

Figure 1  THE MAIN DIAGRAM������������������������������������������������������������������������������������������������������������������ iv


Figure 2  THREE ERAS OF CAD - FLOWS OF INFORMATION���������������������������������������������������������������������� 3
Figure 3  BUILDING INFORMATION MODELING (BIM)����������������������������������������������������������������������������� 5
Figure 4  MATURITY DIAGRAM OF BIM���������������������������������������������������������������������������������������������������� 7
Figure 5  ALGORITHM-AIDED DESIGN (AAD)������������������������������������������������������������������������������������������� 9
Figure 6  ALGORITHM-AIDED BUILDING INFORMATION MODELING (AAB)������������������������������������������ 11
Figure 7  THE MAIN DIAGRAM OF DESIGN���������������������������������������������������������������������������������������������� 16
Figure 8  THE RELATION OF BIM AND AAD��������������������������������������������������������������������������������������������� 18
Figure 9  THE MAIN DIAGRAM����������������������������������������������������������������������������������������������������������������� 20
Figure 10  GEOMETRY AND OBJECT��������������������������������������������������������������������������������������������������������� 22
Figure 11  EMPHASIS OF THEORY AND PRACTICE����������������������������������������������������������������������������������� 22
Figure 12  GRASSHOPPER FOR RHINOCEROS������������������������������������������������������������������������������������������� 24
Figure 13  DYNAMO FOR REVIT��������������������������������������������������������������������������������������������������������������� 24
Figure 14  EXPERIMENTAL HOUSE����������������������������������������������������������������������������������������������������������� 26
Figure 15  THE CONCEPT������������������������������������������������������������������������������������������������������������������������� 36
Figure 16  THE FINISHED PROJECT���������������������������������������������������������������������������������������������������������� 36
Figure 17  PHASES OF MODELING������������������������������������������������������������������������������������������������������������ 38
Figure 18  THE DESIGN PROCESS�������������������������������������������������������������������������������������������������������������� 40
Figure 19  GRASSHOPPER DEFINITION���������������������������������������������������������������������������������������������������� 41
Figure 20  STRUCTURAL SKETCHING ANALYSIS WITH KARAMBA���������������������������������������������������������� 42
Figure 21  STRUCTURAL STEEL FRAME DETAILING IN TEKLA����������������������������������������������������������������� 42
Figure 22  4D BIM (TIME SCHEDULING) IN NAVISWORKS MANAGE�������������������������������������������������������� 42
Figure 23  CLASH DETECTION IN TEKLA BIMSIGHT�������������������������������������������������������������������������������� 42
Figure 24  FONDATION LOUIS VUITTON������������������������������������������������������������������������������������������������� 46

161
Figure 25  CITY-LIFE RESIDENCES������������������������������������������������������������������������������������������������������������ 48
Figure 26  THE ENTRANCE OF LOUISIANA STATE MUSEUM AND SPORTS HALL OF FAME���������������������� 50
Figure 27  DATA EXCHANGES������������������������������������������������������������������������������������������������������������������� 56
Figure 28  DIRECT DATA EXCHANGE������������������������������������������������������������������������������������������������������� 58
Figure 29  INDIRECT DATA EXCHANGE��������������������������������������������������������������������������������������������������� 58
Figure 30  SHARED MAPPING PROCESS���������������������������������������������������������������������������������������������������� 60
Figure 31  MAPPING PROCESS������������������������������������������������������������������������������������������������������������������ 60
Figure 32  MODELING METHODS OF CAD������������������������������������������������������������������������������������������������ 64
Figure 33  BIM PROCESSES������������������������������������������������������������������������������������������������������������������������ 66
Figure 34  AAM PROCESSES���������������������������������������������������������������������������������������������������������������������� 68
Figure 35  AAB PROCESSES����������������������������������������������������������������������������������������������������������������������� 70
Figure 36  COMPARISON CHART BETWEEN AAD AND BIM��������������������������������������������������������������������� 72
Figure 37  DEFINING OBJECTS WITH THEGRASSHOPPER‐ARCHICAD LIVE CONNECTION TOOL���������� 76
Figure 38  DEFINING OBJECTS WITH GREVIT������������������������������������������������������������������������������������������ 77
Figure 39  DEFINING OBJECTS WITH LYREBIRD�������������������������������������������������������������������������������������� 78
Figure 40  DEFINING OBJECTS WITH VISUALARQ����������������������������������������������������������������������������������� 79
Figure 41  DEFINING OBJECTS WITH ELEFRONT������������������������������������������������������������������������������������� 80
Figure 42  DEFINING OBJECTS WITH GEOMETRY GYM��������������������������������������������������������������������������� 81
Figure 43  DEFINING OBJECTS WITH DYNAMO��������������������������������������������������������������������������������������� 82
Figure 44  OBSERVED SIMULATION AND OPTIMIZATION METHODS������������������������������������������������������ 86
Figure 45  MODELING AND SIMULATION PROCESSES IN OOD���������������������������������������������������������������� 88
Figure 46  MODELING AND SIMULATION PROCESSES IN AAD���������������������������������������������������������������� 90
Figure 47  AAB CONNECTS MODELING, SIMULATION AND OPTIMIZATION PROCESSES ����������������������� 92
Figure 48  STRUCTURAL ANALYSIS 1�������������������������������������������������������������������������������������������������������� 97

162
Figure 49  STRUCTURAL ANALYSIS 2�������������������������������������������������������������������������������������������������������� 98
Figure 50  4D BIM SIMULATION��������������������������������������������������������������������������������������������������������������� 99
Figure 51  CLASH DETECTION 1������������������������������������������������������������������������������������������������������������� 100
Figure 52  CLASH DETECTION 2������������������������������������������������������������������������������������������������������������� 101
Figure 53  FORM FINDING 1������������������������������������������������������������������������������������������������������������������� 102
Figure 54  FORM FINDING 2������������������������������������������������������������������������������������������������������������������� 103
Figure 55  SIZING OPTIMIZATION���������������������������������������������������������������������������������������������������������� 104
Figure 56  TOPOLOGY OPTIMIZATION �������������������������������������������������������������������������������������������������� 105
Figure 57  SHAPE OPTIMIZATION���������������������������������������������������������������������������������������������������������� 106
Figure 58  THE BALANCE BETWEEN DESIGN INTENTS�������������������������������������������������������������������������� 112
Figure 59  AAB IN CONCEPTUAL DESIGN PROCESS������������������������������������������������������������������������������� 112
Figure 60  DOUBLE ITERATIVE COLLABORATION��������������������������������������������������������������������������������� 114
Figure 61  THE DESIGN SPACE���������������������������������������������������������������������������������������������������������������� 116
Figure 62  DESIGN PROCESS OF DESIGNER��������������������������������������������������������������������������������������������� 116
Figure 63  THE SOLUTION SPACE����������������������������������������������������������������������������������������������������������� 118
Figure 64  COMPUTATIONAL SEARCH PROCESS������������������������������������������������������������������������������������ 118
Figure 65  DESIGN FREEDOM����������������������������������������������������������������������������������������������������������������� 120
Figure 66  POST-RATIONALIZATION AFTER COMPETITIONS����������������������������������������������������������������� 122
Figure 67  MASS PRODUCTION AND MASS CUSTOMIZATION��������������������������������������������������������������� 124
Figure 68  THE DEVELOPMENT OF STANDARDIZATION������������������������������������������������������������������������ 124
Figure 69  HYBRID DESIGN��������������������������������������������������������������������������������������������������������������������� 126
Figure 70  DIRECTION OF DESIGN AND DEVELOPMENT����������������������������������������������������������������������� 130
Figure 71  COMBINING OOD AND AAD������������������������������������������������������������������������������������������������� 134

163
164

You might also like