You are on page 1of 9

Algorithm-Aided BIM

Harri Humppi1 , Toni Österlund2


1
Tampere University of Technology 2 University of Oulu
1
harri.humppi@gmail.com 2 toni.osterlund@oulu.fi

This paper investigates the relationship of Building Information Modeling (BIM),


Algorithm-Aided Design (AAD) and general Computer-Aided Design (CAD). As
a result of the developments in CAD, new tools have recently emerged that enable
designers to utilize some of the main aspects of BIM and AAD. With this new
modeling approach, called Algorithm-Aided Building Information Modeling
(AAB), designers can use algorithms to generate parametric object models. The
geometric objects contain embedded metadata that can be further utilized in the
design and construction processes. This paper investigates how the new
approach to modeling positions itself in the field CAD. The main result of the
investigation is that the new modeling approach can be seen as a part of larger
assembly that connects two design approaches of Object-Oriented Design and
Algorithm-Aided Design.

Keywords: Algorithm-Aided Design, Building Information Modeling, Digital


Design methodologies

INTRODUCTION defined scripts composed in either textual or graph-


New design methods have emerged through the ical syntax. Scripts allow designers to generate and
transition from analog to digital design. This tran- control outcomes that would otherwise be too com-
sition has started only some decades ago and un- plex to handle by manual modeling.
doubtedly there are still many significant reforms to These two design approaches have been devel-
come. At the moment, Building Information Model- oping rather separately until recent years. Currently,
ing (BIM) is one of the most used digital design ap- new tools have been developed that combine as-
proaches within the building industry. In BIM ap- pects of them both. They enable designers to gener-
proach, design commences by modeling geomet- ate and control object models through scripting and
ric building components with their embedded meta- exchange them with BIM tools (e.g. Mirtschin 2011).
data. The data contained in the objects can be These new tools affirm that advantages of BIM and
used during the design process but also through the AAD can be used in the same design and modeling
whole life cycle of a building. On the other hand, processes.
the use of Algorithm-Aided Design (AAD) methods At the moment, there are only few investigations
are also becoming more common and integrated in to the integration of BIM and AAD on the theo-
into the workflows of design studios. In algorithmic retical level. Most of the research done in this area
processes the form generation is controlled by user- focuses only on a certain narrow field related to the

BIM | Concepts - Volume 2 - eCAADe 34 | 601


topic. A large portion of the current research fo- tion era. The diagram reveals an integrative path
cuses only on developing new software and toolsets. of BIM and Algorithm-Aided Design, called here as
There is a need to investigate the related theoreti- Algorithm-Aided Building Information Modeling (AAB).
cal aspects and ponder upon their implications, be- All of these design approaches are shown as sub-
yond the current tool development. On this basis, sets of Computer-Aided Design. The main data-flows
this paper attempts to clarify the relationship be- from design to construction are shown as separate
tween Algorithm-Aided Design and Building Infor- entities, but in reality the borders between eras and
mation Modeling and the views that they open to- data-flows are blurry and overlapping.
wards new methods of modeling and design. The 2D CAD era started in the 1980s, when com-
This paper builds on the work of Humppi (2015), puters were adopted to present building designs as
which investigates the topic also from practical level. multiple 2D drawings. The only essential change was
The research methodology consists of literature re- that computers partially replaced traditional draw-
view, investigation to the state-of-art in design tools ing on paper. Thus, the CAD era failed to utilize
and software, and case studies which help in building the full creative potential of computers. (Aish 2013)
the theoretical understanding on the subject. The development of BIM started also in the 1980s
(Aish 2013) and nowadays BIM is a commonly used
CONCEPTS OF DESIGN PROCESS method for controlling building design processes. In
Regarding the topic of the research, the meaning of BIM designer can manually adjust parameters that
some basic terms are not yet settled (see Figure 1). define the properties of each object. Thus, a BIM
Here design is seen as an abstract concept con- model behaves like an assembly of rather indepen-
taining all aspects that are related to the design dent objects (Boeykens 2012). Parametric data of
event. Different design approaches are often com- these objects can be exchanged with other object-
bined in the same design process. Design methods (or oriented design software, where the structural com-
methodologies) are different ways to design; they de- ponents are expressed as geometric objects with
scribe the interaction that happens between the de- properties. BIM provides the ability to digitally coor-
signer and the tool. The most essential digital design dinate often complex building processes from design
methods regarding the research topic are modeling, to actual construction (Garber 2014, 14). Nowadays,
analysis, simulation and optimization. BIM has reached a stable position in the field of build-
Design tools are used to generate design solu- ing design and it has also overcome many limitations
tions. In the field of digital design, tools are usually of 2D CAD drawing.
software, add-ons, plugins or custom scripts. All in Lastly, the Design Computing era introduced
all, the design process is a series of methodological the distinction between generative description of
steps that the designer takes in order to find a suffi- the building and the generated model (Aish 2013).
cient design solution. In practice, the design process Scripts and graphs are used to generate buildings,
is nonlinear, so repetition is needed to find the final forms and structures. The repetitive execution of a
solution. script enables the exploration of many design alter-
natives. The main focus in the era has been the use
Three eras of CAD of algorithms to generate geometrical models that
Historical eras of Computer-Aided Design (CAD) are do not contain any data except the geometry itself.
shortly discussed through Figure 2. According to This approach can be called Algorithm-Aided Model-
Aish (2013) the history of practical CAD can be split ing (AAM). Thus, the parametric data and the geome-
into three eras: the 2D Drafting era, the Building try need to be exported and exchanged separately to
Information Modeling era and the Design Computa- other design tools and eventually to fabrication and

602 | eCAADe 34 - BIM | Concepts - Volume 2


Figure 1
Concepts of design
process. The grey
area refers to
design tools as also
in the following
circle diagrams.

construction. generative modeling and generates objects with em-


BIM and AAD are digital design approaches bedded metadata. In research, it has been called
which have been used separately until recent years. A Algorithmic-based BIM (e.g. Feist et al. 2016), Gen-
comparison of BIM and AAD shows that both design erative BIM (e.g. Mirtschin 2011) and Parametric BIM
approaches have many advantages but also draw- (e.g. Aish 2013). Combining aspects from BIM and
backs (e.g. Denis 2014, 14). For example, definition AAD, here the term Algorithm-Aided Building Infor-
of relations between objects is more challenging in mation Modeling (AAB) is proposed and used.
BIM than in AAM. On the other hand, BIM relies on The terms Algorithm-Aided Design and
objects that can contain data facilitating the design Algorithm-Aided Building Information Modeling de-
process. scribe well the main characters of these two design
approaches. These terms can be derived from the
Ambiguous terminology abbreviation CAD, all of them containing the word
The concept related to the design utilizing algorith- 'aided', which refers to the aiding role of computers
mic logic and generative processes can be called and algorithms in design processes.
Algorithm-Aided Design (AAD) (e.g. Österlund 2013; In addition, there is a difference between an ob-
Tedeschi 2014) and is also often called Parametric De- ject and geometry. Geometry refers to an item that
sign (e.g. Boeykens 2012; Harding 2014) or Genera- does not contain any data other than the geometrical
tive Design (e.g. Ferreira and Leitao 2015; Feist et al. representation itself. The term object is used to de-
2016). scribe geometry that contains embedded metadata.
There is no settled term for describing the new Data refers to knowledge that is presented or coded
modeling method that both utilizes algorithms for in some form suitable for better usage or processing

BIM | Concepts - Volume 2 - eCAADe 34 | 603


with computer. would make a difference between them (Weygant
Another important term related to research topic 2011, 25). Also, such terms as 'property' and 'feature'
is 'parametric modeling'. It can refer not only can be used to describe the features of geometry.
to design through scripting but also to modeling
in BIM. For example, parametric modeling can be Relation of BIM, AAD and AAB
divided into 'object modelling', 'associative mod- The main starting point for the research has been
elling', 'dataflow modelling', and 'procedural mod- the awareness that Algorithm-Aided Building Infor-
elling' (Janssen and Stouffs 2015; Janssen 2015). mation Modeling (AAB) tools connect BIM and AAD.
Thus, parametric modeling can be used to describe This paper can be seen as a formal concept analysis
all modeling methods that use parameters to control where the relationships between these concepts are
geometrical and object models, either through algo- investigated.
rithmic processes or object properties. AAB then de- One main notion has been that AAB is not an in-
fines methodology, which combines the aspects of dependent field of modeling. Instead, it is a combi-
parametric, algorithmic and object-oriented model- nation of BIM and AAD. The second notion has been
ing. that BIM refers to modeling and AAD refers to design
The terms 'parameter' and 'attribute' are often approach, so these concepts are not equal. On these
used in literature and in software interchangeably to basis, there exists a need to form a more comprehen-
describe the data that is used to control and define sive theoretical framework of concepts that are re-
geometry or objects. No clear distinction exist be- lated to this new modeling method.
tween their uses, nor is there a settled definition that

Figure 2
Three eras of CAD.
Adapted from (Aish
2013). Sizes of the
grey arrows refer to
current usage of
each modeling
method.

604 | eCAADe 34 - BIM | Concepts - Volume 2


AAB TOOLS AND INTEROPERABILITY jects can be exchanged with external object-oriented
One of the main challenges in digital design pro- performance evaluation software.
cesses is to ensure that the data flows fluently
through the whole design and construction pro- AAB and performance evaluation
cesses. Practically, how to ensure that all needed data As the previous section focused on design tools, this
can be exchanged repeatedly through the process. section investigates design methods. Figure 4 shows
Interoperability refers to the ability to exchange data that AAB tools enable designer to combine two para-
between software, thus eliminating the need to repli- metric modeling methods in the same modeling
cate data (Eastman et al. 2008, 66). A single com- workflow. Therefore, designer can proceed from al-
puter software cannot support all of the tasks that are gorithmic modeling tool to object-oriented model-
needed in the design of a project, so interoperabil- ing tool or do the opposite. On this basis, there are
ity is needed to enable fluent and automated design two perceived main directions for modeling work-
processes. flows.
In algorithmic modeling workflows the interop- Maybe more obvious is the modeling workflow
erability between tools is often lacking because pa- where the designer generates new objects in GAE
rameters and geometry are exchanged separately. and exchanges them to object-oriented tools. In the
Thus modeling methods combining algorithmic pa- opposite workflow the designer defines objects man-
rameter control and object creation can potentially ually in object-oriented modeling tools and then ex-
foster these processes. They offer a possibility for changes them to GAE. Then parameters of objects
linking parameters to geometry. As Figure 3 shows, can be modified algorithmically. Both of the pre-
here AAB tools are seen as a connection between sented workflows can be combined in the same mod-
Object-Oriented Design tools and Algorithm-Aided eling workflow. Shortly, one of the main advantages
Design tools. On both sides of the figure, these tools of AAB is that the designer can simultaneously cre-
can be divided to parametric modeling tools and per- ate or modify large amounts of objects through al-
formance evaluation tools. gorithms defined in the script.
Data-exchange between designs tools are Algorithms can be used to generate object mod-
needed in the design process, so the main data- els on three levels; on model level, object level and ob-
exchange related to research topic are also presented ject joint level. On the model level algorithms can be
in the figure. Data-exchange between tools can be used to define reference geometry for objects. Then
split to direct and indirect file translations (Janssen et the script can be used to control the shape and the
al. 2015). AAB tools produce objects that can also be relations of the model. On the object level parame-
exchanged directly or indirectly with object-oriented ters of objects can be controlled algorithmically. On
design tools. the object joint level algorithms can be used to de-
Figure 3 shows that AAB enables designers to fine joints between objects. All in all, algorithms can
take advantage of both algorithmic and object- make modeling faster and more open for repetitive
oriented modeling tools. For example, designer parametric changes.
can use a Graphical Algorithm Editor (GAE) (e.g. In Figure 4 are also presented performance eval-
Grasshopper, Dynamo) to define the reference ge- uation methods that can be categorized to analysis,
ometry for objects. Then AAB tools (e.g. Geome- simulation and optimization. The main aim of the
try Gym, Grevit, Grasshopper-Archicad Live connec- performance evaluation workflow is to find an opti-
tion) can be used for object creation in GAE. Per- mal result for a design problem. The most optimal re-
formance evaluation plugins can be used simultane- sults can be found by using iterative design process,
ously with these modeling tools. Alternatively, ob- where the evaluation of a current design proposal, a

BIM | Concepts - Volume 2 - eCAADe 34 | 605


proposition for alternatives, solving problems in col- independently. When the sufficient precision level
laboration and determination of the basis for next the of performance is achieved, the process can be con-
loop (Mirtschin 2011). This process can be manual tinued in external performance evaluation software.
or handled automatically through optimization pro- AAB tools generate objects, so secondary simulations
cesses. can be handled in external object-oriented software.
AAB also facilitates double-iterative perfor- All in all, AAB can make external simulation processes
mance evaluation processes. This means that these faster and facilitate iterative workflows because all
processes can be managed in two stages. The pro- parameters related to simulation and objects can be
cess consists of two loops, which are utilized sequen- defined in GAE.
tially. In practice, this means that preliminary eval-
uations can be done in GAEs. These evaluations are AAB as a part of digital design
relatively fast, so the process can contain multiple Previous sections have focused on design tools and
iterations. Designer can manage these simulations methods. Here the main focus is in digital design ap-
Figure 3
Parametric tools
and AAB
data-exchange.

606 | eCAADe 34 - BIM | Concepts - Volume 2


Figure 4
The relation of
Building
Information
Modeling (BIM) and
Algorithm-Aided
Modeling (AAM).

proaches. Figure 5 presents the main diagram, which sign and construction. OOD focuses on standardiza-
shows the discussed concepts together. The figure tion of design, fabrication and assembly, as AAD en-
affirms that AAB is not only connecting design tools ables non-standard procedures in them.
and design methods but also connecting two sepa- As a consequence of industrial revolution, in-
rate digital design approaches called Object-Oriented dustrial processes are commonly based on mass-
Design (OOD) and Algorithm-Aided Design (AAD). This production, on which the OOD approach relies on.
connection opens many new possibilities to CAD by Logics of form-making have been derived from in-
enabling many design processes that have not been dustrial construction standards (Denari 2012). Even
possible before. AAB tools enable designers to use current BIM software are based on these standards.
object-oriented and algorithm-aided design meth- AAD relies on mass-customization, where the effi-
ods in the same design process. In other words, script ciency can emerge in structures and construction
based design methods, such as modeling, analysis, even if all building objects are singular (ibid). Cus-
simulation and optimization, can be combined with tomization was challenging before because there
manual object-oriented design methods. This opens were no tools to fabricate custom objects. However,
new possibilities for collaborative design. technological enhancements in building component
These two digital design approaches have been production have enabled many fabrication processes
rather separate until recent tool developments. Both that were not possible before.
of these sides focus on different aspects of digital de-

BIM | Concepts - Volume 2 - eCAADe 34 | 607


Figure 5
The relation of
Object-Oriented
Design (OOD) and
Algorithm-Aided
Design (AAD).

HYBRID DESIGN PROCESS Hybrid Process refers to design processes that


The use of AAB in design do not exclude other combine a wide range of design methods, strategies
computer-aided modeling methods or craft-based and tools (Sevaldson 2005, 318). This approach is
methods such as drawing and physical model mak- commonly used in architectural design. Then each
ing. For example, both AAM and BIM have some design method, including AAB, can be used cre-
advantages compared to each other, so designers atively and effectively.
should be able to combine them in the same model- AAB facilitates Hybrid Process by connecting two
ing workflow. One design tool can rarely respond to digital design approaches that have been separate
all the needs of designers. Consequently AAB tools before. The development of AAB could possibly lead
can not correspond to all the needs either, so AAB to situation where the relation between OOD and
should be seen as a part of Hybrid Process. AAD is even more integrated than currently. Maybe

608 | eCAADe 34 - BIM | Concepts - Volume 2


then BIM, AAM and AAB could be used simulta- Feist, S, Barreto, G, Ferreira, B and Leitão, A 2016 'Portable
neously to control the same model and workflow. generative design for building information mod-
For example, the same object model that is defined elling', Proceedings of the 21st International Confer-
ence of the Association for Computer-Aided Architec-
through scripting could be simultaneously manually
tural Design Research in Asia CAADRIA 2016, Mel-
modified in BIM software. bourne, pp. 147-156
Ferreira, B and Leitão, A 2015 'Generative Design
Developing AAB for Building Information Modeling', Proceedings of
This paper introduced AAB by proceeding from digi- eCAADe 2015, Vienna, pp. 635-644
tal design tools to digital design approaches. In prac- Garber, R (eds) 2014, BIM Design: Realising the Creative
Potential of Building Information Modelling, John Wi-
tice, a design process of a building has the opposite
ley & Sons, Ltd, Printer Trento Srl
order. Firstly, there is a need to know why we are de- Harding, J 2015, Meta-Parametric Design: Developing a
signing before we can know how to design and what Computational Approach for Early Stage Collabora-
we are designing. In order to develop AAB methods tive Practice, Ph.D. Thesis, University of Bath
and tools we need to know why we are developing Humppi, H 2015, Algorithm-Aided Building Information
them. Design methods and design tools can only be Modeling: Connecting Algorithm-Aided Design and
Object-oriented Design, Master's Thesis, Tampere
developed if the objectives of the design are known.
University of Technology TUT
On this basis, here is suggested that the research and Janssen, P 2015 'Parametric BIM Workflows', Proceedings
development of AAB should also proceed from the- of the 20th International Conference of the Association
ory to practice. As the Figure 5 showed, this paper in- for Computer-Aided Architectural Design Research in
troduced a theoretical framework that positions AAB Asia CAADRIA 2015, Daegu, pp. 437-446
in the field of digital design and CAD. This result can Janssen, P and Stouffs, R 2015 'Types of parametric mod-
be seen as a starting point for further theoretical re- eling', Proceedings of the 20th International Confer-
ence of the Association for Computer-Aided Architec-
search. tural Design Research in Asia CAADRIA 2015, Daegu,
pp. 157-166
REFERENCES Janssen, P, Stouffs, R, Chaszar, A, Boeykens, S and Toth, B
Aish, R 2013, 'First build your tools', in Peters, BP and Pe- 2015, 'Custom Digital Workflows with User-Defined
ters, TP (eds) 2013, Inside Smartgeometry: Expanding Data Transformations Via Property Graphs', in Gero,
the Architectural Possibilities of Computational De- JSG and Hanna, SH (eds) 2015, Design Computing
sign, John Wiley & Sons, Ltd, Printer Trento Srl, pp. and Cognition, Springer, pp. 511-528
36-49 Mirtschin, J 2011 'Engaging Generative BIM Workflows',
Boeykens, S 2012 'Bridging building information mod- Collaborative Design of Lightweight Structures - LSAA
eling and parametric design', Proceedings of the Eu- 2011, Sydney
ropean conference on product and process modelling, Sevaldson, B 2005, Developing Digital Design Techniques.
Reykjavik, pp. 453-458 Investigations on Creative Design, Ph.D. Thesis, Oslo
Denari, N 2012, 'Precise Form For An Imprecise World', in School of Architecture and Design AHO
Marble, SM (eds) 2012, Digital Workflows in Architec- Tedeschi, A 2014, AAD_Algorithms-Aided Design: Para-
ture, Birkhäuser, pp. 28-32 metric Strategies Using Grasshopper, Edizioni Le
Denis, F 2014, Tool for Augmented Parametric Building Penseur
Information Modelling for Transformable Buildings, Weygant, RS 2011, BIM Content Development: Standards,
Master's Thesis, Brussels Faculty of Engineering BR- Strategies, and Best Practices, John Wiley & Sons, Ltd
UFACE – ULB/VUB Österlund, T 2013, 'Design Possibilities of Emergent
Eastman, C, Teicholz, P, Sacks, R and Liston, K 2008, BIM Algorithms for Adaptive Lighting System', Nordisk
Handbook: A Guide to Building Information Modeling arkitekturforskning: Nordic Journal of Architectural
for Owners, Managers, Designers, Engineers and Con- Research, 1, pp. 159-185
tractors, John Wiley & Sons, Ltd, New Jersey

BIM | Concepts - Volume 2 - eCAADe 34 | 609

You might also like