the virtual and the physical - spaeth...

The Virtual and the PhysicalBetween the representation of space and the making of space

A. Benjamin SpaethWassim Jabi

Contents

3 Spatial Representation5 Graphical Representation of Sonic Urban Morphologies

Merate A. Barakat15 FEEL YOUR DESIGN

Maria da Piedade Ferreira, Andreas Kretzer, José Pinto Duarte, Didier Stricker,Benjamin Schenkenberger, Markus Weber, Takumi Toyama

23 Open Virtual RealityValentin Dolhan, Huda S. Salman

33 Computational Thinking35 Qualitative Representation and Spatial Reasoning in a Rule-Based

Computational Design ModelAmer Al-Jokhadar, Wassim Jabi

45 Computation in ArchitectureWerner Lonsing

55 Digitizing the UnexploredSema Alacam, Orkan Zeynel Guzelci

63 Building Performance Computation65 Architectural space quality, from virtual to physical

Aida Siala, Najla Allani, Gilles Halin, Mohamed Bouattour73 Architects’ use of tools for low energy building design

Gabriela Zapata-Lancaster83 Investigating Spatial Configurations of Skycourts as Bu�er Zones in

High-Rise O�ce BuildingsSaba Alnusairat, Shan Shan Hou, Phil Jones

93 Computational Fabrication95 A robotically-driven additive construction planning process using an

ecological materialOdysseas Kontovourkis, Philippos Michael

105 Before and After NURBSSerenay Elmas, Erenalp Saltik

Contents - eCAADe RIS 2017 | 1

115 Visual Thinking Through Model MakingAndrew N.J. McDonagh, Huda S. Salman

125 Modular MyceliaSean Campbell, David Correa, Dylan Wood, AchimMenges

2 | eCAADe RIS 2017- Contents

Spatial Representation

Graphical Representation of Sonic UrbanMorphologies

Simulation Interface for the Design Process

Merate A. Barakat11UWE Bristol - University of the West of [email protected]

Although the interdisciplinary field of Soundscape was originally intended as aparadigm-shift for urban design, the process of designing the aural spatialsignature of urban open spaces has not yet been made readily accessible to thedesigner in modes that align with architectural design procedures. This paper ispart of an on-going investigation (Barakat, 2012; 2015; 2016a; 2016b) with theprimary objective of developing a computational tool that generates qualitativeaural spatial patterns and predicts the perceived aural spaces within an urbanenvironment. The ultimate goals are to provide a spatial designer with 1) goodestimated predictions of the patterns generated by the design, and 2) feedback forpossible spatial alterations. To that end, this paper discusses the graphicalrepresentations of the mapping tool that are mediated by standard architecturalprocesses and visual representation standards. These methods aim to show thespatial formation of the separate acoustic spaces that are centred on possibleforeground signals, occurring within urban spaces.

Keywords: Urban Design, Soundscape, Simulation, Architectural Visualisation ,Computation Design

INTRODUCTIONEvery urban space has a unique aural signature and ispart of a hierarchical meta-system of acoustic spaces.The aural signature is a result of the distribution ofsonic events within a space that create an aggregateof acoustic spaces at different scales, ultimately cre-ating a sound environment. Evidence suggests thatfidelity of aural stimuli is a significant parameter increating an accurate spatial mental map of the sur-rounding environment. The field that is concernedwith the acoustic qualities of urban spaces is the in-terdisciplinary field of Soundscape. Soundscape, as a

field and a term, was originally defined as a relation-ship between the human ear, sonic environment andsociety, and was intended to be an integral part ofurban design. The concept is based on physical pa-rameters as well as on the perceptual and cognitiverestrictions.

Spatial designers lack adequatedesign concepts,metrics and tools to integrate acoustic sensory as-pects into the design process of urban spaces. Prac-ticing architects have not developed a basic intuitionof how sound behaves in a space, and they rely onacoustic specialists when considering sound in the

Spatial Representation - eCAADe RIS 2017 | 5

design. Architectural training does not incorporatean understanding of the fundamentals sound phe-nomenon and cannot predict these aspects duringthepreliminary designprocess. Among these lackingconcepts is the fundamental properties determiningthe behaviour of sound in space, and themeasurableauditory response phenomena.

A number of commercial packages and sounddistribution models are available for acoustic engi-neers. While highly quantitative, the available acous-tic platforms are highly technical and the associatedinterfaces are designed for sound specialists. At thescope of soundscape that integrates multiple spe-cialties, there is an issue of inconsistency. Korn-feld et al. state that there is a “shortage of appro-priate tools supporting an interdisciplinary discourseabout the sonic environment” (2011, p. 29). Multipledomains develop individual means of urban sounddescription that have created different visual “lan-guages.” Kornfeld et al. further point out that thevarious approaches are “[incompatible] means that,when it comes to an exchange of perspectives, interdis-ciplinary discourse is difficult...[However], appropriatetools for supporting this activity do not exist” (2011, p.14). Architects and urban designers use plans, sec-tions and CAD models to coordinate with differentconsultants, and the client, but there are no availableplatforms for integrating soundscape designwithoutentirely referring to a sound specialist.

The majority of the models simulating urbansound propagation consider the acoustic index ofSound Pressure Level (SPL) distribution in the field.This concern is of significance because it disregardsthe frequency dependent indices that are criticalwhen considering more than one source. Indeed, inhis recommendations, Kang indicates that some ofthe “challenges in the simulation of urban sound prop-agation [is] consider[ing] more sound sources. Muchwork has been done for traffic sources, but very limitedwork has been done for other sources, especially posi-tive sources [and] more calculation indices are needed,in addition to simple SPL” (2011, p. 2364).

This paper discusses the in graphical interface

aspect of an on-going investigation (Barakat, 2012;2015; 2016a; 2016b) that has a primary objectiveof developing a computational simulation tool forurban design that generates qualitative aural spa-tial patterns and predicts the perceived aural spaceswithin an urban environment. The investigation isfounded on establishing a relationship between au-ral architecture theories and the urban spatial ex-perience and design. It also explores the mergingof spatial and acoustical computational approaches,through integrating the physical/mathematical rep-resentation of sound to the mapping of the spa-tial envelopes and phenomena of human aural re-sponses. The aim is the development and calibrationof a computational design and decision-aiding toolthat can predict qualitative patterns of aural spatialperception, and translate them into spatial attributeswithin a modelled urban space. The fields of compu-tation simulation, soundscape, and psychoacousticsinform the structure of the tool development.

The tool produces spatial patterns as represen-tations of the distribution of sound energy of pre-dicted acoustic spaces and the intermediary domainsbetween them. The tool structure is built as a JAVA-Based program. The system is designed as a Multi-Agent Based (MAB) deterministic system. The ulti-mate goal of the aural mapping tool is to provide aspatial designer (architect, urban or event designer)with 1) good estimated predictions of the patternsgenerated by the design, and 2) feedback for possi-ble spatial alterations.

To that end, this paper discusses the graphicalrepresentations of the mapping tool that are medi-ated by standard architectural processes and visualrepresentation standards. The discussion presentsthe assimilated nomenclatures, corresponding defi-nitions considered within this scope that is informedby the interdisciplinary psychoacoustic, soundscaperesearch, and architectural practice. The use of thetool is explained in the form of schematic illustra-tions. The explanation also includes how the user,i.e., architect, urban and/or event designer, can op-erate this tool, and what decision-making aspects to

6 | eCAADe RIS 2017 - Spatial Representation

expect.

ASSIMILATED SEMANTICSAs a relatively new domain, the soundscape termi-nology has yet be conclusively defined, necessitatingexplicitly stating which terms pertain to this researchand the associated definitions. The spatial terms areassimilated from soundscape literature (Truax 1999),including soundscape ecology (Farina 2014) and au-ral architecture (Blesser and Salter 2007). Given thespatial character of this research and the need forquantifiable spatial factors, this study is aligned withthe definition provided by Genuit and Fiebig (2006).They define soundscapes as domains that “consist ofa number of spatially distributed sound sources, whichgive the soundscapes their distinctive features” (2006,p. 953).

Genuit and Fiebig’s (2006) definition allows thisresearch to map the aural spatial patterns at a mi-croscale urban space. The focus is on the foregroundsignals that are immediately located on site. Signalsthat are beyond the perimeter of the considered ur-ban space are reduced to background signals with aparticular sound level.

The terms receiver and sensor refer to ahuman lis-tening/perceiving the sonic environment. The map-ping tool designates agent systems to simulate hu-man response, namely the Sensor Agent. The consid-ered foreground signals are denoted as sonic events,sound sources, and signals. These terms refer to theevents emitting a signal that are located within theurban space. The simulation engine reduces theevents (or objects) to point sources, namely SonicEvent Agents. If the distance between a receiver anda sonic event allows the listener to detect and dis-tinguish the signal, then an auditory channel forms.The level of information determines the strength ofthat channel. A receiver can be connected to mul-tiple channels. An auditory channel can be discon-nected for various reasons, including an increase indistance, other competing channels, the presence ofa stronger channel connection, and the diversion ofthe listener’s attention (1999). The simulation engine

is only concerned with the strongest receiver-signalauditory channel.

Truax (1999) presented the term acoustic space.Blesser and Salter (2007) adapted the idiom for auralarchitecture as an acoustic arena. A similar concept isknown in the field of soundscape ecology, known asa patch (Farina 2014). These three terms refer to thearea of space where a sonic event can be cognitivelyde-codified, and it is centredon that event. An acous-tic arena is delineated by the human response to fre-quency dependent parameters, including the audi-tory threshold, and attention and spectral masking.In the presence of multiple adjacent sonic events, anumber of auditory channels are connected to thereceiver, and the associated acoustic arenas ‘over-lap.’ In that intersecting domain two conditions mayoccur: 1) the strongest auditory channel severs theother weaker connections, or 2) all connections areof equal strength negating each other. In the for-mer case, the receiver is considered in the arena as-sociatedwith the strongest auditory channel and theemitting signal is considered dominant. In the lat-ter case, the receiver is not connected to an auditorychannel. The receiver may be counted in the edgedomain. Edge is a term assimilated from the patch-edge epistemological concept (Farina 2014).

As an extensionofGenuit and Fiebig’s (2006) def-inition of soundscape, soundscape patterns may beregarded as the spatial morphologies mediated bythe acoustic arenas (or patches) centred on the spa-tially distributed sonic events, and the edges formingbetween them. Since the event spatial distributiondistinguishes the features of the soundscape, eachsoundscape would have a characteristic morpholog-ical pattern.

GRAPHICAL REPRESENTATIONS ELE-MENTSThe auralmapping tool uses architectural visual sym-bology to represent the resulting sonic morphology.Line-weight, hue, alpha, saturation, and ‘red-lining’are used as analogies for absorption coefficient,spectrum, energy levels, boundaries, and decision-


making feedback, respectively. These methods aimto show the spatial formationof the separate acousticspaces that are centred on possible foreground sig-nals, occurring within urban spaces.

Frequency Spectrum | Hue & Energy Level |Alpha channel“Frequency fofa sound is thenumberof completevibra-tions (cycles) occurring per unit time (seconds).” [Mea-suring unit: Hertz (Hz)] (Cavanaugh and Wilkes 1999,p. 321)

Frequency is one of the wave based propertiesthat distinguish every signal (Turner and Pretlove1991). Hearing adults can detect sounds between 20and 16,000 Hz. Hearing sensitivity is frequency de-pendent. The human response to frequency is ev-ident in multiple psychoacoustic phenomena, suchas loudness and masking, that defines the audiblethreshold of the associated frequencies (Vorländer2008). Two signals sharing the same SPL but havedifferent frequencies will not have equal perceivedloudness (Kuttruff 2007).

The proposed graphics use hue and alpha valuesto represent frequency and detected energy contri-bution, respectively. The colour key in Figure 1 showsthe audible frequency range (20Hz - 16 kHz)mappedlogarithmically along the hue spectrum. The first red(hue= 0˚) and the second (hue = 360˚) represent 20Hz and 16 kHz, respectively. Any parameter relatedto a signal is assigned a colour corresponding to thatfrequency.

Figure 1A hue and alphamapping key offrequency andenergycontribution,respectively. Thekey is generated byan auxiliary JAVAcode using thesame colour codemethodsembedded in thesystem.“The unit Decibel (dB), which is one-tenth of a Bel, is a

logarithmic (to base 10) unit used to measure a ratiothatmaybe representedbypower, soundpressure, volt-age, or intensity.” (Farina 2014, p. 229)

The human internal ear system perceives acous-tic attributes similar to dB logarithmic scale. There-fore, dB is ameasuringunit relative to the humanper-ception of sound energy (Farina 2014). Although hu-man physiology is sensitive to small changes in pres-sure, a linear change in intensity is undetectable (Ca-vanaugh 1999). For example, a difference in SPL of 1dB is barely detectable in a controlled laboratory en-vironment. Onlywith a shift of 3 dB, can a personper-ceive a difference in level.

“Sound pressure level (SPL, Lp) is a logarithmicmeasure of the effective sound pressure of a sound rel-ative to a reference value (2x 10-5 N/m2).” [Measuringunit: decibel] (Farina 2014, p. 227)

Theminimumvalue of SPL on the decibel scale, 0dB, is set as a reference pressure level Pref value of 2x10-5 Pa. This reference pressure level corresponds tothe threshold of hearing for a typical healthy youngperson (Cavanaugh 1999). The decibel scale trans-lates indices to manageable units that are related tohuman perception. This is beneficial for manual cal-culations. In computation, it seems more manage-able to remain within the linear scales and maintainthe input and output interface as converted dB val-ues. Accordingly, the tool computes the energy on alinear scale and uses the decibel scale in the user in-terface, including the graphical representation of the


detected relative energy distribution.Detected relative energies are mapped to the al-

pha channel. An entirely opaque colour that is usu-ally adjacent to a sonic event represents the maxi-mum relative energy contribution. The transparencyis an indication of a decrease in energy distributionwithin the patch domain. Fully transparent (or white)regions represent the edge condition, where the en-ergy valuesmay be below the auditory threshold andambient levels. The edge condition may representa case where two energy levels are detected at thesame magnitude, or the difference between contri-butions is less than 10% of the highest contribution.

It is worth noting here that the absence of colourdoes not mean the absence of sound. It is an indica-tion of absents auditory channels due to conflictingforeground signals. An edge conditions may also oc-cur if the value of energy contribution is lower thanthe energy of the background signals or the audiblethreshold. Figure 2 is a schematic diagram that showsthe attenuation of the relative energy contributionand the maximum distance of energy detection. Theoutermost contour is considered the sphere of influ-ence of the sonic event agent, which is similar to theMAB concept of an acoustic arena.

Absorption Coefficient | Line-weightSoundwave reflection occurs when an acoustic waveimpingesonaboundarybetween the air andanothermedium. When a wave falls on a boundary, the en-ergy is partially reflected, partially absorbed and par-tially transmitted. In the case of a perfectly reflectivewall, all sound energy is reflected with no deprecia-tion, whereas a perfectly absorptive surface reflectsno energy. Perfect reflective and absorptive surfacesdo not exist in reality (Kuttruff 2007).

“Absorption coefficient α is the ratio of sound-absorbing effectiveness (at a specific frequency) of aunit area (m2) of acoustical absorbent to a unit area(m2) of a perfectly absorptive material, usually ex-pressedasadecimal valuebetween1.0 (perfect absorp-tion) and 0 (perfect reflection).” [A non-dimensional aratio] (Cavanaugh and Wilkes 1999, p. 324)

The absorption coefficient is one of the princi-ples governing the behaviour of a sound wave at thesurface of a wall. It is a fraction of the incident energynot reflected fromor transmitted through the surface(Kuttruff, 2007). Sound reflection, transmission, andabsorption are frequency dependent (Kang 2007). Inpractice,materialmanufacturers produce absorptionspecifications manuals for their products in the formof a table of coefficients corresponding to each fre-quency band (Cavanaugh 1999).

Among the graphics that are designed to inte-grate standard architectural language is line-weight.The aural mapping system accepts an architecturalmodel as an input, and the walls are inserted withthe values of the absorption coefficients correspond-ing to each frequency band. The simulation environ-ment is set up to show black walls against a whitebackground. In architecture, the visual opacity of awall is represented as line weights. This simulationengine uses the same concept to translate the reflec-tivity of the wall (1- α); i.e., the higher the absorptioncoefficient (low reflectivity) the thinner the line rep-resenting the wall, as schematically shown in Figure3.

Figure 3Wall line weightrepresentationaccording to theabsorptioncoefficient.


Figure 2Schematic Diagramof therepresentation ofthe relative energycontribution. (Top)A Schematic graphof energyattenuationaccording to theinverse square law.The adjusted datummediates themaximum andminimumcontribution. Theminimumcontributiondelineates thesphere of influence.(Bottom Left)Contours generatedrepresentingattenuation ofdetected energy(Bottom Right).Contourssuperimposed onenergy detectionrepresented usingthe alpha channel(red gradient) andthe lack of energydetection (white).The sphere ofinfluence is theoutermost contour.Different line weights are also used to represent the

acoustic space contour levels but in reverse. Thethickest line represents the outermost contour thatsignifies the lowest detected contribution. Figure 2shows that the exterior contour indicates the ‘wallsurface’ of that acoustic patch; the figurative ‘wall’ isthe patch boundary.

‘Red-lining’ | FeedbackThe concept of ‘redlining’ is a customaryprocedure inarchitectural practices, where an architect indicatesrevisions with a red pen. This simple function is de-signed to perform a similar indication. Although thedesigner would find the graphical representation ofthe energy distribution informative, it can be argued


that only the outermost contour is of significance.This isocline determines the area and configuration ofthe patches and the edges between the considered do-mains, which is the primary goal of this design tool. Inaddition, the redlinemode simulates energydistribu-tion adjusted according to the A-weighting networksthatmimics the frequency human response anddoesnot follow the inverse square profile. The weightingsignificantly changes the contributions detected bythe sensors, and the isoclines graphically overwhelmthe field.

To that end, the principle isocline that is con-sidered in this mode is the outermost isocline. Theareaof that domain ismeasuredandcompared to theinitially designed program area designated for eachsonic event. If there is a discrepancy (i.e., the arenaarea is larger than theprogramarea), the colourof thewalls affecting the patch configuration ‘red line’ (orflag) possible walls to be altered, prompting the de-signer to revise the configuration of the location, ma-teriality, and orientation, as shown in Figure 4 [Right].

ARCHITECTURAL DESIGN APPLICATIONPublic squares are good examples of special useprojects. A client (usually governmental entities orprivate developers) may approach a designer solic-iting recommendations on adding events, providing

entertainment and services for the community, or re-viving the economy of a certain district. These typesof projects are normally complex, and multiple pa-rameters are simultaneously considered to find op-timum solutions. For example, designing the out-side space considers the occupants of the envelop-ing buildings, who have a direct connection with thesquare and are the first to benefit from (or disturbedby) the interventions.

Large projects require a collaborative team oflandscape and urban designers, planners, architectsand event designers. Different specialists are con-sulted and incorporated into the team at later stagesof the project depending on the budget and the typeof client. Unless a large multidisciplinary firm is com-missioned to conduct the project, the team is usuallylead by one of the spatial design specialists. This per-son handles communications with the client and co-ordinations between specialists to meet that client’sgoals.

Figure 5 is a square scheme selected for a se-ries of vignettes that assume the spatial designer isconducting a preliminary design within a small firmbefore potentially taking on the project and assem-bling a team of experts. The scale of the project dic-tates the amount of time the designer is allowed be-fore the client selects a different office that is usu-

Figure 4Calibration test runoutput patterns fora simulation thatconsiderA-Weighted sonicevents in normal(left), and redline(right) mode.


Figure 5Schematic of asquaremorphologicalconfiguration andmateriality andbubble diagramsand spatial andsignal distribution(left). Patternsgenerated at 2ndreflection order fordaytimes whenambient levels arehigh, and 1streflection order forquiet hours whenambient levels arelow, both run inredline mode. And,a schematictimeline for thepossible designprocess cycle(right).ally between 2-3 business weeks. As a general prac-

tice, most projects undergo the same process thatis independent of scale. The materials and geome-try are selected, the site is investigated, and the rela-tion between the programmatic spaces is analysed.Schematic drawings and CAD models are then gen-erated, analysed and adjusted to find an optimumdesign that satisfies requirements such as the client’sneeds and possible neighbouring conflicts. The leaddesigner generates the final schemeas a set of graph-ical documents that the client can understand. Theprocess usually takes a relatively large portion of thelimited timeframe.

Focusing on the use of this tool and on how itcan be used for quick qualitative mapping during anassumed design phase the schemes is set up with:1) spatial morphology (geometry and materiality); 2)site properties; 3) the programmatic use determinedby the client; and 4) the associated occupancy classi-fication.

These four design aspects are progressivelytranslated into a CAD file accompanied by metadataof the spatial configuration and possible sonic eventdistribution, which are input as a comma-separatedvalues (CSV) file. The parameters include the foot-print of the buildings and the 3D location of eachsonic event, along with the estimated mean fre-quency and maximum relative energy levels. Thesite parameters include the estimated ambiancelevel and absorption coefficients of the façades andground that are obtained from general tables or themanufacturer. In the context of setting up the pa-rameters of the pattern simulation tool, the designermay select particular areas (e.g. attraction area andentertainment events) as foreground signals and re-duce other events (e.g. vehicular traffic) to the am-bient levels. The patterning tool is then executed in‘redlining’ mode so that the output simulations flagfaçades that the designer may want to consider ad-justing.


The tool has the capability of creating patternson multiple planes simultaneously. One patterning-plane is sufficient for quick qualitative results. Thiswouldminimise computation timeandallow formul-tiple design iterations. The designers may interpretthese patterns by two methods: 1) checking whichfaçades are flagged, or 2) visually deducing possibleinformation. The designer may consider adjustingthe design based on one or both of these methods.

The tool aims to help the design process inmulti-ple ways. Attraction areas similar to those employedin the square configurations (see Figure 5 ) are ex-amples of reviving a commercial square, where ne-gotiation and cooperation are key when maximisingprofit is the main consideration. The patterns simu-lated for quiet hours showmultiple interferences andlarge edge domains. These areas may be interpretedas zoneswhere none of the contributing sonic eventscan be singularly de-codified. The designermay con-sider consolidating sonic eventsby coordinatingwithshop owners and temporary event performers.

CONCLUSIONArchitects’ training enables them to consider intu-itively most of the projects aspects, e.g. structuralor thermal comfort, and provide consultants work-able models that do not require drastic changesto the client agreed aesthetics and configuration,or other incorporated systems. This training doesnot include aspects pertaining to the phenomena ofsound. By providing the designer a computationalenvironment that provides qualitative aural patterns,the designer can consider sound in urban spaces, theprocess can be shortened and the budget can be ad-equately appropriated.

The tool is intended to be used as a decision-aiding programme that can predict qualitative pat-terns of aural spatial perception, and translate theminto spatial attributeswithin amodelled urban space.The tool is tailored for spatial design processes, ar-chitectural drawingsmediate the inputs andoutputs,and the graphical information provides patterns thatcan be de-codified by an architect. The output is in

the form of geometrical spatial maps of the distri-bution of detected sound energy and isoclines de-lineating the predicted patches of acoustic spaces,and the edges between them. The decision-makingfunction compares the intended architectural pro-gramme area to the delineated spaces and flags pos-sible changes prompting the designer to adjust thedesign (material and geometry) during the prelimi-nary phase. This contributes to a significant reduc-tion in the expenses associated with redesigning in alater phase or, worst case scenario, retrofitting afterthe project is completed.

This paper discusses the graphical interface ofthe tool. The results are spatialmaps that havegraph-ics that follow architectural practices, namely lineweights, hue values, alpha channels, isoclines andredlining that represent reflectivity, frequency, en-ergy attenuation, acoustic spaces peripheries, anddesign amendments, respectively. The visual associ-ation of these acoustic parameters allows the resultsto be an integral part of the design process. The over-all research adopts a number of semantics that de-scribes soundscape in spatial metrics and links theperceptual parameters of context and meaning tothe architectural programme. Although there are ac-counts of considering SPL distribution in simulatingsound in an urban space, the selection of acousticand psychoacoustic frequency dependent indices al-lows this research to simulatemultiple sound sourcessimultaneously.

Of course, there is room for development. Po-tentially, the redlining package prepares the tool tobe developed as a generative decision-making tool.The tool has the capability to generate 3D represen-tations of enveloping acoustic arenas. For complexconfigurations, thenumberof hours required to com-pute multiple horizontal planes simultaneously, maynot be feasible for the schematic phase. If requiredthe algorithm that maps in two-dimensional planescan be adjusted for 3D simulation at low computa-tion cost.


REFERENCESBarakat, Merate A. 2012 ’Urban Acoustic Simulation:

Analysis of Urban Public Spaces through Auditorysenses’, Digital Physicality - Proceedings of the 30theCAADe Conference, Prague, Czech Republic, pp.587-592

Barakat, Merate A. 2015 ’Towards Sonic Urban Mor-phologies: A Modeling Nexus (WIP)’, 2015 SummerSimulation Multi-Conference (SummerSim, Chicago,pp. 389-384

Barakat, Merate A. 2016a, Sonic Urban Morphologies:Towards Modelling Aural Spatial Patterns for UrbanSpace Designers, Ph.D. Thesis, AA School

Barakat, Merate A. 2016b ’Aural spatial mapping tool:constraining the signal sphere of influence vali-dation (WIP)’, SCSC ’16 Proceedings of the SummerComputer Simulation Conference, Montreal, Quebec,Canada, pp. 430-435

Cavanaugh, William J 1999, ’Introduction to Architec-tural Acoustics and Basic Principles’, in Cavanaugh,William J and Wilkes, Jospeph A (eds) 1999, Proffes-sional Development Program, Architectural Acoustics,John Wiley & Sons, Inc., New York, pp. 1-54

Cavanaugh, William J and Wilkes, Jospeh A 1999, Archi-tectural Acoustics, Principles andPractices, JohnWiley& Sons, In., New York

Farina, Almo 2014, Soundscape Ecology: Principles,Patterns, Methods and Applications, Springer Sci-ence+Business Media Dordrecht, New York, London

Kang, Jian 2007,Urban soundenvironment, Taylor & Fran-cis, New York

Kang, Jian 2011, ’Sound propagation at micro-scale inurban areas.’, The Journal of the Acoustical Society ofAmerica, 129(4), pp. 2364-2364

Kornfeld, A-L., Schiewe, Jochen and Dykes, Jason 2011,’Audio cartography: visual encoding of acoustic pa-rameters’, in Ruas, Anne (eds) 2011, Advances in Car-tography and GIScience, Springer Berlin Heidelberg,Paris, pp. 13-31

Kuttruff, Heinrich 2007, Acoustics: An introduction, Taylor& Francis, New York

Turner, John David and Pretlove, A. J. 1991, Acoustics forengineers, Palgrave Macmillan

Vorländer, Michael 2008, Auralization: fundamentalsof acoustics, modelling, simulation, algorithms andacoustic virtual reality, Springer-Verlag, Berlin


FEEL YOUR DESIGNexploring the sensorial experience of Architectural space throughimmersive architecturemodels

Maria da Piedade Ferreira1, Andreas Kretzer2, José Pinto Duarte3,Didier Stricker4, Benjamin Schenkenberger5, Markus Weber6,Takumi Toyama71,2Fachbereich Architektur - Technische Universität Kaiserslautern 3Faculdade deArquitectura da Universidade de Lisboa 4,5,6,7Deutsches Forschungszentrum fürKünstliche Intelligenz GmbH (DFKI)1,2{m.piedade.ferreira|andreas.kretzer}@[email protected],5,6,7{didier.stricker|benjamin.schenkenberger|markus.weber|takumi.toyama}@dfki.de

This paper describes an experiment entitled Experiment 2- ``Feel your Design'',which belongs to a group of experiments undertaken in the context of a PhD inarchitecture. The goal of the experiment described in this paper was to evaluatethe emotional reaction of a viewer to changes in the sensory perception whenbeing stimulated by viewing, listening and smelling immersive architecturalmodels. The Experiment was taken as part of a workshop with students ofarchitecture. The workshop incorporated concepts of ``Sensory Design'' and``Emotional Design''. The task assigned to the students proposed that immersive,atmospheric models were built according to a specific narrative and includedspecific scents and sounds which were supposed to re-enforce such a narrative orinduce a certain mood. The results of the Experiment were evaluated through theuse of a ``Presence Questionnaire'' and a ``SAM chart''. The Experiment had theparticipation of 7 students who produced one model each and served in theExperiment as subjects. Experiment 2 took place on the last day of the ``Feel yourDesign'' workshop. The host institution was Fachbereich Architektur, DigitaleWerkzeuge, TU Kaiserslautern. The experiment had the technical support of theDeutsches Forschungszentrum für Künstliche Intelligenz GmbH (DFKI).

Keywords: Architecture, Immersion, Emotion Measurement, Sensory Design


INTRODUCTIONThis paper describes an experiment which was takenas part of the research work developed for a PhDthesis dedicated to the topic of the relationship be-tween body and architecture (Ferreira, 2016). Fourexperiments were done in the context of the afore-mentioned PhD research. The goal of the experi-ment described in this paper, called Experiment 2 -”Feel your Design”, was to evaluate the emotional re-action of a viewer to changes in the sensory percep-tionwhenbeing stimulated by viewing, listening andsmelling immersive architectural models. This exper-iment took place on the last day of a student work-shop. Theworkshop functioned as an elective coursefor students of architecture and dealt with the phys-ical construction of immersive architectural modelsand the evaluation of subjects’ response to the at-mosphere’s created by the models. The name of theworkshop comes from the assumption that an archi-tectural space can be designed as an experience ifthe architect takes in consideration sensory aspectssuch as haptics, sound, and scent in addition to vi-sion (Damásio, 1999). The idea was inspired by Pal-lasmaa’s (2005) work and his proposal that architectsshould use an embodied, sensorial approach to de-sign as a way of stimulating the body, as a holis-tic sensory system which is not exclusively depen-dent on vision, but also on other senses. The work-shop also incorporated concepts of ”Sensory Design”and ”Emotional Design” (Norman, 2004). The task as-signed to the students proposed that immersive, at-mospheric models were built according to a specificnarrative and included specific scents and soundswhich were supposed to re-enforce such a narrativeor promote a certain mood. The Experiment had theparticipation of 7 students who produced onemodeleach and served in the Experiment as subjects. Ex-periment 2 tookplaceon the last dayof the ”Feel yourDesign” workshop. The host institution was Fach-bereich Architektur, Digitale Werkzeuge, TU Kaiser-slautern. The experiment had the technical supportof the DFKI.

DESCRIPTIONEach participant was given a cubemeasuring 50 x 50x 50 cm and was requested to choose a room typewhich could be any kind of interior space such as anattic, a bedroom, a working room, a library, a mu-seum, or a hospital room. The students were askedto choose a narrative that described amood that wasto be explored through the model and design an im-mersive experience through the combined design ofan interior space, the choice of a scent which accen-tuated the narrative (“scentscape”), and a specificallycreated sound-loop which re-enforced the desiredatmosphere (soundscape). The scale of the interiorspaces was left for each student to choose accord-ing to her/his idea for the model, as well as the loca-tion and dimension of the peeping holes which weremeant to condition the viewer’s gaze.

Figure 1Immersive modelswith peep-in holes

The setup of the Experiment consisted of Architec-tural models placed at seating level, installed withpeep-holes (Fig.1), scent- and soundscape in the in-terior of a box with the same dimensions as an oldphonebooth (Fig.2), with a bench for the viewer to sitand look at the models and a black fabric curtain tokeep the booth in the dark. The tools used for the ex-perimentwereone laptopwithpluggedheadphoneswhich was set at a table next to the booth where themodels were installed. While inside the booth, theviewer was asked to sit on the bench and take a lookinto the peeping hole of themodel, while listening tothe soundscape through the headphones connectedto the laptop outside of the booth (Fig.3). Studentswere advised to install, if possible, small vents inside


the models so that the viewers could feel the chosen”scentscape”more intensely. The students had to de-fine the possible views into the model space on thedisplay side of the cube, to create a soundscape as aloop in mp3-format, and to complete the experienceby adding a specific scent to themodel. The selectedtitle had to describe the desired mood for the archi-tectural space and provide the viewer with the keyto the understanding of the narrative. The goal ofthe exercise was to explore through these immersivemodels themulti-sensory experience of architecturalspace (Pallasmaa, 2011) and evaluate viewer’s emo-tional reactions (Damásio, 1999) to the narrative ex-perienced inside the booth through the analysis ofdata collected andprocessedwith emotionmeasure-ment methodology. To support the design processduring the workshop, students were introduced toartwork on the topics related with the task such ascinema,modelmaking, art installations, aswell as ba-sic notions of architectural archetypes and Plutchik’s(1962) vocabulary of basic emotions. Students werealso introduced to the concepts of emotion mea-surement (Bradley et al., 1994; Kim et al. 2015), im-mersion, presence (Witmer et al., 1998), sensory de-sign and emotional design (Norman, 2004). Studentswere encouraged to use CAD/CAM technologies toaid in the process of manufacturing the models. Theresults of the Experiment were evaluated throughthe use of a ”Presence Questionnaire” (Witmer et al.,1998) and a ”SAM chart” (Bradley et al., 1994). Recallthat the main goal of this experiment was to qualifyusers’ response to immersive architecturemodels, byanalysing sensory data, having inmind that ourmainhypothesis was:

H1 - a user’s emotional response as “compelledor not compelled”, “positive or negative”, “arousedor not aroused” and “dominant or dominated” toan immersive architecture model can be evaluatedthrough objectivemeasurements of emotion using aPresence Questionnaire (PQ) and a SAM chart.

Figure 2Setup with focus onmodel location


Figure 3Diagram of “Feelyour Design”experimental setup

Two secondary hypotheses were then formulated:H2 - architecture is an immersive experience

which can be consciously composed by the architect;the techniques of “emotional design” and “sensorydesign” are an effective strategy to compose specificexperiences of architectural spaces and develop thesensorial awareness of students and designers;

H3 - the feeling of presence and emotional ac-tivation can be induced through the experience ofanalogical models, in this case, immersive architec-tural models;

To verify these hypotheses, the experiment wasdeveloped considering four stages:

1. Identify the design characteristics that aremore suitable to induce certain sensations inthe user, such as “positive, aroused, dominant,compelled”, “negative, not aroused, domi-nated, not compelled”, “joy, sadness, anger,boredom, ecstasy”;

2. Design an immersive model so that those

characteristics are themost important aspectsof the design;

3. Perform experiments with users interactingwith these architectural models and assesstheir emotional experience through theuseofa PQ and a SAM chart;

4. Process and analyse the sensory data col-lected to understand if significant differencescanbe found in the classification anddifferen-tiationbetween a “compelling-positive” expe-rience and a “not compelling-negative” one.

RESULTSExperiment 2 - “Feel your Design” evaluated theemotional experience of architectural models byanalysing changes in the sensorial perception ofthe viewer, while looking at the model, listening tospecifically created sounds (soundscape) and inhal-ing specifically chosen scents (“scentscape”). The re-sults of this experimentwere the answers to the Pres-ence Questionnaire and the SAM chart, where recallof experience and believability of simulation weresystematised. Suchdatadescribes the subject’s phys-iological response and emotional activation, therebyenabling one to evaluate the model’s ability to alterthe subject’s emotional state. The data is organizedaccording to Presence Questionnaire’s “Factors” and“Subscales”, as defined by Witmer and Singer, as wellas SAM’s parameters of Valence, Activation and Con-trol. The final values considered in the analysis of thePQ results were obtained by averaging the ratingsassigned by the subjects to each of the questions,according to Witmer and Singer’s (1995) 1-9 pointscale. The analysis of the data collected throughthe SAM chart also followed the same principle, asthe three parameters of “valence”, “activation” and“control” were rated by the subjects using also a 1-9point scale. After making these calculations, we ob-tained values that qualify the subject’s individual ex-perience of the “Feel your Design” models regardingthe parameters of “presence”, “emotional response”,“valence”, “activation”, and “control”. From the datacollected, we can also qualify the experience of the


groupof subjects as awhole, by averaging the resultsfor the same parameters.

DISCUSSION AND CONCLUSIONSExperiment 2 - “Feel your Design” evaluated theemotional experience of architectural models byanalysing changes in the sensorial perception ofthe viewer, while looking at the model, listening tospecifically created sounds (soundscape) and inhal-ing specifically chosen scents (“scentscape”). Theexperimental results support the main research hy-pothesis H1 - a user’s emotional response as “com-pelled or not compelled”, “positive or negative”,“aroused or not aroused” and “dominant or domi-nated” to an immersive architecture model can beevaluated through objective measurements of emo-tion using Presence Questionnaire and SAM charts.The results collected through thesemeans show that:

Figure 4Interior of model“Psycho”

• - for the immersivemodel “Psycho” (Fig.4), themajority of subjects experienced a high levelof “Presence”, “Arousal” and “pleasure”;

• - for the immersivemodel “Turm”, themajorityof subjects experienced a high level of “Pres-ence”, “Arousal” and “pleasure”;

• - for the immersive model “Surfstation” (Fig.5), the majority of subjects experienced highlevels of “Presence” and “pleasure” and a low

level of “arousal”;

Figure 5Interior of model“Surfstation”

• - for the immersive model “Spiegel” (Fig. 6),themajority of subjects experienced high lev-els of “Presence” and “pleasure” and a lowlevel of “arousal”;

Figure 6Interior of model“Spiegel”


Figure 7Interior of model“Blumenpassage”

• - for the immersivemodel “Blummenpassage”(Fig. 7) , the majority of subjects experiencedhigh levels of “Presence” and “pleasure” and alow level of “arousal”.

All subjects felt they were able to control their emo-tional response to the scenes in the models and alsodescribed them as a positive experience. Most sub-jects described the scenes as “not very dominating”,although the scenes “Turm” and “Surfstation” weredescribed as “dominating”. Interestingly enough,these experiments explored opposite atmospheres,anxiety in the case of “Turm” and relaxation in thecase of “Surfstation”. All subjects reported to have feltcompelled by all the scenes in the experiment, withthe scenes “Psycho”, “Turm”and “Spiegel” being ratedin average as “very stimulating” and the remaining

two “Surfstation” and “Blummenpassage” as “stimu-lating”. The scenes in the models were unanimouslyrated by the subjects as “convincing”, “engaging” and“consistent” with real life experiences in terms of thesensual information. Most subjects reported to havebeen visually involved by all the scenes, as well as bythe corresponding “scentscapes”, soundscapes and,surprisingly, by the haptic aspects as well, since thelatter was not directly explored in this experiment,but the former three were. This suggests that themodels induced a very high level of immersion. Allscenes were rated by the subjects as able to trig-ger the imagination of real actions and the major-ity of subjects reported to have had their attentiondedicated to the scene. This permits us to concludethat Hypothesis H2 is verified - architecture is an im-mersive experience which can be intentionally com-posedby the architect and the simulation techniquesof “emotional design” and “sensory design” are an ef-fective strategy to compose specific experiences ofarchitectural spaces anddevelop the sensorial aware-ness of students/designers. The majority of subjectsreported that they were able to survey inside themodels, were not distracted by the quality of their ex-ecution and couldmove inside the booth andmanip-ulate the interface objects without being distractedby them. Subjects also adjusted quickly to the exper-iment, could concentrate well on the scene and themajority rated it as a good learning experience. Still,except for the case of the scene “Psycho”, which thefinal average score shows that the subjects lost trackof time, this did not happen in the remaining scenes.This suggests that the experience of the model “Psy-cho” was the most immersive. Finally, it can be con-cluded that the illusion of presence and arousal situ-ations can be intentionally induced through immer-sive architectural models, although further researchis necessary to understand which specific design el-ements are responsible for this. Therefore, Hypothe-sis H3, which suggests that the feeling of “presence”and emotional activation of the body of a user canbe intentionally induced through the experience ofimmersive architectural models, also is confirmed.


Experimental results of this experiment show thatPQ and SAM are effective in identifying arousal re-sponses related to “positive” or “negative” emotions,from the neutral condition, when users experienceimmersive architectural models. On-going researchin the fields of IT, psychology and marketing uses anestablished range of values that also were user usedas reference in this experiment. The use of electroen-cephalogram (EEG) and biometric markers is an ad-ditional, interesting method to be used in future ex-periments to observe how the emotions of a user aretriggeredwhile experiencing immersive architecturalmodels. Such technology was unavailable for thisexperiment, but it was used in other experiments inthe context of the aforementionedPhD research (Fer-reira, 2016). The experiment described in this paperwill be repeated and eye-tracking sensing technol-ogy will be incorporated to the experimental setup.We believe this method might be useful to detectwhich points are of most and least interest for theviewer of a scene and give a more comprehensiveanalysis of the subjective experience of looking at anarchitecture scene.

REFERENCESBradley, MMand Lang, PJ 1994, ’Measuring emotion: the

self-assessment manikin and the semantic differen-tial’, Behaviour Therapy and Experimental Psychiatry,25, pp. 49-59

Damásio, A 1999, The Feeling of what happens: Body andEmotions in the Making of Consciousness, HarcourtBooks, Orlando

Ferreira, MP 2016, Embodied Emotions: Observations andExperiments in Architecture and Corporeality, Ph.D.Thesis, Faculdade de Arquitectura Universidade deLisboa (FAUL)

Kim,MJ, Cho,ME andKim, JT 2015, ’Measures of Emotionin Interaction for Health Smart Home’, IACSIT Inter-national Journal of Engineering and Technology, 7(4),pp. 10-12

Lang, PJ, Bradly, MM and Cuthbert, BN 1998, ’Emotion,motivation, and anxiety: brain mechanisms andpsychophysiology’, Biological psychiatry, 44(12), pp.1248-1263

Norman, D 2004, Emotional Design : whywe love (or hate)everyday things, Basic Books, New York

Pallasmaa, J 2005, The Eyes of the Skin: Architecture andthe Senses, John Wiley & Sons, West Sussex

Pallasmaa, J 2011, The Embodied Image: Imagination andImagery inArchitecture, JohnWiley & Sons,West Sus-sex

Plutchik, R 1962, The Emotions: Facts, Theories and a NewModel, Random House, New York

Witmer, B, Singer, G and Michael, J 1998, ’MeasuringPresence in Virtual Environments: A Presence Ques-tionnaire’, Presence, 7(3), p. 225–240


Open Virtual Reality

A new open sourced communicationmedium in Architecture

Valentin Dolhan1, Huda S. Salman21,2Scott Sutherland School of Architecture and Built Environment1,2{v.dolhan|h.salman}@rgu.ac.uk

This study lends itself to recent advances in open source virtual reality technologyand its application within the field of architecture. The research focuses on opensourcing and the relatively inexpensive hardware solutions readily available,such as Google Cardboard. The outline proposal architectural detail level isrelative to the project and architectural practice, but ideally, it is assumed that aninitial proposal presented to the client is based on a considerable amount of workand detail expressed in the form of plans, sections, elevations and variousinternal and external visualisations where materiality is considered using simpletextures and understandable queues to make the material choice clear in order todraw out as much feedback as possible from the client. The paper addresses theproblem of information loss when trying to communicate an architectural designproposal following traditional methods such as information on printed paper,also it compares the subjects impression of static and interactivethree-dimensional panoramic spheres viewed through the pairing of a phone anda cardboard headset. The two representations use the same three-dimensionalmodel of the project. The study experiment with open virtual reality (OVR) showgreat potential for seeing more information that clarifies common problems invisual graphic communication.

Keywords: Virtual reality, Open source, Head mounted display, Perception,Google cardboard , Immersion

RATIONALWith recent breakthroughs in technology, Virtual Re-ality (VR) equipment and interfaces have become af-fordable at a consumer levelwhile also being increas-ingly easy touse andoperate. This emerging technol-ogy startedwith a solid backgroundwithin the enter-tainment industry. However, its applications withinarchitecture cannot be overlooked and it is only re-

cently that the technology has reached a level whereit become appropriate for efficient use, especiallywhen cheap open source hardware alternatives suchas Google Cardboard [2] are making their way on themarket. There is a general consensus that informa-tion is often lostwhena client is beingpresentedwithan architectural proposal. Without much architec-tural knowledge, he can misinterpret the architect’s


intention. Two dimensional drawings like plans or vi-sual perspectives can be limited in their expression.Ultimately, a client may not fully understand an ar-chitectural proposal and might come up with radi-cal changes further down the design process, not be-cause of the inconsistency of the presented propos-als, but rather by his own limited spatial perception.This can ultimately lead to financial loss on both par-ties. A virtual immersion in the critical phases of de-sign such as an outline proposal could help reducethe loss of architectural information, thus creating abridge between the client’s understanding and thearchitect’s vision. This could lead alleviate the designprocess where the architect is not required to investextra effort into project specification. Virtual realitycan become a valuablemedium fromwhich both thearchitect and the client can benefit.

Theoretical BasisIn terms of definition, virtual reality had a rough start.Alternative dimensions have been pondered for cen-turies. An idea came to realisation in the 1950s whena handful of people imagined the scenario of watch-ing a screen that never ceases to broadcast. It is saidthat concept goes even back as the 1860when artistscreated a series of three- dimensional panoramicmu-rals. The 80’s and 90’s revitalised the ideals of a vir-tual reality universe. It was in this period that thepersonal computer became a mass consumer prod-uct and that virtual reality believers were impatientto experience it. However, limitations shadowed theenthusiasm to experience a realistic digital environ-ments. The visionwas still outmatchedby themeans.Later on, in the 1990’s, the technology’s hype faded,makingway for other interestingemergent technolo-gies. Recently, that interest started focusing on vir-tual reality , yet again when Palmer Luckey devel-oped a crude headset which was later named as theOcculus Rift, which was launched in March 2013 [1].However, the virtual reality was still hindered by lim-itedgraphical representation, but theexperiencewasstarting to become believable.

External representationsThe use of external representations (sketch, drawing,model, etc) combines internal, intellectual actionswith external actions and their external result (phys-ical) (Salman 2011). It is claimed that designers pro-duce various types of external representation (virtu-al/physical) to overcome problem-solving and infor-mation uncertainty (Christensen 2005; van der Lugt2000). As such, at a certain point in time, each typeproduced (within the design process) has a differentdegree of information uncertainty, in that it repre-sent a certain level of specificationof theproblemun-der consideration. Stacy and Eckert (2003) empha-sised ”the importance of combining sketches, words,gesture to disambiguate each other”. Externalisationis thought of as an indication of the designers’ levelof information uncertainty (Christensen 2005). Ac-cording to Christensen (2005), different levels of in-formation uncertainty could be assigned to differentexternalisation-types.

Graphic representation continuityIn order to represent, visualise and communicatethree dimensional information, onemust have a firmgrip on principles of graphic language. The elementsthat make up the language are driven by rules whichinfluence information about the three-dimensionalform. They are hard to convey using words, picturesor diagrams. Thus, an issue of continuity arises be-tween three dimensional representation using twodimensional methods (Gill 2004). Current methodsdo not exemplify the relation among observer, ob-ject and representation,while computer visualisationdoes not fully display characteristics of the comput-er/virtual environment. An idea is initially developedby an architect on paper or on a computer. After-wards, it is externalised and communicated throughdrawings, sketches, models or visualisations. Suchtwo-dimensional representations belong to a spe-cific language which conveys information about athree-dimensional object in space. A designer’s suc-cess depends in his ability to efficiently communi-cate and idea. Perspectives and isometric projections


illustrate a relationship between: the observer, anobject, the picture plane, and projection lines. Inturn, they are affected by the distance between ob-ject or projected plane and the observer and ele-ment angles. Therefore, design needs communica-tion tools that focus on continuity within representa-tional methods. Recently developed high end com-puter graphics could be used to illustrate these rela-tionships through immersive environments where auser’s perception is assisted in creating his own un-derstanding and conclusions and being able to iden-tify inconsistent information.The application of vir-tual reality in this casemay influence a user’s capabil-ity todevelophismental ability towards threedimen-sional forms and to relate the traditional two dimen-sional representational methods to his mental pro-jection.This expected availabilitymakes it interestingto explore the potentials of (re)using VR in architec-tural education as well as for professionals communi-cating architectural designs to customers or commu-nity.” (Kreutzberg 2014).

Research AimThis study aims to answer the following question:How efficient is open sourced virtual reality in help-ing the client understand an architectural outlineproposal compared to two dimensional representa-tional methods? from the perspective of final yearMaster of Architecture student, this question led tosetting the following two main objectives:

• To identifywhether open sourcing virtual (VR)reality as technology simplifies the genera-tion and setup of VR scenes and experimentsfor improved visualisation

• To see how efficient open source virtual real-ity is in communicating architectural design ,does it improve client’s spatial perception of aproject?

OPEN SOURCED VIRTUAL REALITYThe wait for the consumer version of the Oculus Riftspurred a need to create alternatives usingDIYmeth-

ods and open sourcing. Such initiatives are enthusedby both creative communities, as well as big compa-nies, such as Google . At Google I/O conference [3],Google announced project Cardboard, a cheap vir-tual reality headset kit that costs approximately $20from various providers or if built independently. Itwas developed as a side project by Google employ-ees as part of a program called “20% project”. In prin-ciple, it is a smartphone mount, only made out ofcardboard and comes in various sizes according tosmartphone size. Additional components beside thepair of 40mmbiconvex lenses thatmakeup thehead-set are: a NFC magnet tag used for virtual naviga-tion input, a hook and loop fastener and a head strap.The virtual reality experience is provided by a sideby side projection capable smartphone. Each im-age corresponds to one eye and are separated visu-ally. The result between the juxtaposition of the sideby side images create one three-dimensional projec-tion. Google’s design ismade by the openly providedschematics, components and instructions for assem-bly. There are no official manufacturers for the head-set and theopen sourcedhardware is onlyoneaspectto project Cardboard. Developers of software for vir-tual reality applications are provided with a standarddevelopment kit from Google’s website. That meansthat users are not only providedwith cheaphardwarealternatives for head mounted displays, but also thetools to create virtual reality experiences. This opensource alternative has proven that virtual reality solu-tions and experiences can be cost effective, provideda user owns an increasingly indispensable tool suchas a smartphone.

Figure 1Google Standarddevelopment kit


Figure 2Open sourcealternative to headmounted displays

EXPERIMENT DESIGNIn order to quantify the differences in perceptionwhen a client might be subjected to a traditionalrepresentational method (in this case, printed ma-terial) and then a virtual reality presentation of thesame project, this paper will analyse an experimentconducted specifically for this purpose. The initialproposal mentioned an experiment which will try toobtain data from the perceptual difference betweentwo dimensional representation and virtual threedimensional representation, made possible also byopen sourced virtual reality. Therefore, the experi-ment aims to prove that virtual reality can be an effi-cient and reliable communication medium betweena client and an architect in the initial phases of archi-tectural design.

The experiment was conducted in two suc-cessive stages of data collection. The first stagewas completed using 2D viualisation of the projectscheme. This was pinned as A0 printed posters(two posters) , participants then were asked to studyand understand them , then answering the assignedquestions.

The second stage was conducted using VR

scenes of the same project scheme. On average, thevirtual tour lasted for approximately eight minutesper student. Then, the students had to complete asecondandfinal questionnairewith similar questionsrelating to their new understanding of the project. .

Figure 3Experiment designand methods

One questionnaire was designed to be used for thetwo successive stages, each addressing the used rep-resentational methods and its effect on understand-ing. Therefore, one rating system was used, consis-tency in the likert scale (rating) and number of ques-tions was important to provide data that is compa-rable to show any differences in subjects response (ifany). The experiment sample consisted of volunteersof stage 1 architecture ( first year’s architecture stu-dents). The choice was meant to simulate a real lifescenario where a client does not possess greatly de-veloped three dimensional vision. On average, thevirtual tour lasted for approximately eight minutesper student. The questions focused on initial impres-sions, the level of architectural detail, the space be-tween the two buildings, heights, interior space, bal-cony sizes and the overall quality of the project.


Figure 4Post impressionquestionnaire A

Technical requirementsThere were a number of requirements that neededto be met in order for the experiment to be carriedout. In order to define the two dimensional repre-sentational medium or the traditional way in whicharchitects present their projects to their clients, aset of two A0 posters had to be printed out. Theycontained a multi-story residential outline proposallevel project of modern Scandinavian style with anappropriate level of detail. The other representa-tional method was the virtual reality presentationwith the help of an open sourced headmounted dis-play, or Cardboard, the open sourced project createdbyGoogle. Theheadsetwas supplementedwithhighresolution display smartphone, the Nexus 5 by LG(researcher’s private smart phone). While the headmounted display was the tool, the presentation it-self had to be created. The virtual presentation con-sisted of several spherical panoramas rendered in-side the three dimensional project model. It repre-sents a 360 degree sphere shaped image where theobserver is placed at its center, able to look aroundjust like in real life at the surrounding environment.Then, the observer is able to fully observe the envi-ronment from a static vantage point (view point) at

the sphere’s core, similar to Google’s Street View.After acquiring theCardboardHeadset andusing

the in-built 360 degree spherical panorama viewer, itwas clear that custom panoramas would have to beadapted in order to be viewedwith the device. Thesecustom panoramas would not make use of the regu-lar camera software capabilities of anAndroid phone.Therefore, theywouldhave tobe imported andmadein a way as tomimic the regular spherical panoramaswhich are the result of a process of photo stitchingand editing being done automatically by the camerasoftware after taking a photo sphere picture.

Figure 5Program used-AutoPano Giga.

The panoramas were created using a real time ren-dering engine plugin called LumenRT(TM) from athree-dimensional model made in SketchUp(TM).The real time rendering engine enables the userto walk inside an enhanced version of the modelwhere quality lighting, shadows and environmentsare added, similar to a game engine.In order to cre-ate the spherical image, screenshots covering a fullspherical field of view had to be taken manuallyand then stitched together with the help of stitchingsoftware such as AutoPano Giga(TM). Vray(TM) en-ables the automatic creation and rendering of suchpanoramas, without the need for manual stitching.However, the rendering quality using Vray was un-satisfactory for the give scenario. Thus, the panora-mas used a real time rendering engine such as Lu-menRT(TM) for a better control over the results. Thepanoramas required post-processing in Adobe Pho-toshop (TM) to clear out any stitching errors. Oncecompleted, additional parameters were added to theimages using the Google Maps Street View Tool in


order to give spatial coordinates along three dimen-sional axes for correct panorama viewing inside a gy-roscopic sensor enabled smartphone. Here the im-agewas split into two small screens for each eye. Thevirtual environment or the person’s view respondedin relation to the movements of the user’s head.

Figure 6Example of thepanoramas werecreated using a realtime renderingengine plugincalledLumenRT(TM)

ResultsA group of 10 (stage one students) students volun-teered to test out the two representations by firstanalysing the architectural posters pinned up on tworeview boards. They would have to imagine them-selves as clients trying to fully understand the projectin terms of size and scale. The average time it took astudent to analyse the twoposters lasted for approxi-mately fiveminutes. After this, they had to assess thelimits of two dimensional representations and reflecton the quality of architectural information by com-pleting the first set of questionnaires (A). This first setwould focus on the project’s exterior space, interiorsand balconies. The subject would have to rate theclarity of information they’ have perceived on a scaleof 1 to 10 within the previously mentioned aspects.Once they completed the questionnaire, they had tolook into the same project using open source virtualreality headset. Here, they had to place the headmounted display over their heads and go throughnine spherical panoramas which consisted of differ-ent exterior, interior or balcony vantagepointswithinthe same project. After this, they had to assess thelimits of VR representations and reflect on the quality

of architectural information by completing the sec-ond set of questionnaires (B).

In order to compare each participant’s’ re-sponses, a graph for the Likert scale data for ques-tionnaire A and B was produced using Excel, eachgraph represents the individual differences with re-gard to clarity of information (figure 7-11), when thetwo representations were used . Then a graph withaverage values of each question was produced for allparticipants which compare the average score for allsubjects (10), figure 12.

Looking at the graphs individually, one can no-tice that there are many differences that a simplesummary would not be appropriate since it wouldmask these individual differences. However, someconsistent results are also apparent, the graphs (fig-ure 7-11) show the difference between the initial im-pression and the virtual reality impression for eachparticipant.

In general participants scored clarity of informa-tion andunderstandinghigher in VR from2Dposters.But there are some lower scores in VR when it com-pare to 2D posters. For example, participant 1 feltthat the overall quality of interior space (Q9) waslower by 1 point after experiencing it in virtual real-ity. participant 2 also felt that the quality of space inthe apartment facing West (Q6) was lower after see-ing the second representation. Participants 5 and 10felt that the quality of exterior space between build-ings (Q3) was lower by 1 point after experiencing thevirtual reality presentation. Participant 5 also hada lower score for the balcony views (Q 12) -( by 2points). Participant 6 scored a general decrease of1 point for all interior spaces (Q5-9). Participant 8seemed to have rated the quality of architectural de-tail (Q2) lower by one point after experiencing spacein virtual reality. Participant 10 felt that the quality ofspace inside the penthouse (Q8)was lower by 1 pointafter experiencing it virtually.

In general the average graph shows that all par-ticipants have rated VR impression higher than twodimensional posters. A detailed description is fol-lowed: with regard to project completeness (Q1),


Figure 7Students 1and 2results


Figure 9Students 5 and 6results



Figure 11Students 9 and 10results


participants’ average score for VR was higher by 1.4points. Also, the level of architectural detail (Q2) seenin VR was higher with an average of 1 point. Theaverage values for the exterior space between thebuildings (Q3) are also higher with an average of 1.1points. However, a significant average value differ-ence is scored with the building heights (Q4) at a 2.1points average. With regard to rating space quali-ties within the three apartments (Q5-8), participantsrated the amount of space in apartment facing north(Q5) in VR at an average point of 1.6. A smaller rat-ing of 0.9 point average increase is recorded for theapartments facing West (Q6). For the apartmentsfacing South East (Q7), the students rated their un-derstanding of it at an average of 1.4 point increase.The smallest difference recorded for an interior spacerefers to the penthouse (Q8) with a 0.3 point average.The overall difference in size of interior spaces (Q9)understanding is rated at an averages 1.5 points in-crease. Another significant difference is notices withcommunicate the overall size of balconies (Q10), adifference value of 2.5 points average is recorded.Also, rating their understanding of the size of pent-house balconies (Q11) recorded a difference of 1.7point average, while the biggest recorded differenceis at 2.7 points average on balcony views (Q12). Theoverall perceptual difference in project size (Q13) issignificant and stands at an average of 2.1 averagepoints. These are total averages of participants.

Figure 12Average score for allparticipants.

To summarise, the most notable perceptual differ-ences can be notices in understanding the followingtwo aspects of a project; (balcony) sizes and views,

then, building heights. Also, the overall project sizehas a significant difference in perception andwas justas important. Interior spaces were also better un-derstood and measured a significant increase in un-derstanding. From the students who rated the qual-ity of spaces interior spaces were the most frequentones which came up. However, The representationsused in the experiments where useful and valid forcommunication. Participants have rated their under-standing of both representations differently, higherrates were given to VR, these responses have demon-strated that student (client) understandingand learn-ing about architectural projects can be enhanced toan extent. Moreover, using new representation tech-nology was welcomed by participating students asbetter tool for them to understand the different as-pects of the project.

OUTCOME AND RECOMMENDATIONSOpen source virtual reality proved especially valu-able in conveying interior space, whether the userthought it was larger or smaller. The difference inperception wasmost obvious in this area and helpedpinpoint where people struggle most in identifyingspace: interiors. The data shows improvement acrossall aspects from the traditional two dimensional rep-resentational method and average values, with theoccasional negative result, which might be causedby each individual’s unique perceptual and cognitiveabilities. The students seemed to get a better graspon the design intent, proving that their understand-ing and judgement is not only improved, but bet-ter equipped in expressing disagreement at such anearly stage within the design process . As such, thisempowers the architect by reducing misinterpreta-tion of information between the two parties and bybetter bridging his intention to that of the client’sperception. Accessible VR technology has come along way since its inception. Today, architects canmake use of cheap open source hardware such asGoogle Cardboard [2] to efficiently and better ex-press space. While the means have become afford-able, it is still up to the architect to create the spaces


for virtual representations. Agoodvirtual representa-tion is dependent on detail and the architect’s threedimensional modelling capability. Virtual reality rep-resentations act as an extension to the existing skillsneeded for good visualisation.

Virtual reality representations are another wayto present and to efficiently communicating design.Quality high resolution panoramas will involve moreeffort and it is up to the architect on how much con-trol they want to have within the process of creat-ing such representations. As mentioned, lower fi-delity breaks immersion and representations can suf-fer. The clearer the information, the better the under-standing. Amanual process of stitching rendered im-ages to form the spherepanoramas is recommended,as it gives more control over the process. One of thestudents made a remark that the virtual representa-tions would be better if the user would be able towalk inside the model in real time. However, opensourced hardwaremaking use of smartphones as dis-plays do not possess the processing power yet to cre-ate real time immersion. The Oculus Rift, HTC Viveand Sony Morpheus are more suited for this kindof immersion, however, they are not open sourced.Real time virtual reality will become one of the mainways in which architects present their projects untila better alternative will replace it. However, thereare many factors which be responsible for a slowadoption. High fidelity real time representations re-quire great amount of graphical processing powerand appropriate hardware is still expensive. Thepricecannot be justified and traditional representationalmethods can relatively express most building mod-elling information at a fraction of the cost. However,adoption of the technology pays back in the longterm when money is saved by reduced design iter-ation. Emergent technologies are usually met withscepticism, especially in the conservative architec-tural profession.

REFERENCESCarreiro, MBT and Pinto, PDL 2013 ’The evolution of

representation in architecture’, 1st eCAADe Regional

International Workshop Proceedings, University ofPorto, Faculty of Architecture, Portugal

Christensen, BT 2005, CreativeCognition: AnalogyAnd In-cubation, Ph.D. Thesis, Department of Psychology,University of Aarhus.

Gill, C 2004 ’Visualizing Continuity between 2D and 3DGraphic Representations.’, Proceedings of the Inter-national Engineering and Product Design Education,Delft University, The Netherlands.

Kreutzberg, A 2014 ’New Virtual Reality for ArchitecturalInvestigations’, Proceedings of the 32nd eCAADe Con-ference - Volume 2, Department of Architecture andBuilt Environment, Faculty of Engineering and En-vironment, Newcastle upon Tyne, England, UK, pp.253-260

Kuksa, I andChilds, M 2014,MakingSenseof Space, Chan-dos Publishing

Van der Lugt, R 2000, ’Developing a graphic tool for cre-ativeproblemsolving in designgroups’,DesignStud-ies, 5(21), pp. 505- 522

Salman, HS 2011, The impact of CAAD on designmethod-ology and visual thinking in architectural education,Ph.D. Thesis, RGU

Stacey, M and Eckert, C 2003, ’Against ambiguity’, Com-puter Supported Cooperative Work, 2(12), pp. 153-183.

[1] http://www.theverge.com/a/virtual-reality[2] https://vr.google.com/cardboard/[3] https://www.google.com/events/io


Computational Thinking

Qualitative Representation and Spatial Reasoning in aRule-Based Computational DesignModel

Amer Al-Jokhadar1, Wassim Jabi21PhD Researcher, Welsh School of Architecture, Cardiff University, UK 2SeniorLecturer, Welsh School of Architecture, Cardiff University, UK1,2{Al-JokhadarA|JabiW}@cardiff.ac.uk

Integrating social and cultural constraints in computational models remains achallenge due to the difficulties in representing them algorithmically. Thisresearch aims to find a mechanism for combining shape grammars and spacesyntax methods for exploring spatial-formal features that affect the social life inresidential buildings. 'Spatial reasoning' as a method for understanding thesocial logic of spaces and the residents' behavior, integrated with 'discursivegrammar' as a method for describing formal and topological relationships, areadopted. Several computational tools are used for analysing qualitative aspects,such as privacy, social interaction, and accessibility. An automated model ofspatial/syntactical analysis, embedded in Rhino/Grasshopper, offers analternative method for extracting topological relations and syntactic calculations.Using this tool, designers can add new aspects to the justified graph of Hiller andHanson, as a representation to formal and social realities, such as orientationand geometric configuration. Results of analysis are transformed into codes,parameters, and descriptions, to be used for designing future developments,inspiring from local traditions. The target is to generate different alternatives forsocially sustainable, and 'contemporary vernacular' buildings, which respect thecontext, and the needs of users.

Keywords: Spatial Reasoning, Syntactic Analysis, Discursive Grammars,Computational Models, Social Sustainability, Visibility Graph Analysis

INTRODUCTIONComputation is about manipulating ideas and solv-ing problems in a clearly defined steps that are rou-tinely made by computer (Terzidis 2006). This ap-proach involves in the thinking process of the archi-tectural design and couldbeused in different phases:analysis, animation, simulation, and formgeneration.Currently, the main focus of computational models

is primarily limited to formal and environmental re-quirements. Yet, qualitative factors, such as socialand cultural constraints, are also important as theylead the building to be in harmony with its contextand theneedsof its users. Integrating these criteria inthe computational process remains a challenge dueto the difficulty of algorithmic representation (Yüksel2014).

Computational Thinking - eCAADe RIS 2017 | 35

This research aims to address such social andspatial issues in the design of vertical residential de-velopments. One approach is to draw inspirationfrom local traditions for generating a ‘contemporaryvernacular’ building. The Middle East and NorthAfrica (MENA) region is selected for the implementa-tion of the model, as most of the current projects ig-nore the specifics of the place and the society (Wood2013). In contrast, the vernacular model of housesshows that these precedents are good examples ofsocially cohesive and healthy environments (Al-Masri2010).

‘Spatial reasoning’ as a method for understand-ing the social logic of spaces, combined with ‘dis-cursive grammar’ as a method for describing for-mal and topological relationships, are adopted. Dif-ferent computational tools are used for analysingsuch qualities. These software include Syntax2D forspecifying spatial elements that affect the visual pri-vacy; AGraph for carrying out syntactic calculations;and DepthmapX for understanding the behaviourof people in relation to the spatial configuration ofthe built environment. Finally, a combined modelof spatial/syntactical analysis, embedded in Rhino/-Grasshopper, is presented.

SPATIAL REASONING AND ‘DESIGN SPACEEXPLORATION’Jerome Bruner, in his studies about the psychologyof knowing, defined ’reasoning’ as ’going beyond theinformation given’ (Bruner 1973). In the field of archi-tecture, spatial reasoning is a logical process of anal-ysis that enables designers’ understanding of the lay-out complexity, and the exploration of features thathave social or experiential significance (Abshirini andKoch 2013). For instance, tracing the visual fieldsfrom a certain location in a building allows a clearevaluation of spatial elements that affect the privacyof occupants. Different methods, such as space syn-tax and shape grammars, could be used for carryingout reasoninganalyses toderive social and spatial pa-rameters that affect the design of the built environ-ment.

Space syntax: a topological analysis ap-proachSpace syntax theory, developed by Hillier and Han-son in 1984, explores social relations implicit in thearchitectural setting. Based on their book ‘The SocialLogic of Space’, this process requires understandingphysical topologies between design elements, tak-ing into account all other spaces in the system (Han-son 1998). These relations are represented visually,through ‘justified node-and-connection’ graphs, toshow the hierarchy of the overall layout. However,these diagrams does not generate descriptions forthe formal reality of the design (Osman and Suliman1994). For instance, spaces couldbe connected indif-ferent alternatives, and have the same justified graph(see Figure 1). Furthermore, functions located on dif-ferent levels/floors need to be identified from othernodes in the system. The next section suggests howthese limitations are addressed.

Hillier and Hanson (1987) also suggested thatspatial relationships can be described mathemati-cally. For example, they proposed different measure-ments, such as connectivity, integration, and controlvalues, to quantify syntactical analyses. Results ex-tracted fromthese tests areuseful for interpreting theoverall configuration and the social life in the build-ing (e.g. high integration values indicate that spacesare busy, more accessible, and less private). In con-trast, low values can mean that these functions aremore segregated and more controlled in movement.However, studies focusing on how such an approachmight be used for generating or inspiring new de-signs remain limited (Lee et al. 2013).

Shape grammars: a formal geometric anal-ysis and generative approachShape grammars, developed by George Stiny andJames Gips in 1970s, is a formal system that workswith geometries rather than symbolic computations(Stiny andMitchell 1978). It allows designers describ-ing and analysing existing forms, and then generat-ing new alternatives based on the original style. Theroots of this approach builds on the work of Christo-

36 | eCAADe RIS 2017 - Computational Thinking

Figure 1A node-and-connectiondiagram that isapplicable for threealternatives ofspatial relations(Adapted byauthors, after(Osman andSuliman 1994))

pher Alexander, ‘A Pattern Language’, in 1977. A ‘pat-tern’, which could vary in its scale from a city to abuilding or a detail, addresses a problem and thenrecognises solutions and design practices that arebalanced within the defined context, in an attemptto reconstruct the knowledge about what makes ar-chitecture beautiful (Alexander 1977).

The framework for developing a shape grammarcould be outlined in four stages. (1) defining vocabu-laries (basic shapes); (2) determining spatial relation-ships; (3) formulating rules to be applied on forms;and (4) combining/articulating shapes through ap-plying rules recursively to define a language of de-sign or to generate new solutions (Eilouti and Al-Jokhadar 2007). After reviewing several cases thatuse this method for generating new spatial config-urations, four types of rules could be applied: (1)additive rules, where geometries are added aroundthe initial shape, such as the grammar of QueenAnne house (Flemming 1987); (2) subdivision rules,which divide the original shape into new zones withnew proportions, such as the grammar of Malague-ria house (Duarte 2005); (3) rules that are based on agrid, which start with defining dimensions and pro-portions for the grid, and then applying transforma-tions, such as Palladian grammar (Stiny and Mitchell1978); and finally (4) transformational rules, whichcombine additional and subdivision rules for creatingopenings, terraces, and roofs (Verkerk 2014).

However, shape grammars do not clarify social,cultural, and environmental dimensions of designs,as they deal only with geometric properties (Co-

lakoglu 2000). Moreover, some design possibilitieshave no architecturalmeaning or are irrelevant to thelocal context (Eilouti and Al-Jokhadar 2007).

Discursive grammarsTo control the process of form generation and theproduction of unique solutions, George Stiny (1980)introduced labels, such as letters; symbols; or points,associated with shapes. Furthermore, specifiedequations, constraints and transformations, such astranslation; rotation; reflection; or scale, could be ap-plied to increase the number of solutions and theflexibility in design.

Another method for controlling the process is toadd descriptions or textual information to the def-inition of rules. Rudi Stouffs (2015) defined threeschemes for these descriptions. The first type is ‘re-flections’, which reveals spatial vocabularies that formthe design. The second category is ‘expressions’ thatdescribe some properties of the design, such as vol-ume, cost or manufacturing plan. The final schemeis ‘design briefs’, which define conditional specifica-tions, or functional zones and their adjacency rela-tions. The followingexample showsdescriptions thatare associatedwith spatial rules for generatingpart ofa courtyard house (see Figure 2).

DEVELOPING A SYNTACTIC-DISCURSIVEMODEL FOR ENCODING SOCIAL AND SPA-TIAL QUALITIES OF DESIGNSThis research aims to build on the benefits of shapegrammar and space syntax methods for exploring


Figure 2Part of a discursivegrammar thatgenerates thesemi-public zone ina courtyard house(Authors)

spatial-formal features that affect the social life of oc-cupants. Moreover, it tries to find solutions for thedifferent limitations facedby these tools, throughde-veloping a syntactic-discursive model that encodesspatial constructs.

Based on a literature review of social sustainabil-ity in residential buildings (Modi 2014; Oldfield 2012),the authors identifies 13 indicators that need to beaddressed in the design process. These indicators in-clude: Population density and crowding; hierarchyof spaces; amount of living spaces that affect socialinteraction; human comfort; accessibility; visual pri-vacy; acoustical privacy; olfactory privacy; spiritual-ity; security and safety; viewing theexterior; availabil-ity of services; and hygiene.

To measure the current needs of residents andcapture such qualities, we conducted a phenomeno-logical study in the Summer of 2016, through dis-tributing a questionnaire to 211 families from 17countries within the study area. Furthermore, we im-plemented a formal-geometric study for extractingtypologies of historical precedents through examin-ing 48 traditional town houses and neighbourhoodsdistributed on MENA region. We then used threecomputational tools for carrying out these typolog-ical and syntactic analyses:

• AGraph: an open-access software, developedby Bendik Manum, Espen Rusten, and PaulBenze, in 2005, at the Oslo School of Architec-ture and Design. This ‘node-and-connectionmodel’ produces syntactic calculations, and

two types of justified graphs: depth of spacesfrom the root space; and integration of func-tions (Manum et al. 2005) (see Figure 3).

• Syntax2D: an open-source software devel-oped in 2007 at the University of Michigan toexecute isovist analysis. It addresses the visualfields of a person at one location of the envi-ronment, and along a movement path (Wine-man et al. 2007) (see Figure 4).

• DepthmapX : a ‘Visibility Graph Analysis (VGA)’tool developed firstly in 2000 by AlasdairTurner at the Space Syntax Laboratory, TheBartlett, University College London (UCL), andthen by Tasos Varoudis (2011-2015). VGA isbased on the reflection of light to understandthe spatial configuration of the environment.VGA includes five types of tests: (a) connec-tivity analysis; which creates visibility connec-tions between all spaces; (b) visual integra-tion, which specifies the degree of privilege ofone point over its immediate neighbours; (c)through-vision analysis, which looks at howvisual fields varies within an environment; (d)depth analysis, which shows changes of direc-tion that would take to get from the selectedlocation to any other locations; and (e) agentanalysis, which indicates patterns of move-ment, and the frequent use of spaces releasedfrom one point (Turner 2001) (see Figure 5).

The developed model of analysis adds new as-pects to the justified graph of Hiller and Hanson, as


Figure 3A justified graphshowing theintegration valuefor each space,produced byAGraph software(Authors)

Figure 4Samples of Isovistanalysis for threetraditionalcourtyard houseslocated in Syria,produced bySyntax2D software(Authors)


Figure 5Samples of VisibilityGraph Analysis(VGA) for threetraditionalcourtyard houseslocated in Syria,produced byDepthmapXsoftware (Authors)

a representation to formal and social realities. Theseissues are (see Figure 6):

1. Patterns of movement, and distances be-tween the centre of the courtyard and thecentre of each space, to analyse accessibilityand security inside houses.

2. The actual geometry of each space ratherthan symbolic nodes. Shapes are arranged toshow:

• Hierarchyof spaces (public, semi-public, semi-private, private, and intimate);

• Orientation (West, East, North, South, North-East, North-West, South-East, and South-West);

• Typeof enclosure (covered, open, semi-open);• Shared surfaces between adjacent spaces;

and• Entry point(s) between spaces.

Moreover, the spatial analysis includes geometricproportions for each space; percentage of space areafrom the overall area of the house; area of the spacein relation to the area of the courtyard; and the dom-inant users for each space (male, female, or both).

A COMPUTATIONAL TOOLKIT FOR PER-FORMING SYNTACTIC-FORMAL ANALYSISUSING GRASSHOPPERThe abovementionedmethod requires from design-ers an extra effort to calculate spatial qualities, suchas areas and proportions of spaces. Moreover, theuse of AGraph software for extracting syntactic val-ues requires drawing the ‘node-and-connection’ jus-tified graph manually. Thus, errors could easily occurduring this process. Therefore, it is useful to developan automated computational tool for analysing floorplans in a short time of execution, and with a highdegree of accuracy that does not require the user topossess an advanced level of knowledge in syntacticanalysis.

For this research, Grasshopper, a plugin forRhinoceros, is used for carrying out the needed anal-ysis. Grasshopper is a visual scripting tool that helpsthe design to process (Fathi et al. 2016). It allows in-put data to be passed from one component to an-other via connecting wires. Several plugins could bedownloaded for executing different utilities withoutleaving the tool itself. The following section illus-trates a detailed workflow of the automated model.


Figure 6Aspects added tothe justified graphas a representationto formal and socialrealities (Authors)

Model workflow and the user interfaceThe model depends on generating the layout of his-torical cases according to a ‘spacepartitioning’mech-anism (Knecht and Konig 2010). It is about splittinga region into sub-spaces (cells). This geometric rep-resentational technique, using non-manifold topol-ogy (NMT), defines topological relations between ad-jacent spaces without any void (Jabi 2016).

The first step requires users to draw the overalllayout boundary for the building (as polyline), inter-nal partitions representing shared surfaces betweenspaces (as lines), and doors (as rectangles). However,thicknesses of walls are ignored. Once these featuresare obtained from a ‘selection’ component, the par-titioning process is executed accordingly using NMT.Aunique legendnumber is assigned automatically toeach cell. This process could be applied on any layoutthat is composed of regular or irregular geometries.

The second step involves typing a function labelfor each cell, and then selecting spaces from lists ac-

cording to two criteria: hierarchy of spaces (public,semi-public, semi-private, private, or intimate zone);and type of enclosure (open or covered area). A tagcomponent is implemented for each cell. Themodel,then, computes the following values, which are alsodelivered on the form of Excel spreadsheet (see Fig-ure 7):

1. The area of each space, and the percentagefrom the total area of the house;

2. The total area for each hierarchical zone, andthe percentage from the total area of the lay-out;

3. The distance from the centre of each cell tothe centre of the main courtyard; and

4. Colour-coded syntactic values for each space,which include:

• Integration value: describes the averagedepthof a space toother spaces in the system;

• Control value: measures the degree to whichan area controls access to its immediate


Figure 7Model workflowand the userinterface, showinginput data from theuser, sample offunctions to carryout the partitioningprocess and theanalysis, and theoutput tables anddiagrams (Authors)

neighbours taking into account the numberof alternative connections that each spacehas;

• Entropy value: the difficulty of reaching otherareas from that space;

• Relative asymmetry: values that are closerto 0, means that spaces are more integrated,more accessible, and less private (minimumdepth of spaces). In contrast, higher valuesthat are closer to 1, indicate that spaces aremore segregated, more controlled in move-ment, and more private (linear sequence ofmovement, and maximum depth of spaces);and

• Difference factor: values that are closer to 0

mean that spaces are more structured, whilehigher values, i.e. closer to 1, indicate thatspaces are more integrated.

Visually, four types of diagrams are generated:

1. Orientation of spaces: based on the coordi-nates of the centre of each cell in relation tothe centre of the layout, a unique coloured cir-cle that indicates the location is assigned toeach space;

2. Node-and-connection syntactic diagram,showing links between spaces. Each linkmeans that there is a door/access betweenthese two cells;

3. Hierarchy of spaces, with a colour code for


each zone; and4. Distances between the centre of the court-

yard and the centre of other spaces.

READINGS FROM SPATIAL AND SYNTAC-TIC ANALYSESWhen the results of syntactical-discursive analyseswere examined, the following spatial features thataffect the social life of occupants were observed.The traditional house consists of different hierarchi-cal zones (public, private, and intimate spaces). Thecourtyard is the largest space in the house and themost accessible and connected function. Other func-tions are controlled and accessed through the court-yard and follow its geometric patternwith a symmet-rical layout arrangement. Moreover, the depth be-tweenprivate areas and the courtyardprovides apro-tected and comfortable atmosphere for family mem-bers to move inside the house easily.

The Isovist analysis and the Visual GraphAnalysis(VGA) show that the privacy of the household is pro-tected from public and semi-public spaces (the en-try hall and the guest room). Different mechanismsare used for achieving this result, and to strictly limitaccess to the courtyard by guests. These features in-clude the bent entrance, the use of partitions in frontof the main entrance, and the size and the locationof windows for guest rooms. Moreover, the spatialconfiguration shows that reception rooms are shal-low spaces that are suited off of the courtyard nextto the entry hall. In contrast, findings reveals that inti-mate spaces (bedrooms), which have a lower integra-tion value, are more integrated with private spaces.

CONCLUSIONAchieving social sustainability in residential build-ings requires a holistic approach for clarifying spa-tial qualities that affect the social life inside houses.A spatial reasoning approach is presented to under-stand such topologies. For instance, studying the lo-cation of each space, and measuring distances be-tween functions, are useful for analysing accessibil-

ity and movement. Moreover, defining connectionsbetween spaces, and describing their geometry andscale, offer information about their hierarchy, thedegree of social interaction that takes place withinthem, and their ability to provide comfort to their oc-cupants. Analysing these factors creates a databasethat can be used to improve the social qualities of fu-ture developments.

The proposed automated model of analysis, of-fers an alternative method for extracting topologicalrelations and syntactic calculations. This tool couldbe easily used to analyse floor layouts that have anysize or geometry, in a short time of execution. More-over, users do not need to know exact proceduresof syntactic calculations, or draw the justified graphfor the system, as themodel provides them automat-ically. Translating such qualities into rules and pa-rameters leads to the construction of a socio-spatialgrammar that is related to the local context. Thisgrammar will be used in the next stages of this on-going research for the emergence of vertical residen-tial developments that are socially sustainable, andrespect the culture and the needs of its users.

ACKNOWLEDGMENTSThe authors would like to acknowledge University ofPetra, Jordan for funding the study. This research issupported by a Leverhulme Trust Research ProjectGrant (Grant No. RPG-2016-016).

REFERENCESAbshirini, E. and Koch, D. 2013 ’Visibility analysis, sim-

ilarity, and dissimilarity in general trends of build-ings layouts and their functions’, In Kim, Y. O. et al.eds. Proceedings of the Ninth International Space Syn-tax Symposium, Seoul: Sejong University

Al-Jokhadar, A. and Jabi, W. 2016a ’Enhancing Social-Cultural Sustainability in Tall Buildings: A Trace fromVernacular Houses’, In: Wood, A. et al. eds. Citiesto Megacities: Shaping Dense Vertical Urbanism: Acollectionof state-of-the-art,multi-disciplinarypaperson urban design, sustainable cities, and tall buildings.Proceedings of the CTBUH-2016 International Confer-ence - China., Chicago: Council on Tall Buildings andUrban Habitat, p. 633–641


Al-Jokhadar, A. and Jabi, W. 2016b ’Integrating Paramet-ric Models with Qualitative Dimensions’, In: Al-Attili,A. et al. eds. Parametricism vs. Materialism: Evolutionof Digital Technologies for Development. Proceedingsof the 8th ASCAAD Conference of the Arab Society forComputer Aided Architectural Design, London: Impe-rial House Publishers

Al-Masri,W. 2010, ’TheCourtyardHouse in Kuwait Today:Design Approaches and Case Studies’, in Rabbat, N.(eds) 2010, TheCourtyardHouse: FromCultural Refer-ence toUniversal Relevance, Ashgate Publishing Lim-ited, in association with the Aga Khan Program forIslamic Architecture, Surrey, England, p. 203–222

Alexander, C. 1977, A Pattern Language, Oxford Univer-sity Press, New York

Bruner, J.S. 1973, Beyond the Information Given: Studiesin the Psychology of Knowing, W W Norton & Co Inc.,New York

Colakoglu, M.B. 2000, Design by Grammar: AlgorithmicDesign in an Architectural Context, Massachusetts In-stitute of Technology

Duarte, J.P. 2005, ’A discursive grammar for customiz-ing mass housing: the case of Siza’s houses atMalagueira’, Automation in Construction, 14(2), p.265–275

Eilouti, B. and Al-Jokhadar, A. 2007, ’A Generative Systemfor Mamluk Madrasa Form-Making’, Nexus NetworkJournal, 9(1), p. 7–30

Fathi, A., Saleh, A. and Hegazy, M. 2016, ’ComputationalDesign as an Approach to Sustainable Regional Ar-chitecture in the Arab World’, Procedia - Social andBehavioral Sciences, 225, pp. 180-190

Flemming, U. 1987, ’More than the sum of parts: thegrammar of Queen Anne houses’, Environment andPlanning B: Planning and Design, 14(3), p. 323–350

Hanson, J. 1998, Decoding Homes and Houses, The PressSyndicate of the University of Cambridge, Cam-bridge

Hillier, B. and Hanson, J. 1984, The Social Logic of Space,Cambridge University Press, Cambridge

Hillier, B., Hanson, J. and Graham, H. 1987, ’Ideas are inthings: an application of the space syntax methodto discovering house genotypes’, Environment andPlanning B: Planning and Design, 14, pp. 363-385

Jabi, W. 2016, ’Linking Design and Simulation UsingNon-manifold Topology’, Architectural Science Re-view, 8628(January), pp. 1-12

Knecht, K. and Konig, R. 2010 ’Generating floor plan lay-outs with k–d trees and evolutionary algorithms’, In:13th Generative Art Conference GA2010, Politecnicodi Milano University, Italy, p. 238–253

Lee, J.H., Ostwald,M.J. andGu, N. 2013 ’CombiningSpaceSyntax and Shape Grammar to Investigate Architec-tural Style: Considering Glenn Murcutt’s domesticdesigns’, In: Kim, Y. O. et al. eds. Proceedings of theNinth International Space Syntax Symposium, Seoul:Sejong University

Manum, B., Rusten, E. and Benze, P. 2005 ’AGRAPH,Software for Drawing and Calculating Space Syntax‘Node-Graphs’ and Space Syntax ‘Axial-Maps’’, Pro-ceedingsof the5th International Space Syntax Sympo-sium, Delft, pp. 97-101

Modi, S. 2014, ’Improving the Social Sustainability ofHigh-rises’, CTBUH Journal (Council on Tall Buildingsand Urban Habitat), Issue 1, pp. 24-30

Oldfield, P.F. 2012, Tall Buildings and Sustainability, Uni-versity of Nottingham

Osman, K.M. and Suliman, M. 1994, ’The Space SyntaxMethodology: Fits and Misfits’, Architecture & Com-portement, 10(2), pp. 189-203

Stiny, G. 1980, ’kindergarten Grammars: designing withForebel’s building gifts’, Environment and Planning B,7, pp. 409-462

Stiny, G. and Gips, J. 1972 ’Shape grammars and the gen-erative specification of painting and sculpture’, Pro-ceedings of the Information Processing Congress 71(Volume 2), p. 1460–1465

Stiny, G. andMitchell, W.J. 1978, ’The Palladian grammar’,Environment and Planning B, 5(2), pp. 5-18

Stouffs, R. 2015 ’Description grammars: an overview’,Proceedings of the 20th International Conference ofthe Association for Computer-Aided Architectural De-sign Research in Asia CAADRIA, Hong Kong, p.137–146

Terzidis, K. 2006, Algorithmic Architecture, Elsevier Ltd,Oxford

Turner, A. 2001 ’Depthmap: a program to perform visi-bility graph analysis’, Proceedings of the 3rd Interna-tional Symposium on Space Syntax, Atlanta, pp. 7-11

Verkerk, N.M. 2014, A general understanding of shapegrammar for the application in architectural design,Master’s Thesis, TU Delft, the Netherlands

Wineman, J., Turner, J., Parra, S., Jung, S.K. and Sense, N.2007 ’Syntax2d: an open source software platformfor space syntax analysis’, In: KUBAT, A. ed. Proceed-ings of the 6th International Space Syntax Symposium,Istanbul, pp. 23-26

Wood, A. 2013, ’The Futureof Tall’, in Parker, D. andWood,A. (eds) 2013, The Tall Buildings Reference Book, Rout-ledge, Oxon, p. 353–361

Yüksel, E. 2014, Architect’s Role in Parametric Design, Mas-ter’s Thesis, TU Delft, the Netherlands


Computation in Architecture

Subtitle: The Handling of Computational Devices in Buildings

Werner Lonsing11Independent [email protected]

The common point of view in architectural design regarding computation states,that in architecture computational aspects of a building's design are ignored oncea building's design is finalized. Computation beyond the completion of a buildingdoes not belong into the realm of architectural practice.The undeniable existenceof computers in buildings should not only be regarded as addition to a building'sdesign. Architects have to deal with computational devices and their functionalityin buildings and they have to design both form and function of them. Anintegrated design approach as part of the holistic concept of designing buildingsneeds to be established, because computation will become an essentialcomponent in the architectural design process.

Keywords: Programming, Computational architecture, smart buildings, cloudcomputing

TOPIC(S)

• Design Concepts &Strategies• Human-Computer Interaction• Smart & Responsive Design• Spatial Reasoning & Ontologies• VR, AR & Visualisation

INTRODUCTIONComputation in architecture, as it is well described inthe conference’s theme, “spans from the virtual rep-resentation of space to its physical making” [Spaeth& Jabi 2016]. This common point of view in archi-tectural design inherently states, that in architecturecomputational aspects of a building’s design are ig-nored once a building’s design is finalized and thebuilding is erected. The finding correspondences

with the typical job profile of an architect that afterthe architectural production has ended the work in-cluding the work the engagement of computers isdone. It looks like a perfect fit.

The shortsighted assumption that computationin architecture is only part of the process to enhanceand ease the design introduces the limitation that af-ter fulfilling the promise of a design the computa-tional elements therein are not any longer part of itsarchitecture, if they have ever been. Computers in-side the finished and established building are usuallynot regarded as part of architecture, because compu-tation beyond the completion of a building does notbelong into the realm of architectural practice.

This is reflectedby the naive attitudes of not han-dling computation by simply ignoring computers inbuildings, like as the conference’s theme, refusing


them, neglecting, denying and even fighting it ina sense, that buildings cannot be ‘smart’ (Koolhaas2015).

It is obvious that these strategies are leading toa dead-end. Theoretical and practical computationalapplications in architecture cannot be left out at will.They are everywhere and buildings are no excep-tion. Instead computers as part of buildings shouldbe considered as valid parts of the architectural de-sign and taken as chance to the ongoing quest of im-proving architecture.

COMPUTERS IN ARCHITECTUREComputationalmachines and applications can be es-sentially divided into two groups, computers in thehand of a user within a building running specific soft-ware, likemobile device with a navigational softwareor an AR-system (Augmented Reality) and comput-ers as part of the building itself, usually small em-bedded computers with limited capabilities and es-pecially no dedicated user interface, commonly as-signed to home automation or ‘smart buildings’.

Mobile devicesFor a decade now mobile devices with their touch-screens have silently infiltrated the perception of ourreality, meaning that the sole focus on our worldis vanishing in favor of a more connected and net-worked surrounding. Any activity like a simple tweetor an SMS on a smart-phone is related to a location,the place where the users gets distracted and shiftsthe focus from the environment to a small display.

Designated applications on mobile devices forcertain buildings or locations are already known, likeapps for museums or exhibitions, often based onaugmented-reality technologies, or specifically de-signed navigational applications for particular locali-ties.

What is lacking is an integrated design of boththe application and the building. By pretending thatprogrammers are not designers and designers cannotwrite code a building’s environment delivers onlythe static context where the app is fitted in. The us-

age of the app should not interfere with the architec-tural design, even though apps are already doing itall the time by mixing realities and changing the ar-chitectural appearance. This ostensible arguing be-comes more and more outdated.

Computersmounted in buildingUnlike walls, doors, windows, handles, locks or elsecomputers are not considered as part of a building.Instead they have become part of them as acces-sories. Computers merged inside building parts, typ-ically some electrical and mechanical items, as em-bedded systems are regarded as component of thatpart of the building. Computerized outlets deploy anintegrated relay as switch, locks aremotorized and soon. The upcoming question here is, if e.g. an auto-mated door is still a door with a lock and somemech-anism or a temporarily opening which appears to bepermanent, because only the absence of users madeit close. Hence in theory nobody never will noticethat particular spatial separation. Just an aspect here,which will not be discussed any further.

The point is that while the relation betweenmobile devices and buildings are only establishedthrough the simultaneous perception of their usersthe relation between computers mounted in build-ings and the building itself are even more tight, andhence is the design.

TERMS ANDDEFINITIONSThe utilization of embedded computers in buildingshas lead to some interesting wordings like ‘home au-tomation’, ‘smart devices’ and IoT (Internet of things).Before discussing them in detail at first these termshave to be acknowledged as technical terms. Oth-erwise philosophical issues and similar will becomeoverwhelming. Neither can a device be taken assmart, nor a complete home automated (What doesit do?) nor reality be augmented and so on.

Smart DevicesAdding an embedded computer, eventual some net-work interface and commonly some sensors and


some actuators to some part of a building assumablymakes them ‘smart’. A window with a computer is a‘smart window’, an outlet with a computer a ‘smartoutlet’ etc. Usually such computerized entities act tosome degree on their own without direct user con-trol, hence sensors and actuators are usually inte-grated.

Besides the inflationary usage of ‘smart’ whichcauses a lot of irritations (Koolhaas, 2015; Lonsing,Smartness and Interactiveness, 2016) the real func-tionality of the computer is disguised. It is not aboutsmartness or intelligence, it is about control. Modernembedded systems establish controlling, timing, su-pervision, regulations etc. in a quality that have beenunthinkable only a few years ago. With servos, color-ful lighting andmodern networking technologies at-tached to it new types of installations can be createdinside computerized buildings making them interac-tive, responsive and more flexible.

HomeAutomationHome automation slightly differs from smart devicesas it includes more sophisticated machines like acoffee-maker or a dish-washer. Already computer-ized those machines have become interconnectedandmay now are messaging their status and also ac-cepting orders from authorized users elsewhere, al-though strictly an Internet connections according tothe concept is not mandatory.

Surprisingly one special consideration in homeautomation is the wide variety of technologies, plat-forms and protocols. Some of themare standardized,some of them are open and a lot of them are still pro-prietary. However, essentially the usage of differentprotocols and frequencies is simply an implementa-tion detail.

IoTThe ‘Internet of Things’ is defined by its status, whereall devices are expected as connected through the In-ternet regardless of their functionality. It can be liter-ally everything.

One point to mention is the ubiquitous usageof speech recognition, which depends on the Inter-

net. All major provider, Apple with Siri, Amazonwith Alexa, Microsoft with Cortana, and Google withits refreshed Assistant are performing cloud-assistedspeech recognition as their services.

Common PatternNeedless to state that the terms and definitions andtheir related technologies are not a major concern.Themain problem is that in architecture a kind of yetunknown flexibility and control of the built environ-ment has been established. With the introductionof computation in buildings, this may include net-worked access, new concepts are requested to inte-grate computers into a building’s designprocess. Theapparent sprawling installations more than indicatethis as a consequence.

A look at the simple example of smart outletsfrom the supermarket, which can be triggered wire-lessly, may demonstrate the issue. There are a lot ofdifferent vendors, each of them uses specific tech-nologies and probably different frequencies, each ofthem has its own cloud and its own app. These de-vice are not designed for integration and their users’experience is less than optimal. But what really mat-ters, even for otherwise technology-resistant archi-tects, is their appearance. Although the look of com-mon outlets is not always pleasant, these kind of ex-tensions are distinguishable well designed in all dif-ferent forms. Plugged into a normal outlet on a wallthey look like pimples, just ugly.

Then there may be different types of outlets, be-cause they were once on sale, with different proto-cols of course. In addition theremaybe are those typ-ical cylindric sometimes blinking devices for speechrecognition, dominating a roombecause they shouldalways listening, and there maybe other devices tocome.

If as architects we otherwise do not care aboutcomputers and have no opinion regarding computa-tion whatsoever, this is a point we have to address.


THE TASKTo prevent a building’s design from electrical clut-ter has become the challenge in modern architec-ture. The related task is to do it by design. It im-plies rethinking the concept of buildings like it hasbeen done before when new technologies were in-troduced like e.g. the usage of steel or glass.

Therefor the computational infrastructure aspart of buildings should be laid out during the de-signprocess. A special typeof installations are centralcomputers or anetwork thereof inbuildings. Theyes-tablish an informational grid and a local cloud boundto the location eventually effecting the ownership ofdata, reliability, security and other issues of moderncomputation. A holistic view of a network as inte-grated part and as a specific element of the construc-tion of a building, when a building provides connec-tivity like shelter, is the focal point of this task. De-signing a layout for computation needs special con-sideration on its own during the design process.

Computers in public spacesArchitecture is not limited to buildings and there-for computation not bound to their physical pres-ence. Public spaces as part of urban planning arealso about to be computerized. One example aremuch more flexible public lighting concepts basedon smart street lights.

Some ProjectsInstead of theoretically elaborating on design issuessome projects are introduced. They have in commonthat they are all deploying a micro-controller basedon the Arduino-platform. While the genuine Arduinodoes not provide connectivity the specific type ofcontroller from Particle here in use are connected.Their different types of boards are interchangeableand can offer connection either through WiFi, GSMor Bluetooth with the same code-base. In additionthey can access either the company’s global cloud ora cloud locally installed on a small server, typically aRaspberry Pi.

Figure 1Lighting with acolor changinglamp.

The project related to deliver some new computa-tional supported functionality in buildings are moreor less all related to lighting. As technology artificiallighting simply provides the best cost-benefits ratio.User interfaces on the other hand may already cutcosts and do provide some savings. By substitutingwall switches and their really expensive high-currentpower lines with connected and programmable cus-tom made switches the switch panels on walls couldbe elimanated altogetherwith something like e.g. in-tegrated touch areas made from concrete.

Empathic Lighting. Empathic lighting as project(Lonsing 2016, Empathic Lighting) explores the us-age of computers, sensors and actors to illuminatea scene. With moving RGB-LEDs a fast respondinglighting system is developed to not only defy theimpact of changing daylight conditions but to an-swer the environmental light intake on the fly andeven anticipate a viewer’s acting. Still at a modelscale theworking systemcomprises two specially de-signed lighting units and a smart phone as both ascamera input and user interface.

Figure 2Concept ofempathic lighting,the lamps aremoving accordingto the position of auser.


Figure 3User-interface on asmart-phone tocontrol a singlelighting unit.

The lighting units are spot-lights with RGB-LEDsmounted on pan-tilt brackets. They are individu-ally controllable: color, brightness and the directioncan be changed on demand. Each lighting unit de-ploys its own computational controller, a commonArduino-board with an Ethernet-shield for connec-tions.

To determine the actual illumination a camerais used, which is already part of a mobile device.By means of image processing the immediate visualappearance is analyzed like the presence of usersand the ambient lighting. An overall color value isdetected by summarizing all pixels and calculate amean value thereof which is used for corrections.Then the detected color components of the LEDare lowered and the opposite color components areraised. As result the scene should always appear in

a constant light regardless of environmental inputswhatsoever.

An additional feature is the usage of the built-inface detection to anticipate the viewing direction ofa user in order to illuminate only the area of intereststhe user is looking at.

Figure 4Complete system,colored LEDsmounted onpan-tilt-bracketswith servos andmicro-controllers.

Fado Lamp. The Fado-lamp is a simple self-madeRGB-lamp similar to common mood lamps. It utilizes3D-printed parts to mount the LEDs and the micro-controller on top of the socket of the then removedlight bulb. The chosen Pixie-LEDs have the advan-tage, that they can be daisy-changed, hence a variantwith three LEDs is already projected.

Figure 5Projected socketwith designedparts.


Figure 6Printed parts andthe real lamp.

To create the printed parts the socket was measuredandvirtually rebuilt todesign themounts. Themicro-controller then were virtually placed as winglets andmounted using the physical pin layout of the boards.Due to the small size of the pins the fitting of thewinglets still has to be worked out.

Outlet Spider. The outlet spider is a smart re-mote switch which is built around a micro-controllermounted on a power shield. The shield includingAC/DC converter is inserted inside a typical junctionbox from a hardware store in order to simulate atypical building situation, where the installations aresunk into walls.

Figure 7Outlet spider withmicro-controllerand relais inside ajunction box.

The concept avoids any unnecessary devices, likesmart power bars or smart outlets to be plugged intoconventional outlets.

Smart Street Light. The Smart Street Light project(Lonsing 2016, Smart Street Light) was designed asa common street lamp with added colored lightingallowing for new concepts of public spaceswith inte-grated lighting installations.

Figure 8Concept of a smartstreet light.

As project it enters the realm of urban planning.The prototype of a street light is made from a com-monmushroom shaped enclosure, wherein differenttypes of light sources as high power LEDs, one RGB-and three white LEDs, sensors and the electronic de-vices, micro controller, shields and a power converterare integrated. It combines ambient and directedlighting as result of environmental stimuli detectionand an animated installation as well.


Figure 9Inside of the streetlight.

As a street light the power can be supplied throughthe pole. A remote control is not yet installed. Theusage of higher voltages here simply requires moreattention and extra care.

User Interfaces in BuildingsRegardlessof the functionalityof their correspondingcomputational devices common user interfaces inbuildings are following only two patterns, and bothare relying on the Internet comprising remote dataprocessing and storage.

Either connected embedded computers are con-trolled through mobile devices or by speech recog-nition. Both use cases are not as intuitive as theycould be. Focussing on some small display and wip-ing through screens is always distracting and almostas cumbersome as the time consuming verbal com-munication is .

Flexible Switch Pad. The switch pad combines anOLED-display, a push-rotary-button and a micro-controller. The button provides the input to display alittlemenu on the the display, walk through it and se-lect some functions, while the push-button is an on-off switch. One application is to control the colors ofthe LED of the Fado-Lamp and simply switch it on oroff.

Figure 10Projected switch tocontrol a complexlamp with a rotarybutton and a smalldisplay as UI.

Figure 11The same functionalswitch withmicro-controller ona breadboard.

The provisional implementation is hard coded, butit is contemplated, that all devices, once connected,may deliver theirmenus and related functions, hencethis small pad (Weiser 1991) is capable of controllingmore than one and different types of devices.

Wand.Thewand is a coloredball ona stick,which canbe visually traced. A camera with integrated imageprocessing detects the ball and its color and sendsthe detected features as values to the controller likee.g. what color has been chosen, is the ball raised orlowered and so on. Various colors may invoke differ-


ent functions.

Figure 12Camera withmicro-controllerand two coloredballs

The controller then acts according to the installedfunctionality.

PushButton.Thepushbutton is a simple all-purposebutton with a big push button, a rotary switch to se-lect a specific functionality and the micro-controller.Based on the idea to create a simple and easy touse emergency button the actual implementation ismounted inside an off-the-shelf button, which hasbeen produced to some office entertainment appli-cation making random comments.

Figure 13Simple push-buttonwith integratedrotary switch.

The button itself offers little degree of freedom. Oncethe up to twelve different functions are assigned,the button can be placed anywhere, and it can bepressed by hand or foot.

Touchable Pad from Concrete. The concrete pad isstill highly experimental. Goal is to establish a touch-

able surface made out of concrete without any me-chanical components. Using concrete as materialwith some electrically conductive additive and prob-ably some reinforcement permits almost any formand kind of integration in buildings made from con-crete.

Right now the pad can execute only one func-tion, because there is only one single touch area. Itis connected to the micro-controller by only one sin-gle wire.

It is contemplated that more touchable surfacescan provide more functionality. Then every touch-able area is connected by a dedicated wire., increas-ing their number significantly.

Other parameters may be arranged similar toother usages of concrete, it can be colored by usingpigments, the surface can be to some degree struc-tured and so on, as long as the electrically function-ality is maintatined.

Figure 14Pad made fromconcrete with wire.

COMPUTATION AS PART OF THE DESIGNIn architecture computers are already bound to thedesign process: whatever kind of building and con-struction is imagined, it is developed as data, at firsta virtual model, than as integrated building model.This widely accepted approach ignores the compu-tational statements done within a building. In otherwords, handling the undeniable existence of com-puters in buildings is not the task of an architect. This


has to be changed.At first, it has to be acknowledged that archi-

tects have to deal with computational devices andtheir functionality in buildings. As consequence theyshould secondly design both form and function ofthose computational devices and finally their shouldbe an integrated design of them as part of a holisticconcept of computation in buildings.

The role of programmingThe question is, how and to what degree architectshould and canhandle computers in buildings. Thereis no answer yet except that someprogramming skillsare required, because programming is an essentialpart of computation and no computer is runningwithout software. Programming will become part ofthe design process the one way or the other.

Then a question remains: Can architects acquireandmaintain an in-depth knowledge of some specif-ically designed architectural SDK (software develop-ment kit), programming language included?

This question is not simple to answer, but thereare some aspects worthwhile to mention:

• Architects are already programming. Theyare writing C-sharp or Python scripts as exe-cutable code for computers to enhance theirdrawing software-packages.

• Programming itself is changing dramaticallyand has become diversified. The old model ofa desktop-computer running some softwarenow competes with mobile-, cloud- and em-bedded computation and even more.

• New technologies are evolving. Tools like vi-sual programming or cloud-compilation de-tach the programmer from a specific platformand its IDE.

Other issues, like if there will be any support fora standardization from an industry or a major sup-porting governmental entity like ANSI (American Na-tional Standards Institute) or ISO (International Orga-nization for Standardization) then will be resolved al-most automatically based on needs.

CONCLUSION, KIND OFComputation will become an essential componentof architecture, there is no question about it. Bycloser looking at buildings it becomes evident, thatthe upcoming task here, if taken seriously, is nothingless than injecting a nervous system into pre-existingstructures, like themodernization of buildings, or his-toric concepts as building designs.

The standard phrase, that the introduction ofnew technologies will sooner or later lead to newforms of design applies here, too. It has happenedbefore. The difference to those events from the pastwith the introductionof newmaterials and structuresis only, that structures and materials are already inplace as static elements. With computation the nextkind of architectural style will become dynamic, flexi-ble and responsive. Buildings will change from an al-most immutable state to a collection of mutable ob-jects in control by the buildings’ inhabitants.

REFERENCESAchten, H 2015 ’Closing the loop for interactive architec-

ture’, Real Time-Proceedings of the 33rd eCAADe Con-ference, pp. 623-632

Al-Haddad, T 2008, ’Parametricmodulations inMasonry’,CAADRIA 2008 Proceedings of, 1, pp. 221-228

Anders, P 1999, Envisioning cyberspace: Designing 3Delectronic spaces, McGraw-Hill Professional

Beorkrem, C 2013, Material strategies in digital fabrica-tion, Routledge

Borchers, JO 2001, ’A pattern approach to interaction de-sign’, Ai \& Society, 15(4), pp. 359-376

Burry, M 2011, Scripting cultures: Architectural design andprogramming, John Wiley \& Sons

Dougherty, D 2012, ’The maker movement’, innovations,7(3), pp. 11-14

Fricker, P, Wartmann, C and Hovestadt, L 2008, ’Process-ing: Programming Instead of Drawing’, Proceedingsof eCAADe 2008, 2, pp. 525-530

Gershenfeld, N, Krikorian, R and Cohen, D 2004, ’The In-ternet of Things-The principles that run the Internetare now creating a new kind of network of everydaydevices, an“ Internet-O.”’, Scientific American, 291(4),pp. 46-51

Hertz, G 2011, ’Arduino Microcontrollers and TheQueen’sHamlet: Utilitarian andHedonizedDIY Prac-tices in Contemporary Electronic Culture’, Proceed-


ings of the 31st Annual Conference of the Associationfor Computer Aided Design in Architecture, 1, pp. 44-47

Hymers, W 1980, ’Electrically conductive concrete’, Con-crete construction, 25(5), pp. 411-415

Koolhaas, R 2015, ’The Smart Landscape: Intelligent Ar-chitecture’, Artforum international, 53(8), pp. 212-217

Lee, S, Tewolde, GS, Lim, J and Kwon, J 2015 ’Visionbased localization for multiple mobile robots usinglow-cost vision sensor’, Electro/Information Technol-ogy (EIT), 2015 IEEE International Conference on, pp.280-285

Lonsing, W 2013 ’Introducing aWorkshop to Build an Af-fordable 3D–Scanner: Presenting a Variety of Com-putational Concepts to Novice Students of Archi-tecture’, Knowledge-based design: proceedings of the17th conference of the Iberoamerican society of digitalgraphics. Valparaiso, Chile: SiGraDi, pp. 475-478

Lonsing, W 2015, ’Beyond Smart Remote Controls-Developing a More Integrated and CustomizableImplementation of Automation in a Building by Uti-lizing Tools and Concepts from Makers, Real Time-Proceedings of the 33rd eCAADe Conference’, Vi-enna, Austria, 2, pp. 679-686

Lonsing, W 2016 ’Empathic Lighting’, Design \& DecisionSupport Systems

Lonsing, W 2016, ’Smartness and Interactiveness in Ar-chitecture’, BlucherDesignProceedings, 3(1), pp. 157-164

Lonsing, W 2016, ’Smart Street Light with Capabilitiesbeyond Industrial Standards.’, Education forResearch,Research for Creativity, 1, pp. 215-220

Malinowski, A and Yu, H 2011, ’Comparison of embed-ded system design for industrial applications’, IEEEtransactions on industrial informatics, 7(2), pp. 244-254

O’Sullivan, D and Igoe, T 2004, Physical computing: sens-ingand controlling thephysicalworldwith computers,Course Technology Press

Spaeth, B and Jabi, WM 2016 ’The Virtual and the Physi-cal’, (Conference Theme)[1], Cardiff, Wales, UK

Teymourzadeh, R, KokWai andHoong, Mok Vee, Ahmed,SA, Chan, KW and Hoong, MV 2013, ’Smart gsmbased home automation system’, Systems, Process &Control (ICSPC), 2013 IEEE Conference, 1, pp. 306-309

Weiser, M 1991, ’The computer for the 21st century’, Sci-entific american, 265(3), pp. 94-104

Wiener, N 1961, Cybernetics or Control and Communica-tion in the Animal and theMachine, MIT press

[1] http://sites.cardiff.ac.uk/architecture/ecaade-

ris2017/


Digitizing the Unexplored

Bridging the Concrete Experience and Abstract Geometry Through PhysicalExperiments

Sema Alacam1, Orkan Zeynel Guzelci21Istanbul Technical University 2Istanbul Kultur University1,2{semosphere|orkanguzelci}@gmail.com

Ways of promoting creative approaches in the context of embedding digital designapproaches in education are increasingly becoming a popular topic inarchitecture schools. On one hand, digital design methods, approaches and toolsare diversified, while on the other hand, the danger for students losing controlover the design process by appears as a new phenomenon. We argue thatexperiential dimensions of empirical observation conducted with physicalexperimentation might be helpful for students to comprehend digital topicscreatively. With a particular emphasis on engagement of physicalexperimentation and computational thinking, this study aims to present apedagogical model which has been applied from 2014 to 2016 within apostgraduate course. The potentials and limitations of the introducedpedagogical model are investigated through students' bottom-up and top-downalgorithm development processes and the outcomes of the digital modellingexercises.

Keywords: pedagogical model, dynamic process, physical experiment,computational thinking

INTRODUCTIONTheoretical debates on chaos, complexity and non-linearity have resulted in a series of well-establishedcomputational models of dynamic processes suchas computational fluid dynamics, self-organisationof living organisms, swarm behaviour, various opti-mization models and physics engines. As a conse-quence, emerging models and approaches were in-troduced to researchers of different fields such as en-gineering, natural sciences, design, art and architec-ture.

As the basis or components of these computa-tional models; mathematics, geometry, algorithms,programming languages or representation of spacein digital environment can be all considered as aproduct and reflection of human thought. Regard-ing the relationship between reality and abstraction,Einstein (1921) used the term “axiomatic”. In the firstemergence of concepts, theories and axioms, theremight still be relationship among the experiencedworld, inference and assumptions. After gaining alogical structure, axioms and signs start to perform


autonomously and independently of their sources.When it comes to computational models, they usu-ally depart from original sources at least two times,being interpreted and reduced by the theoreticiansandby thecomputer scientist. This reductionprocessmakes it difficult for architects to understand, com-prehend and generate new assumptions in relationwith the original intuitive content.

Although architects have adapted various com-putational models to design processes, many archi-tects still tend to approach CAAD possibilities via as-sumptions of Cartesian space in relation to the on-tological crises of the digital age. There is a neces-sity to explore new relationships in the transitionsamong 2D, 3D and 4D information beyond Alberti’sperspective. With Lonsway’s (2002) words, the com-putational complexity is reduced to theMinkowskianSpace in the usage of CAD software. Kolarevic (2000)also emphasizes the dynamic, open-ended and un-predictable characteristics of digitally driven archi-tecture as a new potentiality. However, in differenceto other debates, Lonsway (2002) clearly pointedout the differences between visual complexity andcomputational complexity by asserting Greg Lynn’s(1998)blob still doesnot embodycomplexitybeyondMinkowskian Space conception.

Ononehand,ways of thinking, concepts, and vo-cabulary of conventional approaches dominate howonemight comprehend and approach the potentialsof digital tools. However, new concepts have beenemerging while they effect the improvement of un-derstandings. As a consequence, the emphasis onthe concepts such as typology, form and geometryhas been replaced by the focus on topology, forma-tion and geometrical relations in the field of archi-tecture. Moreover, new concepts and results are ex-panding the researcharea and innovativedesignpro-cesses.

In this study, we focus on how and in whichways itmightbepossible to introduce computationalthinking logic to design students as away of thinkingbeyond well-defined software packages and algo-rithms. With this concern inmind, we examined both

bottom-up and top-down approaches of algorithmdevelopment. What we mean with bottom-up is re-lated to generating algorithms based on an iterativeabstraction and diagramming process. In contrast tothis, the top-down algorithm development processrefers to the demystification of well-structured algo-rithms by changing parameters or components.

This study aims to propose a pedagogical modelfor introducing digital modelling approaches to ar-chitecture students with the focus on developinga better understanding on computational thinking.The scope of this study will examine the proposedpedagogical model which has been applied to post-graduate students from 2014 to 2016.

ALGORITHMS IN DESIGN - DESIGNING AL-GORITHMS“Algorithms, software, hardware, and digital manufac-turing tools are the new standards that determine notonly the general aspect of all objects in a nonstandardseries, but also the aspects of each individual prod-uct, whichmay change randomly or by design” (Carpo,2011, p.99).

In a world in which algorithms replace the visu-als and physicals with an increasing speed, it is un-avoidable for architects to engage with theses onnew faces of representation. Carpo (2011) definedthis transformation as “shift from mechanical to al-gorithmic reproduction” which affect both the wayarchitects design and the role of architects in designprocesses. In this case, thequestionof howarchitectsemploy invisible and immaterial algorithms in designarise, while constitution of physical spaces and quali-ties has beenmost closely concerning architects. Theencounter of architects with algorithms leads a dia-logue crisis between a structured language and theperceived and designed world of the architect.

The compilers execute programming codes lineby line, in a linear order. The expressions, and the re-lations among different expressions enable a recur-sive flow, in other words syntactic and semantic re-lations make a linear text work as hypertext. Devel-oping an algorithm is similar to an attempt to trans-


late any process including nonlinear ones into a fi-nite number string-like lines. As Terzidis (2004) un-derlines, an algorithm as a computational procedurecovers processes such as deduction, induction, ab-straction, generalization, and structured logic. In thissense, developing algorithms involve several multi-plicities of translation, transformation and conver-sion of different information types amongst eachother, and constitution of new relationships amongdifferent domains and dimensions.

In a broader sense, this study focuses on seek-ing pedagogical approaches which might acceleratestudents’ usage of digital modelling environmentsmore creatively with respect to gaining computa-tional thinking skills. This focus brings out anotherembedded contradiction: whether it is possible toencourage algorithmic ways of thinking for architec-ture students while keeping a critical distance to theexisting commercial software and tools.

The pedagogical approaches to facilitate cre-ativity in computational thinking, and the tool-independent teaching strategies have been on theagenda of many educators from different fields (Na-tional Research Council, 2011; Nickerson et al., 2015).One of the related topics which has been mentionedin “Pedagogical Aspects of Computational Thinking”Workshop was that: “Learning and teaching compu-tational thinking should be contextualized” (Wolz inNational Research Council, 2011).

In our study, physical experimentation processis considered as a source of intuitive context andcontextualizationof variables, expressions, loops andfunctions. Exploration new relations based on stu-dents’ empirical observations and constitution ofnew links between the explored relations and ex-isting computational models, in other words defin-ing new problems took priority in our study differ-ent than other problem solving based approacheswhich has revealed problem solving based strategiesmight invoke students learning motivation. More-over, findings similar with the literature (National Re-search Council, 2011), working collaboratively had anadvantage over individual learning environments.

We aim to make a particular emphasis on ex-panding and utilizing the potentials of computa-tional complexity towards a dynamic understand-ing of the things and the world instead of merelyrepresenting cartesian space. In this regard, we at-tempted to constitute a cyclic connection betweenwell-established computational models on dynamicsystems and the empirical research based of simpledynamic and changing environments. Related no-tions of flow such as flow as an abstract concept,flow as a particular feature of elements, flow as anobservable phenomenon of dynamic systems havetriggered theemergenceof various researchenviron-ments in the field of computational design. Usageof dynamic systems as a source for intuitive explo-rations is not new in architecture.

Figure 1Examples fromstudies oncondensation( Hill,2006, p.101); fluiddynamics Kaijima etal. 2013, p.121);magnetic fields(Bello Diaz & Dubor,2013, p.239); fluidcrystallization(Url-1)

In Figure 1, top-left, an image of condensation isshown, which was introduced by Hill (2006), as apart of immaterial aspects of computational design.Dubor and Diaz (2013) presented a material behav-ior driven form-finding approach based on obser-vation of magnetic behavior in different scales andscopes (Dubor&Diaz, 2013). A part of Dubor andDiaz’s (2013) experiment setup is shown in Figure 1,bottom-left. One of their remarkable contributions isthe workflow associated with digital fabrication pro-cess (Dubor&Diaz, 2013). In Figure 1, bottom-right anempirical study focusing on crystallization of fluids is


shown (Url-1). Implementation of a well-establishedcomputational fluid dynamics in architectural designis shown in Figure 1, top-right (Kajima et al. 2013).

In his 2006 book entitled ”Immaterial Architec-ture”, Hill pointed out the potentials of the dia-logue between material and immaterial, concreteand ephemeral, conception and perception. Hill(2006) contributed to theoretical debates on explor-ing new meanings for creativity in design. Hill (2006)quoted from Moholy-Nagy that: ’A definition ofspace which may at least be taken as a point of de-parture is found in physics- ”space is the relation be-tween the position of bodies” (Moholy-Nagy, 1947,p.57; Hill, 2006, p.67). Giving the experience its duein early 20th century, Moholy-Nagy’s expression canbe considered as a manifestation of space-self-timeinterrelationship towards a new and dynamic under-standing of geometry. In respect to the priority of ex-perience, this study is an experimental processwhichinvolves attempts to multiplicate the viewpoints ofthe perceiving subject looking to static geometryvia one-to-many model. In addition, many-to-manymodel which is explained in detail in the next sectionaims to accumulate the singular abstractions of thesubject who observes dynamic process

PEDAGOGICALMODELIn the scope of the study, two models that we callone-to-many andmany-to-manywere tested in post-graduate courses. Durationof the implicationof eachmodel was 4 weeks. The course involved introduc-tion of the sample studies, theoretical discussions,and readings separate from the experimental pro-cess. The implementation was repeated 3 times indifferent semesters. The one-to-many and many-to-many models included explorative, experimentaland structured exercises. The physical experimenta-tion, digitizationof theobserved informationderivedfrom experiments, and the outcomes of the mod-elling exercises were evaluated.

One-to-ManyThis model was considered to depart from an initialstatic geometry. The source of the initial geometrywas changed in different semesters, and the changesincluded things such as a part of a bodyor a part fromthe built environment preferably involving rich topo-logical relations or complex geometrical information.The students were expected to depart from the ini-tial geometry and gain experience in exploring gen-erative and behavioural relations while generating aseries of digital and physical models.The four-weekmodule included but was not limited to the follow-ing phases:

• Measurement and 3D scanning. Extraction ofgeometrical relations from a given context.

• Generation of the same geometry in digitalenvironment through a systematic topologi-cal investigation (Eg: sequentiality, repetition,in/out, near/far, etc.)

• Exploration of performative relations basedon fabrication with different materials.

The modelling exercise shown in Figure 2 consistedof 3 steps. First, the data were collected with a 3Dscanner. Subsequently, the lines of the layers weremanipulated and surfaces were created inside thelayer curves. Finally, joints’ details were optimizedand a 1:1 scale model was prototyped.


Figure 2Digital and physicalhead models byAren Semerci

Figure 3 illustrates a process in which the columnheads can be varied by different modelling tech-niques.

Figure 3Generation ofcolumn heads withfolding andsubtractionoperations by GülceKırdar

In Figure 4, an existing muqarnas geometry is mod-elled with a conventional 3D modelling tool. Repro-ducing the geometry with a parametric tessellation.A non-continuous formwas generated with Voronoi.The new form was filled with a simulated cloth. Thedigital model was prototyped with the combinationof plaster and cloth. Hang points and length of thestrings were calculated in Grasshopper.

Figure 4Muqarnas analysisand computationalinterpretation byJohn Bogren

Many-to-ManyThe main difference between one-to-many andmany-to-many models was the assumption of initialsource. Instead of a static geometry, a dynamic phys-ical, biological or chemical process was selected forexamination. This model included:

• Explorative experimentation: Selection of adynamic process such as movement-basedbehaviour (magnetism), or particle behaviour(liquid behaviour - fluid dynamics, crystalliza-tion, etc.).

• Trial-Error: Repeating the same experimentthrough a series of structured experiments.Therefore, this phase involved a design exper-iment setup. The emphasis was on the explo-ration of qualities and parameters.

• Documentation: (Video record and time-based framing, phenomenon-based framing,etc.)

• Analysis: The students were asked to ex-tract information from the experiments thatthey did. Although extracting information inmost cases depended on visual analysis of thevideo records, this phase might cover variousanalyses such as phenomenon-based, event-based, and threshold-based ones.


• Developing algorithms for digital modelling:This phase involved both bottom-up algo-rithm development and top-down mergingand adaptation of the existing algorithms. Inthe first case, the processwasmore importantthan the end results. The final outcome didnot need tobe apreciseworkingmodel. How-ever, it needed to be a new insight which canbe improved by others in a long term. Thesecond case focused on merging, manipulat-ing, adapting and re-compilingdifferent algo-rithms and samples on purpose.

Figure 5Flocculation of ironballs related to thegeneratedmagnetic field byYiğit Can Ülkücü

Figure 6Analysis andinterpretation ofsugar crystalizationby John Bogren

Table 1Selected studentexperiments ofmany-to-manyapproach

FINDINGS, OUTCOMES AND EVALUATIONThis part evaluates the common aspects and thedifferences of the student works, which were de-rived from implementation of the proposed peda-gogical model between 2014 and 2016 in postgrad-uate courses. The selected student works of themany-to-many model were based on experimenta-tion of viscosity, magnetism, ferro-fluid behaviour,cracking, diffusion and crystallization processes. Thegaps and connections between the information ex-tracted from the physical experimentation and thetop-down examination of the existing algorithms areexplained.

CONCLUDING REMARKSThe foundations of computational thinking arerooted in abstraction, translation and transformationprocesses. These processes conducted with physicalexperimentation might be considered as one of thecore experiences for gaining a better understand-ing on the potentials of computational complex-ity. In other words, physical experimentation, doc-umentation, abstraction and digitizingwould lead toperceptual knowledge, which is different than well-established closed computational models. There-fore, we argue that revisiting the empirical researchby architects will be important in the digital age toexplore new and open-ended relations beyond well-established algorithms. In this sense, this study maybe considered as an attempt to explore new ways of


Figure 7Flow diagrams ofexperimentationand images fromexperiments byTaylan Kılıç.

introducing computational models to architecturestudents with the aim of supporting students in de-velopingpersonal reflections on the existing and alsounexplored computational design models.

One of the findings of this study is that, in thecase of the bottom-up processes where algorithmsintroduced explicitly as a foreign language involvingvocabulary elements, syntactic rules and logical op-erations, it might be difficult to integrate the experi-ence of developing algorithms with different scopesand contexts. On the other hand, gaining experiencein algorithmic thinking depending on particular anddigestible contexts might lead to tacit learning. Weargue that the context-based tacitly acquired knowl-edge has the potential to be transformed into differ-ent problems.

Figure 8Regeneration of thepattern ofcondensed wateron glass surface viaimage sampler byAhmet Onkas


Figure 9Sequential photosof experiment anddiagramming theoutcomes byBegum Moralioglu

REFERENCESCarpo, M 2011, The alphabet and the algorithm, MIT PressNational Research Council, , 2011, Report of a Workshop

on the Pedagogical Aspects of Computational Think-ing, National Academies Press, Washington, D.C.

Bello Diaz, G and Dubor, A 2013 ’Magnetic architecture.A new order in design’, 1st eCAADe Regional Inter-national Workshop Proceedings, University of Porto,Faculty of Architecture (Portugal), pp. 237-245

Einstein, A 1921, ’Geometry and experience’, in Einstein,A (eds) 1921, Ideas and Opinions, Random House,New York, pp. 232-246

Hill, J 2006, Immaterial Architecture, Routledge, NewYorkKaijima, S, Bouffanais, R and Willcox, K 2013 ’Computa-

tional Fluid Dynamics For Architectural Design’, Pro-ceedings of the 18th International Conference of theAssociation of Computer-Aided Architectural DesignResearch in Asia CAADRIA, Hong Kong, pp. 169-178

Kolarevic, B 2000 ’Digital architectures’, Proceedings ofthe ACADIA 2000 Conference, Washington, pp. 251-256

Lonsway, B 2002, ’The Mistaken Dimensionality of CAD’,Journal of Architectural Education, 56, pp. 23-25

Lynn, G 1999, Animate form, Princeton ArchitecturalPress, New York

Moholy-Nagy, L 1928/1947(4th edit.), The New Vision,George Wittenborn, New York

Nickerson, H, Brand, C and Repenning, A 2015 ’Ground-ing Computational Thinking Skill Acquisition

Through Contextualized Instruction’, Proceedingsof the eleventh annual International Conference on In-ternational Computing Education Research, Omaha,NE, USA, pp. 207-216

Oxman, N 2007 ’Digital craft: Fabrication-based designin the age of digital production’, Workshop Proceed-ings for Ubicomp 2007: International Conference onUbiquitous Computing, Innsbruck, Austria , pp. 534-538

Terzidis, K 2004 ’Algorithmic Design: A Paradigm Shift inArchitecture?’, 22nd eCAADeConference Proceedings,Copenhagen (Denmark) , pp. 201-207

[1] http://www.archdaily.com/407750/fluid-crystallization-skylar-tibbits-arthur-olson


Building Performance Computation

Architectural space quality, from virtual to physical

Aida Siala1, Najla Allani2, Gilles Halin3, Mohamed Bouattour41MMRCA ENAU, Tunisia & MAP/CRAI ENSA Nancy, France 2,4MMRCA ENAU,Tunisia 3MAP/CRAI ENSA Nancy, France1,3{siala|halin}@crai.archi.fr [email protected]@hotmail.fr

The adoption of digital design has offered new computational methods improvingand deepening the morphological dimension of architectural production.Algorithmic and parametric design associated with new fabrication technologieshave made generating increasingly complex shapes possible, forming the basis ofnon-standard architecture. Such approaches raise many interesting questions likethe qualitative and functional dimensions of the provided architectural spaces,which remains a very fiercely debated topic in the field of CAAD. This paperlooks for the qualitative and functional spatial requirements that ought to betaken into account in these practices to accompany the quantitative data on whichmodels are currently founded, thus providing the possibility of a more completevision of new space, recognising that space is perceived is a reading of manyfactors, both measurable and un-measurable.

Keywords: BIM, Architectural space, Requirements, Spatial quality, Topology

INTRODUCTIONThe continuous development of digital design toolshas been concentrated in recent years, especiallyon the morphological side of architectural produc-tion, leaving the functional and qualitative aspectsof provided spaces barely explored. Architecturalproduction is based on spatial requirements deter-mined at an early stage of the design process. Theserequirements are given in the form of geometricalconstraints (e.g. length, width, ceiling height, etc.),but also and mainly in the form of non-geometricalones (e.g. ambiance, accessibility, relation betweenspaces, visibility, communication, etc.). Current BIMpractices do not take those non-geometrical spatialrequirements into consideration since they are basedon the use of measured standards and codifications,

which transform all building information into quan-titative data. Thus, only constructive information istaken into account during the design phases (Sialaand al. 2016). The qualitative and functional dimen-sions of architecture, on which the whole design de-pends in most cases, are not considered. The mainquestionwhich therefore arises is howexactly to takeinto account and represent information about spaceswithin BIM tools. This concerns information on thescale of the space itself as an empty, invisible andimmaterial building entity. It also concerns informa-tion on the scale of the properties of space, especiallynon-geometrical ones, like qualitative and topolog-ical (or functional) requirements. Non-geometricalinformation can ensure the transition of the designmodel, from a set of design requirements expressed

Building Performance Computation - eCAADe RIS 2017 | 65

by the future user (owner) of the building (describedspace) to a finally constructed space (physical space),going through the design phases (virtual space).

In our approach, we discuss the concept of ar-chitectural space as a key concept allowing the inte-gration of qualitative and topological spatial require-ments in current collaborativepractices. If thosenon-geometrical parameters could influence the evolu-tion of the design model according to the futureuser’s needs, their presence in BIM tools would becrucial. So which are the relevant non-geometricalrequirements that should be taken into account andwhat could be their form of representation in BIMtools? Regarding these matters and based on an in-vestigative and analysing work, we propose identi-fying and representing relevant non-geometrical re-quirements necessary to describe spaces during thedesign phases. This restructuring approach aims totake into account non-geometrical spatial require-ments, especially qualitative and topological ones, incurrent collaborative usages. The overall goal of thisstudy is to enable designers to measure, predict, as-sess, comprehend and manage the layout of spacesand their quality, respectiveof new requirements, de-fined at an early stage of the design process.

In this paper, we first focus on the status of archi-tectural space information in recent research worksto identify the need for space representation. Then,we look for the relevant non-geometrical require-ments that should be taken into account in currentcollaborative uses. Afterwards, we try to provide a re-structuring approach for the identified requirementsto allow for their integration and management incurrent BIM tools as spatial parameters. By the endof this paper, our space model inclusive of relevantqualitative and topological spatial requirements willbe demonstrated. A discussion dealing with our fu-ture work will explain how to guide current digitaldesign tools towards the qualitative and functionaldimensions of architecture.

INFORMATION ABOUT SPACES IN RECENTRESEARCHWORKSWhen performing design tasks, architects give spacenot only a form (geometrical properties), but also asense (non-geometrical information like ambiance,quality, relation between spaces, etc). Based on a re-quested program, they provide design models thattend to satisfy the future user requirements. User re-quirements are usually given in terms of spatial qual-ity and topology (e.g. the intentions required by ahouse owner might be ”The living room has to en-joy a maximum of sunshine with visibility on the gar-den. Itmust be close to thebathroomandcommunicat-ing directly with the kitchen”). To identify how spaceis described in recent research works, several worksdealing with space data modelling have been inves-tigated. Between the investigated models, we em-phasise the following:

• The IFC model [1]• KIM model (KIM 2015)• Ekholm model (Ekholm 2000)• Bjork model (Bjork 1992)

Based on this analysis work, we noticed that most ofthe investigated space models focus on the quanti-tative properties of spaces (e.g. geometric proper-ties, position, quantity, etc. IFC, Ekholm and Bjorkmodels include partial descriptions of topology ofspaces, since they allow a space to be decomposedinto sub-spaces and admit that a space can belongto a larger spatial structure (zone, floor, etc.). How-ever, none of the analysedmodels dealt with the pos-sible topological requirements on a space. Proposinga method for automatic updates of spatial require-ments, Kim focuses especially on quantitative ones(geometrical requirements, quantity requirements,etc.). Between all the observed works, only the IFCmodel includes an aspect of qualitative spatial re-quirements (e.g. whether the space requires artifi-cial lighting, whether an external view is desirable,the activity type of the space, etc). All the remainingqualitative spatial requirements possible are not sup-ported by all investigated models. Qualitative spa-

66 | eCAADe RIS 2017 - Building Performance Computation

tial requirements consist of descriptions about theintended whole ambiance of a space (e.g. acoustic,thermal, hygrometric, lighting, safety and coating re-quirements).

IDENTIFICATION OF SPATIAL REQUIRE-MENTSOur purpose in this section is not only to illustratethe different requirement types on a space, but alsoto discern relevant ones to integrate into current BIMusages. Our method is based on the content analysisof written descriptions of design requirements. Thisanalysis work was organised around three chrono-logical phases, as structuredbyWanlin (Wanlin 2007),the pre-analysis work, the tools exploitation, as wellas the results treatment, inference and interpreta-tions.

The pre-analysis workThepreliminarywork concerns the first phase of intu-ition and organisation in order to operationalise andsystematise the initial ideas to arrive at an analysisschema. To represent the possible concepts describ-ing requirements on spaces, we focused on architec-tural programming studies developed by specialisedagencies in architectural programming. From theavailable and accessible ones, those having themostdetailed descriptions of spatial requirements wereprivileged. To collect the maximum amount of data,we selected several representative types of real pub-lic construction works. Among investigated publicworks, we emphasise the following:

• The schooler and extracurricular group ofVany, Fance (2013)

• The elementary school of Saint Antonin,France (2016)

• Elsa Triolet College of Thaon Les Vosges,France (2017)

• The emergency Center of Dieulouard, France(2013)

• The nursery andparental care center of Laxou,France (2015)

• The multidisciplinary health center of

Montchanin, France (2015)

In the first empirical step, we started by acquaint-ing ourselves with the collected documents. This al-lowed us to limit the investigation scope letting im-pressions and certain orientations come about dur-ing the reading and re-reading of each program-ming study. Spatial requirements are described inexplanatory texts, tables and diagrams. We pro-ceeded by gathering the close or similar words (lex-ical words) used to qualify spatial requirements andjoining them with their relating descriptions. Theidentified words and descriptions have constitutedthe basis of the second analysis step.

This study focuses only on objective spatial re-quirements, which have factual referents. Require-ments like ‘pleasant space,’ ‘cheerful space,’ and‘warm space’ are completely subjective requirementsand cannot be considered in this study.

Table 1Example ofAccessibilityRequirementsCategorisation

The hardware operationThis second phase consistsmainly in carrying out thecoding, counting or enumeration operations accord-ing to the specifications previously formulated. It in-volves in turn two steps:

The first step concern the categorisation opera-tions, in which we elaborated a grid of categories, isthe first step. We started associating each identifiedword used to qualify a space with the related adjec-tives. For example, the word lighting was associatedwith natural or artificial, direct or indirect, zenithal,controlled, etc. Then, words belonging to the samefamily (having a common cardinal) were groupedunder a generic heading. For example, the wordsAccess, Accessible, Accessibility, etc. were grouped


Table 2Topologicalrequirementscondensed grid ofcategories

under the generic heading Access (table 1). Next,headingswere gathered in turn, by synonyms and/orantonyms under the same title, in order to provide acondensed and simplified representation of the rawdata. For example, the headings Access, Entrance,exit, etc. were classified under the title Accessibil-ity. Finally, titles were classified by type. For exam-ple, titles Accessibility and Relation and arrangementwere classified as Topological requirements, whereAccessibility gathers requirements concerning the ac-cess to space and the exits, Relation groups require-ments about connection between spaces (relation-ship, pathways, traffic, distribution, communication,separation, etc.) and arrangement gathers the adja-cency requirements (proximity, layout, etc.).

This step has allowed us to identify and appre-hend the different types of spatial requirements. Il-

lustrated spatial requirements are classified into fivetypes, namely: geometrical requirements (shapeand dimensions), quantity requirements, occu-pancy requirements (Equipment, number of oc-cupants, occupancy duration, activity type, etc.),topological requirements andqualitative require-ments. Among these requirement types, we areprimarily concerned with non-geometrical require-ments, in particular topological and qualitative ones.

The second step involves filling the grid accord-ing to the identified words and their descriptionswith the related usage frequency. Software support-ingqualitative andcombined researchmethodswereused to systematise the coding and counting works(NVivo [2] ). The topological anand qualitative re-quirements condensed grids of categories are shownin Tables 2 and 3.


Interpretation, inference and validationIn this last phase, the raw data is treated using sim-ple statistical operations (based on percentages) tobe meaningful. The results are summarised in di-agrams (Figure 1) that condense and highlight theinformation provided by the analysis work. A finalstep in this analysis work focused on the results ver-ification. Our approach was verified by statisticaltests on a new real public work programming study.This verification was carried out on a project thatfalls within the educational framework since half ofthe analysed programming studies belongs to edu-cational institution projects. The selected program-ming study concerns the kindergarten of Jean Jau-res, Libourne -France (2012). We noticed in this final

step that results were very close and in some caseseven statistically identical. It should, however, benoted that slight differenceswhich do not exceed 6%were found in certain descriptions . The verificationwork has enabled this study to identify new familiesof words synonymous with those presented in thisanalysis work.

DISCUSSIONTopological requirements are described in texts, di-agrams and functional flow charts, which explainthe required arrangement between spaces, the rela-tionship between them and accessibility constraints.Topological descriptions are almost fixed and recur-rent in all analysed programming studies. Figure

Table 3An excerpt fromqualitativerequirementscondensed grid ofcategories


Figure 1Topological (1-a)and qualitative(1-b) requirements,descriptions andrelated usefrequency.

1.a shows that the must used topological require-ments are: the arrangement of spaces (close to, nextto, near, etc.), the type of communication relationbetween spaces (visual or physical), the accessibil-ity type (main or secondary) and constraints (width,height, etc.), the type of connection relationbetweenspaces (direct or indirect), etc.

Figure 1.b shows that much of the qualitativedescriptions are indicated to raise awareness aboutthe importanceof certainqualitative aspects of space(e.g. air-conditioning, heating, ventilation, noise,etc). For example, 42% of acoustic requirement de-scriptions revert to existing standards and regula-tions (complieswith, conforms to, standardised, etc.).25% of these descriptions revert to the space typeof the activity type in which they will be housed (anadapted treatment, the attenuation will be adaptedaccordingly, etc.) and 14% describe the desiredacoustic quality (reinforced, optimal, of quality, etc).Only 20% of the acoustic description are quantita-tive (attenuation: 60 dB, 43 dB, etc). Thus, 80% ofthe acoustic requirements refer to standards and en-

gineering.Qualitative requirements gather descriptions

that define the overall ambiance of a space. This an-alytic work has enabled this study to identify quali-tative descriptions very commonly used in the pro-gramming phase. These descriptions concern: thecoating type of a space or space-type (marble, tiles,parquet, etc.), the acoustic attenuation (which mustcomplies with standards), the thermal needs (mustbe air conditioned, will be heated, etc.), the re-quired ventilation (natural or mechanical), amongother things.

It should be noted that qualitative and topolog-ical requirements include essentially qualitative de-scriptions, which are generally given in binary andopposite representations (e.g. interior exterior, mainsecondary, direct indirect, artificial natural, etc).

Topological requirements include only qualita-tive descriptions. Qualitative requirements are moreimportant then topological ones as 71% of the de-scriptions relate to spatial quality. They include notonly qualitative description, but also some quanti-


tative ones (e.g. the required temperature (°C) in aspace, the ventilation flow (m3/h), the sound attenu-ation (dB), the solar radiation (wh/m²), etc) alongwithother typological descriptions, like types of coatings,spot lights, stores, among others.

All these descriptions are useful during the de-sign phases to assist the design evolution processin taking into account both measurable and non-measurable requirements. To further illustrate ourresults, the diagrams presented above were sum-marised in a single diagram (Figure. 2) based onlyon significant percentages. To descry the relevantdescriptions of spatial requirements, we have de-fined a percentage threshold (10%) below which de-scriptions were not taken into consideration. Basedon this final diagram, the next section consists ofthe representation of the identified relevant require-ments in a space data model.

Figure 2Topological andqualitativerequirementssummariseddiagram

A SPACE MODEL INCLUDING NON-GEOMETRICAL QUALITATIVE AND TOPO-LOGICAL REQUIREMENTSThis section presents our view of how relevant spa-tial requirements can be accommodated in the same

schema and how can they be related to spaces.In addition to <Geometrical> spatial require-

ments (area, width, ceiling height, etc.), <quantity>requirements and <Occupancy> requirements, ourspace model (Figure 3) includes also <Qualitative>and <Topological> requirements as classified in theprevious section. Proposed space model consists oftwo instances: the Requirements Model and the De-sign Model.

In the Requirements Model, a <Space> shouldsatisfy a set of requirements according to its <SpaceType>, <Space SubType>,<Activity type> or <Activ-ity SubType>. For example, in a university, the space“Pedagogical depot-003” should satisfy topologicalqualitative descriptions imposed by its <space type>as a “Depot” (All depots will have direct access fromthe common traffic). It should also satisfy those re-quired by its <Space SubType> as a “Pedagogical de-pot” (Pedagogical depots for the storage of teachingmaterials must be contiguous to classrooms and ac-cessible directly from them). Thus, the space “Ped-agogical depot-001” must satisfy both of those re-quirements. Its final topological requirements wouldbe as follows: contagious to a classroom, access fromthe classroom, and access from the common traffic.Thus, requirements with qualitative descriptions arecumulative.

On the other hand, a <Space> should also sat-isfy a set of requirements according to the <Activ-ity Type> or <Activity SubType>. For the same ex-ample of a college, the <SpaceType> “Classroom”should satisfy requirements imposed by its <Activitytype> “Having Class” (a natural lighting is required)and those required by its <Activity SubType> “watch-ing projection” that necessitates stores to minimisethe natural lighting. The resulting requirements herewill be: natural lighting, equipped with stores.

In the Design Model, architectural <Space> hasthree types of data classified as follows:<General>data, which indicates general properties of space(e.g. ID, GUID, name, etc.).<Geometrical> data, whichgroups geometrical properties of space (e.g. length,width, ceiling height, position, etc.) and <Quantity>


Figure 3Schema of spacedata model,including identifiedqualitative andtopologicalrequirements

data, which gives the location of space in the entiremodel.

CONCLUSIONThe overall goal of this research is to assist de-signers in providing architecture that satisfy non-geometrical requirements described during the pro-gramming phase, especially topological and qualita-tive ones. Those requirements are implicitly knownby designers and architects in current collaborativeusages, since the absence of qualitative informationwithin BIM models is currently evident.

We have attempted in this paper to provide astructural approach of the relevant topological andqualitative requirements necessary to be taken intoconsideration in collaborative usages. This approachhad allowed us to propose a data modelling all infor-mation about spaces, their requirements and proper-ties. A futureworkwill focus on new collaborative us-ages allowingdesigners to addandmanage topolog-ical and qualitative spatial requirements during thedesign phases.

REFERENCESAnders, Ekholm and Sverker, Fridqvist 2000, ’A concept

of space for building classification, product mod-elling and design’, Automation In Construction, 9, pp.315-328

Bo-Crister, Bjork 1992, ’A conceptual model of spaces,space boundaries and enclosing structures’, Au-tomation In Construction, 1(3), pp. 193-214

Kim, Tae Wan, Kim, Youngchul, Cha, Seung Hyun andFischer, Martin 2015, ’Automated updating of spacedesign requirements connecting user activities andspace types’, Automation In Construction, 50, pp.102-110

Siala, Aida, Allani, Najla, Halin, Gilles and Bouattour, Mo-hamed 2016 ’Toward space oriented BIM practices’,Proceedings of eCAADe 2016, Oulu, Finland, pp. 653-662

Wanlin, Philippe 2007, ’L’analyse de contenu commeméthode d’analyse qualitative d’entretiens : unecomparaison entre les traitements manuels etl’utilisation de logiciels’, in Recherches qualitatives,ARQ (eds) 2007, Bilan et prospectives de la recherchequalitative, Hors série

[1] http://www.buildingsmart-tech.org[2] http://www.qsrinternational.com


Architects’ use of tools for low energy building design

Methodological reflections from Ethnography and Philosophy ofTechnology

Gabriela Zapata-Lancaster11Welsh School of Architecture, Cardiff [email protected]

Design practitioners face an increased pressure to design low energy buildingsbecause of the need to reduce the carbon emissions of the built environment. As aresponse, building performance simulation tools (BPS) have been created fordesigners to facilitate the decision-making and help them to propose low energybuildings. This paper is based on a research that adopted ethnographic researchto conduct a case study comparison and explore how BPS tools were deployed bydesigners during real-time design process. The research adopted a constructivistapproach informed by philosophy of technology and human computer interactiontheories to reveal what designers were doing during the design process asopposed to what they should be doing according to best practice advice. Thispaper focuses on the application of ethnographic methods and brings attention tothe advantages, challenges and limitations of adopting ethnographic research toinvestigate the `context of use of tools'. The discussion of the method bringsattention to the context of use of tools as the departure point to develop a range ofsolutions for design support.

Keywords: building performance simulation, design tool, design process, energyperformance, ethnography, social context

INTRODUCTIONThere has been a rapid growth in the number of BPStools that fit all the stages of the design process (fromearly to advanced design), to be deployed by differ-ent practitioners: architects, building services engi-neers andenergy specialists. While thebenefits of us-ing BPS tools in the design process are widely recog-nised, building designers face several challenges toincorporate BPS tools in the design process (Zapata-Lancaster and Tweed 2016; Alsaadani and Bleil de

Souza 2012). The use of BPS tools in routine de-sign process is likely to be limited and their poten-tial impact as design-aid remains latent (Clarke et al.,2008), suggesting a gap in the practical use of BPStools by designers (Macdonald et al., 2005). Researchon performance gaps (discrepancy between the as-designed intentions and the actual building perfor-mance during operation) suggest the failure of as-designed performance calculations to represent theperformance of the building in use anddiscrepancies


due to different factors (Van Drokenlaar et al 2016).The study of design support tools for practitionershas been carried using questionnaires, interviewsand focus groups. Despite the important contribu-tions of those approaches, they are limited in provid-ing an in-depth analysis of the context where design-ers operate. The context of use is likely to affect theway BPS tools are invoked in the design process, ashighlightedby theZeroCarbonHub (2010): “...perfor-mance issues are more much more concerned withthe processes and cultures within the industry thanwith the model that is used to predict carbon emis-sions”. Therefore, the purpose of this ethnographicstudy was to reveal in detail the context where de-signers operate and how they inform the low energybuilding design; including design process aspects,knowledge aspects and tool deployment aspects.The study aimed to investigate in detail how BPStools were used in the real-time routine design pro-cess (dodesign teamsuseBPS tools as recommendedby best practice advice, as expected by the BPS tooldevelopers ?) and more importantly, why were BPStools used that way (understanding that can informthe development and improvement of tools, identifi-cation of design support needs that could translateinto tool improvement and/or the identification offeatures and capabilities as perceived by the users).This paper discusses the use of ethnographic meth-ods to investigate the ‘context of use of BPS tools’.The reflection about the application of ethnographicmethods as a way to engage with the users aims toencourage the debate about how the researcher andtool development community engagewith the antic-ipated users to understand their needs and solutionsfor design support.

LITERATURE REVIEWPhilosophy of technology and human-computerinteraction perspectives question tool-centric ap-proaches that reduce human action and problem-solving to deterministic models where the tools fitinto processes regardless of the features of the con-text where they are incorporated. Winograd and

Flores (1987) argue that technologies become partof the pre-existing network of human interactions.Therefore, the context should be considered to un-derstand how technologies fit and change the net-work where they are used (Winograd and Flores,1987). Conventional ways of thinking that assumethat technologies fit as forecasted by the tool devel-opers are likely to be misleading because they ne-glect the interaction between tool and user in thecontext of use (Borgman 2004). In this sense, Gib-son’s concept of affordances is relevant to highlightthe potential of an object to enable action (Gibson1979). Affordances relate to the inherent character-istics of objects, to the ways that users perceive thepotential of objects to enable or deter action and theenactment of different sets of actions. Affordancesare latent and their realisation requires the ‘associa-tion of objects and situatedness’ (Hodder, 2012). Inother words, whilst objects have the potential to fa-cilitate action; the user and the context of use play arole in the ways objects are deployed.

The affordance of objects and the situatednessare part of the relativity of the relations human-technology and culture-technology (Ihde, 2004).While technologies provide a ‘framework for action’;they are defined by existing patterns of use, inten-tions and preferences. Ihde (2004) uses the term‘double ambiguity’ to refer to the dichotomy be-tween ‘trajectories of development’ and the ‘instru-mental intentionalities’ of objects. The trajectoriesof development are the ways how users use the toolwhile the instrumental intentionalities are the usesanticipated by tool developers. Ihde argues that theintended uses of tools outlined by their developershave little effect on the subsequent history of the toolbecause theuser candevelop a variety of uses includ-ing ones thatmight not have been anticipated by thedevelopers (Ihde, 2004). Idhe’s argument does notnegate the technical properties of objects; it simplydraws attention to the idea that tool’s properties arepart of relativity of the human-technology relation-ship.

In architectural and engineering research, a


number of precedents have addressed the use oftools from a perspective that acknowledges thehuman-tool juncture. For instance, Bucciarelli (2002)suggest that artefacts facilitate the communication,negotiation, learningand living the languageof engi-neering. Bucciarelli (2002) argues that tools are partof the construction of common understandings andpraxis. Similarly, Berente et al. (2010) point out thatnew tools are introduced to old systems and rou-tines, leading to new configurations of practice whileCoyne et al (2002) use the ’evolutionarymetaphor’ tohighlight thedynamicnatureofdevices and the com-plex ecology of devices to refer to notions such asderivation, improvement, survival, suitability to pur-pose, adaptation, inheritance of features, and recom-bination. Those arguments suggest that tools arepart of complex processes and the relationships theyestablish as individuals within groups. Tweed (2001)looks at the early introduction of CAD in design prac-tice and suggests that the stereotypical and totalis-ing view that represents the end-user as a single typeof designer fails to consider the beliefs, norms, val-ues andhistory of end-users that fall outside the idealtype. Simplified assumptions about the way prac-titioners engage in problem-solving may be limitedbecause they are prone to focus on a typical or ra-tional ’ways of doing’ that could overlook (tacit) fea-tures of practice. This aspect has been addressed byChastain et al (2002) and Harty (2008). Chastain etal (2002 p.239) suggest that the properties of toolsare inferred from the tool developers’ assumptions ofpraxis so when a new technology is adopted, a dys-functional relationship might emerge between toolsand tasks. Tool developers embed assumptions andconstraints about the intended users which enableand deter the ways how tools are used (Harty, 2008).

These literatures contest the ideaof ‘ready-made’tools and suggest that tools are used in relation totheir cultural and social dimensions beyond their in-herent capabilities. In spite of the artefactual proper-ties of tools, the users are prone to manipulate theirproperties in use, deploying the tools for purposesother than those anticipated by the tool developer.

Tools for low carbon design are in this sense no dif-ferent to any other tools and could be similarly af-fected by the social context. The term ‘social context’refers to the physical and social location where peo-ple interact and develop as part of the group. It com-prises the beliefs, paradigms, motivations, attitudes,habits, repeated patterns of action that unfold dur-ing the interactions between individuals (Berger andLuckmann 1967). This research acknowledges thatthe concerns emerging in the design process couldinfluence the potential and perceived affordances oftools, their roles and their patterns of use as com-pared towhat is expectedbybest practice advice andby models that recommend the ‘correct’ use of BPStools by designers.

RESEARCHMETHODOLOGYIn order to investigate what people do in real timedesign process, the study adopted a theoretic frame-work based on propositions fromphilosophy of tech-nology and human-computer interaction theories(section above) and develop empirical work using anethnographic approach. Ethnography is a qualita-tive researchmethod that enables the studyofmean-ings and experience engendered in the social milieuof a group (Hammersley and Atkinson 1995; Silver-man 2005; Bryman 2008). The locus of the ethno-graphic analysis is culture. It allows the explorationof ‘the social issues and the behaviours that are notclearly understood’ (Angrosino 2007 p.26) by consid-ering the influence of the social context in the cre-ation of meanings, attitudes and beliefs held by agroup (Lecompte and Schensul 1999 p.58).

In design research, ethnography has been usedas methodological approach to study practitionersin design and construction; for example, Bucciarelli1998; Lloyd and Deasley 1998; Baird et al 2000; Balland Ormerod 2000; Button 2000; Gorse and Emmitt2007, to cite few. In the ethnographic study aboutmigrant constructionworkers, Pink et al (2010 p. 658)argue that ‘ethnographic research can make visibleinformal (or unofficial) worlds of action, interactionsand ways of knowing that can easily slip under the


industry (or official) horizons of notice’.The study analysed in-depth the design pro-

cesses of six non-domestic buildings located in Eng-land andWales by four sustainable architecture prac-tices. The study investigated how designers usedtools to embed energy performance during routinedesign process. Building upon the principles of qual-itative research (Denzin and Lincoln 1998; 2008; Sil-verman 2006; Creswell 1998), purposeful samplingwas done based on communities of practice criteria. The investigation analysed the conceptual and de-tailed design phases. The delivery and constructionphaseswere outside the scope. However, the aspectsrelated to delivery, construction and operation wereconsidered in that they overlap to thedesignprocess.The main research participants were the architects.Other design teammemberswere included to depictthe dynamics and richness of the process.

The ethnography study was conducted fortwelve to twenty-onemonths per case study. It com-prised an average of seventy five hours per architec-ture firm distributed in eighteen to twenty five visitsper firm. The data collectionmethods included inter-views, non-participant observation (ie in meetings,design team exercises), document analysis, shadow-ing of work, visits of construction sites. The perfor-mance targets and BREEAM credentials of the casestudies are described in table 1.

ETHNOGRAPHIC FINDING RELATED TOBPS TOOL USE DURING THE DESIGN PRO-CESSCOMPLIANCE V PERFORMANCE-BASEDPROCESSThe use of ethnographic research enabled the iden-tification of challenges faced by designers to use BPStools as recommended by best practice advice (fur-ther reading: BSRIA Building Services Research andInformation Association 2009. BSRIA BG4. The Soft-landings framework for better briefing, design, han-dover and building performance in-use. Retrievedfrom http:www.bsria.co.uk/services/design/soft-landings/s) . The data suggests that BPS tools were

not consistently deployed as recommended by bestpractice advice. BPS tools were not consistently de-ployed as expected by BPS tool developers (ie inearly design stage, to compare design options, etc)(For further details, see Zapata-Lancaster and Tweed2016). All of the research participants were work-ing in sustainable architecture firms and designinglow energy buildings; yet, the project drivers suchas cost, construction time and buildability featuredas priorities for decision-making, compromising theenergy performance intentions of design teams insome of the case studies. The project drivers of thewider stakeholders (ie. client and construction team)affected the prevalence of the energy target and theuse of BPS tools during the design process. As illus-trated, by the following quotes, energy performanceis unlikely to be an explicit requirement even in lowenergy buildings:

Building Services Engineer 2: ‘the client couldnot confirmwhether cost, quality or timewas ofmostimportance’

The client referring to low energy design targets:Client 1: ‘if it is not a compulsory requirement, we donot want it’

The ethnographic work enabled the continuousobservationsof thedifferentways that BPS toolswereinvoke in the design process in the context of rou-tine design process. In the case studies (projects de-signed by architects with experience in sustainabil-ity), there were two types of engagement with BPStools for energy purposes: use of BPS tools promptedby compliance (namely, compliance-only process)and performance-based use of BPS tools (pervasiveand consistent use for design decision-making). Thelevel of engagement was observed in the long-termnon-participant observation and discussed with theparticipants during follow-up interviews. The real-time use of BPS tools in relation to the deliverablesand the project drivers was a key ethnographic find-ing that illustrates a facet of the processes and cul-tures within the building industry.


Table 1Energyperformancetargets andBREEAM credentialsin the case studies

BPS IN COMPLIANCE-ONLY PROCESS-CHALLENGESThe aim of the compliance-only process was mainlyto produce as-designed estimation models todemonstrate that the energy regulatory require-ments were met. In this situation, BPS tools wereinvoked only in the light of policy requirements(planning application, building control application)to produce the evidence to estimate the as-designedperformance. In that situation, BPS tools were notused to inform the decision making. BPS tools wereused fine-tune an already designed building. Inthe compliance-only circumstance, the use of userfriendly design tools was irrelevant because rapiddecision making during early design was done onthe basis of experience and heuristics (ie rules ofthumb, adopting design strategies that were used inprevious designs). The main challenges in the use ofBPS in the context of compliance only, as perceivedby research participants:

1) BPS tools did not afford quick estimation inearly design.

2) BPS toolswere too time consuming in detaileddesign to be deployed in parallel to decision mak-

ing (ie. effort for modelling input and computationaltime for calculation).

3) BPS tools were perceive as an ‘investigationexercise of energy performance’ rather than an aidfor decision making when the priorities are reducingcapital cost and time on site.

BPS, A COMMUNICATION TOOL IN THEPERFORMANCE DIALOGUEIn the performance-based process, BPS tools wereused to inform the performance dialogue as recom-mended by best practice advice. BPS tools and theirresults were part of the design negotiations and thedecision-making throughout thedesignprocess. Theas-designed estimation informed the design strate-gies from early design to detailed design.

Interestingly, there was a communication roleplayed by BPS tools to facilitate the dialogue aboutenergy performance within the design team and be-tween the design team members and other stake-holders and decision-makers in the process:

1) Communication within the design team:the energy expert/simulationist conducted the as-designed estimation and took the results to codesign


sessions with other design team members ie. archi-tects, acoustic engineers, architectural technologistin order to decide how different strategies affecteddifferent requirements (ie specificationofwindows inrelation to glare, sound protection, heat gains)

2) Communication (education) of the widerstakeholders: the design teamsused the as-designedestimations to illustrate the potential benefits of lowenergy strategies in relation to metrics/aspects thatclients were concerned about for example, opera-tional cost. The following excerpt illustrates the BPSuse to communicate with the client:

Building Services Engineer 3a: ‘They [the clients]wanted to have the PVs on and all those things. They[the clients] wanted to knowhowmuch it would savethempotentially off their bills because they get just aset budget every year to run andmaintain the school,so obviously themorewe can reduce the energy con-sumption, the more they can spend on text booksand things like that for the kids.’

This communication aspect with the clients, sug-gests that the energy metrics obtained by BPS toolsare ‘translated’ to concerns that are more relevant tothe stakeholder. For example, the designers in one ofthe case studies had developed a tool, the energy billsaving estimator. It was a spreadsheet that calculatesthe weekly costs associated to the energy consump-tion for space heating, water heating and lighting.The weekly costs were aggregated and compared tothe basic income of the user to determine the per-centage of basic income likely to be used to pay theenergybills. The clientsweremore compelled to sup-port lowenergy strategies if they could link their ben-efits to potential savings during operation.

3) Communication between the simulation-ist/energy expert/design teams and the contractor:to link the as-designed performance intentions to ac-tions required from the other teammembers to facil-itate the achievement of as-designed intentions. TheBPS results were used to produce evidence about theneed of certain performance specifications in build-ing materials and standards on site:

Architect 4b: ‘There’s a reason why the building

is like that, you know. And I think [achieving a lowcarbon building] it’s about educating people. I feelthat contractors don’t understand what [architects]we’ve been doing, to the depths that we’ve taken it.So you have someone, like well, we can do a framein steel because it’s cheaper. Hang on a sec, no, no,the frame isn’t just there to support the building, theframeactually has other parts toplay in thewhole en-vironmental model of what the building is. Yeah, andI think that’s shown when we get the questions go-ing, can we use this as an alternative? And you go,well it doesn’t work for this reason, that reason andthat reason. I can see you want to use it because itlooks the same, but that’s where the similarity ends.So it’s about educating them [the contractors] to re-ally understand ... beingmore attentive to things likethat.’

USERS OF BPS TOOLSIn alignment with Lawson’s notion of the de-signer’s problem solving preferences (Lawson 1997)-preference for quick understanding over numbers-,this work found that no architect in any of the casestudies used BPS tools or any quantifiable methodto estimate performance. User friendly ‘early design’tools were not used to inform decision making:

Architect: ‘It’s about basic principles; do youneed software to tell you that?’

Even the Building Services Engineers acknowl-edged using experience and heuristics for earlydecision-making:

Building Services Engineer: ‘We sort of starttalking about the low carbon strategies quite earlywith them without any calculation so it is quiteexperience-based in a way. We look at the orienta-tion, the form, themassing, the things that you coulddowithout the calculations, the things that you knowthat will work. It is done in that way, it is more quali-tative than quantitative...’

Another interesting aspect is that the estimationof energyperformancewasperceived tobe aduty forthe energy specialist. Generally, there was an energyspecialist or simulationist in the design teamwhode-


ployed BPS tools as requested (either only for compli-ance or throughout the design process for decision-making). Even within the building services engineer-ing field, it was suggested that the increasingly ambi-tious energy goals required by legislation were lead-ing to the creation of a ‘sub-expertise’ in BPS tools:

Building Services Engineer 2: ‘Even in our dis-cipline [building services engineering], there is sub-expertise with people who can produce models andworry about Part L and the other engineerswhohavenot been trained in producing models who will relyon the modelling group...’

These aspects bring attention as to whether themove towards the widespread use of BPS tools bypractitioners is bringing to question the knowledgeand expertise of different practitionerswithin the de-sign team . (further reference, see Dreyfus (2004)

AS-DESIGNED PERFORMANCE ESTIMA-TIONBYBPS TOOLSAFTERDESIGNPHASEThereweremixed views about the extent towhat theresults of BPS tools represented the performance offuture buildings. There was a generalised concernamong research participants that the as-designedestimations did not reflect the in-use performanceand that there were connections between the as-designed estimation and the building performancein further stages of the building lifecycle process(construction -as-built performance; operation -in-use performance). This resulted in conflicting viewswhere the designers perceived that BPS tools weretheoretical or compliance-only instruments. The as-designed performance estimation was regarded asan uncertain representation of performance unlikelyto be met during construction and operation. Inother words, the BPS tools were seen as limited forthe appraisal of performance due to the uncertaintyembedded in the as-designed evaluation in relationto as-built and in-use performance.

Architect 2: ‘How reliable is the use of advancedsimulation tools?’

Architect 4b: ‘You’ll never reach the [energy] tar-get; it’s some kind of false target really. It’s uncon-

trollable, because you can’t control people. You cancontrol lights to a degree; you can obviously controlheating and ventilation. But on the modeling we’vemanaged to reach the target and that’s completelyright if no one is going to switch a plug.’

Building Services Engineer 2: ‘Sometimes Istrongly disagree with simulation and its pretty pic-tures... is this helping to understand or just generat-ing pretty pictures, images of performance that can’tbe mapped against actual energy use?’

REFLECTION OF THEMETHODThe discussion few of the ethnographic findings ofthe main study presented in the previous section re-veals the challenges in adopting BPS tools in the re-lation to the context of use ie. project drivers (reduc-ing cost more important than energy performanceduring operation). In situationswhere energy perfor-mance was not an explicit project requirement, thedesigners could use BPS tools for compliance-onlypurpose (no/minimum use of BPS tools to informdecision-making) or to support the performance dia-logue between stakeholders (using BPS tools to com-municate within the design team and beyond). Theethnographic research enabled to identify some ofthe solutions devised by designers to ’align visions’of different stakeholders in performance-based pro-cess. In cases where BPS tools were used to mediatethe communicationwith the ’non-energy expert’, BPStoolswere used to provide evidence and to ’translate’energy metrics to concerns that were relevant to dif-ferent stakeholders who participated in the decision-making process ie. cost savings in operation canbe invested in other expenditures (client-user ’s con-cern); buildability and standards site needed tomeetas-built energy performance (contractor’s concern).Yet, there is a generalised concern about the credi-bility of as-designed estimations (and BPS results) inrelation to discrepancies with as-built and in-use per-formance.

The key advantages of the ethnographic ap-proach is that it enabled the comparisonof case stud-ies of design in action to identify the commonali-


ties and differences in the design process enactedby designers, including how BPS tools were used ineveryday design process (bottom up understandingof the circumstances that facilitate and that inter-fere with the use of BPS tools by designers). Theuse of theoretical informed propositions fromphilos-ophy of technology and human-computer interac-tion allowed the focused investigation of design pro-cesses without imposing any research agenda. Yet,the use of these theories enabled a focused construc-tivist investigation to identify the participant’s viewsand experiences in real-time design process. Theethnographic engagement helped the researcher toavoid out-of-context interviews, observations, retro-spective accounts, self-accounts and snapshots ofthe process so the use of BPS tools was linked to theconcerns and drivers of the different stakeholders inthe process (within and beyond the design team)-cultures and processes-.

Thedisadvantageof ethnographicwork is the re-sources need to conduct the investigation: lengthof the immersion, access to the research setting,time/budget constraints, commitment of the re-search participants, asymmetry of data collectedacross the case studies (specific to comparativeethnographic studies). Qualitative researchmethodswhere the researcher creates scenarios and promptsproblem-solving situations which prompt compar-isons and consideration of real circumstances are al-ternatives to ethnographic work; for example boardgames, scenario setting during focus groups. Thesemethods can help to identify situations, challengesand solutions that build upon the participants’ expe-riences, requiring less resources in terms of time andaccess to participants than ethnography.

CONCLUSIONSThe ethnographic method enables an in-depth un-derstanding of how processes unfold (behaviours,actions) as experienced by participants; and, moreimportantly, why that happens. In relation to thiswork, the detailed field data is relevant to illustratethe complex practices that surround the adoption of

BPS tools in building design. The field data suggestthat the project-specific circumstances, the driversof the process, the regulatory panorama determinehowthedesignprocessdevelopsover time; and, howthe design teams engage with BPS tools during thedesign process.

However, the reader should be cautious in trans-posing these findings to other situations. Transfer-ability of the field finding is one of the potential lim-itations of ethnographic work. The specific circum-stances and processes are not prescriptive or predic-tive of other case studies. The findings are specificto the context of analysis, to the specific case stud-ies. Yet, in reporting indetail the circumstances of thecase studies, participants, research design and anal-ysis, the reader can understand the research circum-stances and what it means to the wider context. Ul-timately, the detailed descriptions of how BPS toolsare used and why it happens in such way have aglobal relevance ingenerating insights aligned toHCIpropositions: BPS tools are not ‘ready-made tools’,the designers use tools in unexpected ways in rela-tion to the context of use. Ultimately, this ethno-graphic work has identified some of the challengesfaced by design teams. It illustrates existing prob-lems and solutions which can be the source of inspi-ration for the provision of design support, whetherthat is realised by low tech or high solutions.

ACKNOWLEDGEMENTSThis work was supported by a Building Research Es-tablishment (BRE) PhD. The author thanks the re-search participants for their support to this work.

REFERENCESAlsaadani, S and Souza, CD 2012 ’The Social Component

of Building Performance Simulation: Understand-ing Architects’, Proceedings of the Building SimulationandOptimizationConference. Loughborough, UK, pp.332-339

Angrosino, M 2007, Doing ethnographic and observa-tional research, Sage, London

Baird, F, Moore, CJ and Jagodzinski, AP 2000, ’An ethno-graphic study of engineering design teams at Rolls-


Royce Aerospace’, Design Studies, 21, pp. 333-355Ball, LJ and Ormerod, TC 2000, ’Applying ethnography in

the analysis and support of expertise in engineeringdesign’, Design Studies, 21, pp. 403-421

Berente, N, Baxter, R and Lyytinen, K 2010, ’Dynamicsof inter-organizational knowledge creation and in-formation technology use across object worlds: thecaseof an innovative constructionproject’,Construc-tionManagement and Economics, 28, pp. 569-588

Berger, P and Luckmann, T 1967, The Social Construc-tionofReality-A treatise in the sociologyofknowledge,Penguin, Middlesex, England, Harmondsworth

Borgman, A 2004, ’Focal things and practices’, in Kaplan,D (eds) 2004, Readings in the philosophy of technol-ogy, Rowman and Littlefield Publishers, Maryland,pp. 114-136

Bryman, A 2008, Social researchmethods, Oxford Univer-sity Press, Oxford ; New York

Buciarelli, LL 1988, ’An ethnographic perspective on en-gineering design’, Design Studies, 9, pp. 159-168

Buciarelli, LL 2002, ’Between thought and object in engi-neering design’, Design Studies, 23, pp. 219-231

Buciarelli, LL 2003, ’Designing and learning: a disjunc-tion in contexts’, Design Studies, 24, pp. 295-311

Buciarelli, L 2001, ’Design Knowing & Learning: A So-cially Mediated Activity’, in C, M, surname missing,initials missing and Newstetter, W (eds) 2001, East-man, Elsevier, Design knowing and learning cogni-tion in design education, pp. 297-314

Button, G 2000, ’The ethnographic tradition and design’,Design Studies, 21, pp. 319-332

Chastain, T, Kalay, YE and Peri, C 2002, ’Square peg in around hole or horseless carriage?’, Reflections on theuse of computing in architecture, 11, pp. 237-248

Clarke, JA, Johstone, CM, Kelly, NJ, Strachan, PA andTuohy, P 2008, ’The role of built environment energyefficiency in a sustainable UK energy economy’, En-ergy Policy, 36, pp. 4605-4609

Coyne, R, Park, H and Wiszniewski, D 2002, ’Design de-vices: digital drawing and the pursuit of difference’,Design Studies, 23, pp. 263-286

Creswell, JW 1998, Qualitative inquiry and research de-sign. Choosing amongst five traditions, Sage, London

Denzin, N and Lincoln, Y (eds) 1998, Strategies ofQualita-tive Inquiry, Sage, California

Denzin, N and Lincoln, Y (eds) 2008, Landscape of quali-tatite research, Sage, Los Angeles

Van Drokenlaar, C, Dowson, M, Spataru, C and Mu-movic, D 2016, ’Review of the regulatory energyperformance gap and its underlying causes in non-domestic buildings’, Front. Mech. Eng, 1(17)

Gibson, JJ 1979, The ecological approach to visual percep-tion, Houghton Mifflin, Boston; London

Gorse, CA and Emmitt, S 2007, ’Communication be-haviour during management and design teammeetings: a comparison of group interaction’, Con-struction Management and Economics, 25, pp. 1197-1213

Hammersley, M and Atkinson, P 1995, Ethnography. Prin-ciples in Practice, Routledge

Harty, C 2008, ’Implementing innovation in construc-tion: contexts, relative boundedness and actor-network theory’, ConstructionManagement and Eco-nomics, 26, pp. 1029-1041

Hodder, I 2012, Entangled: an archaeology of the relation-ships between humans and things, Wiley Blackwell,Sussex

Ihde, D 2004, ’Phenomenology of Technics’, in Kaplan,D (eds) 2004, Readings in the philosophy of technol-ogy, Rowman and Littlefield Publishers, Maryland,pp. 137-160

Lawson, B 1997, Howdesigners think, The design processdemystified, Oxford, Architectural Press

Lloyd, P and Deasley, P 1998, ’Ethnographic descriptionof design networks’, Automation in Construction, 7,pp. 101-110

MacDonald, IA, McElroy, LB, Hand, JW and Clarke, JA2005 ’Transferring simulation from specialists intodesignpractice’, Proceedingsof International BuildingPerformance Simulation Association, pp. 657-662

Pink, S, Tutt, D, Dainty, A andGibb, A 2010, ’Ethnographicmethodologies for construction research: knowing,practice and interventions’, Building Research and In-formation, 38, pp. 647-659

Silverman, D 2005, Doing qualitative research: A practicalhandbook, Sage, London

Silverman, D 2006, Interpreting qualitative data. Methodsfor analysing talk, text and interaction., Sage

Tweed, C 2001, ’The social context of CAAD in practice’,Automation in Construction, 10, pp. 617-629

Wenger, E, McDermontt, R and Snyder, W 2002, Culti-vating communities of practice: a guide to managingknowledge, Harvard Business School Press, Boston

Winograd, T and Flores, F 1987, Understanding computerand cognition. A new foundation for design, Addison-Wesley, Reading, Massachusetts,

Zapata-Lancaster, G 2010, Zero Carbon Hub. Closing thegapbetweendesignedandbuilt performance. Carboncompliance for tomorrow’s new homes., Zero CarbonHub, London

Zapata-Lancaster, G and Tweed, C 2016, ’Tools for lowenergy building design: an exploratory study of the


design process in action’, Architectural Engineeringand DesignManagement, 12(4), pp. 279-295


Investigating Spatial Configurations of Skycourts as BufferZones in High-Rise Office Buildings

Coupling building energy simulation (BES) and computational fluiddynamic (CFD)

Saba Alnusairat1, Shan Shan Hou2, Phil Jones31,2,3Welsh School of Architecture, Cardiff University, UK1,2,3{AlnusairatSF|HouS1|JonesP}@cardiff.ac.uk

Skycourts are attracting widespread interest in the contemporary construction ofhigh-rise commercial buildings. Due to their remarkable function as publicrealms and transitional nodes, those spaces could introduce alternatives to thevernacular courtyards/atriums in high-rise buildings. This paper outlines themethodology that was adopted to examine the performance of basic spatialconfigurations of skycourt in high-rise office buildings in a temperate climate.Significantly, these areas accomplished as buffer zones that are non-ventilatedand unheated while accommodating combined ventilation strategy. The study isprocessed via coupled simulation between a building energy simulation (BES)and a computational fluid dynamic (CFD). The developed coupling approachaims to improve the prediction for skycourt performance that is essential for theassessment of thermal comfort conditions and energy consumption. Results tonominate the optimum spatial configuration of skycourt along the vertical sectionof the high-rise office buildings are discussed briefly. However, the focus is on themethodology.

Keywords: Skycourt, Spatial Configuration, Coupling Simulation, ThermalComfort, Energy Efficiency

INTRODUCTIONThe skycourt, by its various configurations, is recog-nised as communal areas in high-rise commercialbuildings, established from the concept of the court-yards in low-rise buildings. They function as publicsocial void spaces which can provide leisure; wellbe-ing for workers, obvious connections and social in-teraction as hubs. They could perform such as areas

for circulation and transition. Moreover, they couldperform as buffer zones between the indoor and theoutdoor thus could mediate the climate conditions,provide thermal and acoustic protection to the in-terior and reduce heat loss. Furthermore, skycourtcan offer a contemporary alternative to the vernac-ular courtyard or atrium in high-rise buildings due toits potential to allow natural light and ventilation to


enter deeper into the interior of the high-rise build-ing and avoid unwanted solar gain (Goncalves andUmakoshi, 2010; Pomeroy, 2008, 2007; Yeang, 1999).

Therefore, these areas could play a promisingrole in conserving energy and improving the ther-mal comfort for buildings (Alnusairat andElsharkawy,2015; Pomeroy, 2014). The perspectives of obtain-ing significant reductions in energy consumption be-sides enhancing the thermal comfort of users in theseareas are considered in this study. The paper is partof research seeks to investigate the energy savingpo-tentials associated with the modification of the sky-court design. This objective could be achieved bydemonstrating skycourt as an integrated buffer ele-ment in high-rise office buildings and exploring itsconsequence on reducing demand for heating andcooling, besides potential advantages on occupants’thermal comfort at the skycourt, in temperate cli-mates such as London.

Figure 1Diagrams for thespatialconfiguration ofskycourt inhigh-rise officebuildings (the whitecoloured zonerepresents theskycourt)

Skycourt may be classified on the basis of posi-tion in the midst of the high-rise building into sky-entrance, sky-terrace, sky-court and sky-roof. Thesky-entrance is the open or void space located at thelower floor(s), whereas the sky-roof is located at thetop of the building. The sky court is the open orvoid spacebetweenfloors, andfinally, the sky-terraceis the space located at the corner(s) of the building.These spaces are two or more floors height linked di-rectlywith the surrounding indoor andoutdoor areasby open walls or indirectly through enclosed walls.

However, the space configuration or form ge-ometry of skycourt can be divided into several pro-

totypes: hollowed-out space, corner space, sidedspace, infill space, interstitial space and chimneyspace (Pomeroy, 2014). See figure 1. This paperfocuses on the first three configurations becausethese spaces are widely constructed in the studycontext. Also, these configurations reveal usefulmodels to test the research’s hypothesis that sky-court could function as a buffer zone that is non-ventilated and unheated space intermediate be-tween the inside -the office zones that combinedcontrolled air temperature- and the outside- the ex-ternal environment-. These glazed void areas couldbe connected with the outdoor by one edge, twoedges and three edges. See figure 2. The hollowed-out prototype (A) represents the one edge connec-tion. The corner prototype (B) displays the two edgesconnection and finally the three edges connection il-lustrated by the sided prototype (C).

COUPLING STRATEGY: BUILDING ENERGYSIMULATION (BES) AND COMPUTATIONALFLUID DYNAMIC (CFD)A large and growing body of literature has arguedthat simulationplays an important role as an interme-diate point of knowledge for developing design solu-tions in construction. It is a useful technique for de-veloping and testing theory, in addition, it could beused as a design tool for generating design alterna-tives, predicting performance and defining the opti-mum solution that improves performance. This tech-nique could be conducted for analysing the effect ofspace(s), the system(s) or device(s) at several scales toinform more details (Groat and Wang, 2013).

However, methods for modelling the ambientconditions concerned for investigating thermal com-fort, air containment and energy efficiency could bedivided into two groups: physical models and nu-merical models. These differ in the base of mod-elling techniques and level of details for the inputdata. The physical (reduced scale) models could rep-resent or reproduce the characteristics of the full-scale physical context or system, such as materialsand products that are readily manipulated by exper-


Figure 2The spatialconfigurations ofskycourt (the whiteshaded zone) floorplans considered inthe study: (A)hollowed-out, (B)corner, and (C)sided prototypes

imenters and difficult to be obtained in full scale. Ontheother hand, numerical (mathematical)models arerecommended when dealing with questions of scaleand complexity. These use numerical approximationto predict the thermal, airflow performance insideand outside the buildings, and energy consumption,in various constructions such as atrium and glazedbuildings due to cost-effective and efficiency (Wangand Wong, 2008). In this model, mathematical prob-lems are formulated so that they can be solved withfundamental arithmetic equations of heat transferand fluid dynamics (Prajongsan and Sharples, 2012).However, the most powerful modelling is the com-puter simulation, which offers a useful tactical toolthat could produce a large body of accurate infor-mation in short periods for determining the thermaland energy performance (Ali and Armstrong, 2008;Groat and Wang, 2013). Computer programs couldreplicate the real-world contexts or events (‘’virtualworld‘’) for the purpose of studying dynamic inter-action (‘’synthetic elements‘’) that resulted of manip-ulated factors within the setting (Groat and Wang2013). Therefore, numerical simulation model tech-nique was selected to investigate the thermal, airperformance inside the skycourt, and encounter themost affordable spatial configuration of skycourts inhigh-rise office buildings.

Recently, the new direction of building ther-mal and energy simulation has been established is

in which, two models could be interrelated. Thismethod is known as the ”coupling models” that iswidely used in ventilation studies. Simulation meth-ods in the construction are classified into two ma-jor modules: the building energy simulation and theairflow method. In recent years, many literature hasrecognised the Computational Fluid Dynamics (CFD)as the most accurate and detailed model among theairflow models (Zhai and Yan, 2003). Barbason andReiter (2014) concluded from reviewing several simu-lation studies that computational fluid dynamics sim-ulation has been accepted as an appropriate simu-lation to investigate all kinds of aero-thermal phe-nomena in buildings. For example, it can predictthe full distribution of air velocity, air temperatureand air quality. In addition, it can inform the perfor-mance of both the natural and mechanical ventila-tion, the contaminant dispersion, the internal and ex-ternal airflow, and the heat islands. Moreover, CFDis sophisticated for the current architectural style,which characterised by glazed facades and atriumconfigurations (Barbason and Reiter, 2014). On theother hand, fully CFD simulation requires long calcu-lation times. Further, airflow models require thermaland flow boundary conditions that can be obtainedfrom the BES (Zhai and Yan, 2003). CFD stands onnumerical techniques to solve the equations for thefluid flow, themass of containment species, the ther-mal comfort and indoor air quality analysis. It can


solve the equations by dividing the spatial contin-uum into cells among grid, which requires iterationsto achieve a converged solution (Zhai et al., 2002). Incontrast, the Building Energy Simulation (BES) standson the principles of energy (heat) balance equationsthat considers the internal heat transfer between thespace air and surface. These include energy balanceequations for a space air, for a surface (e.g. wall andwindow) and for the radiative heat flux (Zhai et al.2002). Therefore, BES could provide thermal, energyanalysis for the whole building and the HVAC sys-tems. The output of this simulation includes mean(average) air temperature, heating, cooling, ventila-tion, solar gain, fabric and incidental loads. BES couldbe obtained on an hourly basis for the whole year.Unfortunately, this type of simulation assumes air aswell-mixed. Therefore, it is unable to provide de-tailed predictions of the spaces’ indoor air proper-ties such as the distributions of air velocity and tem-perature, the relative humidity and the contaminantconcentrations (Zhai et al. 2002; Zhai and Yan 2003).Therefore, it is argued that integrating BES and CFDtogether can produce complementary informationfor the energy consumption and the indoor thermalcomfort for buildings. Furthermore, it is agreed thatthe coupled simulation can predict more accurate,detailed and quick results compared to the separatesimulation (Barbason and Reiter, 2014; John and Yan,2005; Wang and Wong, 2008; Zhai et al., 2002; Zhaiand Yan, 2003). The coupling approach stands onproviding the interior surface temperatures and theheat extraction rate that are obtained from BES tothe CFDmodel so the airflow simulation could calcu-late specific air thermal conditions (Zhai et al. 2002).Therefore, theCFDmodel can receivemore exact andreal-time internal thermal conditions thus can pre-dict the dynamic indoor thermal conditions. This sig-nificance is essential for the assessment of indoor airquality and thermal comfort. Moreover, the BES canobtainmore accurate convection heat transfer coeffi-cient from the boundary envelope. This process pro-duces a more precise calculation of energy demandand full thermal behaviours of the building enclosure

(John and Yan 2005). In addition, using this mech-anism of integration can eliminate few assumptionsthat handled via each separate application, reducethe computation timeofCFD (WangandWong, 2009;Zhai and Yan, 2003).

There are two major approaches for couplingthermal and CFD simulation to reduce computingtime-cost: the static coupling and the dynamic cou-pling. However, Zhai et al. (2002) distinguished athird key strategy for coupling simulation. That is thebin coupling. The static coupling process includesone-step or two-step data exchange between BESand CFD programs. The process can be performedmanually with a few coupling iterations and does notrequire hard modifications of individual ES and CFDprograms. While, the dynamic coupling process re-quires continuous coupling between the BES and theCFD at each time step. This method may occur inone-time step, quasi-dynamic or full-dynamic. Theone-time-step focuses on the coupling at one spe-cific time step of interest. At that point step, the itera-tionbetweenESandCFD is carriedout to reacha con-verged solution. However, coupling might happenwithout iteration at each time step in a period such asin the quasi-dynamic. That is the CFD simulation ob-tains the boundary conditions from the previous BEScalculation at the specific time step, then returns thethermal information of indoor air to BES of the nexttime step. The full dynamic coupling involves itera-tion between BES and CFD to reach a converged so-lution at each coupling time step before moving onto the next step. In the bin coupling process, the BESreceives info that is pre-calculated by the CFD andsaved it in the bins to be used for subsequent energycomputation (Zhai et al. 2002).

Generally, the approaches of exchanging databetween the BES and theCFDmodulesmaybe classi-fieddepending on the type of data transfer into threemethods. In the first method, the indoor surfacetemperatures transfer from BES to CFD, then convec-tive heat coefficient and indoor air temperature fromCFD to BES. The second approach considers trans-ferring indoor surface temperature from BES to CFD


and then convective heat flux from CFD to BES. Thethird method includes transferring interior convec-tive heat flux from BES to CFD and then returns con-vective heat coefficient and indoor air temperaturegradients fromCFD to BES.Method one is consideredthe most appropriate one due to its stability. How-ever, method two is the most expensive one since itrequires explicit in BES and implicit in CFD. Whereas,method three is not recommended since it is unableto control air temperature (John and Yan, 2005; Zhaiand Yan, 2003).

The next section describes the coupling simula-tion approach that is adopted in the study.

INTEGRATINGHTB2ANDWINAIRFORTHESKYCOURT ANALYSISHTB2 and WinAir are used in this study. The HTB2software version 10 is used to inform the thermal per-formance and energy efficiency while WinAir Version4 is adopted as a CFD simulation to inform the ven-tilation performance inside the skycourt. The inputdata required for the HTB2 comprises information ofboth the regional climate data and the building, in-cluding info regarding the building size, construc-tion materials, small power, building services (heat-ing, lighting, ventilation and occupancy) during theoccupation and avocation periods and the diary ofapplication. The output data includes thermal con-ditions represented by air temperature, air humid-ity, mean radiant temperature, element surface tem-perature, and mean surface temperature. Further-more, energy performance embodied by space gainsand losses from heating, cooling, incidental, solar,ventilation and fabric loads. The output informationcould be in the form of power (W) or energy (kWh).This information could be based on the hourly, daily,monthly or yearly database. On the other hand, theWinAir software requires knowledge of the buildingsize, inflowandoutflow rate in case of fixed flow rates(mechanical ventilation), pressure boundaries in caseof varying flow rates (openings -natural ventilation),internal heat gain and heat loss and pollutant con-ditions in the event of source of contaminants. See

figure 3. The climatic data considers specific time atthe specific date. The study investigates the perfor-mance of skycourt in summer, winter and transitionalseasons emplacing the hottest hour, the coldest hourand mid-temperature hour. The output data com-prises graduating information of air temperature, airvelocity and air concentration showing the airflowpattern.

The paper aims to predict the indoor air temper-ature, and the air velocity at the occupancy level ofthe skycourt under the assumption that these zonesare buffer areas do not consume energy for heatingneither cooling. However, constant air supply thatexhausts from the adjacent offices‘ zones is assumedto modify the internal environment of the skycourt.Similar conditions are conducted for the three spatialconfigurations to perform the fair comparison. Exter-nal coupling is adopted in this study to ensure ac-curate prediction of the indoor environment for theskycourt, and to eliminate time cost. Therefore, twomodels should be built separately in the HTB2 andtheWinAir. A schematic model was developed in theenergy simulation to predict the thermal conditionsinside the skycourt space and the energy consump-tion for heating, cooling and ventilation, consider-ing that the skycourt space consists of three zones;lower, middle and upper zone. Furthermore, a gridmodel was built in the WinAir to investigate in de-tails the airflow phenomena - air temperature andvelocity. Data exchange for boundary conditions isneeded to bridge the two programs. The static cou-pling strategy is chosen to couple the two simula-tions. The thermal conditions for the CFD (WinAir)simulations are obtained from previously calculatedvalues from the energy modelling software HTB2.These include the surfaces’ heat transfer, the inlet -airsupply, the outlet -air exhaust and the internal heatgains involved inside the skycourt. Then, the resultedtemperature from the CFD simulation was comparedwith the average skycourt temperature from the BESto find the predicted temperature difference. Thetemperature difference was small (approximately 1°C). That little difference is usually accepted for venti-


Figure 3Diagram shows theHTB2 and WinAircoupling models

lation cases to continue the simulation for the nexttime step (Wang and Wong 2008). Therefore, one-step data exchange was adopted in the study.

RESULTS AND COMPARATIVE ANALYSISResults from the energy simulation for the annual en-ergy demand for heating and cooling the buildingsare represented in Figure 4. The difference of en-ergy demand for heating and cooling in the build-ing between the selected prototypes is small. De-tailed monthly heating, cooling, solar, fabric, ventila-tion and power loads for the skycourts are shown inFigure 5.

Figure 4Results from BES ofannual energydemand for heatingand cooling for thebuildings

Figure 6 provides an overview of the thermal condi-tions including air temperature, mean radiant tem-perature and the external air temperature, further, itpresents the heating, cooling, ventilation, incidental,solar and fabric loads at the lower zone of the sky-court in thehottestweek in summer for the threepro-totypes (A), (B) and (C).


Figure 5Results from BES ofmonthly heating,cooling, solar,fabric, ventilationand power loads forskycourts

Figure 7 illustrates the specific gradient thermal con-ditions -air temperature and air velocity obtainedfrom the CFD simulation for one case simulation at14.00, 28 June.

The results from the BES and CFD simulation arehighlighted in Table 1. It is apparent from this tablethat CFD simulation can provide accurate informa-tion at the occupancy level of the skycourt related tothe comparison criteria, while the BESprovides an av-erage temperature for thewhole skycourt space. Thetable shows that there is a small temperature differ-

ence (nearly 1 ° C) between the twomodels. This pro-vides strong evidence of the efficiency of integratingboth the HTB2 and the WinAir programmes. Overall,the results indicate that skycourt prototype A is thebest among the three prototypes in terms of thermalcomfort- air temperature and air velocity and energyconsumption reduction for the building.

CONCLUSIONSThe paper has described a method to explore thethermal performance and the energy consumptionof several spatial configurations of skycourt in high-rise office buildings in a temperate climate. The cou-pling simulation system of integrating BES and CFDis recommended for studies that examine the ther-mal performance of spaces in details at specific zoneswithin the whole space. Significantly, studies thatuse simulation as a design tool. It can be seen fromthe simulation results that the method shows effi-ciency to study the thermal conditions at the sky-court space. Therefore, this approach could be ap-plied to investigate spaces that are humongous andtall such as skycourt, atriums, courtyards and plazas.However, to further improve the accuracy of the re-sults, segmentation of the skycourt space in the BESmodel into more than one space to obtain morespecific results to feed to the CFD model is recom-mended.

It is anticipated that the coupling of HTB2 andWinAir programmes produce minimum temperaturedifference (nearly 1 ° C). This result acknowledges thecorresponding and compatibility between the twoprogrammes. Therefore, the technique described inthis paper to couple HTB2 and WinAir programmescould be applied to predicting the indoor environ-ment of other spaces.

The comparison between the three spatial con-figurations of skycourt (A, B and C) regarding thermalcomfort -air temperature and air velocity- and energyloads shows that prototype A is the recommendedconfiguration.


Figure 6Results from theBES of thermalconditions andloads in skycourt -A,B and C models atsummer


Figure 7Results from CFD ofthe thermalconditions -airtemperaturegradient (° C) andair velocity (m/s) inskycourt -A, B and Cmodels at 14.00, 28June, summer,External airtemperature is 28.3°C, RH is 42%.Location of crosssections is shown infigure 2


Table 1Summary ofcoupling resultsfrom BES and CFD

ACKNOWLEDGEMENTThe authors would like to thank Al-Ahliyya AmmanUniversity, Jordan for funding this research

REFERENCESAli, Mir M. and Armstrong School, Paul J. 2008 ’Overview

of Sustainable Design Factors in High-Rise BuildingsMir’, Proceedings of CTBUH8thWorld Congress, Dubai

Alnusairat, Saba and Elsharkawy, Heba 2015 ’Passive de-sign approach for high-rise buildings: from court-yards to skycourts’, Proceedings of ZEMCH 2015 Inter-national Conference, Lecce , Italy, pp. 739-748

Barbason, Mathieu and Reiter, Sigrid 2014, ’Couplingbuilding energy simulation and computational fluiddynamics: Application to a two-storey house in atemperate climate’, Building and Environment, 75,pp. 30-39

Goncalves, Joana Carla Soares and Umakoshi, Erica Mi-tie 2010, TheEnvironmental Performanceof Tall Build-ings, Earthscan

Groat, Linda and Wang, David 2013, Architectural Re-searchMethods, John Wiley & Sons

Pomeroy, Jason2007, ’The sky court - A viable alternativecivic space for the 21st century?’, CTBUH Journal, Fall2007, pp. 14-19

Pomeroy, Jason 2014, The Skycourt and Skygarden:Greening the urban habitat, Routledge

Pomeroy„ Jason 2008 ’Sky Courts as Transitional Space:Using Space Syntax as a Predictive Theory’, Proceed-ings of CTBUH 8thWorld Congress, Dubai

Prajongsan, Pimolsiri and Sharples, Steve 2012, ’Enhanc-ing natural ventilation, thermal comfort and energysavings in high-rise residential buildings in Bangkokthrough the use of ventilation shafts’, Building andEnvironment, 50, pp. 104-113

Wang, Liping and Wong, Nyuk Hien 2008, ’Coupled sim-ulations for naturally ventilated residential build-ings’, Automation in Construction, 17, pp. 386-398

Yeang, Ken 1999, The Green Skyscraper: The Basis for De-signing Sustainable Intensive Buildings, Prestel

Zhai, Zhiqiang 2003, ’Solution characters of iterativecoupling between energy simulation and CFD pro-grms’, Building and Environment, 35, pp. 493-505

Zhai, Zhiqiang John and Chen, Qingyan Yan 2005, ’Per-formance of coupled building energy and CFD sim-ulations’, Energy and Buildings, 37, pp. 333-344

Zhai, Zhiqiang, Chena, Qingyan, Haves, Philip andKlems,Joseph H. 2002, ’On approaches to couple energysimulation and computational ?uid dynamics pro-grams Zhiqiang’, Building and Environment, 37, pp.857-864


Computational Fabrication

A robotically-driven additive construction planningprocess using an ecological material

The introduction of 3D clay printing for large scale construction

Odysseas Kontovourkis1, Philippos Michael21,2Department of Architecture, University of [email protected] [email protected]

This work presents an additive construction approach based on roboticmanufacturing and ecological material principles. Through a robotic automatedprocess, the construction planning of a small size shelter is developed in respectto functional, environmental and structural design criteria. The latter playsfundamental role in the process of construction planning since it influencesdecisions in regard to the way the overall structure is erected using roboticmechanisms. Within this frame, clay is used as the ecological material,influencing the design development of tool paths followed by the robot duringconstruction. In parallel, the functional and environmental criteria are directlyrelated to the construction planning aspect of the process and are cross-examinedto adequately respond to the initial design parameters of control. Aim of thiswork is to achieve the additive construction planning of the proposed design andto discuss limitations and potentials of the application of 3D clay printingtechnology in the computational design and construction of large scale structures.

Keywords: Additive construction, ecological material, robotic technology,computational design, 3D clay printing.

INTRODUCTIONThe latest developments in the field of rapid proto-typing especially towards the establishment of Ad-ditive Manufacturing (AM) techniques for three di-mensional production of objects in large scale, leadto the introduction of different procedures that areimplemented in various industries including archi-tectural and construction. In general, AM proce-duresmake use of gantry andmounted nozzle, to de-posit ready-made mixtures through computer con-trolled processes. The methodology requires the de-

velopment of a three-dimensional model, which isthen split into several layers executed by the addi-tive mechanism. Following this process, specific toolpath is created that automatically adjusts the noz-zle position, movement, and material flow (Lim et al,2009) according to the material capabilities. In theliterature review a number of AM processes targetedat the design and construction of architectural prod-ucts in large scale through additive layering (Wuet al,2016) can be found, pioneered by the Contour Craft-ing technique (Khoshnevis and Dutton, 1998), and

Computational Fabrication - eCAADe RIS 2017 | 95

followed by other approaches like D-Shape [1], Con-crete Printing (Lim et al, 2012), Additive manufactur-ing of concrete (Bos et al, 2016), etc. Such processesare based on similar principles however differencescan be found concerning the use of material, theirmanufacturing logic and their applications. For in-stance, Contour Crafting (CC) is a layered fabricationtechnology that uses computer control to exploit thesuperior surface-forming capability of trowelling tocreate smooth andaccurateplanar and free-form sur-faces (Khoshnevis and Dutton, 1998; Khoshnevis etal., 2006). Similarly, Concrete Printing is a procedurewhich makes use of a gantry and a mounted nozzle,to deposit a readymade mixture using a computercontrolled algorithm (Lim et al, 2012). In comparison,D-Shape has a different approach to the constructionprocedure. Similar to inkjet powderprinting, this pro-cess uses a powder deposition method and a binderto harden the selected area of each layer [1]. In all thelarge scale AM processes, the nozzle responsible forthematerial deposition ismounted either on a crane,a frame (gantry) or a robot. In the cases under consid-eration all nozzles are mounted on frame structure,allowing the deposition ofmaterial to bemade freelyin x-y-z axes. The selection of themachinery used forthe construction procedure is directly correlated tothe other factors of the process, such as the toolpathfollowed for the completion of the product, the scaleof the finished product as well as the material used.

The use of AM techniques in architecture andconstruction industry could potentially give rise tonew construction capabilities that might include theresolving of complex design issues and complex as-semblies. Also,might offer thepossibilities for on-sitemanufacturing, less labour and construction time aswell as might promote the local and ecological as-pect of material use. According to specific projectrequirements, appropriate technique can be used.Concerning issues such asmateriality, product shapeor speed any of the discussed AM procedures couldbe further explored to be implemented, or even anew concept technique could be studied.

Thiswork, utilizes an AMprocedure similar to the

Contour Crafting and Concrete Printing techniqueswhere layer by layer construction in actual scale to-gether with the deposition of specificmixture are im-plemented. The proposed methodology employs astandard industrial robotic arm instead of a gantry asthe deliveringmechanism, with a custom-made noz-zle mounted on it, and it uses clay instead of con-crete. The application of robotic technology and theuse of clay material formulate the two important as-pects of current research investigation. Within thisframe, an approach for the robotically-driven con-struction planning of a shelter is suggested. Func-tionality along with the environmental aspect andthe ecological behavior of the structure, all are exam-ined simultaneouslyofferinganadditivemanufactur-ing planning solution.

At first, the research is focusedonmateriality andexperimentation on the clay mixture. In the secondstage of analysis, the selected material is tested insmall scale physical prototypes, with mixture, shapeand deposition affecting one another, in order toreach a desired and complete outcome. To add an-other layer of sophistication to the structure, an in-novative depositionmethod is utilized in order to re-duce time and material used during the large scaleconstruction process by creating gaps in the thick-ness of the shell, which at the same time provide airgaps to help insulate the interior. The capabilities ofthe robotic arm are cross-examined by toolpath in-vestigation that is carried out in respect to the geom-etry of prototypes and the material limitations. Fi-nally, the additive construction planning of a small-scale structure is presented as a case study.

PROPOSED METHODOLOGY AND EXPERI-MENTSThe core of investigation lies on the tool path plan-ning development that formulates the overall struc-ture driven by functional and environmental param-eters. In order to achieve an effective methodologyfor the robotically-driven additive constructionof theshelter, the material mixture and its static behavioras well as the tool path planning are investigated

96 | eCAADe RIS 2017 - Computational Fabrication

Figure 1Initialexperimentation ondifferent mixtures

Table 1Materialcomponents ofclay-like mixtures

through extensive experimentation. Aim is to findthe limitations and potentials of selectedmaterial to-wards appropriate pre-design and pre-constructiondecisions for the development of structure in actualscale.

Experiments on material properties, staticbehavior and tool path planning processIn order to successfully support our hypothesis an ini-tial and ongoing investigation into material proper-ties, static behavior and tool path planning process is

Figure 2Vertical stacking oflayers of filament


exemplified in this part of the paper. The materialityaspect is directly connectedwith thedecisions foron-site construction and the use of localmaterial, aimingto promote an ecological approach to the process.Using a local material with no or as few as possibleadditives for construction, the process enforces min-imum disruption to the environment. After the lifecycle of the constructed structure the material disin-tegrates and returns once again to its original state.That being set, the base material is a granular mate-rial found in the ground, in the form of sand or soil.

Different mixtures are created and tested fortheir efficiency (Figure 1). In order to satisfy the stabil-ity of theproduct, aswell as the ecological aspect andthe extrusion process requirements, both sand andsoil are combined with epoxy glue, water, clay, ash,or even concrete. Several quantities and analogiesare made and documented. Multiple combinationsgive rise to various differentmixtures, but noneof theresults are able to provide the necessary propertiesthat can fit the aim of this case study. For example,while giving stable results, the mixture with epoxyglue is hard to handle during the extrusion process.Likewise, further experimentation proved that con-crete or gypsum do not merge very well with sand,although they are strong binding agents. On a finalnote, none of the above is able to hold the mixturetogether, destroying the product.

These initial observations lead to the selectionof water and soil as the main materials. The mixturecontainsmostly soil (80%), whilewater acts as a bind-ing agent. Although, this combination is easily pre-pared and dried, it produces unstable results and forthis reason themixture is combinedwith straw to cre-ate a mud-like mixture, mostly used in the creationof adobe bricks (Table 1). Water acts as the bindingagent and straw is used for the consistency of mix-ture. The result is a clay-like mixture, easily extrudedthrough a nozzle in the form of a hose with squareopening 4cm x 4cm. Important aspect taken intoconsideration is the stability ofmaterial so that shapeandoverall design canbe kept steadyduring layeringprocess. After an appropriate curing time period (30

minutes approximately), the material dries enoughand is able to withstand the addition of another layeron top of it. As long as the environmental conditionsare kept constant, it takes about three days to drycompletely, leading to its hardening. In this work, anumber of in-house clay mixtures and material com-positions arepresentedand their static properties arefurther discussed.

Figure 3Deposition ofmaterial in theexample ofhorizontal buildingelements

The mixture 1 is almost in liquid form and assuch is easily extruded from the nozzle. However, ithas limited structural abilities and it hardly keeps itssquare form and for this reason is rejected. Mixture2 has better characteristics and is able to remain in asquare form after extrusion. Nevertheless, the anal-ogy of the straw is very low and the material startedcrackingduring the curingperiod. Mixture3presentsthe best quality result due to the relatively easy ex-trusion through the nozzle and the adequate struc-tural capabilities, strengthening through the curingperiod into a solid mass. Mixture 4, due to its low


Figure 4Thirteen layers offilament for verticalelementsconstruction

moisture, becomes very hard and rigid, making it al-most impossible to be extruded. Consequently, mix-ture 3 is selected as the construction mixture to beused for further experimentation (Table 1).

By using the custom-made hose, the selectedmixture is deposited in layers of filament. The verti-cal stacking of layers of filaments are cross linked andremain for certain curing time period, giving enoughtime to the previous layers to dry and solidify so thatstresses of the next layer ofmaterial can bepreserved(Figure 2). Specifically, in this work the additive con-struction of three representative building prototypesis investigated, focusing on the tool path develop-ment in one or two axes, togetherwith the respectivestatic behavior of each component.

The prototypes reflect the morphology exam-ined in the part of case study and include horizontalsurfaces, inclined surfaces and vertical surfaces withdual-aspect openings. In addition, the prototypesare designed to incorporate gaps in the interior ofthe walls to minimize material and also provide unitswith as less weight as possible, without compromis-ing stability and structure.

Firstly, the 3D clay printing of horizontal buildingelements is created by a double pyramid depositionof material in a number of vertical layers of filamentwith one less stripe in each layering. At the middleof the structure, the reversed procedure is being fol-lowed, until the prototype is completed (Figure 3).

Then, the vertical elements with dual-aspectopenings are defined in a similar way using the dou-ble pyramid construction logic but with an eccentric-ity in the far corner. Two of these shapes, in symmet-

rical position, create the diamond-like shape of theopening. In general, the prototype is developed bythirteen layers through the deposition of material ina form of a square-like shape, whereas in each layer,the length of the perimeter is reduced by the thick-ness of a single stripe. This continues until the mid-dle of the structure, when the procedure is reversed(Figure 4).

Finally, the inclined surface is developed usingagain a double pyramid shape, yet in a form of can-tilever. The pyramid is created in a ninety-degrees ro-tation in comparison to the other prototypes and inan offset shape. In this way, the connection to theexterior and interior surface is maintained, althoughan inclined logic is followed. Similarly, with the hori-zontal element, the inclined surfaces are consisted ofparallel stripes of material deposition.

The three prototypes determine the limitationsand potentials of material and static behavior as wellas the effectiveness of tool paths design. This is donein a bidirectional manner from design to physical ad-ditive construction and vice versa until the appropri-ate design and construction rules are defined.

Figure 5Initial numericalsimulation resultsof the secondprototype

In addition, preliminary studies in regard to the static


behavior of the second prototype is conducted. Fig-ure 5 shows a finite element analysis example of thesecond prototype that is simulated in Abaqus [2], us-ing information derived from the work done by (Il-lampas et al, 2014; Illampas et al, 2017) in the area ofdesign and production optimization of adobe bricksin Cyprus. Models show maximum principle stressand displacement distribution in the z-axis. The re-sults are used as the foundation for the next phase ofexperimentation that is the robotic execution.

Experiments on robotics and automationWithin the framework of robotically-driven construc-tion and in an attempt to proceed with the actualconstruction of the suggested prototypes in futurestage, initial experiments in regard to the use ofrobotic andautomationmechanismsandparticularlythe applicationof appropriatematerial extruders andend-effector tools are conducted. In regard to thematerial extruders, air pressure machines, plungersand augers are investigated. A difficult issue un-der consideration is the continuous flow of material,which is correlated with the mixture’s properties; thesofter the mixture is, the more constant the flow. Onthe other hand, the softer the material is, the less itcould hold its extrusion form. So, the most adequatebalance is found that would allow steady flow andeven form to be developed. The machine extruderselected for further experimentation is based on anauger propeller that pushes material through a hoseand off the nozzle.

In regard to the end-effector tool applied, a cus-tom made end-effector tool made of square open-ing (4cm x 4cm) is used following the manual exper-imentation. For study purposes the end effector isattached to an industrial ABB robot, model IRB2600-20/1.65 (Figure 6). In comparison with the manualconstruction of prototypes, the robotic automationrequires theestablishmentof aprocedure to incorpo-rate an interrupted flow of material but also its con-trolwhen this is necessary. Towards this direction, ini-tial experiments to identify appropriate mechanismsand material control are examined.

Figure 6Material extruderand end-effectortool mounted onthe robot

CASE STUDYBased on the results derived from the experimentsas regard the material and static behavior as wellas the tool path planning process, in this sectiona comprehensive design of the robotic tool path isdescribed, providing an additive construction plan-ning solution for the development of the suggestedsmall size shelter in actual scale. The pre-designandpre-construction development is also influencedby functional and environmental criteria formulatingthe overall morphology of the structure.

Figure 7Conceptual study ofnatural lightingentering the shelter

The shelter is a symmetrical, rectangular structure,4.50m width and 4.50m height. A hole 1.50m widthon the rooftop aims to enhance the environmentalconditions in the interior space.

As regard environmental aspects involved in thedesign of the shelter, openings at a height of 1.08mprovide ventilation and allow the light to enter the


interior space. Even though these openings are rela-tively small, their number and repetition pattern pro-vide adequate environmental conditions in the inte-rior and prevent the direct insolation and wind. Incombinationwith the hole on the rooftop, a breeze iscreated, providing enough sunshine to penetrate theroom (Figure7). Thehole at the topcreates a vacuum,strong enough to push the warm air from the insideand prevent exterior wind from directly penetratingthe interior (Figure 8). The shelter creates a small in-terior, sufficient for a couple of people to temporaryinhabit the space for a short time period. It providesbed, sitting spaces, table and fire pit, which acts bothas a heater and a stove. As shown in Figure 7 and Fig-ure 8, everything revolves around the fire pit in themiddle, creating four different functional areas in theinterior: resting area, sleeping area, eating area, andthe entrance.

Figure 8Conceptual study ofwind flow in theexterior and interiorspace

The construction system is distinguished into threebasic parts, regarding the material depositionmethod (Figure 9). These parts are related with thethree representativebuildingelements studied in themethodology section of this paper. In the lower part,horizontal construction logic is applied to serve theneed for floor and functional area creation. Adjustto that, the vertical construction logic is examined,aiming to provide adequate space in terms of height

and to host the openings of the structure. Finally,the inclined construction logic is applied, in order todevelop the roof of the shelter.

Figure 9Three basic parts ofthe shelteraccording to theconstruction logicapplied

For the additive construction planning of thehorizontal part, which functions as the foundationand the floor of the structure, the first prototype isused as reference. In the same way, two parallel hor-izontal surfaces in the logic of incorporating a dou-ble pyramid system is applied. Keeping the designtechnique same, the scale is modified to fit the oc-casion. Nine different layers of material patterns arecompleted in order to fully construct the floor. Thepattern creates a shape resembling of two opposingpyramids, which form air gaps between the upperand the lower part of the floor. The height of the flooris 50cm, sufficient to createdistance fromtheground,providing in parallel structural stability to the wholesystem. In the upper level where functional areas ex-ist, the technique remains the same and the proce-dure is repeated due to the same horizontal orienta-tion of construction system. Directly above the hori-zontal part, the additive construction planning of thevertical part is implemented. In this part, the depo-sition process is adjusted to follow a rotated double


Figure 10Three characteristiclayers of theadditiveconstruction logicthat correspond tothe horizontal,vertical and inclinedparts of theconstructionsystem

Table 2Parameters for theadditiveconstructionplanning of thestructure

pyramid system, as achieved based on the secondphysical prototype. Using this method, the open-ings and the extended vertical wall in the height of0.74m are developed, allowing enough height for anaccommodating space in the interior. Following thevertical additive construction planning, an inclinedpart is created to act as a roof of the shelter, using theinclined construction logic. Five layers in a repeatedpattern provide an adequate height of the shelter tobe covered (Figure 10).

The shelter consists of 50 topologically similarsegments that are repeated in a polar array aroundthe central point. Due to the curvilinear design, eachsegment is partially distorted. The distortion occursin one direction, which help to maintain a topologi-cal continuity of the structure. The width and lengthare the only parameters affected, which can be com-pensated by using longer filaments (length) or morerows of filament (width). A parametric algorithm us-ing Grasshopper software [3] (plug-in for Rhino [4])controls the maximum distortion, checking the pa-

rameters of transposition between layers. The seg-ment with the largest distortion reaches dimensionsof 40cm x 25cm, while the smaller one is 25cm x10cm. For the construction of each segment, firstly,the lower layer is created, with as many depositionfilaments as necessary, and then, for each layer theamount of filaments is reduced by one. When theprocedure reached the middle stage is reversed withtwo filaments the minimum. Table 2 demonstratesparameters of additive construction planning in re-spect to the different parts of construction system.

An indispensable part of additive constructionplanning process is the robotic scenario imple-mented. The robotic construction system consists oftwo separated parts, the service robot and the mainrobot, which are detached upon reaching the con-struction site. The service robot is deposited, remain-ing on-site until the project is completed. The pur-pose of the service robot is to provide support to themain robot by collecting resources from the environ-ment; it collectsmaterial (soil) and power (sun). Thus,


Figure 11Mobility capabilitiesof the robot duringadditiveconstructionexecution

the main robot is lighter and more flexible to carryout the given tasks and to complete the construc-tion of the shelter. To accomplish this, themain robothas toovercomecertainobstacles that include the re-quiredheight andwidthof structure. The robotic armIRB4600 can reach up to 2.5m, which is by no meansenough to cover the 4.5m height of the structure.Certain limitations in regard to the capability of mainrobot (IRB4600) to reach vertically the total height(4.5m) of structure are overcome by a hydraulic liftsystem that is implemented under the base of therobot, adding an extra 1.5m height. In terms of hor-izontal reach construction capability, the main robotis also limited and for this reason is programmed tomove around the shelter, alternating between 4 dif-ferent positions. Mobility of the main robot ensureseffective construction of the whole area. In paral-lel, the deployment of flexible cables and pipes con-nected with service robot allows uninterrupted con-struction process.

An example of robotic construction scenario isdemonstrated in Figure11. In case of floor construc-tion, the robot is placed at the middle of each sidewith the capability to reach from one edge to the

other. As the structure is divided into 50 segments,each side consists of 12-13 divided parts, that arecompleted one layer each time. The robot has to al-ternate 9 times between 4 positions (RP), before 50segments of the floor are completely constructed.The height of the floor is within the main robot’sreach, and therefore the lift hinge mechanism is notnecessary. The same procedure is repeated until all50 segments of all 9 layers are completed. In a sim-ilar way, all the different parts of structure and thelayers of shelter are constructed, each time by liftingthemain robotupwardswhen the structure is gettinghigher. Figure 12 shows three representative steps ofthe construction simulation. Through the digital sim-ulation, one can comprehend the construction pro-cedure and the continuum between layers, startingfrom the first layer at the bottom of the shelter andgradually reaching the top at a height of 4.5m.

CONCLUSIONThis work is concentrated towards the investigationof the appropriate material mixture and propertiesso that morphological and construction demandscan be effectively executed using robotic technol-

Figure 12Threecharacteristics stepsin the simulation ofconstructionsequence


ogy. Aim is to development an additive construc-tion planning process, implemented through 3D clayprinting in actual scale. In addition, the work aims toraise awareness in regard to the limitations and po-tentials of the suggested technology, offering a firstinside into the feasibility of additive construction so-lutions in the production of large scale objects.

This initial experimentation and respective re-sults show possible limitations and potentials of 3Dclay printing technology applied within the generalframework of additive manufacturing process. Con-sidering the large amount of parameter and criteriathat are involved in the suggestedmethodology, fur-ther investigation and deepening into different di-rections is necessary to be developed. Materiality,static behavior, design of toolpath, robotic and au-tomation mechanisms applied, all are involved andare necessary aspects for further examination. In thispaper, an introductory and a general overview of oursuggested 3D clay printing process has been demon-strated, opening the ground for future experimen-tation and development that will take into accountpossible criteria involved.

REFERENCESBos, F.P., Wolfs, R.J.M., Ahmed, Z.Y. and Salet, T.A.M.

2016, ’Additive manufacturing of concrete in con-struction: potentials and challenges of 3D concreteprinting’, Virtual and Physical Prototyping, 11(3), pp.209-225

Illampas, R, Charmpis, DC and Ioannou, I 2014, ’Labo-ratory testing and finite element simulation of thestructural response of an adobe masonry buildingunder horizontal loading’, Engineering Structures, 80,pp. 362-376

Illampas, R, Loizou, VG and Ioannou, I 2017 ’Effectof straw fiber reinforcement on the mechanicalproperties of adobe bricks’, 6th Biot Conference onPoromechanics, Paris [In press]

Khoshnevis, B. 2006, ’Mega-scale fabrication by contourcrafting’, Int. J. Industrial and Systems Engineering,1(3), pp. 300-320

Khoshnevis, B. and Dutton, R. 1998, ’Innovative rapidprototyping process makes large sized, smooth sur-faced complex shapes in awide variety ofmaterials.’,Materials Technology, 13:2, pp. 53-56

Lim, S., Buswell, R.A., Le, T.T., Austin, S.A., Gibb, A.G.F.and Thorpe, T. 2012, ’Developments in construction-scale additive manufacturing process’, Automationin Construction, 21, pp. 262-268

Lim, S., Le, T., Buswell, R., Austin, S., Gibb, A. and Thorpe,T. 2009 ’Developments in construction-scale addi-tive manufacturing processes’, Proceedings of theConferenceGlobal Innovation inConstruction, Lough-borough University, Civil and Building Engineering,pp. 512-520

Wu, P., Wang, J. and Wang, X. 2016, ’A critical review ofthe use of 3-D printing in the construction industry’,Automation in Construction, 68, pp. 21-31

[1] http://d-shape.com/[2] https://academy.3ds.com/en/software/abaqus-stud

ent-edition[3] http://www.grasshopper3d.com/[4] https://www.rhino3d.com/


Before and After NURBSInvestigation of Changes in the Material and Fabrication Processes of theMonocoque Structures

Serenay Elmas1, Erenalp Saltik21,2Istanbul Technical University1,2{serenelmas|erenalpsaltik}@gmail.com

It is possible to define monocoque structure as a system which the structure andthe skin work together. The earlier examples of monocoque structures, whicheffects of form and material on structural behaviours can be clearly observed, areseen in the automotive, ship and aircraft industries. From the early 20th centuryonward, many pioneering construction engineers and architects have worked onmonocoque structures to seek alternative paths to space construction.With the useof NURBS (Non-Uniform Rational B-Splines) based programs, free forms ofcurvilinear and complex surfaces are facilitated both in the virtual and physicalenvironment. At this point, relatively basic geometries have left their place tomathematically more complex geometries. In this study, an interpretation hasmade based on the analysis and comparison of selected samples in the context ofmaterial and fabrication strategies. This study can be considered as an attempt tounderstand the impacts of NURBS on monocoque structures.

Keywords: monocoque structure, structural skin, form-material-structurerelationship, digital fabrication

IntroductionWhen the monocoque word is examined etymo-logically, it is seen that it was comprised of usingthe mono (single) and coque (shell) words together.Thus, it is possible to infer that the bearing featureis integrated into the shell and there is no need fora separate load-bearing system. From the early 20thcentury onward, monocoque systems, which are fre-quently encountered in the automotive, ship and air-craft industries, where form and material directly de-termine structural behavior, have been worked bymany pioneering engineers and architects.

In the 1950s, work on the NURBS curves was ini-tiated in order to make the curvilinear bodies of carsand boats identifiable (Burry and Burry 2010). Thesedevelopments for monocoque systems in other in-dustries also affected architects, and NURBS curvesstarted to use effectively in modeling programs forarchitecture since thebeginningof the1990s. NURBScurves allow curvilinear surfaces with continuity tobeeditedwith control points in certain fracture zonesso that all curves forming the surface are not identi-fied as a point sequence. Furthermore, the data loadis minimized to create surfaces with an exact mathe-


Figure 1The timeline ofconsidered samples

matical definition. In this case, as much as the com-puter representations of the structureswere affected,the production logic was also affected. Therefore, Itis possible to say that the communication betweenthe design tools and the production tools is strength-ened. The constructability need of designed struc-tures and their communicationwith fabrication toolshave laid the groundwork for interrogating existingfabrication methods, the use of materials with differ-ent performance characteristics in line with new fab-rication tools and structural needs have also come tothe fore simultaneously.

The impact of NURBS-based programs on thedesign and production phase of monocoque struc-tures has been tried to be understood and discussedthrough selected samples which were constructedbefore and after the early 1990s. The period dur-ing which the NURBS-based programs were startedto be used in the architectural drawing programswas determined as a threshold. Since the new dig-ital production tools and designed free form struc-tures refers the search for new materials, the sam-ples were handled in terms of materials and produc-tion systems in the working process. Because of thesymbiotic interaction of form, material and structuralrequirements with each other, and the relationshipwith NURBS curves can be clearly observed, mono-coque structures were decided to work on.

In the scope of the study, 6 samples were se-

lected considering the criteria such as completiondate, material variety, production methods and form(Figure 1). These examples are, Algeciras Market Hallby Eduardo Torroja (1933), Restaurante Los Manan-tiales by Felix Candela (1958), Raybould House byKolatan-Mac Donald (2003), Beast by Neri Oxman(2010), Porsche Pavilion by Henn (2012), ICD/ITKEPavillion by Achim Menges and Jan Knippers (2015).

Differentation in ProductionMethodsThe creation of free-formed curved surfaces in thedigital environment has become possible with theuse of NURBS-based programs in design. In progressof time, the relationship of the designed form withthe structure became analysable with simulations orrelated plugins installed in the programs, and at thesame time, the feedback received could be interferedwith. While the use of NURBS-based programs al-lows flexibility in the design process and support fornew geometry searches, this freedom has comewithvarious challenges in the construction phase. Forinstance, sometimes subdivision surface modellingis preferred to NURBS due to the necessity of post-rationalizing process for panelization (Hudson et al.2015). Nowadays, as constructability is consideredone of the primary objectives of computational de-sign, the need for realising designed free-forms ge-ometries in a digital environment, has led to a ques-tioning of existing production methods. Pottmann


Figure 2Porsche Pavilion2012 perspectiveview and twistingdetail [1].

(2007), also mentioned that complicated freeformshapes require advanced methods of geometric def-inition to make them constructible. Over time, theproduction methods of curved surfaces that are fre-quently used in the automotive, ship and aircraft in-dustries have begun to be used in architectural con-struction scales. In addition to the questioning ofexisting production techniques and adaptation toarchitectural scales, suitability of NURBS-based pro-grams for the operation of computer-aided produc-tion tools has led to developing these tools, producenew tools, and use digital tools with different func-tions at a later stage.

It is possible to say that the involvement of thedesigner in the whole process from the conceptphase to the production of the design is supportedby the research groups formed in the design offices.With these interdisciplinarygroups,material researchand the use of digital production systems have be-come part of the design. In the case of the PorschePavilion (2012), designed by Henn, the design andproduction process was similarly completed underthe control of the architectural office (Figure 2). The

shell, which is structurally challenged by being a can-tilever, is constructed in two layers using stainlesssteel plates, which then work together structurally,without losing the monocoque character [1].

“ Generating doubly-curved structures from ini-tially planar elements is of major interest in the fieldof architecture, as double curvature is highly bene-ficial to the structural behaviour, while the planarityof the elements facilitates fabrication” (Bechert et al.2016). In the case of the Porsche pavilion, metalplates were cut directly by transferring the digitalmodel into digital production machines and thenbent at thedesired angle at the shipyard. Theprocesswas completed bymounting themodules created bycombining the layers (Figure 3). By using prefabrica-tionpossibilities productionprocesswas accelerated,at the same time, planar steel plates were bent at dif-ferent angles. Likewise, most of Coop Himmelb(l)au’ssteel parts of buildings were produced in shipyards,because geometrically complex steel plates could bemade directly from the computer model efficientlyand precisely [2]. Also, the thickness of shell wascontrolledbyusing structural simulations and related

Figure 3Porsche Pavilion 3Dmodel and theproduction phaseof units [1].


plugins. Thus, considering the tensile and compres-sive forces at different values applied to differentzones on the shell surface, the shell thickness variedthroughout the structure.

The shell thickness is fixed in the early 20th cen-tury examples of monocoque structures, which wereconstructed with reinforced concrete. Because themain purpose was reaching the larger spans withthinner structures than the traditionalmasonry struc-tures by taking advantage of the mechanical proper-ties of thematerial. In this period, the design and theconstructability of the curvilinear shells were testedwith physical models over a long period of time, be-cause the related digital tools have not yet been de-veloped. In the direction of the material, the con-struction of shells was completed by in-situ moldmethod. It can be interpreted that the thickness ofthe material is kept constant throughout the entireshell, in order not to extend the long constructionpe-riod any longer by changing the molds, and to avoidthe disrupt total behavior of the system against theloads. The implementation of a new material in cer-tain forms with a new technique has created a pro-cess to experience the direct structural behavior ofthe material.

Although Candela placed great importance onform studies, in most of his projects hyperbolicparaboloid was selected as a base unit and 90% ofhis works are the results of rotation and intersec-tion of this geometry in diversely (Cassinello et al.2010 ). It can be seen that the shell samples de-signed by the use of hyperbolic paraboloids, which

form a more dynamic and complex perception thanthe domed forms, are considered as structural art inmany sources. Contrary to the visual impression thatit has created, symmetrical forms and mold makingsuitability have caused these forms tobe repeated fora long time. Restaurante LosManantiales (1958), oneof the most significant works of Candela, is an exam-ple of amonocoque structure formed at the intersec-tion of the centers of four circularly arranged hyper-bolic paraboloids. The preferred symmetry to easeconstruction process can be also clearly observedfrom drawings (Figure 4).

William Mitchell observed that architects candraw what they can build and build what they candraw, and that observation now seems to be valid, itis possible to come to the conclusion that the possi-bilities offered by new digital production techniquesand the capacities of these tools are beginning to beinfluential in shaping the designed forms (Kolarevic2003). The ICD / ITKE Pavilion (2015), designed byAchim Menges and Jan Knippers, can also be tracedas a final product of a process between digital designand production tools.

In ICD / ITKE Pavilion, air supported ETFE wasused as a mold and the carbon fibers were adheredto the inner surface of the mold by the robotic arm,which is left in the air supportedmold (Figure 5). Theform has been developed using structural simula-tions and computational design tools in line with theconstraints imposed by the robotic arm to be used inproduction. An agent-based program has been de-veloped so that thepath tobe followedby the robots,

Figure 4Restaurante LosManantiales formdiagram and roofplan 1958 [3].


Figure 5ICD/ITKE Pavilion2014-2015production process[4].

and therefore the locations where the fibers will beplaced canbedetermined. Thedeformations that oc-cur in the form during bonding are also included asa parameter and feedback is continuously obtainedby using an embedded sensor system. In this re-spect, the production of new robot control codes isallowed digitally in accordance with the notificationsreceived simultaneously with the actual productionconditions [4]. When the whole process is consid-ered, it can be deduced that the design and the pro-duction are integrated and the density of the usedmaterial, turns the system into a monocoque struc-turebygaininga load-bearing character step-by-stepduring construction.

Figure 6Material variety inmonocoquestructuresbefore/after NURBS

Both the conditions of constructibility and the re-sponses of continuous surfaces to loads and stresses,lay the groundwork for the re-evaluation of the struc-tural shells and tectonic concepts. While the load-bearing feature of the shell provided the environ-ment for the production of new forms, these new

forms referred to search for new materials (Kolarevic2003).

Material ProcessHistorically, NURBS invented for progression require-ment in computational surfaces for CNC tools. Thus,it can be deduced that the roots of NURBS surfaces liein its connection to material systems, different thanother approaches act without consideration of ma-terial characteristics (Cabrinha 2005). As a result ofthe use of NURBS-based programs in architecture,the complex curvilinear surfaces became analysableand transformable with featured plugins. These fea-tures encouraged architects to test new materials’structural behaviors in different form alternatives be-fore construction phase. Besides, new fabricationtechniques and enhanced communication with pro-duction tools broadenedmaterial variety that can beused with these tools (Figure 6).

The materials used in the monocoque structuresamples were investigated under the headings of (i)single, (ii) layered and (iii) compound in the study.In the first examples of structural shells, concretewas used as a single material. At the beginning ofthe 20th century, the structural behavior of the newbuilding material has been emphasized for a longtime, and it has become possible to construct verythin structural shells compared to the traditional ma-sonry construction. Putting it another way, since theconcrete was a new building material, the possibili-ties offered by material and new construction tech-


Figure 7Algeciras MarketHall 1933 [5].

niques have been evaluated during this period. Also,it proceeds from the possibilities and limitations of-fered by the material, rather than form search. Ad-ditionally, considering the duration of static analysisand construction phase, a large number of reinforcedconcrete shells have been constructed for quite along time without searching for different materials.

Forms, that were generated by simple combi-nations and transformations of defined geometries,had been tried to improve and complexify in time.However, It is seen that basic defined geometrieswere not disappeared entirely in reinforced concreteshells. Eduardo Torroja designed a shell, formed bythe intersection of a hemisphere, and seven smallerspheres, for the Algeciras Market Hall (1933) (Figure7). Although such a form is quite simple and hasa very basic mathematical description, it has beendistinguished from the conventional masonry by us-ing concrete and it has been possible to constructa very thin shell of 8 cm, compared to the masonry.Then the reinforced concrete shell thickness was de-creased to 4cm, and the minimum thickness wasreached in Restaurante Los Manantiales (Figure 4),designed by Candela.

This process followed by the structural shellsof Heinz Isler. By experiencing the compressivestrength and weakness against tensile loads of theconcrete and tensile strength and weakness againstcompression loads of the textile materials, Isler hasa new approach to form search and constructability.

At this point, it is possible to deduce that the perfor-mance characteristics of the materials are beginningto play an active role in form generation as well asthe production process. At the same time, it can besaid that such experimental processes and the prob-lems faced by them constitute a basis for digital sim-ulations used today. At the International Associationfor Shell Structures (IAAS) conference in 1959, Islerpresented 39 applicable form alternatives for struc-tural shells, tested with textile materials, and statedthat these forms are only selected from the infinitevariations possible to produce (Chilton 2000). Therequirement to build reinforced concrete structuresin-situ, prolongs workmanship and production pro-cesses. Therefore, when considering the advantageof prefabrication, a material which is totally work-ing in unit logic as compared to the construction ofa holistic material with a mold, is decreased assem-bly and accordingly construction duration. At thesame time, the operations that can be applied to theseparately produced units will be much more practi-cal than intervening with the mold to create a con-tinuous surface. When considered in this context, itis possible to investigate the monocoque structuresformed by using various metal sheets and woodenplates also under the title of the single material.

From another point of view, the need to meet astructurally different tension and compression valueat each point on the shell surface can be considereda sign of the heterogeneity of the nature of the sys-


Figure 8Beast 2010 3Dmodel and material- pattern detail [7].

tem. When this heterogeneity is interpreted throughthe material, it can be concluded that the structurecan be formed by using different materials togetherinstead of a single material. Raybould House (2003),designedbyKolatanMacDonald Studio, is a proposalof using polyurethane foam to cover a prefabricatedcage, preferably a layeredmaterial instead of a singlematerial. Unlike concrete, the cage has a digital pre-fabrication process. At this point, in order to preservemonocoque character, it is important that the layersthat make up the structure act like a single element,against the loads on it.

Accelerating in the search for new materials, it ispossible to say that composite materials are handledwith a different approach in terms of layering proper-ties, and they are one step further from the single andlayered materials. Neri Oxman’s experimental stud-ies onmonocoque structures in 2007using themulti-material 3d printer, which can combine multiple ma-terials, can be seen as the first examples of the use ofa compound material in this area [6]. Oxman (2015)mentions a variety of methods and tools that incor-porate computational design, material science, en-gineering, and advanced digital production systemsthat they have developed in their research. One ofthe products of such interdisciplinary processes canbe seen as Beast (2010), the next step in experimentalwork onmonocoque structures. In the project, an or-ganic form consisting of curvilinear surfaceswas pro-duced by using a continuous surface consisting of acombination of five materials and allowing this skin

to act as the structure of the form (Figure 8) [7]. Thematerials that differ according to their rigid / elastic-ity levels, which aremarked with colors from black towhite. Then, structural requirementsweremet by ad-justing the frequency, flexibility, and strength of thecompound material. The design has advanced overthepossibilitiesprovidedby the combineduseofma-terials in different flexibility, besides generating andstructuring the formhas been completedbyprioritis-ing the material (Oxman 2012).

The material-fabrication-design (MFD) modeldescribed by Rivka Oxman (2015) is defined as thedesign of the tectonic relation between the structureand thematerial within the framework of productionlogic. When assessed in the context of this model,the mentioned integration and process orientationof production can be clearly observed both in Beastand ICD / ITKE Pavilion examples.

Concluding RemarksIn all of the samples considered in the study, mate-rial behavior and productionmethods are seen to beprioritised as an effect of the being monocoque ofthe case studies (Table 1). Therefore, the triggeringbetween new materials and new production tech-niques can also be read through examples. Consid-ering the reinforced concrete shells studied, it is pos-sible to talk about the continuation of a long periodof experimentation before the construction phase. Inthese examples, problems and difficulties that mayarise in the construction process have led to the pref-


Table 1The informationabout materials andfabricationprocesses ofconsideredsamples.

erence of simple geometries and symmetry, eventhough the form decisions have been made accord-ing to load-bearing characteristics andmaterial prop-erties. The thickness of the shell, which is held con-stant along the form, can alsobe read as an indicationof this condition. With the use of NURBS-based pro-grams, surfaces can bemathematically defined in thelevel that they can communicate with the produc-tion tools. In other words, instead of applying basicmathematical forms due to production constraints,complex forms have beenmathematically explained.By solving the problem of communication with themeans of production, the generated forms are ableto be freed from the basic geometry and became ed-itable freely. Nevertheless, in the case of reinforcedconcrete shell case studies investigated, the experi-

mental progression can be interpreted as proximityto the present computational design process basiclogic, by shifting the focal point to the behavior ofmaterials and the structural requirements rather thanthe form search. Especially in Beast and ICD / ITKEpavilion samples, it is seen that the design and pro-ductionprocess is integratedwith each other and theform has become an outcome of this integrated pro-cess. Therefore, production methods, as well as ma-terials, continue to be one of the leading factors ofthe process. It can be said that minimizing the la-bor hours and workmanship faults in the concreteshells became possible by replacing the material inthe Porsche Pavilion, on the other hand similar re-quirements have been provided by the selected pro-duction method for the ICD / ITKE pavilion.


It is now possible to observe that the produc-tion methods and materials used to participate inthe design process and play an active role in shap-ing the design instead of gaining importance dur-ing the construction phase of the structure. Begin-ning to use NURBS-based programs and support-ing these programswith related add-ons and simula-tions over time allowed for a more detailed accountof the relationship between material, structure, andform. The demand to produce digitally designedforms required improvement in production tools, onthe other hand, thematerials used and the structuralrequirements have begun to influence form simulta-neously. Although all studied examples have contin-uous experimental processes as a common feature,the concern of form, which is formed by the con-straints of constructability, has gradually begun toleave its place to thedevelopingprocess in the centreof specific materials and fabrication tools.

ACKNOWLEDGEMENTSWe would like to thank Meltem Aksoy, Ethem Gürerand Sema Alaçam for their constructive commentson earlier drafts of this study.

REFERENCESBechert, S, Knippers, J, Krieg, O D, Sonntag, D, Menges,

A and Schwinn, T 2016, ’Textile Fabrication Tech-niques for Timber Shells - Elastic Bendingof Custom-Laminated Veneer for Segmented Shell Construc-tion Systems’, in Adriaenssens, S, Gramazio, F,Kohler, M, Menges, A and Pauly, M (eds) 2016, Ad-vances in Architectural Geometry, vdf Hochschulver-lag AG an der ETH Zürich, pp. 154-169

Antuña Bernardo, J 2003, ’Gran mercado de Algeciras’, inSambricio, C (eds) 2003, Manuel Sánchez Arcas : Ar-quitecto, Fundación Caja de Arquitectos : FundaciónCOAM, Madrid, Spain, pp. 214-219

Burry, J and Burry, M 2010, The New Mathematics of Ar-chitecture, Thames & Hudson, London, UK

Cabrinha, M 2005 ’From Bézier to NURBS: IntegratingMaterial and Digital Techniques through a PlywoodShell’, Proceedings of the 2005 Annual Conference oftheAssociation forComputerAidedDesign inArchitec-ture, Savannah, Georgia, pp. 156-169

Casinello, P, Schlaich, M and Torroja, J. A. 2010, ’Félix Can-dela. In memorian (1910-1997). From thin concreteshells to the 21st century’s lightweight structures’,Informes de la Construcción, 62, pp. 5-26

Chilton, J 2010, ’Heinz Isler’s Infinite Spectrum FormFinding in Design’, Architectural Design, 80, pp. 64-71

Hudson, R, Schaefer, G, Kroeker, R, Forest, N and Bur-nay, D 2015 ’Subdivision SurfaceModeling to FosterResponsive Design Solutions in an IntegratedMulti-disciplinary Team’, Proceedings of the 33rd eCAADeConference, Vienna, Austria, pp. 403-413

Kolarevic, B (eds) 2003, Architecture in theDigital Age: De-sign andManufacturing, Taylor & Francis

Oxman, R 2012, ’Informed tectonics in material-baseddesign’, The International Journal of Design Studies,33, pp. 427-455

Oxman, R 2015 ’MFD: Material-Fabrication-Design: AClassification ofModels fromPrototyping toDesign’,Proceedings of the International Association for ShellAND Spatial Structures (IASS), Amsterdam, Nether-lands

Oxman, N, Soldevila, L M and Duro-Royo, J 2015 ’FormFollows Flow: A Material-driven ComputationalWorkflow For Digital Fabrication of Large-Scale Hi-erarchically Structured Objects’, Proceedings of theACADIA 2015 Conference – Computational Ecologies:Design in the Anthropocene., Cincinnati, USA

Pottmann, H, Asperl, A, Hofer, M and Kilian, A 2007, Ar-chitectural Geometry, Bentley Institute Press

[1] http://www.henn.com/en/research/monocoque-shell

s[2] https://www.iconeye.com/architecture/news/item/

9667-martin-luther-church-by-coop-himmelb-l-au

[3] http://www.archdaily.com/496202/ad-classics-los-manantiales-felix-candela/5340c230c07a809fab0000cd-ad-classics-los-manantiales-felix-candela-plan

[4] http://icd.uni-stuttgart.de/?p=12965[5] http://www.ce.jhu.edu/perspectives/protected/id

s/Index.php?location=Algeciras%20Market%20Hall

[6] http://web.media.mit.edu/˜neri/site/projects/monocoque1/monocoque1.html

[7] http://www.materialecology.com/projects/details/beast


Visual Thinking ThroughModel Making

Between the preferred and desired

Andrew N.J. McDonagh1, Huda S. Salman21,2The Scott Sutherland School of Architecture and Built Environment, The RobertGordon University, Aberdeen, UK.1,2{a.n.j.mcdonagh|h.salman}@rgu.ac.uk

Emerging technologies of the twenty-first century are drastically changing theway designs are produced, and adjusting the very processes of commonarchitectural practice. This creates uncertainty amongst architects as to theircurrent methods of designing and the effectiveness of those methods in aprofession now driven by technology. This paper gives context to such designmethods from a visual thinking perspective, addressing digital and physicalmodel making and various other design tools.Previous studies have investigatedsketching and its use by many designers as a tool for conceptualisation, beforedeveloping that concept through digital models.

Keywords:Working model, 3D Physical model, Digital modelling, Visualthinking

INTRODUCTIONThrough the current ‘technology age’, the place of ar-chitects in the construction industry is ever-shifting -perhaps due to our accessibility to digital modellingsoftware, perhaps due to its generally clear user in-terface and ease of use. In certain areas, it is believedthat many people have lost sight of the importanceof the architect in design. This, alongsidemany otherfactors, has led to an increase in the amount of self-build projects, and large-scale development compa-nies and volume housebuilders ‘designing’ our builtenvironment. From a personal point of view, the ar-chitectural model should be a rough tool for discov-ery, and there is only so much discovering that canbe done of a 3Dmodel on a computer screen, demar-cated by dimensions (Lawson and Loke, 1997). Thereare, however, two types of architectural model: “the

study model and the presentation model” (Suther-land, 1999). These two types compose a certain craftin architecture. Yet has this craft of physical modelmaking in the architect’s practice been replaced bya new digital one? The answer to that question isessential in defining the role of craft in practices to-day. On that point, this paper will consider the archi-tect’s place in the construction industry with regardsto their skills in model making and staying relevantin a world of ever- advancing technologies. This pa-per intends to reinstate the importanceof thatphrase‘design process’, and redirect the focus of designersback to that very process amidst the current upsurgeof technology. This continued rise in new technol-ogypresents thegreat variety ofmodellingprocessesthat impact the design process today. A study intothis topic would delineate the place of models in the


21st century design studio and discuss the impor-tance of visual thinking through physical as well asdigital modelling processes. “The variety of designprocesses that inform the fabrication of architectureis nowgreater thanever” (Dunn, 2012). Ahugeplayerin that is Building Information Modelling (BIM). This -alongside various other types of digital media - havethe potential to radically shift the way designs arecreated in the industry, not to mention the ways inwhich they could greatly support and extend creativ-ity in architecture. “As a term and method that israpidly gaining popularity, BIM is under the scrutinyof many building professionals questioning its po-tential benefits on their projects” (Barlish and Sulli-van, 2012). Thus, this paper will provide a balanceof both physical modelling and digital modelling, in-forming current and future decisions of whether ornot to adopt such digital design tools.

Aim andObjectivesA computer screen is only ever 2½D. So when youmake a 3D model from it, there is always somethingyou are not expecting. ”(Løddesøl, 2014). SnøhettaArchitects utilise a constant process of 3D modellingand 3D printing as an approach to visual designthinking: discussing, sketching and investigatingspatial form. Dophysicalmodels allowgreater auton-omy in this design development than 3D models? Istheremore freedom to think visually through a roughconceptual model before it has dimensions? Thisstudy aims to investigate these questions throughthe subject of ’visual thinking’ whilst examining var-ious processes of model making. For clarity, this pa-per will refer to handmade models as ”physical mod-els” and computer generatedmodels as ”digitalmod-els”. First the research will cover physical models, de-scribed as ”a vehicle for process, idea and reflection”(Voulgarelis and Morkel, 2010), of which there aretwo genres: the ”working model” and the ”presenta-tion model” (Sutherland, 1999). This study will theninvestigatedigital (computer-generated)models, be-fore comparing the two types and discussing whichis more effective with regards to that which is of con-

cern to this paper: the ”working model”. As a subjectof great personal interest, a study into architecturalmodel making would delineate the place of work-ing models in the design process. The time-frameof this paper also lent itself to the completion of astudy in the Stage 5 Architectural Design Studio atRobert Gordon University investigating various stu-dents’ use of this and other mediums in their designprocess. The overall aim of the paper is to comparethese two methods of modelling with regards to de-sign development at a conceptual level. This carriestwo objectives, each of which has an element in bothphysical modelling and in digital modelling. The ob-jectives are:

• to investigate, through a process of designreflection and visual thinking, which onemethod ofmodelmaking ismore pertinent todesigners (final year students) at a conceptuallevel.

• to establish towhat degree alternative designtools are being used in design conceptualisa-tion before being presented through physicaland computer-generated models.

Both objectives will be addressed initially throughthe studio- based project study and practice basedsurvey.

REVIEWWhat is visual design thinking?The creative process described in this section sug-gests a back and forth movement between the brainand the medium, and is highly reminiscent of Don-ald Schön’s description of the way an architect ‘holdsa conversation with the drawing’ (Schön, 1983). Inthe Review of Educational Research journal, a use-ful description is given: “design thinking is generallydefined as an analytic and creative process that en-gages a person in opportunities to experiment, cre-ate and prototype models, gather feedback, and re-design” (Razzouk and Shute, 2012). This activity isfurther emphasised in the paper ‘Problems, framesand design perspectives on designing’, where Schön


states that “when [design] moves function in an ex-ploratory way, the designer allows the situation to‘talk back’ to him, causing him to see things in a newway - to construct new meanings and intentions”(Schön, 1984). Offered here is an essential processof the designer, which engages the brain througha critical progression of design and redesign, but isthis process being used in twenty-first centurymodelmaking?Moore (2003) laments a similar idea that “theconcept of the visual [also] hinges on the idea thatthere is a direct connection between what we arelooking at and the manner in which we think, that‘external things’ are the causes of our ‘inner’ experi-ences” as “the mind takes the form of the object per-ceived, without its matter” (Putnam, 1999). Salman,Laing and Conniff (2006) “A designer’s ability to solvedesign problems depends on his/her ability to cre-ate a virtual world where visual thinking becomespossible and helps to externalise ideas of the dif-ferent design situations. This design world includessketches, diagrams, drawings and physical models”.When tackling a design problem, it is possible to con-nect with this world through a design conversation(Schön, 1991) with one’s preferred mediums and de-sign tools. Thus, through the designer’s use of physi-cal models to help solve that design problem, he/sheinteracts with a 3D world (Mitchell, 1990).

Visual Thinking in Physical ModellingWhat exactly is the craft of physical modelling? phys-ical models and their description act as “a vehicle forprocess, idea and reflection”. Voulgarelis and Morkel(2010) also observe that “the cognitive focus of thephysical model is that it enhances dialogue. Not onlydoes the model talk back, but it is an easy graphicform to access visually and verbally for both stu-dent and lecturer”. This creates a clear contrast tomodel making on the computer, and its lack of ‘ac-cessibility’ and engagement of the senses. Although3D printing has more recently changed this, digitalmodelling was previously only accessible “within a‘live’ computer where students can show the wholemodelwith ease. However, students tend toprint out

views that hide problematic issues - this is not possi-ble to achieve with a physical model.” These viewsestablished by Vougarelis and Morkel reminisce thework of Schön, regarding the ‘conversation’ that thedesigner holds with the medium. It is recognisedthat all mediums ‘talk back’ in different ways (Schön,1989; Breen et al, 2003; Lawson, 2015; Mitchell, 1990;Salman, Laing and Conniff 2014).Another observa-tional study (Akalin, 2003), stated that “whenconceptmodels or muck-ups are considered, such tasks in-volved inmodellingmay be regarded as away of rea-soning just like sketching”. This quotation suggestsan analogous back and forth motion between mindand medium. There is no function of a design tool,unless it aids critical analysis and communicates tothe brain feedback of the design created with thattool. This in turn begs the question of the effective-ness ofmodelmaking as a creative tool for designde-velopment. Undoubtedly it will depend on the stu-dio, and the designer (Lawson and Loke, 1997), butis there a better balance that can be found in thearchitecture profession today than sketching a con-cept before developing and representing it throughphysical and computer-generated models? This pa-per aims to act as an instrument in striking that bal-ance, but research must first be done into the digitalrealm.

Visual Thinking in Digital Modelling“Visual design thinking is performed through three-dimensional digital models. Designing in a three-dimensional digital environmentmight be describedas sketching in space” (Abdelhameed, 2004). Al-though this type ofmodelmaking suggests onewithmore complexities, it also suggests one with lessboundaries andmore liberty.Is there anythingwe cantake from physical modelling into the new digitalcraft? Jordan Brandt writes in the book ‘PersistentModelling’ that “craft is marked by the mind relatingthe purpose of the work to the motions of the hand”(Ayres, 2012) and concludes by saying “the use andcreation of our technology could well be informedby the process of craft. So instead of authoring a


deterministic model, how do we craft the modellingprocess to accommodate a persistent continuum?”This suggests that digital modelling systems shouldbe ever-advancing, meaning at a design level such asystem should enable us to expand, through a pro-cess of imagining, what we can do in the digital mod-elling realm.“Tools in the more literal sense, like pen-cil, compasses and ruler, say next to nothing aboutthe designs devised with their assistance” (Gänshirt,2007). The same could be said about Building In-formation Modelling (BIM). It could be said that atthe end of the day it is the end product that deter-mines the effectiveness of that tool, not how it wasarrived at. There is, however, an evident increase inthe various avenues that a design may take, and thatis where potential lies for further exploration of vi-sual design thinking. Bearing in mind that each de-signer designsmore effectively usingparticular tools,a notable area for investigation is whether or notthis digital exploration is, or can be, for everyone.Itis argued by Lawson and Loke (1997), however, that“present CAD tools do not support the kind of vague-ness and uncertainty that those manual conceptualsketches allow and thus often prematurely fixate orcrystallise developing design concepts.” Or, in usingthe language of Donald Schön, “such CAD drawingsare insufficiently conversational but seem more likeimperative statements made by the computer leav-ing little or no room for further contributions fromthe designer” (Lawson and Loke, 1997).With regardsto design exploration, Peter Rowe (1987) notes that“there are many different styles of decision-making,each with individual quirks as well as manifestationsof common characteristics”. Goel (1995) further de-scribes the design process as being a task of com-plex cognition. It is evident then, that this processis different for every designer and his or her under-standing of themedium, as Abdelhameed (2004) ob-serves that “what an architect can conceive and com-prehend depends on what this architect can visuallyperceive through the media used”. Designers haveand will always have different skills and capacities;therefore it is unlikely that digital modelling will be-

come customary to architectural design. However, inan attempt to accommodate personal preference, al-ternativemediumsofdesignareoffered throughpro-grams such as Google SketchUp, Adobe Photoshop,andAutodeskAutoCAD. Breen (2004) informsus suchsoftware has allowed students andpractices alike theopportunity to personify working methods (Achten2003; Achten and Reymen 2005) according to theirstyles, preferences and abilities.

Craft and Technology“Computer-assisted design might serve as an em-blem of a large challenge faced by modern society:how to think like craftsmen in making good use oftechnology” (Sennett, 2008). There are such archi-tects as Greg Lynn, who seek to expand the use ofcomputers in design, through research, educationand practice. Such acts are essential to the existenceof a ‘digi-craft’, and perhaps also essential to the rele-vance of architects throughout the 21st century. Inhis article, ‘Design Course Goes Digital’, John Marxemphasises that “recent advances in computer hard-ware and software have opened opportunities fora digital design process that does not diminish butrather enhances creativity” (Marx, 2000). This out-lines a point essential to the formation of a digitalcraft; creativity. But is that creativity using physicalor digital design tools¿‘The word ‘tool’ might invokethe mechanical, rather than the digital age. Yet ‘tool’was used for representing both action and thinking,as in ‘thinking tool’, and therefore the notion of con-ceptual tools covered both physical and cognitive ac-tivities” (Jonson, 2005). These conceptual tools havealways stretched design exploration, whether physi-cal or cognitive. The architect started out designingusing the pencil (the design tool) to create sketches(the medium), forming the gesture that produces aconcept for the rest of the design to follow. Then astechnology developed, his design tools changed toperspectivedrawings, tophysicalmodels, andnowto3Dmodelling. Regardless of thedevelopmentof newdesign tools, software etc., the architecture industrywould still be designing andmaking, using whatever


tool or media is available to them.With regard to therange of new design tools available to architects to-day, Nick Dunn (2012) offers a strong background tohistoric digital modelling processes and their morerecent proliferation in architecture in his book titled‘Digital Fabrication in Architecture’. “CAD/CAM pro-cesses have been used in engineering and indus-trial design for over 50 years in the developmentand fabrication of cars, aeroplanes and smaller con-sumer goods. Components are usually designed anddeveloped with three-dimensional modelling soft-ware, and then scale models are produced using arapid prototyping process that translates digital in-formation into physical object” (Dunn, 2012). How-ever, with regard to these computer-generatedmod-elswhichwerediscussed earlier as beingdemarcatedby dimensions, and hence too rigid to fulfil its func-tion as a development model. It is argued by Law-son and Loke (1997), that “present CAD tools do notsupport the kind of vagueness and uncertainty thatthosemanual conceptual sketches allow and thus of-ten prematurely fixate or crystallise developing de-sign concepts.” Or, in using the language of DonaldSchön, “such CAD drawings are insufficiently conver-sational but seem more like imperative statementsmade by the computer leaving little or no roomfor further contributions from the designer” (Lawsonand Loke, 1997). How is it that one model inspiresa thought in the mind, which in turn creates a ‘de-velopment’ of models?The question is then, are themedia and tools in common use today allowing thatprecision and elegant design required in order to en-dorse this newdigital craft, or are they just equippingus to get the job done faster? It is deemed by somethat computing as a tool is unsuitable for conceptual-isation (Lawson and Loke, 1997). It has been shown,however, that CAD is actually a conceptual tool usedto foster new methods in the perception and con-ceivability of design (Jonson, 2005). “Arguably, then,the view that CAD is inappropriate for conceptualis-ing seems to be based on a preconception of con-ceptual tools as surface, rather than deep structures”(Jonson, 2005).In the methodology, various concep-

tual tools will be investigated with relation to craftin order to determine that which is most appropri-ate to design ideation. In related studies, sketchinghas traditionally been considered a core conceptualtool (Schön, 1983; Bilda and Demirkan, 2003; Jonson,2005). However, in the last ten years, significant in-crease has been evident in the use of digital modelmaking at both university level and practice level de-sign studios. Thus, as a new digital craft matures,what is now the norm for conceptual ideation maybe significantly different in the architectural designstudio than what it was a decade ago.

METHODOLOGYThis paper research methodology aims to observethe conversational aspect of a design “process” (iter-ation cycles, ideation). Design process acts as a richsource for reflection, which pertains to the reflectiveaspect of design (referring to Schön’s 1987) theoryof conversation with the material. At the same time,reflection has a methodological aspect and can beused as a research method in design as well as otherfields, whether the investigated setting is practice oreducation (Salman 2011). Reflection can occur alongwith designing (concurrently) or after reaching a sat-isfied result (retrospectively). Design process studiesinspired the empirical element of this research to em-phasise two paradigms of design research method-ology: the constructive and the reflective-practice.Case study research is widespread in social sciencesand education (Scott and Morrison 2006). Accord-ing to Robson (2002) a case study is: “a strategy fordoing research which involves an empirical investi-gation of a particular phenomenon within its real-life context using multiple sources of evidence”. An-other mentioned feature was its flexibility as an ap-proach to combine quantitative methods and qual-itative methods in one study. Moreover, Scott andMorrison (2006) state that the main focus of a casestudy in educational research is to give the case ac-tors a ‘voice’ in such a way that the researcher’s rolemay sound passive to the benefit of the case actors.


Figure 1StudentParticipation inStudio Study

SampleWith respect to the study context, the choice of sam-ple was (Master of Architecture) final year students.The choicewasmade to reduce the differences in de-sign cognitive abilities and to have cohesive under-standing of the skill level and its impact on the usedmedia (Salman2011). Usingdesignmediumswithnoexperience seems to reduce the possibilities for de-signing (Coyne, Park and Wiszniewski 2002; Jonson2005). However, there is a high probability that thisstudy sample may represent part of a bigger popula-tion within UK schools of architecture that embracea similar pedagogical approach. At the same time,the results would deviate if the same study were tobe employed in schools of architecture which em-brace digital theories and computation. The samplewas final year students (Master of Architecture) co-hort 2014-2015, who studied architecture for at leastfour years and used physical modelling and digitalmodelling for problem solving and presentation.

Data CollectionMethodDesign studies previously have maintained a focuson sketching, its impact on designing and have en-gaged less emphasis on reflection and visual think-ing with regard to digital and physical model mak-ing, with little concern for design tools in the cur-rent digital age. The research instrument was usedto collect such data was adapted from Jonson (2005)self reporting study. The study matrix was designedin a similar manner to that of Jonson study (2005),

with students being instructed to select the medi-ums (or design tools) used in their conceptual stagedesign process from: sketching [S], words (spokenor written) [W], physical modelling [PM] and digitalmodelling [DM]. Each time the participant switchedto a different medium, a new box was ticked, fol-lowed by a brief description of (a) the task under-taken through that medium, and (b) any notable ob-servations, landmark events, or “Aha!” moments ex-perienced through this medium. Any other designtools used, of which there were none, were to beincluded as a numbered footnote at the bottom ofthe page. Students were also advised that, due tothe nature of the model making study, this matrixwas to be completed only when using either physi-cal or digital modelling (in conjunction with anothermedium). An example is shown in Figure 2. The in-formantswere also asked to tick abox indicating theirpreferences for sketching (either 2D or 3D), andMod-elling (either physical or digital). The data was gath-ered through their daily studio practices by jottingdown the most frequent conceptual tool they hadused.Students who made notable observations withrelevance to the literature, or had particularly inter-esting visual design processes were then questionedthrough follow-up interviews. The collection of theirthoughts after the studio study regarding the designprocess, or “reflection-on-action” (Schön, 1983), pro-vided a rationale for the quantitative data received.In these instances, the self-report became a prelimi-nary to interviewing (Burgess, 1981), revealing pointsfor speculation and helping to regulate the authen-ticity of the results (Jonson, 2005). These interviewswill be semi-structured, using several guided ques-tions. The students selected for these interviews in-clude one preferring digital model making and onewho was favourable of physical model making, of-fering a balanced discussion of the subject area andthose mediums concerned within it.


Figure 2Studio Study Matrix(Adapted fromJonson 2005)

StudioMatrix Findings andDiscussionA total of 10 students volunteered in the self-reporting study, providing a round number for sub-sequent analysis. The studywas undertaken over thecourse of a three-week period, and responses com-pleted by participants when using digital or physicalmodelling over that time-frame. Then, the self re-porting matrix and semi structured reflections wereanalysed using descriptive statistics and frequencyanalysis (Field 2005), this is presented in percentagesand reported in this section.

Self-reporting analysis in Figure 3 shows thatmore students (60%) preferred to use 2D sketchingover 3D sketching in exploring their early ideas. Simi-larly, analysis also shows that 60%of students in pref-erence of physicalmodelmaking over digital. This ar-guably shows that the majority of final year studentsare still using a combination of 2D and 3D in physi-cal format rather than digital. But others (40%) have

shifted this trend towards 3D digital tools for designthinking. On the level of the individual skills, the find-ings refer dominantly to participants’ habitual designmethods as one constraint that affected their use oftheir digital skills in its full computational capacity.

Figure 3Sample’s preferedmediums

In brief, whether the produced artefacts (model,drawing) would affect the student design conceptu-ally, students separated what is analytical from whatis regarded as a complementary act (drawing activ-ity). In other words, 2D sketching and 3D physicalmodelling has a strong relation with authorship andidentity, bearing in mind that drawing brings under-standing, and from understanding students can de-velop more ideas. However the matrix analysis alsorevealed that sketching [S] has been identified bythe students as the primary conceptual tool . Thesefindings also challenged the view that digital mod-elling is an inappropriatemedium for conceptual de-sign. What has been observed historically in con-ceptualisation studies is that verbalisation [W], “on itsown or in combination with other conceptual tools,emerged as the prime mover for getting started”(Jonson, 2005). It is apparent in this research, how-ever, that conceptualisation in recent years has beensomewhat different.

A number of tasks were undertaken by thestudents who used digital modelling -Sketchup,which was used more prominently for testing de-sign ideas,generating site models and also to try andachieve a sense of scale. On the other hand, phys-ical modelling was most commonly used for creat-ing site models and performing massing studies ofexisting surroundings and also proposed develop-ments; most seemingly with presentation in mind asopposed to development. The only other instance ofits use was in exploring spatial relationships for one


student’s precedent study.

Figure 4Tasks UndertakenThrough DigitalModel Making

It was apparent that more subjects preferring physi-cal modelling used digital modelling than those pre-ferring digital modelling using physical modelling.This trend, backed up by student feedback, suggestspressures of cost and time upon the students, influ-encing their decision to use digital modelling, whichis said by students to be “quicker and easier to vi-sualise”. Various disadvantages of digital modellingwere also noted by the students, with one student re-marking with regards to speed, “it takes time to com-pile enough information to gain a sense of characteruntil themodel is more developed”. Another studentalso noted that it wasn’t very helpful in developing aconceptual idea.

Practitioners Survey Findings and Discus-sionThis section contained a brief background to thestudy andadescriptionof visual design thinking. Par-ticipants were asked to indicate their model makingpreference for the conceptual stage of design. It wasshown in this study that 70% of the practitioners sur-veyed preferred digital model making for design inthe conceptual stage, and four participants preferrednot to answer due to such modelling processes notbeing used in their practice. Despite the results infavour of digitalmodelmaking, seventeenpractition-ers felt that physical models were more assistive fordesign development , but still used digital due totime and cost constraints, as well as flexibility furtheralong the design process.

A common view was that physical models aregreater for conceptual massing, form design andease of communication to others, whilst digital mod-els have their advantage when it comes to modifi-cation, contained information and the exploration ofmultiple solutions.

Figure 5PractitionersPrefered Mediums

With regard to the frequency of modelling use inpractice; the first relating to digital (3D CAD) mod-els in the early stages of the design process and thesecond relating to physical (handmade)modelling. Itcan be seen how drastically the design processes ofarchitecture practices have changed in recent years,showing that not a single practice in the sample al-ways create physical models, however, 37.5% (andever-increasing) percentage of practitioners alwaysutilise digital modelling software in the early stages.The two data sets are almost symmetric suggestingthat through the increase of digital modelling usage,there is a subsequent decrease in physical modellingusage, showing that practices are perhaps utilisingindividual specific mediums, rather than a range ofdesign tools.

Figure 6Digital & PhysicalModelling Usage inEarly Stages ofDesign Process

The findings show that there has been a significantneglect of the craft of physical modelling in the ar-chitecture practice in recent years. Respondents of


this question outlined similar reasons to those of the‘Frequency of Use’ section. The recurring theme oftimewas recordedprimarily, but secondly reasons in-cluded factors such as lack of flexibility, cost, lack ofportability and storage, as well as featuring a higherquality and more resourceful output.

CONCLUSIONSA challenge of the future, however, will be in retain-ing focus on actual design. Undoubtedly many stu-dents are pursuing an architecture course with greatfamiliarity of computers, and aspirations of 3D print-ers and technology, therefore the task for such edu-cational bodies will be to try and develop a hungerfor good design first, retaining this as the central-ity of the architecture course in the 21st century. Insummary, it is important that architects must alwaysprogress with ever-advancing technology, but neversubstitute it for the brain. As a subsequent of thedaily challenges faced by designers, their problemsolving and resourceful strategic thinking skills aresecond to none. Therefore, we must use those pro-ficiencies to benefit the world around us today inwhatever way we can. By a process of visual thinkingand design reflection, we as an industry must con-tinue to imagine and create great spaces and build-ings using a range of design tools, and whatevertechnology helps us do so most efficiently.

Some architecture schools are advancing fasterthan others, but all operate differently. Therefore,further research should be focused on a wider pop-ulation to gain more representative inferences andconclusions, perhaps including architecture schoolsacross the UK. This would minimise the subjectivityof such research by allowing comparative studies totake place. A more detailed study may also be ben-eficial in establishing what must be done in adapt-ing education systems in accordance with changesemerging throughout the new ‘digital era’ facing ar-chitectural practice.

REFERENCESAyres, P (eds) 2012, Persistent Modelling: Extending the

Role of Architectural Representation., RoutledgeBarlish, K and Sullivan, K 2012, ’How tomeasure the ben-

efits of BIM – A case study approach’, Automation inConstruction, 24, pp. 149-159

Bilda, Z and Demirkan, H 2003, ’An insight on designers’sketching activities in traditional versus digital me-dia’, Design Studies, 24, p. 27–50

Breen, J, Nottrot, R and Stellingwerff, M 2003, ’Tangi-ble virtuality—perceptions of computer-aided andphysical modelling’, Automation in Construction,6(12), pp. 649-653.

Coyne, R, Park, H and Wiszniewski, D 2002, ’Design de-vices: digital drawing and the pursuit of difference’,Design Studies, 3(23), pp. pp. 263-286.

Dunn, N 2012, Digital Fabrication in Architecture, Lau-rence King Publishing Limited

Field, A 2005, Discovering statistics using SPSS, SageJonson, B 2005, ’Design ideation: the conceptual sketch

in the digital age.’, Design Studies, 6(26), pp. 613-624Lawson, B 2015,What Designers Know, RoutledgeLawson, B and Loke, SM 1997, ’Computers, words and

pictures.’, Design Studies, 2(18), p. 171–184.McDonagh, ANJ 2015, Visual Thinking Through Model

Making, Master’s Thesis, RGUMitchell, WJ 1990, The Logic of Architecture: Design, Com-

putation, and Cognition, MIT Press.Moore, K 2003, ’Overlooking the visual’, The Journal of Ar-

chitecture, 8, pp. 25-40Putnam, H 1999, The Threefold Cord; Mind, Body and

World, Colombia University Press, New York, Chich-ester, Surrey

Razzouk, R and Shute, V 2012, ’What is design thinkingand why is it important?’, Review of Educational Re-search, 82, pp. 330-348

Robson, C 2002, Real world research: a resource for socialscientists andpractitioner researchers, Blackwell pub-lishing, Oxford

Salman, HS 2011, The impact of CAAD on design method-ology and visual thinking in architectural education,Ph.D. Thesis, RGU

Salman, HS, Laing, R and Conniff, A 2006 ’CAAD Visual-ization Techniques Mediate the Conceptual DesignProcess as a Thinking Tool.’, Proceedings of eCAADe2006, p. 701

Salman, HS, Laing, R and Conniff, A 2014, ’The impact ofcomputer aided architectural design programs onconceptual design in an educational context.’, De-sign Studies, 35, pp. 412-439.

Schön, DA1983, The reflectivepractitioner, Temple Smith,London

Schön, DA 1984, ’Problems, frames and perspectives on


designing’, Design Studies, 5 , pp. 132-136Schön, DA 1991, The Reflective Practitioner: how profes-

sionals think in action, Harper Collins., New YorkScott, D and Morrison, M 2006, Key ideas in educational

research, Continuum, UKSutherland, M 1999,Modelmaking., Norton., New YorkVoulgarelis, H and Morkel, J 2010 ’The Importance of

Physically Built Working Models in Design Teach-ing of Undergraduate Architectural Students’, Con-nected 2010, 2nd International Conference on DesignEducation, University of New South Wales, Sydney,Australia, p. 2


Modular Mycelia

Scaling Fungal Growth for Architectural Assembly

Sean Campbell1, David Correa2, Dylan Wood3, Achim Menges41,2,3,4ICD, University of [email protected] [email protected],4{dylan-marx.wood|achim.menges}@icd.uni-stuttgart.de

This paper covers the findings of a master's level investigation into the potentialof fungal-based architecture. The overarching hypothesis of the research is thatfungal growth can be scaled for architectural application through thedevelopment of a multi-scalar assembly system. This investigation was conductedin three stages. In the first stage, fungal growth patterns were tested on a varietyof substrates, at the petri dish scale. In the second stage, findings from thesegrowth tests were used as drivers for the development of an assembly unit. In thethird stage, combinations of these units were arranged to produce largerelements. These elements were then used to build a proof of concept prototype.The final prototype demonstrates the strengths of the multi-scalar system, whilefurther outlining the technical challenges of using fungal growth as a means ofarchitectural assembly.

Keywords: fungal-based architecture, modular mycelia, synthetic biology

INTRODUCTION:There are a number of companies currently produc-ing fungus-based materials as sustainable alterna-tives to conventional building materials. Fungi’s ca-pacity to grow on agricultural waste, it’s low-energyand carbon neutral requirements for production, aswell as, its inherent biodegradability make it ideal asa sustainable alternative to current building materi-als.(Travaglini et al. 2016) Fungi also have a rangeof mechanical properties that make them potentiallywell-suited for building technology. (Travaglini et al.2016)Despite the sustainable, economical, and struc-tural benefits, there is an inherent problem in us-ing fungal-based materials simply as an alternativeto standard building materials. The problem with

this approach is that fungus-based materials retainthe assembly techniques of their traditional counter-parts. Fungal-based bricks are stacked in the sameway as conventional bricks; they require just asmuchlabour and just as much mortar. A new method ofconstruction couldemerge simplyby leaving the fun-gal bricks alive during assembly, rather than killingthe organism after the brick is formed, which rendersthe material inert. While this approach adds a levelof control and consistency to thematerial, it neglectsfungi’s capacity for self-organized growth. (Websterand Weber 2007) And, more importantly, it neglectsfungi’s potential to challenge conventional methodsof construction.


Figure 2Different Stages ofFungal-BasedMulti-ScalarAssembly System

The aim of this research project is to investigate thepotential of fungal-based architecture through thedevelopmentof amulti-scalar assembly system. Scal-ability is a principal characteristic of fungal growth.The processes that determine the interconnectiv-

ity of a fungi’s mycelium at the micrometer are thesame processes that determine the form of a fungithat stretches kilometres.(Davidson 2007) When in-corporated into an assembly system, these samepro-cesses couldbeexploited tomanufacture the individ-ual elements of a structure, then exploited again tobind the overall structure together. Despite the scal-ability of fungal growth, challenges emerge whenfungi are grown in large volumes of substrate. Main-taining optimal environmental conditions(moisture,temperature, nutrition) and preventing contamina-tion become more difficult as the volume of growthincreases. Given the heterogenous nature of fun-gal colonization, it is also difficult to maintain con-sistent growth throughout large volumes of sub-strate.(Boswell et al. 2007) The overarching hypothe-sis of the research is that a multi-scalar assembly sys-tem could resolve the challenges inherent in exploit-ing fungal growth for architectural application.

There are a number of opportunities presentedby embedding biology processes into architecture.

Figure 1Process Diagram ofMulti-ScalarAssembly System


Natural systems exhibit a range of functions thatare not found in current construction. These func-tions include adaptation, resilience, self-repair, self-organization.(Imhof and Gruber 2015) Each functionrepresents a novel way to engage the built environ-ment. It is important to note that the benefits of im-buing architecture with natural functions go muchfurther than novelty. Natural systems tend to bevery efficient in their use of energy and resources.(Imhof and Gruber 2015) The potential benefits forarchitecture range from the economical to the sus-tainable. There are a number of researchers who arecurrently exploring the potential of engineering bi-ological systems for architectural application. Theseresearchers include Rachel Armstrong, who is cred-ited with projects such as Hylozoic Ground and Fu-ture Venice, as well as, Barbara Imhof and Petra Gru-ber, who direct GrAB (Growing as Building) an on-going research project that uses patterns and dy-namics found in nature and applies them to architec-ture. Through their work, these researchers are effec-tively applying the tools of synthetic biology to archi-tecture. Synthetic biology is an interdisciplinary ap-proach to designing biological processes for practi-cal applications, which”offers new ways to combinethe advantages of living systemswith the robustnessof traditionalmaterials to produce genuinely sustain-able and environmentally responsive architecture.”(Armstrong and Spiller 2010)

The Hy-Fi Tower, by The Living (David Ben-jamin), Polyominoes, by Mycoworks (Phillip Ross),and Mycelium Tectonics, by Gianluca Tabellini, arethree precedents of the current research project. Attwelve meters high, the Hy-Fi tower is one of thelargest andmost resolved fungus-basedarchitecturalproject to date. 10,000 fungal-based bricks weregrown by Evocative, a biomaterials company, for theproject.(Holcim 2015) The Hy-Fi tower is an exampleof a fungal-based project that engages constructionat the material, manufacturing, and waste stages ofa building life cycle. At the material stage, local agri-cultural waste (corn stalks) was chosen as a substrate.The shredded corn stalks were the inoculated with

mycelia and packed into moulds.(Holcim 2015) Afterthe mycelia had fully grown through the substrate,it was killed, producing strong, but inert bricks. Thebrickswere then assembled into the tower using tim-ber scaffolding as form work and mortar to hold thebricks together. The bricks were then composted af-ter the tower was disassembled. (Holcim 2015)

In contrast we have Polyominoes, an ongoing re-search project by Mycoworks. From an architecturalperspective, Polyominoes is a less resolved projectthan the HyFi tower. As the project title suggests,these blocks take the form of a polyomino, which is ashapemadeby combining a number of squares edgeto edge. Polyominoes are often used for problems orpuzzles that involve tiling a plane. Tetris is a popularexample. In this project, the blocks are made from amix of mycelium and sawdust, which is packed intoa mould. The shape and scale of the mould is de-termined by the dimensions of standardized lumber.After the mycelium grows through the sawdust, theblocks are removed from their moulds and are put inclose physical contact.(Mycoworks 2015) The blocksthen fuse through hyphal bonding, and a continuousobject is formed.

The third precedent is a thesis project titledMycelium Tectonics, which was conducted at theUniversity of Bologna by Gianluca Tabellini. Similarto Mycowork’s Polyominoes, Tabellini’s research ex-plored fungi’s potential to create structures. For theproject’s final prototypes, Tabellini designed hang-ingmodelsmadewithmycelia infusedhemp strands.The strands are held by a small framemade from ply-wood and threaded rods. Cotton balls are also addedto the frame after they are soaked in water.(Tabellini2015) After the cotton balls and hemp fibres are se-cured in the frame, the prototype is wrapped in plas-tic to maintain a high moisture level. (Tabellini 2015)As the fungi’s mycelia grows through the hemp fi-bres, the hanging strands become structural. Afterthe mycelium has fully grown through the strands,the supporting rods are removed. Unlike the previ-ous precedents, which tended to grow themyceliumas a homogenous mass, here, fungi’s natural growth


processes are used to create a structure with variablematerial densities. It is also important to note thata mould was not used to determine the final shapeof the prototype. Instead, the prototype’s final formwas determined by the arrangement of the substrate(hemp strands).

These precedents represent three approaches totwo important variables in fungal-basedarchitecture:architectural viability andbiological utility. The archi-tectural viability of a project ismeasuredby its capac-ity to function as a inhabitable structure. Whereas,the biological utility of a project is measured by itscapacity to exploit the biological processes of fungi.At one end of the scale is the HyFi Tower, which hasa high score for architectural viability due to its scaleand its capacity to function as an architectural struc-ture for three months before it was disassembled .However, it has a relatively low score for biologicalutility because the role of fungi in theprojectwas lim-ited to producing a new kind of compostable brick.The biological processes of the fungi did not con-tribute to the form of the tower or the method of as-sembly. At the other end of the scale is MyceliumTectonics, which has a high score for biological util-ity because fungal growth is used to add structure toa hanging model. However, due the scale of the fi-nal prototypes and, more importantly, the inherentissues of scaling the process, its architectural viabil-ity is relatively low. Polyominoes falls between thetwo other precedents in terms of these two variables.While it has not produced an inhabitable structure,the mycelial blocks could easily be incorporated intoa functional architectural element. For biological util-ity, theproject is a clear demonstrationof howhyphalbonding can be used as a means of assembly, eventhough the design of the blocks themselves are de-termined by the dimensions of standardized lumber.The current research will attempt to balance thesetwo variables in a similar fashion. It will also attemptto reach a similar level of resolution. The goal of thecurrent research is not be to produce functional ar-chitecturemade from fungus. Instead, thegoal of theresearch is to demonstrate the strengths of a multi-

scalar assembly system, while outlining the technicalchallenges of using fungal growth as a means of ar-chitectural assembly.

STAGE ONE: GROWTH EXPERIMENTS:

Figure 3Fungal GrowthExperiments on aVariety ofSubstrates

The first stage consisted of a series of growth experi-ments. These tests were used to gain amore intimateunderstanding of fungal growth processes. Pleuro-tus ostreatus, a species of basidiomycetes, was cho-sen for these experiments. Pleurotus ostreatus waschosen for its growth speed, its resilience to environ-mental condition and its adaptability to various sub-strates. The colonization patterns of Pleurotus os-treatus were tested on a variety of substrates. Thesesubstrates included organic material, such as agar,seed, straw and wood, and inorganic materials, suchas sand and plastic, as well as mixtures of both or-ganic and inorganic materials. The substrates werechosen to test how key parameters, such as nutrientcontent, moisture content, density and geometry,would affect the colonization patterns of the fungus.Due to the challenge presented by contaminants, af-ter an appropriate substrate has been chosen, itmustbe pasteurized in an autoclave. The requirements oftemperature, pressure and time will all be specific tothe substrate chosen. It is important to note that the


substrate will not be fully sterile after it is pasteurizedin the autoclave. It will, however, reduce the amountof competing organisms, which will increase the suc-cess rate of colonization for the chosen fungus. Afterthe substrate is pasteurized, it is put into a petri dishand inoculated with Pleurotus ostreatus at a centralpoint. A HEPA filter was used to remove the contam-inants in the air so they will not be introduced to thesubstrate during the inoculation process. The inoc-ulated petri dishes were then put into a grow cham-ber. The temperature of the grow chamber is set to22 degrees Celsius. Though the fungi would growmore quickly at a higher temperature, it would alsoincrease the chances of contamination. The results ofthe growth experiments were used as design driversfor the assembly unit. The seed substratewas chosenamong the other substrates because it produced thethickest fungal growth the quickest. An importantobservation fromthese experimentswas that despitethe differences in geometry and nutritional contentof the substrates used, the fungi maintained a radialgrowthpattern. A spherical shapewas chosen for theassembly unit to match these radial growth patternsand to maximize the efficiency of colonization.

STAGE TWO: FUNGAL ASSEMBLY UNITS:

Figure 4Growth Containerof the AssemblyUnit

Figure 5Assembly Unit AtThree DifferentScales

A second set of experiments was used to determinethe the method for production, the design of thegrowth container and the size of the assembly unit,.The inoculation process for the substrate is the sameas the process used in the first experiment. Thespherical containers were made from a thin clearplastic that was not autoclavable, so the substratehad to be pasteurized separately, then added to thecontainers once it had cooled. The containers weresterilized separately usingethanol anda cotton swab.Once the substrate was added to the spherical con-tainer, it was then inoculated with the fungi froma central point. A hole was added to both ends ofthe container for gas exchange. The containers werethen wrapped in parafilm to prevent competing or-ganisms from contaminating the substrate. The in-oculated containers were then put into a grow cham-ber, whose temperature was set to 22 degrees Cel-sius. The assembly unit was tested at three differ-ent diameters (40, 90 and 220millimetres). Althoughassembly units were successfully grown at all threesizes, a diameter of 40 mm was chosen. This choiceof scale was predicated by the space available in thegrow chamber.


STAGE THREE: LARGER ASSEMBLY:

Figure 6Fused AssemblyElements

Figure 7Stack of AssemblyElements

In the third stage of the research project, the assem-bly units were combined in different arrangementsto produce larger assembly elements. The benefit ofcreating assembly elements from the individual unitsis that they produce different functions. For example,where individual units could be used as a pourableinfill, a triangular element made of three combinedunits couldbe stacked. Theuseof stackable elements

would remove the need for rigid formworkwhen cre-ating larger assemblies, as in the case of a wall. Toproduce these larger elements, the spherical assem-bly units were first removed from their containers. AHEPA filter was used as a precaution to minimize air-borne contaminants, though the assembly units areresilient to competing organisms once they are fullycolonized. After they are removed from their contain-ers, the assembly units are arranged on acrylic traysthat have been sterilized. The acrylic trays are cutwith rows of circles, which are used to hold the spher-ical assembly units in place while they fuse. After theassembly units are arranged, the trays are wrappedin plastic to create amoisture sealing envelope. Thenthe trays were stored in a grow chamber (22 degreesCelsius).

DISCUSSIONThe speed of growth is pivotal in the colonizationof a substrate. Even after pasteurization, substrateswill often contain minute amounts of competing or-ganisms. If the chosen fungi grows too slowly, thecompeting organisms can overrun the substrate. Animportant factor in the success of the seed sub-stratewas its geometry. When aggregated, the seedswould produce gaps that promoted the transfer ofmoisture and oxygen through the substrate. Thesegaps also provide the fungi with a growth matrix.In contrast, the fungi had difficulty growing throughdenser substrates, like fine sawdust, which became athick paste after pasteurization.


Figure 8Acrylic Tray with aCollection ofAssembly Elements

Figure 9Collection ofAssembly Units

The use of assembly units effectively discretizes fun-gal growth. This approachmay be a suitable methodfor up-scaling fungal growth processes for architec-tural applications. One of the challenges that thisapproach addresses is the heterogenous nature offungal growth. Since each assembly unit is inocu-latedwith fungus froma central point, the funguswillbe more evenly distributed throughout a larger as-sembly when the units are aggregated. Much moreevenly, at least, than if an aggregation of substratewas inoculated at a single point. Also, by discretizingfungal growth, contaminated spheres can be easily

removed and cleaned. It would be very difficult tocontrol the growth of a contaminant in a larger as-sembly. Another benefit to discretizing the fungalgrowth process is that the different growth speedscan be accounted for. The spherical shape of the as-sembly unit provide a specific benefit for use in largerassemblies. The gaps provided by their geometry actas a growth matrix (much like the seeds did at thesmaller scale). These gaps also allowmoisture and airto be distributed more evenly throughout the largerassembly. The fungi’s ability to grow across thesegaps solves the potential problemofminimal surfacecontact.

It is important to note that the larger elementsareproducedby the samegrowthprocesses that pro-duced the individual units. The network of hyphaethat binds the individual unit is the same networkthat holds the larger elements together. The largerelements are produced without adding new mate-rial and with only minimal formwork. Once the in-dividual units are fused into larger elements (whichonly requires a few days) they have to be removedfrom the tray or themyceliawill overrun the tray. Thisexcessive growth will make the larger elements dif-ficult to remove from the tray cleanly. The require-ment for the prompt removal of the larger elementsworks to the advantage of the overall assembly sys-tem. The need to continually remove fused elements


creates space for new elements on the tray. As newelements are added to the tray, the spherical contain-ers can be reused to produce newunits. The requiredremoval of the elements essentially turns the multi-scalar assembly system into a iterative process. (asseen in the overall process diagram) This iterative na-ture also proved to bean advantage in the last step ofthe assembly system, when the elements are stackedto form an architectural component, like awall. Sincethe elements are stacked gradually, they have timeto fuse, making them more stable and solid, beforenewelements are added. This iterative stacking elim-inates the need for rigid formwork. However, a mois-ture barrier is still required tomaintain a high level ofhumidity.

Contamination proved to be an issue when as-sembling the final prototype. To add new elementsto the prototype, the moisture barrier had to beopened, which created an opportunity for compet-ing organisms to be introduced. The unintentionalintroduction of competing organisms is a constantchallenge when working with fungi. Contaminantswill grow in identical conditions, and will generallygrow much faster than the intended fungi. (Websterand Weber 2007) Since the assembly elements werefully colonized when they were added to the finalprototype, the assumption was that they would beable to fend off competing organisms. And they did,insofar as the contaminants were only able to growalong the surface of the elements. An attempt wasmade to remove the contaminants using ethanol anda wipe, but it was unsuccessful. Contaminants arealso very difficult to contain or remove once they areintroduced. In hindsight, this challenge could havebeen overcome through the design of a more func-tional moisture barrier.

OUTLOOK:

Figure 10Seed SubstrateMixed WithSilica/Hydrogel

Fungi have a number of properties that make thema great candidate for a building material. They canabsorb nutrients from a variety of substrates, whichallows them fungi to convert agricultural waste intoa number of useful products. They are biodegrad-able, once the material has served its purpose, it canbe quickly returned to the soil where it will be ab-sorbed and potentially used to create more of thesame material. The energy input required for fun-gal growth is minimal. Fungi will essentially growin the dark at room temperature with little more in-put. These characteristics of fungal-based materi-als make them sustainable and economical. Fungialso have a number of material properties that areparticularly suited for building technology. They arelightweight, which increases the ease of transporta-tion and assembly. They will also have lower deadloads. They are durable, and have the capacity to lo-calize vibrations. (Mycoworks 2015) This capacity in-creases the fungi’s ability to absorb impact loads andfail non-catastrophically, unlikemasonry or concrete.(Mycoworks 2015) They are flame resistant and self-extinguishing. (Mycoworks 2015) Despite all of thesebeneficial properties, fungi’s true potential as a build-ing material lies in their capacity for capacity for en-


vironmentally responsive, self-organized growth.This research project explored the potential of

fungal growth as a means of architectural assembly.Despite the inherent scalability of fungal growth, anumber of challenges emerge when fungi are grownin large volumes. These challenges include: preserv-ing an optimal growth environment, preventing con-tamination, and maintaining consistent growth. Theoverarching hypothesis of the research was that theintroduction of a multi-scalar assembly system couldnot only resolve these challenges, but also guide fun-gal growth in such a way that allowed for the in-troduction of design intention. The multi-scalar as-sembly system thatwas developedduring the courseof this research was successful at the levels of as-sembly unit and assembly element insofar as it bal-anced the biological utility and architectural viabilityof the fungi. Even though contamination was an is-sue that emerged when assembling the final proto-type, it does not invalidate the multi-scalar assemblysystem. Since, it is an issue that could have been re-solved through the design of amore functionalmois-ture barrier.

Therewereanumberofpotentially fruitful explo-rations that were abandoned because they did not fitwithin the stated objectives of this research project.The first experiment dealt with the effects of mois-ture on fungal growth. In it, silica gel was added tothe seed substrate. The addition or removal of mois-ture had amuchmore drastic effect on fungal growththan changing the type of substrate. The precision ofcontrol over the fungal growth is pretty remarkable.As can be seen on the figure at the top of the follow-ing page, the fungus grows toward the hydro gel onthe left side, while the initial chunk of mycelia on thedry side never starts growing. Another fruitful explo-ration was the use of mixed substrates. This kind ofexperimentation would exploit the fungi’s ability togrow on a variety of substrates. The physical prop-erties of a fungal-based material will depend bothon the fungi used; as well as, the type, and arrange-ment of the substrate. There are a lot of opportuni-ties to create differentmaterials properties bymixing

different kinds of substrates. These properties couldthen be combined to create new materials with inte-grated functionality, particularly in dealing with dif-ferent structural conditions.

Figure 11Fungi Grown on aMixed Substrate

Figure 12Using MixedSubstrates to CreateFunctional Grading

Another area of experimentation for fungal based-materials is usingmixed types of substrates. This kindof experimentation would exploit the fungi’s abilityto grow on a variety of substrates. The physical prop-erties of a fungal-based material will depend bothon the fungi used; as well as, the type, and arrange-ment of the substrate. There are a lot of opportuni-ties to create differentmaterials properties bymixing


different kinds of substrates. These properties couldthen be combined to create new materials with inte-grated functionality, particularly in dealing with dif-ferent structural conditions. The reason for choosingstructural conditions as the criteria of fitness is be-cause this is the area where the greatest challengesof using mycelia-based materials at the architecturalscale currently fall.

ACKNOWLEDGEMENTS:The biological research for this project was con-ducted at the Department of Plant Biotechnology,University of Stuttgart under the supervision Dr.Arnd G. Heyer.

REFERENCESArmstrong, R. and Spiller, N. 2010, ’Synthetic biology:

Living quarters’, Nature, 467, p. 916–918Boswell, G.P., Jacobs, H., Ritz, K., Gadd, G.M. and David-

son, F.A. 2007, ’The Development of Fungal Net-works in Complex Environments’, Bulletin of Mathe-matical Biology, 69, p. 605–634

Davidson, F.A. 2007, ’Mathematical modelling ofmycelia: a question of scale’, Fungal Biology Reviews,21, p. 30–41

Imhof, B. and Gruber, P. (eds) 2015, Built to Grow - Blend-ing architecture and biology, Birkhäuser

Travaglini, S., Noble, J., Ross, P.G. andDharan, C.K.H. 2013’Mycology Matrix Composites’, American Society forComposites—Twenty-Eighth Technical Conference

[1] http://www.ecovativedesign.com/[2] http://src.lafargeholcim-foundation.org/flip/A

15/A15Book/HTML/files/assets/basic-html/page-60.html

[3] http://www.mycoworks.com/portfolio/polyominoes/[4] http://www.mycoworks.com/page/2/[5] http://mycelium-tectonics.com/


Index of AuthorsAAl-Jokhadar, Amer 35Alacam, Sema 55Allani, Najla 65Alnusairat, Saba 83

BBarakat, Merate A. 5Bouattour, Mohamed 65

CCampbell, Sean 125Correa, David 125

DDolhan, Valentin 23

EElmas, Serenay 105

FFerreira,

Maria da Piedade 15

GGuzelci, Orkan Zeynel 55

HHalin, Gilles 65Hou, Shan Shan 83

JJabi, Wassim 35Jones, Phil 83

KKontovourkis, Odysseas 95Kretzer, Andreas 15

LLonsing, Werner 45

MMcDonagh, Andrew N.J. 115Menges, Achim 125Michael, Philippos 95

PPinto Duarte, José 15

SSalman, Huda S. 23, 115Saltik, Erenalp 105Schenkenberger, Benjamin 15Siala, Aida 65Stricker, Didier 15

TToyama, Takumi 15

WWeber, Markus 15Wood, Dylan 125

ZZapata-Lancaster,

Gabriela 73

Authors - eCAADe | 135

136 | eCAADe - Authors

the virtual and the physical - spaeth...

Documents