Specifications for computer-aided conceptual building design

1 Howard, H C, Levitt, R E,Paulson, B C, Pohl, J G andTatum, C B ‘Computer Inte-gration: Reducing fragmentationin AEC industry’ Journal of Com-puting in Civil Engineering Vol 3No 1 (1989) 18–212 Bedard, C ‘Automating build-ing design process with KBES’ASCE Journal of Computing inCivil Engineering Vol 4 No 2(1990) 69–83

www.elsevier.com/locate/destud0142-694X/03 $ - see front matterDesign Studies 24 (2003) 51–71PII: S0142-694X(02)00009-1 51 2002 Elsevier Science Ltd All rights reserved Printed in Great Britain

Specifications for computer-aidedconceptual building design

K. Meniru, H. Rivard and C. Bedard, Center for Building Studies,Department of Building, Civil and Environmental Engineering,Concordia University, 1455 Maisonneuve Blvd. West, Montre´al, CanadaH3G 1M8

The relative absence of computer support in conceptual building designdeprives designers of advanced capabilities for the manipulation,organization and representation of design data. Such capabilities wouldbe highly beneficial to the design process, particularly when designersfrom different specialties must work together to provide the best possibledesign solution. We look into how designers work during conceptualdesign with particular attention on how information is manipulated andorganized. We then draw specifications for the creation of a computertool to support such an early design stage.�c 2002 Elsevier Science Ltd. All rights reserved.

Keywords: architectural design, computer supported design, conceptualdesign, design research, protocol analysis

Buildings represent a complex assembly of many different systemssuch as the enclosure, structure and ventilation systems. Designersdetermine the scope of construction in the early design stage by

introducing the initial data that all other stages are completed against. Theearly design effort starts with a minimum number of professionals whomake relatively isolated decisions on important parts of the design prob-lem1. The success of the final design solution therefore depends not onlyon the successful interpretation of these decisions by other professionalswho become involved at different times in the process, but also on howeasily all designers can coordinate/combine knowledge at the earliest poss-ible time2.

One of our goals at the Center for Building Studies (CBS), is to encouragethe collaboration of building designers in maximizing the generation andsharing of knowledge and expertise for the design and construction ofbuildings. Spurred on by a recent survey showing increasing use and

3 Rivard, H ‘A survey on theimpact of information technologyon the Canadian architecture,engineering and constructionindustry’ Electronic Journal ofInformation Technology in Con-struction Vol 5 (May 2000) 37–564 Baya, V and Leifer, L J‘Understanding information man-agement in conceptual design’ inN Cross, H Christiaans and KDorst (eds) Analysing designactivity, Ch. 7, John Wiley &Sons, Chichester, UK (1996) pp151–1685 Akin, O Psychology of archi-tectural design, Pion Limited,London, UK (1986)6 Fazio, P ‘A case study in inte-grated building design’ CanadianJournal for Civil Engineering Vol17 (1990) 636–6427 Bedard, C, Gowri, K andFazio, P ‘Integration in the build-ing industry’ Journal of Comput-ing in Civil Engineering Vol 5 No5 (October 1991) 333–3358 Rush, R D (ed.) The buildingsystems integration handbook,Butterworth Heinemann, Oxford,UK (1986)9 Fricke, G ‘Successfulapproaches in dealing with differ-ent precise design problems’Design Studies Vol 20 No 5(1999) 417–42910 Atman, C J, Chimka, J R,Bursic, K M and Nachtmann, HL ‘A comparison of freshmanand senior engineering pro-cesses’ Design Studies Vol 20No 2 (1999) 131–15211 Casakin, H and GoldSch-midt, G ‘Expertise and use ofvisual anology: Implications fordesign education’ Design Stud-ies Vol 20 No 2 (1999) 153–17512 Dorner, D ‘Approachingdesign thinking research’ DesignStudies Vol 20 No 5 (1999)407–415a13 Lipson, H and Shpitalni, M‘‘Conceptual design and analysisby sketching’ Artificial Intelli-gence for Engineering Design,Analysis and Manufacturing Vol14 No 5 (2000) 391–40114 Purcell, A T and Gero, J S‘Drawings and the design pro-cess’ Design Studies Vol 19 No4 (1998) 389–430

52 Design Studies Vol 24 No. 1 January 2003

demand of computers at the early stages in design firms3, we investigatewhat specifications would be required for an early design computer-basedtool. Two important issues must be addressed in the development of sucha tool. The first issue can be derived from the reasoning that design is aprocess of capturing and organizing design information in such a mannerthat it can be retrieved and reused at a later time. The success of a designproject depends largely on how well the designer manages such infor-mation4. The second issue is concerned with the need for an integrateddesign process, i.e. where technical knowledge is brought into the buildingdesign process as early as possible in order to improve the overall buildingdelivery process5–8.

Our plan for an early design computer tool is therefore two-fold. Primarily,we want to help the designer (architect) manage the data that comes outof the early design process without disrupting their preferred method ofdesign. Such a software tool shall only manage the data produced by thedesigner and not necessarily redefine the manner in which data is produced.For this reason our focus in studying the design process remains mainlyon the workspace activity, not in explaining how the designer is thinkingabout design. Second, we want to integrate the workspaces of architectsand engineers so that early design data created by the architect is accessibleto the engineer without loss in meaning or context. Although veryimportant to the overall goal of this research, the focus of this paper is onthe first issue which is to capture the architect’s early design data manipu-lation and organization methods.

We provide the results of a design protocol study of eight individualdesigners in which we concentrate on the activities and events over timeon the workspace. We consider how or why the designer operates to enableus to understand what the drawing depicts and identify the means requiredto support its development. Combining literature survey and observationswe then draw specifications for developing a computer tool that wouldprovide valuable assistance at the conceptual building design stage.

1 Review of the design processEarly design for buildings invariably starts with a single designer(architect) who works using brief notes of requirements and sketches9. Alot of sketches are used which provide a means with which the designerclarifies the characteristics of the design, logs their design process, storesideas and reveals the mechanics of their thinking process10–12. In particular,the sketch is the most important tool13 for communicating design at theearly stage and have associations with hidden meanings in the designer’simagination14 which most likely will not be fully or easily understoodby others.

15 Goldschmidt, G ‘On visualdesign thinking: The vis kids ofarchitecture’ Design Studies Vol15 No 2 (1994) 158–17416 McGown, A, Green, G andRodgers, P A ‘Visible ideas:Information patterns of concep-tual sketch activity’ Design Stud-ies Vol 19 No 4 (1998) 431–45317 Gero, J S and McNeil, T‘An approach to the analysis ofdesign protocols’ Design StudiesVol 19 No 1 (1998) 21–6118 Goldschmidt, G ‘The dia-lectics of sketching’ Design Stud-ies Vol 4 (1991) 123–14319 Heylighen, A, Neuckerm-ans, H and Bouwen, J E ‘Walk-ing a thin line—between passiveknowledge and active Knowingof components and concepts inarchitectural design’ DesignStudies Vol 20 No 2 (1999)211–23520 Cross, N (ed.) The auto-mated architect, Pion Ltd, Lon-don UK (1977)21 Cross, N ‘Natural intelli-gence in design’ Design StudiesVol 20 (1999) 25–3922 Goel, V Sketches ofThought The MIT Press, Cam-bridge, MA (1995)23 Verstijnen, I M, Hennes-sey, J M, van Leeuwen, C andHamel, R ‘Sketching and cre-ative discovery’ Design StudiesVol 19 No 4 (1998) 519–54624 Rodgers, P A, Green, Gand McGown, A ‘Using conceptsketches to track design pro-gress’ Design Studies Vol 21 No5 (2000) 451–46425 Gunther, J and Ehrlen-spiel, K ‘Comparing designersfrom practice and designers withsystematic design education’Design Studies Vol 20 No 5(1999) 439–451

53Computer-aided conceptual building design

The use of such brevity enhances the designer’s imagination and alsoallows the process to forge ahead. This can be seen in design sessionswhere the designer begins to draw by making simple random lines onpaper15 iterating between sequences of externalization andcriticism/analysis11,16. During externalization the designer sketches mostlywithout regard for technical details while during criticism/analysis thesketch is reviewed as designer restructures or reinterprets17 their mentalidea/picture. This continues in the form of a dialogue14,18 between thedesigner and the sketch until the designer moves on to other areas in thedesign. When this move occurs, the solution for the present problem areais not necessarily complete or satisfactory. The designer keeps a sequentialdevelopment process but this occurs on several fronts simultaneously19,often leaving behind partial solutions to return at a later time.

Although the designer’s sketch may be thought of as consisting of a hier-archical arrangement of components, such moves as described above leavesthe design process seemingly as a jump from one point to the other resem-bling a non-linear procession that is very difficult to predict10,20,21. Thisunpredictability applies even to seemingly finished sketches as the designermay come back with changes at a later time. This move can be attributedto the designer’s need to subdivide the problem into many different sub-categories where their attention shifts from high level overall views to lowlevel details of the design problem17.

As a designer draws and reviews a sketch, it undergoes many duplicationsand transformations16,22–24. Duplications allow the designer to reproduceideas or sketches already existing in or outside the workspace. There aretwo types of transformations—vertical and lateral. Vertical transformationoccurs in successive detailing of design data while lateral transformationallows for the creation or development of new data.

Designers use different views when sketching e.g. plans, elevations and3D. They use these, sometimes simultaneously, to think of and managethe design in a way that fully represents the overall view of the emerg-ing solution21.

The sketch is not only a tool that helps the designer to solve problems. Itis also used to understand issues concerning requirements and other designinquiry that lead to a better knowledge of the design problem and/or therequirements that are being considered25.

One of the important characteristics of early design is the pursuit of morethan one solution9. This may happen consecutively or concurrently and is

26 Bhavnani, S K and Ghar-rett, J H Jr ‘Leading indicators ofCAD experience’ in U Flemmingand S Van Wyk (eds) CAADFutures ‘93. Proceedings of theFifth International Conference onComputer-Aided ArchitecturalDesign Futures, Elsevier, NewYork (1993) pp 313–33427 Mazijoglou, M, Scrivener,S and Clark, S ‘Representingdesign workspace activity’ in NCross, H Christiaans and KDorst (eds) Analysing designactivity, ch. 18, John Wiley &Sons, Christchurch, UK (1996)pp 389–41528 Dorst, K and Dijkhuis, J‘Comparing paradigms fordescribing design activity’ in NCross, H Christiaans and KDorst (eds) Analysing designactivity, ch. 12, John Wiley &Sons, Christchurch, UK (1996)pp 253–270


not just a way to expand the possibilities of a successful solution but it alsointroduces ‘spare parts’ from which the designer can consolidate desirablesolutions into a single ‘best’ possible solution for that particular design ses-sion.

The review of the design process provides us with important characteristicsof early design. However it is missing a detailed picture of the workspaceactivity concerning the sequence and output of the designer’s actions. Forexample, how is data added to a sketch to cause transformations? It isimportant for us to see the drawing steps in order to capture solutions suchas those created in lateral transformations. Also, observing the workspaceactivity will help in improving the user interface for early design as wedo not have existing CAD tools that provide adequate interfaces for manag-ing early design information4,13,26.

To acquire this knowledge we conduct our own study of designers at workwith a goal to observe the early design process, how the designer interactswith the sketches, when these interactions take place and finally the charac-teristics of the sketch with regard to its development that should be savedfor the success of the design process bearing in mind that more than onetype of designer may access such information.

2 Study of designersThere are many ways to study the designer’s work process namely throughinterviews, observations, case studies, protocol studies, simulation trials,reflections and theorizing. In the following pages, we present the resultsof a protocol study. Protocol studies make it possible to reduce the com-plexity of the design process4,27 allowing the researcher to collect data ata very fine level of detail depending on their unique perspective and inter-est4,28. In an attempt to provide a good cross section of data, we study howeight individual designers manipulate and organize conceptual design data.

2.1 The protocol studyThe protocol study is divided into two sections consisting of an interview(15 min) and a design session (45 min) and lasts a total of one hour. Bothsections of the study are videotaped and transcribed.

The study involves eight architects. Six are from Canada while two arefrom the United States. For the purposes of anonymity, each designer isreferred to by code C0x where ‘x’ ranges from 1 to 8. Professional experi-ence ranges from 3 to 20 years (see Table 3).

The study begins with a set of documents sent to the designer prior to the

29 Newman, W M and Lam-ming, M G Interactive SystemDesign Addison-Wesley, Boston,MA (1995)30 Holtzblatt, K and Jones, S‘Contextual inquiry: a partici-patory technique for systemdesign’ in D Schuler and ANamioka (eds) Participatorydesign, ch. 9, LawrenceEarlbaum, Hillsdale, NJ (1993)pp 177–21031 Laurel, B (ed.) The art ofhuman–computer interfacedesign, Addison-Wesley, Bos-ton, MA (1990) pp. 85–9032 Cross, N, Christiaans, Hand Dorst, K ‘Introduction: theDelft protocols workshop’ in NCross, H Christiaans and KDorst (eds) Analysing designactivity, John Wiley & Sons,Chichester, UK (1996) pp 1–16


study date which includes an introduction, a consent form, a set of samplequestions and the sample design problem. The introduction provides adescription of the research project, its status, the goals of the study as wellas information about the researchers and the funding agency. It informsthe designer of the intent of the study to respect the designer’s privacywith regard to the dissemination of knowledge obtained from the study.The consent form advises the designer of their right to ask questions or todiscontinue at anytime during the study. The designer is required to signthe consent form to continue the study. The questions to be asked in thestudy are made available to prepare the designer.

The sample design problem describes the need for a residence on a 9×31m (30×100-feet) of land. Spatial requirements consist of a minimum ofthree bedrooms, two bathrooms, living and family rooms, kitchen, externalplay area, two car garage, workshop and office. The structure can be upto three levels in height, typical in the neighborhood. Other residences arelocated on all sides except the front, which serves as an entrance. Thedesigner is provided with setbacks and required to separate entrances tothe living and working areas. A budget and a site plan is provided as wellas information about the client. At the end of the interview, questions thatthe designer may have regarding the documents are answered before pro-ceeding to the design session.

The interview consists of a mixture of structured and unstructured questionand answer style meeting29. We formulate our questions (see Fig. 1) basedon our goal to develop an early design computer tool according to thefollowing headings (with the number corresponding to the question);enumerating activities (1), work methods (2,3,4,5), tracing connections (6)and performance issues (7). We conduct the study in each designer’s workenvironment as information relating to how they work is richest in theiractual work environment30. We allow them freedom to elaborate and wefollow up any interesting pointers.

In the design session, we ask the designer to treat the interviewer as theclient or representative and provide a design solution as they would nor-mally do in such situations. Ignoring some possible drawbacks (see design-er’s comment, Fig. 2), we requested that they speak their thoughtsaloud31,32 as this will not only be crucial in transcribing the work but itwill also tell us more about the designer’s view of the workspace andhow/what the designer is interacting with. The designer’s commentreinforces that talking affected the results of the design session but not thegoals of this research as alluded by the comments in italics.

We transcribe the interviews using three columns. The first column (see


Figure 1 Interview questions

Figure 2 Designer’s comment on speaking aloud

Table 1) records the sequence of the questions. Each question can bereferred to by a local or global time. We use two six-hour video cassettesshowing the tapes duration in minutes (global time), we then add a localtime for each study for easy reference. A second column records the ques-tions asked by the interviewer while a third records the answers fromthe designer.

The transcription of the design session (see Table 2) follows the samestructure as the interviews in the first column but without the sequencenumbering. The second column records the designer’s actions while thethird records the designer’s comments.


Table 1 Sample transcribed interview

Sequence local Questions Designer’s answersglobal

012 If you were… I know you have not Maybe there is another way to extract stuff… I haveused computers to design but I am got this shape and I draw this line and then I continuesure you have heard a lot about what as I draw with tracing paper [I keep some parts andthey can do and you have seen some maybe I have other ideas but they] fade away but yousimulations. If you were to say still have them very faint [C03008RS]. I think itsomething that is very crucial or that should be on the screen. I think that if you can sayyou feel is missing from computers. that this is a wall and this is a wall, then it gives youWhat would it be? a perspective and then I have got these two walls, I

continue to play with this perspective, I got thiswindow here and it shows there… I just have neverthought of that before. If you say this is 1:50, it shouldbe easy to get other information about theenvironment, then you can get the perspective of thebuilding and the street.

0:11.072:15.45

During transcription, we follow the designer’s progress and mark the areasthat we feel are relevant to our purposes. Each mark or sketch by thedesigner on the drawing sheets are labeled with a unique id for that sheet.The id ‘C0401A’ for example means mark A on sheet 01 from designerC04’s design session. Each action references the point in the designer’scomments when it was performed. An example of a reference is [∗1]. Thismark shows a reference to number 1 action in that transcription row.

Finally, we go through the transcriptions identifying requirements we feelshould be supported using computers. These requirements are labeled withcodes that identify the designer in whose session it was discovered (C04),the number of it in sequence to the overall requirements noticed for thatdesigner (005) and two letters to show that it is a requirement specification(RS). These examples will appear as C04005RS. After the transcriptions,a set of unique requirements are compiled through numerous iterations ofcomparisons and discussions.

2.2 Protocol study resultsThe design sessions show the designers going through some stages in thedesign process which we will call activities (effort towards an intermediategoal) and using tactics which we will call events (actions at particular timeswithin an activity) that guided or influenced the activity. In this sectionwe will discuss each designer, in brief, with a diagram showing their activi-ties and events (see Figs. 3–10). The following is a description of theactivities identified.


Table 2 Sample transcribed design session

Time: Designer’s actions Designer’s commentsLocalGlobal

0:15.45 1. Designer reads the requirements carefully [∗1] Obviously it is impossible to start the computerbecause the first question [it] is asking you is traceline from one point to another… [meanwhile] youhae no idea [where the lines are going to be yet]. SoI will start with a sketch paper. When you designthe first stage of a project the scale is veryimportant… so you must work at a very small scale.An 11x17 size sheet will be enough. Because therequirements are very precise, I can begin to tracemy lot. What is very important is to work at thesame time in plan as well as in elevation andsection, [C04005RS] because you are missingopportunities. 1/8 [scale] is alright because this isnot an urban [design] project [∗2]. It is only oneplot. [∗3]. So you need a fine and rigid base. [∗4].Usually you are working on different layers [on thecomputer it is very easy…]

3:19.45 2. Reads the requirements. Using the scaleruler to draw a scaled outline of the siteat C0401A.

3. Designer uses very dark lines for theoutline of the site for visual cues.

4. Designer reads the requirements againand marks off some requirements on thedrawing such as set-backs, accesses andorientation at C0401A. [C04006RS]

0:21.39 So you can start straight away to do this on thecomputer?

3:25.39 Yes you can do that but it could be done laterbecause the next thing now is to put my tracingpaper over it and start drawing freehand[C04021RS]. You cannot design with the computer[at least with the systems we have used]…

Table 3 Designers’ experience

ID Professional experience (years)

C01 3C02 5C03 10C04 15C05 10C06 5C07 20C08 20


Figure 3 C01 - Design pattern




Figure 6 C04 - Design pat-

tern


tern


tern


tern


Figure 10 C08 - Design

pattern

2.2.1 Design briefThe designer tries to understand the nature of the client’s needs and theresources that are available. The requirements are organized in an access-ible manner depending on the designer’s style and this information isreferred to numerous times during the process.

2.2.2 Site preparationThe site is determined by establishing setbacks, adjacent construction, sunpath and available footprint.

2.2.3 Building spaceThe relationship between spaces and elements within the building areresolved often using simple geometry such as bubble diagrams. This iswhere the designer produces possible solutions for the design problem.The designer combines several types of visual aides such as diagrammaticelements, plans, sections, elevations and volumetric studies. This activityconsiders abstract items such as spaces, circulation, relationships, percep-tions, proximity and character.

2.2.4 Building elementsThis activity considers more tangible items such as walls, floors and furni-ture and places them in the design.

Table 4 describes the recognized events and their codes, while Table 5shows how codes may be combined to show simultaneous or closelysequenced events. The following are summaries of each designer’s session.

C01 (see Fig. 3) spends a few minutes tabulating the requirements providedin the design brief. These tabulations bring together all the important dataand provide an easy reference during design. The designer establishes theavailable footprint on the site before starting the building space activity.Building elements such as windows, doors and furniture are included at


Table 4 Designers’ event codes

Code Description

3 Three-dimension drawings for studying or exploring ideasA Designer makes an alternate sketchB Designer backtracks to a previous pointE Designer creates an elevationG Designer groups spaces or requirements using bubblesL Designer labels drawingsN Designer begins drawing on a new sheetS Designer shades or renders drawingsX Designer works in a section

Table 5 Example combinations of event codes

Code Description

B~N Designer backtracks to a previous point using a new sheetB~N~A Designer backtracks, uses a new sheet and starts alternative idea

the end of the session and the designer makes two major changes in designdevelopment at about 25 and 35 minutes in the session.

C02 (see Fig. 4) spends a few minutes asking questions about the designbrief before going into building space activity. Some relevant data is jotteddown on paper as a reminder, however these are random, uncoordinatedevents. Building space design proceeds in plan and section simultaneouslyand the designer spends time to configure the building elements in someof the spaces. Some time is spent at the end of the session on the overallelevation and its building elements such as doors and windows. Thedesigner works on a single solution although there is one major change indevelopment at about 33 minutes into the session.

C03 (see Fig. 5) reads the brief very quickly before beginning to work onthe site preparation. The brief is referenced continuously as the availablebuilding footprint and setbacks are established. The building space activitybegins with plans, sections, grouping of spaces using bubbles in sectionaldiagrams and volumetric studies. After a satisfactory configuration of aparticular space, the designer spends a little time elaborating on the build-ing elements for that space. The designer works on a single solutionalthough many options are created for different parts of the design.

C04 (see Fig. 6) begins by reviewing the bylaws and the brief to collectrequirements. Data from the brief is then used in the site preparation where


the designer identifies the available footprint and renders the sun paths(winter and summer). After configuring each building space, the designerillustrates the building elements found in the spaces. This serves mainlyto validate the decisions surrounding the building space design. Thedesigner works on a new design alternative after the completion of the firstalternative, starting at 38 minutes. The designer concludes the session with3D volumetric study of facades.

C05 (see Fig. 7) takes the time to understand the design brief and discussesthe requirements thoroughly with the client ( i.e. the researcher playing theclient) using a combination of labels and bubbles. A quick site configur-ation is achieved with more bubbles and shades, which are elaborated inthe building space activity. At certain times the designer designates orintroduces building elements like a stair, a door or a window within theboundaries of a bubble. These act as icons or reminders for a later stagewhen the sketch will be drafted. This designer prefers a verbal andextremely abstract mode of initial design. Such a method starts with vagueissues, which are detailed in subsequent iterations.

C06 (see Fig. 8) extracts data for the site preparation from the design briefand moves onto design space activity where most of the session time isspent. In the building space activity several references to building elementsare made in an effort to elaborate the spaces with items such as furniture.The designer explores a second alternative at about 33 minutes in the ses-sion because of unresolved problems.

C07 (see Fig. 9) starts with a systematic analysis of the design brief. Allrequirements are tabulated and summarized. Maximum footprint as wellas floor area, possible levels and total cost of building are estimated fromthe brief. The designer then provides isolated configurations for each build-ing space by combining information obtained from queries to the client,references to the site and other requirements. This allows the designer tocollect a snapshot of all spaces with their interior organization and finishes.This also provides an early transition to the building elements activity incomparison to other designers in the study. These spaces are thenabstracted for easier manipulation. The designer spends a few minutesillustrating the available footprint before starting the building space activityusing the abstractions in the form of a combination of quadrilaterals, circlesor ellipses. After a successful design is selected, the abstract forms areexchanged or substituted for their detailed version from the earlier isolatedconfigurations. This illustrates a unique approach to all other designersespecially C05 and resembles a bottom-up design approach.

C08 (see Fig. 10) starts by studying the brief before preparing the site for


building space configurations. The designer remains mainly in the buildingspace activity but references to the building elements and design brief aremade, as parts of the design are resolved. The designer reverses directiondue to a misunderstood requirement at about 24 minutes and creates analternative to a part of the building at about 36 minutes. This designer hasused the greatest number of new sheets compared to the others due to thepreference of problem resolution using layers. This provides ample flexi-bility in creating alternatives and an easy way of introducing new ideas.

2.3 ObservationsThe diagrams do not show all of the events in the activities due to a limi-tation of space. The following enumerates our observations that will enableus to derive specifications for the possible development of a computer toolfor early design.

(1) The designers performed all four activities. In the first two (designbrief and site preparation), the designer defines and sets the boundariesfor the design problem. In the next two (building space and buildingelements), potential solutions are developed. The designer may referto more than one activity in order to create a design item, such aslooking at the site (site preparation) to determine the noise sourcesand the design brief for owners requirements when locating and confi-guring a bedroom space (building space). Table 6 shows that mostreferences are made to the design brief activity and to the buildingelements activity. This is consistent for all designers except C07 whodid the opposite i.e. most references were made to Building Spacefrom Design Brief, showing a bottom-up approach.

(2) The events show, in more detail, the actions the designer takes duringthe activities and may give hints to the type of tool or assistance thatthe designer requires to accomplish the activity. From Table 7, themost frequent event is the use of new sheets (N) followed by thegrouping of spaces or requirements (G), use of labels on sketches (L)and shading or rendering of drawing element (S).

(3) All designers (except C07) adopted a top-down design style i.e. start-ing with high level (abstract) design decisions, then providing details

Table 6 References between activities

Activity References to [frequency (%)] References from [frequency (%)]

Design brief 12 (52) 3 (13)Site preparation 1 (4) 5 (22)Building space 2 (8) 14 (61)Building elements 8 (35) 1 (4)

Kene Meniru

Highlight

33 Eisentraut, R ‘Styles ofproblem solving and their influ-ence on the design process’Design Studies Vol 20 No 5(1999) 431–437


Table 7 Sketch Elevation (E)

Event Occurrences Design brief Site preparation Building Building[frequency (%)] space elements

Volumetric study (3) 10 (9) 0 0 8 2Alternate sketch (A) 4 (3) 0 0 4 0Backtracks (B) 8 (7) 0 0 8 0Sketch elevation (£) 9 (8) 1 0 4 4Group spacing (G) 18 (15) 3 1 14 0Label sketch (L) 17 (14) 7 3 5 2New sheet (N) 27 (23) 1 0 24 2Shade/render (S) 15 (13) 2 3 10 0Sketch section (X) 10 (9) 0 0 10 0

for these decisions as the design session progresses. C07, on the otherhand, has used a bottom-up approach that begins with detaileddecisions, which were then used as references or templates in thedevelopment of design options.

(4) Five designers (C01, C04, C05, C07, C08) began by scrutinizing thebrief and usually rearranged or reorganized requirements into anaccessible format for easy reference during the design session.

(5) All designers extracted the building footprint, bylaws and other para-meters from the design brief to determine the orientation of the pro-posed structure.

(6) All designers used multiple views for building space configuration.The most common of these views are plans and sections. In additionfour designers (C01, C02, C03, C04) used elevation drawings whiletwo (C03, C04) explored with volumetric diagrams.

(7) All designers used simple diagrams to show design items. For instancebubbles or other geometric forms were used to depict spaces.

(8) All designers followed a design process that displayed several solutionpath branches. For instance, a stair design at point X (see Fig. 11),may branch into the development of a straight staircase at Xa2, or acircular version at Xb2.

(9) Some designers (C04, C06, C08) relied on the development of alterna-tive solutions, parts of which may be later merged to produce a suc-cessful solution. For instance, version Xa2 and Xb2 in Fig. 11 arealternate solutions.

(10)Designers approached the design process differently33 because of per-ceptions and experience. They may utilize similar tactics but the resultis always different, therefore the tools they use or need requiremaximum flexibility and minimum effort to customize.

Kene Meniru

Highlight


Figure 11 Branching in the design process

3 Specifications for conceptual design supportCombining knowledge from literature and the protocol study, we arrive ata list of specifications for a computer tool (from now on referred to as thesystem) to assist designers in the early stages of design. Each specificationis presented under two headings, scope (provides background andjustification) and specification.

(1) Requirements Repository

Scope: The design brief describes the design problem, goals and avail-able resources. All designers rearrange this information for easy refer-ence which is kept readily accessible throughout the entire design pro-cess (see items 1 and 4 in subsection 2.3).Specification:The system should:

� Provide the ability to enter requirements from the design brief, stan-dards, codes and other sources.

� Allow designers to organize these requirements in a format that isconvenient to them (e.g. as a list or a hierarchy).

� Provide easy access to the requirements so that the designer can referto them throughout the design process.

� Allow the designer to establish a relationship between two or morerequirements. For instance, the desire to maintain the same floor lev-els between two spaces.

(2) Application of Requirements


Scope: Designers C01, C03, C04, C07 and C08 use information fromthe design brief to apply in the design. For instance C07 worked outminimum areas for each required space and used this information asa guide in creating the forms for both Site Preparation (availablefootprint) and Building Space activities. As such, adherence to theright scales and sizes is maintained during configuration without theuse of detailed and cumbersome diagrams (see items 5 and 7 in subsec-tion 2.3)Specification: The system should:

� Allow the designer to quickly extract the site parameters and isolatethe minimum areas available for design. To do this, it should bepossible for the designer to scan the site or input the boundaries ina convenient scale to which the system could apply information suchas setbacks, automatically.

� Apply, check or monitor parts of the requirements against the designsuch as preliminary total cost of materials or total size of circulationspace. Requirements that change such as later addition or subtractionof spaces in the design, should dynamically be reported and updatedin the requirement repository and vice versa.

(3) Multiple Levels of Abstraction

Scope: C03 uses different design representations. For instance a planview may be drawn with furniture while at other times empty spaceswith only walls might be more adequate14,15,33. Also different designerslike architects and engineers make use of different views. For instance,the structural engineer may want to focus on the structure only andabstract all other non-structural items out of the view. Extra infor-mation only tends to distract and complicate the design environment(see items 6 and 10 in subsection 2.3).Specification:

� The system should provide the ability to present design views atdifferent levels of abstraction (for example, walls can be representedwith single lines even though the system understands that this is awall and therefore can represent both) and for different disciplines(for example, the architect wants to see architectural elements suchas spaces, walls and furniture while the engineer may prefer walls,slabs and reinforcement).

(4) Version Management

Scope: Designers C02, C03, C05, C07 and C08 keep records of thedesign progress so that they can backtrack10 to an earlier point in thedesign session or branch off towards a different goal. This action is


desirable and creates alternative design solutions16,23. Designers do noteliminate these alternatives but rather keep them so that they may berevisited at some later stages in the design session (see items 8 and 9in subsection 2.3).Specification: The system should:

� Save every important step (or version) of the design evolution aswell as any branching that occurred to facilitate backtracking.

� Provide a way to book mark steps in the solution path for later refer-ence, comparison or backtracking.

(5) Element Interaction

Scope: All designers interact with design items in dynamic ways asthey create, adjust and manipulate them. A typical procedure for con-figuring an element is to draw the outline of the object as it is scaledand located accordingly. The concern is primarily with the size andprofile before any specifics such as length and width. Even then, suchspecifics are tentative and changes occur numerous times. When inter-acting with design items, the designers think in three-dimensionsalthough the item may be in two-dimensions, therefore the systemshould assist in simple, fast and intuitive three-dimensional elementconfiguration (see items 2, 3 and 7 in subsection 2.3).Specification: The system should:

� Provide an approximate way of sizing (drawing using dimensions),scaling (adjusting sizes) and locating/positioning items in a fast andintuitive manner. During the creation of design items, the designermay not be sure of the dimensions, scale or location. The systemshould provide visual means by which the designer can make areasonable decision about these parameters. A report about the sizeand location of design items should be constantly available or access-ible to the designer.

� Provide a fast and intuitive way of assisting the designer not onlyin converting from two to three-dimensional views but also in cre-ating quick three-dimensional views from the start. Such implemen-tations should support design items that might not have closedboundaries (as needed by current software in order to process a sub-traction, for instance).

� Neither limit the designer nor perform actions that change the design-ers configurations. For instance, preventing the projection of a three-dimensional item due to an incomplete profile definition or the auto-matic redefinition of such a profile so as to accomplish the three-dimensional projection.

(6) Automatic Feedback

34 Haptic Technologies, Inc.Technology for your fingertipsMontreal, Canada. 1999.


Scope: All designers remain alert to clues and opportunities in thedesign. They also try to anticipate limitations and problems that mayhinder the success of the design. One way in which they do this is bychecking the design at intervals and performing “what if...” scenariosevery now and then. For example if the designer wishes to know howmany items would occupy a particular surface area, it should be poss-ible to automatically replicate these items visually. Designers can vis-ualize scenarios in their mind but often prefer to make investigativesketches away from the main design area. These separate visualizationsprovide a richer feedback because of the isolation of the designitem/scenario, the freedom to make new design items at new viewangles and the possibility of using three-dimensional views. These vis-ualizations are usually saved and situated in prominent places such ason walls, where the designer often refers to them as design continues(C03, C04, C05 and C07 during interviews).Specification: The system should:

� Provide an interface for performing “what if...” scenarios and inte-grating the results back into the design, or storing the results forlater reference.

� Provide means through which the consequences of design decisionsare shown, such as daylight values in a space given a window size,or the path of a shadow over the course of time.

� Provide direct tactile interaction with design items in order toenhance the designers perception. For instance Haptic34 offers “ theCAT (Computer-Assisted-Touch)” force feedback technologywhich provides real-time feedback of the form, texture and reactionsto external forces from digital objects.

� Provide the designer with non-intrusive feedback when errors orseemingly impossible actions or manipulations are detected.

(7) Design Overview

Scope: All designers produced many sheets of drawings and referredto them frequently. Computer monitors provide smaller view screensin comparison to drawing sheets and designers often find it difficultto locate and orient their focus (See items 1 and 6 in subsection 2.3).Specification:

� The system should provide the designer with a persistent bird’s eyeview of the design. The computer screen is limited in size but anoverview could allow the designer to navigate to parts of the hiddenscreen without being disoriented or lost. The overview should beindependent of the other view(s) used for active design.

(8) Design Liability

35 Emkin, L Z ‘Misuse of Com-puters by Structural Engineers: Aclear and Present Danger’ in NK Srivastava (ed) StructuralEngineering World Wide. Pro-ceedings of the Structural Engin-eering World Congress. ElsevierScience Ltd, San Francisco, July19–23, 1998, Paper Reference:P331-3, 8 pages.


Scope: Designers assume full responsibility for design decisions35 asthese decisions originate from professional/personal experiences andstyle. No designer is willing to delegate power to the computer accord-ing to the answers given by all the designers to the last question ofthe interview (see Fig. 1).Specification:

� The system must not “design” for the designer. All changes or ideasintroduced automatically by the system must be examined andacknowledged by the designer before inclusion in the design.

� The system should provide a means for the designer to compile alist of “ to-do” items possibly with extended notes.

� Designs are sometimes driven by experience and styles. A systemshould be customizable to allow the designer to specify how certainactions or details should be presented or allowed to come together.In this way the designer’s personal style and preferences may be pre-served.

� The system should have an automatic logging process that keepstrack of the designer’s identity for any work or changes made. Accessto these logs should be protected and the activity should be cumulat-ive.

4 ConclusionEarly design is a complex activity and computers have potential in provid-ing assistance. In this paper we review some knowledge of the designprocess and describe a study of eight designers at work in order to under-stand the characteristics of design information on the designer’s workspaceand how the designer tries to manipulate, organize and store this infor-mation.

Studying the designers’ processes reveal that there are four main activitiesusually performed in sequence—design brief, site preparation, buildingspace and building elements. These activities require different design capa-bilities and types of interaction. The end of each activity produces infor-mation that is referenced by other activities as the process continues. Theactivities are populated by events, which represent different actions orfunctions used in these activities.

We have drawn from this study a set of specifications that can guide theimplementation of a computer tool for supporting early design. Thesespecifications are not meant to be exhaustive but rather to be a first draftwhich hopefully will evolve with time.

We must also draw the attention of the reader to the fact that these specifi-


cations are neither solutions nor do they suggest how the system shouldbe implemented. For instance the specification under Application ofRequirements and Automatic Feedback (see section 3) regarding providingquick cost estimates and daylight data feedback does not suggest or assumeany specific or existing method of input or feedback available insoftware today.

These are implementation issues and are not addressed in this paper. Wehave noticed these requirements based on our study and we do not proposehow they should be solved except that in the studies we performed, thedesigners needed or used (manually) such a capability to make their designsession more successful. Obviously, there needs to be a better way to solveit than what is presently available in computer tools for early design. Sucha tool will certainly bring some engineering knowledge earlier in the designprocess which is desirable.

For the next step of this research we have selected some specifications andare developing a prototype system that can help the architect to organizeand manipulate early design information. We hope to limit the amount ofinformation the designer has to provide or manage during early design. Inthis way, we hope to increase the time and effort that the designer spendsin the design thinking stages.

AcknowledgementsThe authors wish to thank the Natural Sciences and Engineering ResearchCouncil of Canada as well as Concordia University and its FacultyResearch Development Program for financially supporting this research.The authors are grateful for the effort and time from the designers in partic-ipating in the interviews and protocol study.

Specifications for computer-aided conceptual building design

Documents

Transcript of Specifications for computer-aided conceptual building design