Parametric design

Tool, medium or new paradigm?

Aleksander Asanowicz11Bialystok University of Technology1asanowicz@gmail.com

Parametric design is an emerging research issue in the design domain. However,discussions about the creative process in parametric design are limited. What ismore, despite the passing of 57 years of parametric design's existence we still donot know what parametric design is. Is it a simple tool, which is useful in somekind of optimization of the architectural form, or it is a medium, which helpsarchitects develop unexpected solutions, and perhaps this is already a new designparadigm? The presented paper will contain general considerations relating tothe nature of parametric design, the history of which starts in 1960, when D.T.Ross has formulated the thesis that our main objective is to formulate constrainsand all needed parameters of the solved problem.Please write your abstract hereby clicking this paragraph.

Keywords: optimisation, parametric design, design tool, design media

HISTORICAL BACKGROUND OF THE PARA-METRIC DESIGNAs a result, the development of a design methodol-ogy, caused by the increasing complexity of both theobject and the design process, as the main principleof improvement of architectural design approachhasfocused on the objectivization and rationalization ofthe design process. The main principle was definedas follows: “When we know exactly what is and alsoexactly what ought to be we are able to establisha direct efficiency relation. By appropriate compar-isons of what is and what ought to be, efficiencies,both ideal and practical, can be established.” (Emer-son 1964, p. 23)The first attempt of applying some-thing like parametric designwas undertaken in 1960,when D.T. Ross in Computer-Aided Design: a state-ment of objectives, has formulated the thesis that

“...we now declaim that our main objective is not tosolve problems, but to state problems”, which meantto formulate constrains and all needed parameters ofthe solved problem. (Ross and Mann 1960) The firstcomputer program conducted under the Computer-Aided Design Project was written by Ivan Sutherlandin 1963 in the course of his PhD thesis at MIT.In the following years research has focused on theproblems of how computers can aid the facility lay-out process and the designer’s interaction with mul-tiple design databases. The classical layout programsare: block diagramming software - CRAFT (minimizesnonadjacent cost and usedwhen quantitative data isavailable) and relationship diagramming - CORELAP(based on location preference between areas andused when quantitative data is not available). (Lee1967).CRAFT: Computerized relative allocation of fa-

PARAMETRIC AND GENERATIVE DESIGN - Volume 2 - eCAADe 35 | 379

cilities’ technique (Buffa, Armour andVollmann1964)was one of the most popular computer based lay-out procedures. Algorithms elaborated by Buffa et.all, based on the heuristic approaches, were used toplace factory spaces in a way which would reducethe total material transportation cost. The cost was aparameter defined by a function of the distance be-tween work centers, the frequency of movement ofmaterial between the centers, and the cost involvedin each move. The algorithm starts with an initiallayout and proceeds to improve it by interchangingspaces in pairs to achieve its goal. The combinationalproblem lengthens the solution process as the num-ber of work centers increases. If designing layout for20work centers, exchanging twowork centers simul-taneously, would require 190 evaluations. Exchang-ing three work centers simultaneously increases thenumber of evaluations to 1140, and if we would liketo change 20 centers it would need n! - 20! evalu-ations. With this amount of points, making changesmanually is practically impossible.The time requirement for a computer is trivial for anevaluation of these possibilities. This technique per-mits a considerably larger number of evaluations ina short time. It does not give the optimal layout;but the results are good and near optimal, whichcan be later corrected to suit the need of the layoutplanner.CORELAP: Computerized Relationship Lay-out Planning is the oldest Layout Planning routineand was developed by Lee and Moore. CORELAPgenerates a layout on the basis of the total closenessrating (TCR) for each department and beginwith twomajor inputs: a relationship chart and space require-ments. The starting point is creation of the Muther’sgrid table which displays designer’s preferences forrelative (pair-wise) department locations. Muther hasproposed a six points scale: A - absolutely necessary,especially important, I - important, O - OK, U - unim-portant, X - undesirable. To obtain a layout, the useris required to input the following: number of depart-ments, relationships weights, relationships cut-offs,partial adjacency value, and relationship. Then thecomputer generates the layout. The total closeness

rating, the order of entry of departments in the lay-out, the numerical closeness value and distance be-tween the departments are also shown. CORELAP ac-cepts relationships between as many as 40 depart-ments. (Lee and Moore 1967) The new version of theCORELAP algorithm can be obtained from Tompkinset al (2003) where a new interactive version of thesoftware is presented. The user interface in this im-plementation of CORELAP is the spreadsheet and theuser inputs data through Microsoft Forms.In 1980,the author has in his diploma work used the routine,a modification of CORELAP, developed by Maria Os-trowska in 1977. The diploma subject was a PrimarySchool. As the first step the functional program wasdefined and it was decided that the school shouldconsist of 9 areas: A - Library, B - Classes 1-4, C -Math-ematics and Environment, D - Humanities, E - Tech-nical Lab (DIY), F - Art, G - Sport, H - Recreation, I- Administration. In each area, different functionalblocks were placed, and as a result a matrix of 40 x 40elements was obtained. Connections between theelements were evaluated on the basis of a 4-pointscale: 1 - direct connection, 2 - strong connection, 3-medium connection, 4 - weak connection.These values were the design parameters. Due tohardware limitations, this large matrix was simpli-fied to a symmetrical matrix of 9 x 9 elements, whichmeant that firstly the location of the 9 school ar-eas was calculated - the computer produced a graphwhich graphically represented the links betweenmain functional elements. The same procedure wasthen repeated for areas A, B, C, H and I. As a resulta comprehensive scheme of functional links was cre-ated. On the basis of the schemes the building planwas created manually. To get many variants of theplan, values of the chosen parameterswere changed,and matrix was calculated again. (see Figure 1) Prac-tice has shown little usefulness of thesemethods, es-pecially if the laboriousness of preparing data at theinitial preparatory stages is taken into account. It wasnot possible to apply subjective parameters, whichled to oversimplification of the plan. These programswere characterizedby aprimitive input andoutput as

380 | eCAADe 35 - PARAMETRIC AND GENERATIVE DESIGN - Volume 2

Figure 1Primary School(modification ofCORELAP)

they worked on the principle of batch processing.The old programs were very user unfriendly. De-spite the difficulties caused by the imperfection ofcomputer hardware (time consuming batch process-ing) and simple software, the relational method washelpful in designing the building plan.It should benoted that with the development of computer tech-nology, these programs have become one of thebasic tools for industrial facility planners. This hasprofound effects on organizational productivity andprofitability. Optimal layouts reduce materials han-dling costs, help streamline all operations in a facil-ity, and reduce energy costs. With large amountsof money being spent on new facilities each year,it is natural that industrial facility planners, design-ers, and architects long for a superior Facility Lay-out Optimization software. On the market, we mayfind many professional computer programs for Lay-out Planning.One of the interesting layout optimiza-tion tool is VIP-PLANOPT (formerly known as Layopt).This software is an improvement algorithm for devel-oping alternative and efficient block layouts from aninitial block layout provided by the user. In the ab-sence of an initial layout, one may be also randomlygenerated by the program. The algorithm used in

the program is based on the algorithm developed bythe Bozer, Meller and Erlenbacher. It extends a well-known facility layout algorithm (CRAFT) to facilitieswith multiple floors. It also enhances CRAFT by con-trolling department shapes by allowing flexible de-partmental area requirements. (Bozer, Meller and Er-lebacher 1994)BLOCPLAN is an interactive programdeveloped by Donaghey and Pire. Quantitive andqualitative data can be used as input. It can developa single story ormulti-story layout. As the BLOCPLANgenerates an initial layout andmakes enhancementsof this layout, it can be explained as both a construc-tion and an improvement method. The user may op-tionally choose the random construction and auto-matic search. (Donaghey and Pire 1991)Micro CRAFT(MCRAFT) is anextendedversionofCRAFT, presentedby Hosni, Whitehouse and Atkins. MCRAFT dividesthe plant area into bands and assign these bandsto one or more facilities. Moreover, MCRAFT elim-inates the pair-wise exchange limitations. By usingMCRAFT, all pairs can be tried with the pair-wise ex-change method, which makes a large contributionto finding an optimum solution. (Hosni, Whitehouseand Atkins 1980)

PARAMETRIC AND GENERATIVE DESIGN - Volume 2 - eCAADe 35 | 381

PARAMETRIC DESIGN AS AN “OPTIMIZA-TION TOOL”All computer programs discussed above are a kindof an optimization tool, the goal of which is design-ing a cost-effective product in minimum time. In or-der to achieve this goal, the requirements of opti-mum designs are becoming more important. Para-metric design modelling platforms and scripting en-vironments allows for rapid generation of 3Dmodelsand enable multilevel evaluation of parametrically-driven design alternatives. The key to understand-ing what is parametric architecture, is the word “pa-rameter”. Design is based on carefully described pa-rameters, which are used by computer programs forgenerating original, unusual and difficult to describespatial forms with mathematical precision, optimiz-ing them mostly in terms of environmental or func-tional conditions.Parametric design is treated as an“optimization tool”.In architectural design, differentparameters were used to “optimise” the architecturalform. Each parametric design starts with a paramet-ric variation, which can be employed for the differen-tiation of a field, layer or subsystem. The most com-mon are the parameters associated with the struc-ture of the building, energy efficiency, sun exposure,location, acoustics or aerodynamics.This is fully un-derstandable because, for example, in the design ofhigh-rise buildings the cost of “architecture” is only40%of the total cost,MEP (mechanical, electrical, andplumbing) - 25%, structure - 30% and elevator sys-tem is 5%. (Nicknam and Elnimeiri 2011)In the cur-rent design practice, issues described above are typ-ically left to be dealt with after the architectural formis well articulated. This approach is time consumingwhen the architect proposes multiple options. Thesolution is a full integrationof optimization toolswiththe CAD system, where all drawings can be automat-ically updated after achieving the optimum and sat-isfying all imposed constraints. The advantage of thiswayofworking is a shorteningof theoptimization cy-cle time and radical reduction design time. Anotheradvantage of this approach is also the ability to re-duce investment costs, as themost important design

decisions, which have the most significant cost im-pacts, are made at the concept stage of a project.Theserious disadvantage is that parametric methods aremostly used at the stage of detailing of the projectwhen the designer may update the CAD model onlyafter receiving anoptimumdesign fromoptimizationtools. Consequently, in traditional Computer AidedDesign the main consuming factor is the design op-timization cycle.

PARAMETRIC DESIGN AS A DESIGNMEDIUMIn 1994 Ranulph Glanville has asked: “What is the dif-ference between a tool (or toolkit) and a medium?”and then answered: “The difference is that the tooldoes what we want, amplifying, in some sense, ournatural abilities. It applicable within the field of itsintention. The medium, on the other hand, ”bitesback“. It suggests other uses, may not quite work aswewant (...) with amedium, the side-effects may be-come more important than the original intentions.”(Glanville 1994, p.11)In this chapter, we will discussparametric design as a design medium which is im-measurable in potency and in its ability to help ourthinking - and thus can take a role as a partner in en-hancing our creativity.The second thesis is that theserendipity is an early, especially turning point of thevery process of evolution. Serendipity is the effectby which one accidentally discovers something for-tunate, especially while looking for something elseentirely. As Lawrence Block said: “Serendipity. Lookfor something, find something else, and realize thatwhat you’ve found is more suited to your needs thanwhat you thought you were looking for.” (Parker andTalbott 2008)Traditional CAAD systems limit design-ers‘ creativity by constraining them towork with pro-totypes provided by the system’s knowledge base.The design is creative if it cannot be composed fromthe prototypes in the system’s knowledge base. ACAAD system supports creative design if it allows thedesigner to define novel prototypes to cover theseideas. It is creative if it discovers new prototypesby itself. If we analyse the process and results of

382 | eCAADe 35 - PARAMETRIC AND GENERATIVE DESIGN - Volume 2

Figure 2Parametric designWorkshop at BUT -Painting by sound,A. Krzywinska andA. Bentkowska

the implementation of parametric methods, we canconclude that most of the obtained forms totally dif-fer from designers’ expectations. (see Figure 2) Asubjective/emotional factor has a great effect on thedecision-making process in designing. Intuition, un-predictability and no logic are the essence of creativ-ity. In design, often an inappropriate use of toolsgivesbetter results than theproper. Thiswayofwork-ing with media may surprise and stimulate the de-signer, offering himor her unforeseen shapes and so-

lutions. The goal is to change creative boundariesof contemporary design tools.In many publicationson architectural parametric design we may read thestatement that Architecture is simply the collectionof principles and operational requirements that arebeing applied in order to solve design problem. Thisleads to the conclusion that architectural design is anobjective based activity whose primary purpose is todefine what is required to provide a solution. If ar-chitecture is an objective-based activity, then what

PARAMETRIC AND GENERATIVE DESIGN - Volume 2 - eCAADe 35 | 383

Figure 3Parametric designWorkshop at BUT -Marching Cubes -jewellery, M. Cycholand M. Bernecki

place do emotions and subjective meanings take inthis activity?We need the theories and methods forinnovation and creation. If we treat parametric de-sign as a medium, we should formulate new param-eters for aesthetic design on the conceptual designlevel. These parameters usually result from subjec-tive assumptions. (see Figure 3) To connect para-metric design and creativity, Lee proposed to adoptthe concepts of divergent and convergent thinking,two critical factors in the creativity model, to under-

stand parametric design. In parametric design, diver-gent thinking generates a variety of solutions withthe parameters as the “potential answer” to a ques-tion, while convergent thinking identifies the mostappropriate solution as the “right answer” to a ques-tion with rules. (Lee 2014, p. 266)An interesting ap-proach, in which the subjectively selected parame-ters and the optimization process were connected,was undertaken at the Faculty of Architecture, Bi-alystok University of Technology. The goal was to

384 | eCAADe 35 - PARAMETRIC AND GENERATIVE DESIGN - Volume 2

Figure 4Mobile structure fordividing the spaceof an auditorium, M.Cychol, M. Berneckiand E. Marcinowska

create space elements which may be used as a mo-bile partition for dividing the space of an audito-rium. The auditorium is a multifunctional space forlectures, consultation, meetings, and exhibitions. Or-ganisation of such different activities simultaneouslyneeds mobile dividing partitions, while until now alarge black box was used. Students wanted to makethe space more attractive by introducing a new ele-ment. An additional function of this element shouldbe a space for leaving information (“a mailbox”) anda bench for sitting. Because these elements shouldcontrast with the boxes, it was decided to use curvedlines. The first stepwas tomodel the black boxes and

then draw curves which defined multi-curved vol-ume. In the next step, in the Grasshopper, this vol-ume was divided onto vertical and horizontal parts.Then the algorithm for creation of frame joints wascreated. As a result, it was possible to generate theoptimum contours of the ribs and prepare informa-tion for CNCmachine. The next stage of optimizationwasperformedafter determining the thicknessof theparticle board plate. After design, the elements werecut and assembledwithout the need to use any tools.(see Figure 4)

PARAMETRIC AND GENERATIVE DESIGN - Volume 2 - eCAADe 35 | 385

CONCLUSION - IS PARAMETRIC DESIGN ANEWDESIGN PARADIGM?The Oxford English Dictionary defines the basicmeaning of the term paradigm as “a typical exam-ple or pattern of something; a pattern or model”. Thehistorian of science Thomas Kuhn in his book TheStructure of Scientific Revolutions has defined a sci-entific paradigm as: “universally recognized scientificachievements that, for a time, provide model prob-lems and solutions for a community of practitioners”(Khun 1996, p.10). On this basis, wemay assume thatthis is the replacement of one methodology over an-other on the basis of a consensus of the majority.However, in the case of architectural design it seemsto be a premature statement. It is difficult to ac-cept the idea of a “majority consensus” in the case ofreplacing traditional computer-aided design meth-ods by parametric design. Opinions about the ar-chitectural revolution causedbyparametricmethodsare exaggerated. The history of architecture showsthat designs have always been created in relation tochanging factors - climate, technology, usage, envi-ronment, culture, andeven the character of thebuild-ing, in other words: they have always been designedparametrically. As was shown in Chapter 1, para-metric architecture was the subject of architecturalconsiderations in the 1960s - a few decades beforethe digital revolution.Nowadays, parametric designis one of many design methods. If we analyse theprevalence of the use of this method, we can notethat it is only used in a small number of prestigiousdesign offices and innovative schools of architecture.We can only hope that this method will be appliedmore widely in the future.The research was carriedout in the framework of statutory work S/WA/1/2016and financed by the Ministry of Science and HigherEducation

REFERENCESBozer, YA, Meller, RD and Erlebacher, SJ 1994, ’An

Improvement-Type Layout Algorithm for Single andMultiple-Floor Facilities’,Management Science, 40(7),pp. 918-932

Buffa, ES, Armour, GC and Vollmann, TE 1964, ’Allocat-ing facilities with CRAFT’, Harvard Business Review,42, No. 2, pp. 136-158

Coons, SA andMann, RW 1960, ’Computer-AidedDesignRelated to the Engineering Design Process’, Techni-cal Memorandum Electronic Systems Laboratory, De-partment of Electrical Engineering MIT, 8436-TM-5,pp. 1-29

Donaghey, CE and Pire, VF 1991, BLOCPLAN-90, User’sManual, Industrial EngineeringDepartment, Univer-sity of Houston

Emerson, H 1911, ’Philosophy of efficiency’, The Engi-neering Magazine, 41(Reprinted in classics in Indus-trial Engineering, Prairie Publishing Co., Delphi, Ind.1964), pp. 23-26

Hosni, YA, Whitehouse, GE and Atkins, TS 1980, MICRO-CRAFT Program documentation, Institute of Indus-trial Engineers

Jabi, Wassim 2013, Parametric Design for Architecture,Laurence King, London

Kuhn, TS 1996, The Structure of Scientific Revolutions, Uni-versity of Chicago Press, Chicago

Lee, Ju-Hyun, Gu, Ning and Williams, AP 2014, ’Paramet-ric design strategies for the generation of creativedesigns’, International Journal of Architectural Com-puting, 12(3), pp. 263-282

Lee, R and Moore, J 1967, ’CORELAP – computerized Re-lationship Layout Planning’, The Journal of IndustrialEngineering, 18(3), p. 195–200

Nicknam, M and Elnimeiri, MM 2011 ’A design optimiza-tion workflow for tall buildings using parametric Al-goritm’, CTBUH Conference, Seul, pp. 561-569

Parker, M and Talbott, D 2008, Silver Linings: Meditationson Finding Joy and Beauty in Unexpected Places, RedWheel/Weiser

Ross, DT 1960, ’Computer-Aided Design: a statementof objectives’, Technical Memorandum Electronic Sys-tems Laboratory, Department of Electrical Engineer-ingMIT, 8436-TM-4, pp. 1-26

Sheng-Fen, Chien and Yee-Tai, Yeh 2012 ’On Creativityand Parametric Design: A preliminary study of de-signer’s behaviour when employing parametric de-sign tools’, Proceedings of the 30th eCAADe Volume 1,Prague, pp. 245-253

Tompkins, J,White, J, Bozer, Y, Frazelle, E, Tanchoco, J andTrevino, J 2003, Facilities Planning, JohnWiley & SonsInc.

386 | eCAADe 35 - PARAMETRIC AND GENERATIVE DESIGN - Volume 2