COGNITIVE PRODUCT DEVELOPMENT: A METHOD FOR CONTINUOUS IMPROVEMENT OF PRODUCTS AND PROCESSES

385

�� !"��"#$

��

�� !"��"#$

%&' ��(�

��

��

)�*+��

&��%�� ,��-��,!.��,��'��!-��/��0

��

�� !��

�� "�� #��

�� $ �� "��

�� %��"��

��& �� #��

��

�� !��

�� "�� !�� !��

� ��& �� "��

�� "��!�� "��

�� & ��

��

1��(2�� 0(

��

��

3�4��05��(�(��

3�4��4��056�(�(��

*Corr. Author’s Address: &��%�� ,��-��,!.��,��'��!��7��0��3�� !��8��6!� ��'/&��!-��/��0 !8��9�� (��

1 INTRODUCTION

1.1 The High Costs and Risks of ProductDevelopment

The development of complex systems suchas buildings, plants, infrastructures, off-shorestructures and aircrafts, has a high risk of budgetand time overruns. In the construction sector, forexample, budget overruns between 10% and 30%happen frequently. But overruns of 80% or moreare not exceptional [9]. Also the aerospace,automotive and railway industries are often plaguedby serious budget and time overruns. These costsand risks make many enterprises reluctant with theintroduction of new products or the investment innew projects.

Although many different factors maycontribute to these overruns, two factors appear tobe essential for most cases: (1) problems causedby high complexity, and (2) the unpredictability ofconsequences of ‘new’ knowledge.

The complexity of a system can be definedas the total number of interactions or

interdependencies between components of asystem [13]. Complexity depends on the numberof components, but tends to grow more thanproportional to this number. Also the number ofinteraction- or dependency-types may increasecomplexity. Examples of interaction types aremechanical interaction (such as mechanicalfixation), electrical-, chemical-, and controlinteraction. If a system has n components and iinteraction-types, it may have maximally i.n.(n-1) interactions with other components.Complexity can thus be reduced by reducing thenumber of components or by reducing thenumber of interactions or dependencies. Thelatter can be accomplished by modularizing adesign such that each module behaves as a ‘blackbox’ that has minimum interactions with itsenvironment.

On the other hand, the trend towards ‘mass-customization’, i.e. the offering of client specificsolutions based on a generic design, increasesproduct complexity because designers have to keepall variant-solutions and their consequences inmind.

Gielingh.pmd 10. 06. 08, 17:05385

386

�� !"��"#$

%��'

A new product is the result of ‘new knowledge’.The consequences of this knowledge are oftenunpredictable. That is why a design has to beverified thoroughly through modeling, analysis,simulation and testing, before the actual product isbeing realised. For the automotive sector it isestimated that design changes make up 75% of totalproduct development time and costs [1].

Product development may thus be improvedin terms of costs, risks and quality if the problemscaused by high complexity and the consequencesof “new knowledge” can be reduced. This subjectwill be addressed by combining principles forsystems modelling with a theory on cognitivepsychology that is generalized and extended fordesign, engineering and production.

1.2 Origin of This Theory

The methodology that is described in thispaper finds its origin in a theory for productmodelling developed by the author in the 1980-ies. After several implementations, applications andfurther refinements it evolved into a theory andmethodology for the acquisition, organization anduse of product and process knowledge. Milestonepublications were: (a) the General AEC ReferenceModel [3] which became part of the Initial Draftof ISO 10303 STEP; (b) the IMPPACT ReferenceModel [4], which was implemented for integrateddesign and manufacturing of ship propellers atLIPS and the design and manufacturing of aircraftcomponents at HAI; (c) the PISA Reference Model[5], which was implemented to support the earlydesign process of cars at BMW; and (d) the Theoryof Cognitive Engineering [6] which was partiallyapplied in a large Design-Build-Maintain projectin the oil and gas sector. This paper presents acomprehensive summary of the last - most recent -theory.

1.3 Structure of This Paper

This paper starts with an introduction of theTheory of Cognition which is founded on a wellknown theory for cognitive psychology (chapter2). Next, this theory is generalized and extendedfor design, production and product lifecycle support(chapter 3). Chapter 4 discusses in brief a Theoryof Systems. It addresses in particular the principlesof modularization of systems. Subsequently chapter

5 combines the theories unfolded in 3 and 4 into atheory for cognitive design, production and supportof complex systems. Chapter 6 describes one ofthe cases where these principles are applied: a largedesign-build-maintain project for the oil and gassector. Chapter 7, finally, draws conclusions.

2 KNOWLEDGE CREATION ANDCONTINUOUS IMPROVEMENT

2.1 Continuous Improvement

Continuous Improvement (CI) is a widelyused concept with a variety of meanings. For manyit is synonymous to innovation, for others it is acorner stone of total quality management.

CI is a basic element of LeanManufacturing, where it is known as the kaizenprinciple [16]. In this context CI aims primarily atquality management. It is implemented here as anorganizational solution, by making teams of peopleresponsible for improving their own part of theproduction process.

Bessant and Caffyn [2] define CI as ‘anorganization-wide process of focused and sustainedincremental innovation’. Lindberg and Berger [8]propose a model, identifying five types of CIorganization, which is based on two dimensions.The first dimension addresses whether CI is partof ordinary tasks or not, the second makes adistinction between group tasks and individualtasks. In practically all studies, focus is on thehuman aspect of Continuous Improvement.

2.2 A Theory of the Learning Organization

Also theories about learning organizations arehuman centered. Probably the most well known oneis the SECI model of Nonaka et.al. [11] and [12].

According to this theory, design knowledgewithin an organization is developed according to aspiraling process that crosses two arrays ofapparently opposite values, such as chaos and order,micro and macro, part and whole, implicit andexplicit, body and mind, deduction and induction,emotion and logic. According to these authors, thekey for successful knowledge management is tomanage dialectic thinking in order to resolveapparent conflicts in design objectives.

The spiral develops in an interactionbetween implicit (i.e. residing in the heads of

Gielingh.pmd 10. 06. 08, 17:05386

387

�� !"��"#$

��

human individuals) and explicit (i.e. recorded,transferable and shareable) knowledge.

Four stages of knowledge transformationare recognized: Socialization, Externalization,Combination and Internalization, to be abbreviatedas SECI.

Socialization is a process in whichindividual implicit knowledge is transformed intoshared implicit knowledge, for example through(informal) meetings, discussions and other formsof human interaction.

Externalization is the expression and/orrecording of knowledge with the objective tobecome a shared resource for the organization.

Combination is a process aiming atunification, integration and/or generalization ofindividual pieces of shared knowledge, whichresults in knowledge of higher value for theorganization.

Internalization, finally, is the process inwhich individual members of the organization pickup shared explicit knowledge, for example throughreading, learning, training and experiencing.

The process of sharing depends heavily onthe existence of a platform of common experiences,values and concepts. Such a platform providescontext that is necessary for the understanding ofwords and gestures. It forms the ‘commonality’ ofa shared domain. This platform is called ‘ba’ byShimizu [14]: ‘ba’ is the Japanese word for ‘place’or ‘location’. ‘Ba’ can be a physical or a virtualplace, and represents not only a location in spacebut also a location in time.

Nonaka’s theory comprises further thenotion of Knowledge Assets (KA’s): the different

forms in which knowledge is encapsulated.Examples of KA’s are experiences (implicitknowledge developed through practicing), routines(business practices that may exist in implicit orexplicit form), concepts (common ideas) andsystems (explicit knowledge, recorded in the formof documents, databases and models).

2.3 An Assessment of the SECI Model

Nonaka’s theory focuses largely on (a) theinteraction of human individuals with otherindividuals through socialization and (b) the sharingof knowledge in explicit (i.e. documented) form. Itdoes not incorporate feedback from real worldexperiences. Also, it does not address the possiblerole of more advanced forms of knowledgerepresentation, such as in the form of product models.

Product models that are used for simulation,such as structural analysis, virtual reality and thedigital mock-up, may play an essential role in thelearning cycle of a design and engineering team.

Product modelling requires a paradigm shiftin industrial production, because its nature is sodifferent from document based working. A productmodel is a near-to-reality ‘image’ of a product thatmay exist on different levels of concreteness andcompleteness. Product models result in (virtual)experiences for the human beings that work withthem. In contrast, documents must be read in orderto provide knowledge for the reader and are henceonly accessible for those who master the languagein which they are written.

Consequently, there is a need to make KMtheory consistent with modern scientific principles,including feedback from virtual or real experiences.

2.4 Theory of Cognition

The missing link between KM theory andfeedback from reality is provided by a moderntheory on cognitive psychology [10]. Neisserdefines Cognition as ‘ the acquisition, organizationand use of knowledge’. As his theory originated inthe context of psychology, it is focused on thehuman individual. This part of the theory will bediscussed first; in section 3 it will be generalizedand adapted to learning organizations for designand engineering.

Experiences in the immediate and remotepast influence human cognition. Experiences are

Fig. 1. The SECI process as proposed by Nonakaet al.

socialisation

individual -> community

externalisation

informational express ion

internalisation

informational impress ion

combination

partial -> integral

implicitknowledge

explicitknowledge

socialisation

individual -> community

externalisation

informational express ion

internalisation

informational impress ion

combination

partial -> integral

implicitknowledge

explicitknowledge

Gielingh.pmd 10. 06. 08, 17:05387

388

�� !"��"#$

%��'

memorized and organized by the human brain.Similar experiences confirm and reinforce eachother, and result in abstract structures in the humanmind that are called schemata. For example, aperson who has seen dozens of dogs will developan abstract idea that combines the observed featuresthat all dogs have in common. This abstract ideabecomes a concept [15]. Concepts may becomeindependent sources of knowledge. By associatingconcepts with symbols such as words, knowledgecan be communicated to other people.

Concepts and conceptual structures have abiological origin: they enable a human being toanticipate and act more effectively in newsituations. A child that has touched a hot stove onceor twice will associate the two concepts ‘stove’ and‘hot’, and will be more cautious in next encounterswith stoves. Similarly, once we have eaten manyapples, we associate the shape and colour of appleswith their taste: we know that small green applescan be hard and sour, and that yellow or red onesare mostly sweet and soft.

The human senses provide an enormousamount and a continuous stream of sensory stimuli.Only some of these are really of importance. Theextraction of useful stimuli from the irrelevant onesis called perception [10].

As the life-experiences of individual peoplediffer, the understanding and interpretation ofsensory stimuli, in the form of new experiences,will also differ. Hence, two people may act andreact differently if they are confronted with oneand the same situation. For example, a person whois once attacked by an aggressive dog will have adifferent concept, and may act differently, than aperson who never had such an experience.

From the above it can be concluded that ‘oldknowledge’ plays an essential role in humanperception, an thereby affects the creation of ‘newknowledge’. Existing knowledge determines hownew information is interpreted and valued. Theentire process of sensing and the interpretation ofsensory stimuli by a human being will be calledimpression.

Learning is not just a passive process thatis based on the observation of physical reality.Much can be learned by exploration andexperimentation. Baby’s learn by touching things,by putting them in their mouth, and by throwingthem away. Children learn by playing, which is acombination of action and observation. Action

partially affects and changes the physical realitythat surrounds us. The whole of activitiesperformed by human beings that affect physicalreality will be called expression.

The learning process of human individualscan thus be depicted by a circle that is intersectedby two orthogonal axes (Fig. 2). The vertical axisrepresents physical reality (top) versus knowledgeabout reality (bottom). The horizontal axisrepresents impression, which includes processessuch as sensing, observing, interpreting andperceiving (left) and the process of expression,which includes various forms of acting (right).

The cognitive process that is depicted inFigure 2 applies to individual people but it can alsobe applied to organizations, such as industrialenterprises. For enterprises, physical reality mayform a market. Market analysis, or theinterpretation of client needs, is a form ofimpression. To serve this market or these clientswith products and/or services is a form ofexpression. Once these products and/or services areconsumed or applied, physical reality has changed.These changes may form input for product orprocess innovation.

The cognitive cycle may also be applied toelectronic knowledge processing systems.Knowledge about existing reality may be obtainedvia input devices, such as sensors or measuringequipment, or through human observations that aredocumented. And existing reality can be changedthrough output devices such as CNC machinery,process control systems or documentedinstructions.

The cognitive cycle forms thus the basis fora theory about learning enterprises and learning

Fig.2. The Cognitive Cycle

physicalreality

knowledge

expression

directs

modifies

addsexperiences

to

impression

isobserved

through

Gielingh.pmd 10. 06. 08, 17:05388

389

�� !"��"#$

��

information systems that, in contrast to Nonaka’stheory, involves reality.

3 RETHINKING DESIGN AND PRODUCTIONAS A SPECIAL FORM OF COGNITION

Concepts that result from experiences withreal phenomena can be analyzed and decomposedinto conceptual primitives. These primitives - orfeatures - can be manipulated in the human mindto form new concepts. A simple but illustrativeexample is the mermaid, which originated in themind of Hans Christian Andersen. This imaginarycreature is partially a girl and partially a fish. Themermaid is an example of an imaginary concept:although mermaids do not exist and cannot beobserved, it is ‘assembled’ from features that canbe sensed. As a concept it can also be visualized inthe form of a naturalistic expression, such as apainting or a statue.

A new product results also fromimagination. Designs of new products areassemblies of features that are extracted fromexisting reality. These features comprise knowledgeabout shapes, materials, techniques andtechnologies, and are manipulated in the mind: theycan be resized, reshaped or re-arranged.

If the cognitive cycle is applied to industrialenterprises, then “physical reality” includesphysical products, clients and markets,“impression” includes analysis, “knowledge”includes technological and production knowledgeas well as design, and “expression” includesproduction and servicing. The cognitive cycle forproduct creation is shown in Figure 3.

The design of a new product usually has tobe verified before it can be produced. Such averification can be done by making physicalprototypes, or by asking experts to analyze andapprove the design. With the advent of CA-technologies, it is now also possible to verify adesign in virtual reality through product modelsand simulation technologies (Fig. 4).

As a model is an abstraction of reality, it isused to check some - but not all - properties of theanticipated product. In early design phases, only afew properties are checked, and the more designprogresses into detailed design, the more propertiesare added.

Hence, the cognitive cycle is not traversedonly once for the development of a product, but manytimes. Each successive traversal of the cognitive loopadds more detail to the product specification, up toa point where sufficient knowledge is acquired sothat a safe, error-free production process can beexpected, and the resulting product is likely to meetclient and market expectations. This iterative processcan be depicted by expanding the cognitive cycleinto a spiral (Fig. 5).

In this figure, the design process is supposedto start with an initial idea (Product Concept L0),which is modelled and simulated in virtual realitythrough Model L0, and which is subsequentlyanalyzed. Based on the outcome of this analysis amodified and/or more detailed specification is made(Product Concept Ln), modelled and simulated, andso on. This process ends once the final specification

Fig. 3. The cognitive cycle for product creation

physical reality

knowledge

production(expression)

directs

creates

addsexperiences

to

analysis(impression)

isobserved

through

feed

bac

kkn

ow

led

ge

Imag

inat

ion

(des

ign

)

exis

tin

gp

rod

uct

s

new

pro

du

ct

Fig. 4. To improve speed and quality of a design,physical reality can be simulated through virtual

reality as part of the cognitive productdevelopment cycle.

physical reality

productionanalysis

exis

tin

gp

rod

uct

s

new

pro

du

ct

virtual reality

imag

inar

yp

rod

uct

modelling

knowledge

feed

bac

kkn

ow

led

ge

Imag

inat

ion

(des

ign

)

physical reality

productionanalysis

exis

tin

gp

rod

uct

s

new

pro

du

ct

virtual reality

imag

inar

yp

rod

uct

modelling

knowledge

feed

bac

kkn

ow

led

ge

Imag

inat

ion

(des

ign

)

Gielingh.pmd 10. 06. 08, 17:05389

390

�� !"��"#$

%��'

is ready (Product Concept Lp), after which productioncan start. Production is the final stage of expression,resulting in the intended physical product.

The Figure 5 shows that the cognitive loopdoesn’t stop there. The physical product, once inuse, can be considered as a prototype for a futureproduct. Knowledge that is acquired from theproduct in actual use can be very helpful for thecreation of new products, for example for analysisand simulation purposes.

A product concept is not limited to a staticdescription of the product. It may also apply toprocesses, such as production, maintenance andoperation processes.

The whole of models, analysis results,performance data and other information forms aninterrelated structure of product and processknowledge that can be used as a basis for newdesigns. This structure will be briefly described inthe next chapter.

4 THEORY OF SYSTEMS

Modern products can be seen as complexsystems, consisting of objects, where each objectinteracts with one or more other objects to form a

functional whole. Hence, because of theirinteraction, the whole is more than the sum ofobjects that form its parts. An object can be a systemby itself, in which case it is called a sub-system.Complex systems may have multiple levels of sub-and sub-sub-systems. And as the performance of aproduct is determined by its behaviour in itsenvironment, the product itself can also beconsidered as a sub-system.

Systems are often modelled and depictedgraphically by means of an inverted tree structure.The top (or root) of this inverted tree depicts the whole;the branches depict the parts; see also Figure 6.

The terms ‘whole’, ‘system’, ‘sub-system’and ‘part’ have no absolute meaning. Parts maybe seen as ‘wholes’ or ‘systems’ in their own right.Any object of which a model is made, is part of alarger whole: buildings are part of cities, whilecities are part of regions or nations, and so on. Onthe other side, even the smallest object that ismodelled consists of things that are smaller.Hence, no absolute dividing line can be drawnbetween the model of a system and the context inwhich it is placed. For this reason, the presentedtheory does not use the terms ‘system’ or ‘part’.These terms are only used for explanatory reasons,

Fig. 5. A top-down design process can be depicted in the Cognitive Cycle by a spiral

�

Conceptual World

��

��

��

��

��

��

Physical Reality

More detail, m

ore specific

��

��

Expression

Impression

More properties, m

ore realistic

��

��

��

��

��

��

��

��

PerformanceMeasurement

��

��

Feedback forFuture products

Virtual Reality

Gielingh.pmd 10. 06. 08, 17:06390

391

�� !"��"#$

��

and are therefore placed between parentheses inFigure 6.

The present theory is about knowledge ofsystems, and considers therefore knowledge as asystem by itself. The circles in Figure 6 refer toUnits of Knowledge (UoK’s). These Units maycomprise any kind of knowledge about anysubject. For reasons of comprehensibility, Unitsof Knowledge will also be referred to as‘knowledge objects’, or simply ‘objects’. Aknowledge object may refer by itself to a physicalobject, a (physical) process, a feature of a physicalobject, or other phenomena that are subject ofinterest.

Modern enterprises operate today incollaborative networks. A vehicle, for example, isnot designed and built by a single company, but bymany companies that together form a supply chain.As a consequence, a part or subsystem within avehicle may have two intellectual owners: the OEMand the supplier. The OEM defines requirementsand boundary conditions for the part or subsystem,while the supplier proposes a solution for it. Thesupplier may, on his turn, have its own supply-chain.

This idea is depicted in Figure 7 by splittingeach circle in two halves. The upper halverepresents requirements and boundary conditions,the lower halve the proposed solution. This ideawas first proposed as part of the General AECReference Model [3], where the upper halve wascalled ‘Functional Unit’ and the lower halve‘Technical Solution’.

The importance of this concept is thatknowledge about objects in a system can now be

modularized, where modules of higher aggregationsubsume modules of lower aggregation.

The split between Functional Units andTechnical Solutions does not have to be restrictedto knowledge transactions between differentcompanies. Also a single company can benefit fromthe modularization of a system model.

The Functional Units in a system model mayform network relations with other Functional Unitswithin the same module, or with other modules onthe same level of aggregation.

Each module can be described at twodistinctive levels: (1) generic, parametricdescription, and (2) specific description, where allparameter values are defined or where objects aredescribed in explicit non-parametric form.

The specific description is split into threesub-levels: (2a) Lot (i.e. one or more identicalobjects), (2b) Individual (a single individual object)and (2c) Occurrence (a moment in the life of anindividual).

The modules should not be made too largeso that the number of interactions or dependenciesinside a module remain limited. Complexityreduction makes it possible to describe all moduleswith parametric technology. The resulting hierarchyof modules is capable to represent any system,regardless its overall complexity, using parametrictechnology.

A top-down oriented design process usuallystops at a point where pre-existing solutions existthat fulfill the requirements of correspondingFunctional Units.

More details about the theory of lifecyclemodeling are given in [6] and [7].

(system)

(sub-systems)--

(sub-sub-systems)

(system)

(sub-systems)--

(sub-sub-systems)

F

S

F F F

FFFFFFFF

SS S

SSSSSSSS

Fig. 6. System composition can be modelled inthe form of an upside-down tree

Fig. 7. Modular system decomposition

Gielingh.pmd 10. 06. 08, 17:06391

392

�� !"��"#$

%��'

Conceptual World

��

��

��

��

Physical Reality

��

��

Expression

Impression

More properties, m

ore realistic

��

��

��

��

��

��



Virtual Reality

Conceptual World

��

��

��

��

Physical Reality

��

��

Expression

Impression

More properties, m

ore realistic

��

��

��

��

��

��



Virtual Reality

5 PRODUCT CREATION AS A COGNITIVEPROCESS

The system model that is described inchapter 4 fits into the cognitive productdevelopment process such as described in chapter3. While a top-down design process progresses, ituncovers several layers of detail, each layercorresponding with a level of the compositionhierarchy of the system. This continues until pre-existing solutions are found that meet therequirements of corresponding Functional Units.These final (pre-existing) solutions are depicted bythe halve circles marked F at the bottom of thisfigure.

The system modules that are depicted byrectangles with rounded ends in Figure 7 may alsobe based on pre-existing solutions. These solutionsmay be reused in parametric form, so that parametervalues still have to be defined, or in explicit, non-parametric form. In either case it will be possible

to replace older solutions that were chosen inprevious designs by new solutions. This principleis shown in Figure 9.

It shows on the left the configurationhierarchy of the design of an initial product, andon the right of a revised design, possibly of a nextversion of this product. The solutions colored whiteare reused without modification. The solutionscolored black are new. The solutions colored greyare reused but with some modification. The lattergroup may make use of the same generic(parametric) template, but with different parametervalues.

This idea can be remotely compared withthe configuration of a personal computer. At thetop-level of a computer model, a computer has aprocessor, primary memory and secondarymemory.

The architecture of a computer is such thatfor each functional unit different solutions can beinstalled: a computer may use different processors,

Fig. 8. Modular system decomposition as an integral part of the cognitive design process

Gielingh.pmd 10. 06. 08, 17:06392

393

�� !"��"#$

��

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F F

S

F

S

F

S

Product Version 1 Product Version 2

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F F

S

F

S

F

S

Product Version 1 Product Version 2

S

�

F

S

F

S

F

S

F

S

SSSSS

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

FFFFFFF

F

S

F

S

F

S

F

S

F

S

F

S

F

S

SSSSS

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

FFFFFFF

ConceptualHierarchy

S pecificationHierarchy

Ins tallationHierarchy

F

S

F

S

F

S

F

S

SSSSS

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

FFFFFFF

F

S

F

S

F

S

F

S

F

S

F

S

F

S

SSSSS

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

F

S

FFFFFFF

ImaginaryPhase(design,planning)

ActualPhase(production,cons truction,operation,maintenance,disposal)

specification

installationa b

realworld

experience

combinationgeneralisation

knowledgefeedback

ed

cf

different primary memories and different secondarymemories. To replace one by another is often assimple as pulling out one card or device andplugging in another. Although two primary memorycards may have different capacity and be producedby different vendors, the chips or other electroniccomponents may be obtained from the same sub-

supplier. Hence, re-use of solutions may occur onany level.

Using this principle, design becomesbasically a process of configuring solutions.

If each module – i.e. each solution – is tracedfrom design to subsequent lifecycle phases,lifecycle performance knowledge becomes

Fig. 9. Solutions (or specifications) are marked with an S. Reused solutions that are not changed arecoloured white, the ones that are changed but based on the same generic template are coloured grey,

and fully new solutions are coloured black

Fig. 10. The three types of hierarchy, filled with imaginary or actual data, form the basis for cognitiveproduct development. It starts at the lower left corner (a) with the specification of a product using thegeneric design objects. During and after the realization of the product, real world data are collected,

analyzed, combined and generalized, and made available for use and re-use in future projects.

Gielingh.pmd 10. 06. 08, 17:06393

394

�� !"��"#$

%��'

available for each object. This knowledge will beacquired for individuals at different occurrences,but can be generalized and, after statisticalprocessing, be made available to generic(parametric) descriptions. This means that for allobjects marked white or grey in Figure 9 productlifecycle knowledge is available for design. Thisknowledge can be used for gradual improvementof a design.

The result is that the generic parametricdesign objects in design systems provide access toa huge amount of lifecycle data associated withprevious applications and implementations. Themore this system is used, the more experiencesbecome available for design. Design then becomesa cognitive process, where generic, parametrictemplates built on top of the knowledge acquiredby other experts in the product lifecycle play therole of cognitive schemes.

Figure 10 shows horizontally the three typesof hierarchy (i.e. the conceptual hierarchy ofimplicit parametric objects, the specificationhierarchy of explicit objects, and the installationhierarchy of individual objects) and verticallyactual versus imaginary data.

The cognitive cycle starts with thespecification of a new product (10a, lower leftcorner). This specification can be either in explicitor implicit form. In the latter case, use is made of ageneric parametric model. The result is anImaginary Specification Hierarchy. The individualcomponents are derived from the specification,resulting in an Imaginary Installation Hierarchy (b).After realization of the product, real worldexperiences are collected first on an individual level(c) and then combined via statistical analysis (d).This knowledge can then be generalized and addedto a generic (parametric) product model. From thenon it will be available at the start-up of new designprojects.

After several traversals the solution basebecomes richer and offers increasing levels ofhistoric life-cycle knowledge to the designer.

6 APPLICATIONS

6.1 Early Applications Based on ParametricTechnology

Most principles described in this paper havebeen applied, implemented and improved in

projects of different kind and for different industrialsectors, such as mechanical products, ship-building,electronics, automotive and construction.

The first detailed software implementationwas for a manufacturer of interior walls in 1982,and is described in some detail in [7].

A second case was the implementation of afeature based, fully integrated CAD/CAM solutionfor a manufacturer of ship propellers. It is describedin [4].

Both cases led to significant efficiency gainsin the overall production process. But they lackedknowledge feedback to support the cognitiveprocess such as described in this paper.

6.2 Application in a Large Project for the Oiland Gas Sector

The case that will be described in moredetail here did address the latter subject. Forpractical reasons it was not based on parametrictechnology but on a data warehouse supported bya PDM system. This case concerns the realizationof 29 almost identical plants in a serial constructionprocess.

Below the surface of Groningen, a provincein the northern part of the Netherlands, lies one ofthe largest reservoirs of natural gas in the world.This reservoir is exploited since 1958 and is nowmore than half empty. As the natural gas-pressurehas dropped there was recently a need to installcompression units. Also, as the installations werenearing the end of their lifetime and had highoperational costs, there was a need to renovate theinstallations. The company that exploits this gas-reservoir - NAM, a joint venture of Shell and Exxon- decided to contract this huge effort as an integratedDesign-Build-Maintain project. The project startedin 1996 and has a duration of at least 25 years.

The natural gas is exploited via hundredsof pipelines that reach the surface of the Earth on29 locations, called clusters, distributed over a largearea of land (Fig. 11). Each cluster is equipped witha small plant for the drying and cleaning of thegas, and for the separation and processing ofpollutants. An aerial photo of one of these clustersis shown in Figure 12.

The 29 plants cannot be constructed all atonce. In the most favorable scheme between 2 and3 plants per year would be constructed.Consequently, construction of the last plant would

Gielingh.pmd 10. 06. 08, 17:06394

395

�� !"��"#$

��

start 12 to 15 years after the first one. In these years,construction-, operation- and maintenance personnelgain a lot of experience that can be used to improvethe quality of the design, the quality of processes,and the reduction of overall lifecycle costs.

Furthermore, technology and science willcontinue to develop. Equipment such as pumps andvalves, measuring devices and control systems willbe improved. It makes therefore sense toincorporate the potential of new knowledge into adesign. But, on the other hand, it may be adisadvantage for operation and maintenance if allplants become different. Hence, it was decided tostrive for an optimum between functionaluniformity and technical differentiation.

This problem was solved by organizing thedesign as a modular system, in line with theprinciples described in this paper. Whereverpossible, the same solutions would be chosen foreach plant, resulting in uniformity. This made italso possible to track lifecycle experiences,enabling building and maintenance processes to befurther optimized. The goal was to build the lastplant for 70% of the costs of the first. Newknowledge that could improve the design wouldbe incorporated in new modules that replace oldermodules. Application of modularization principleswas essential here: it had to be avoided that a simpledesign change would propagate too far in otherplaces of a design.

An important tool for the realization of thisconcept was the development of the knowledge

feedback system, see Figure 14. Knowledge createdin each process would be used by that process forcontinuous improvement. But this knowledge alsohad to be made available to the design and planningdisciplines, so that design and planning could befurther optimized. The latter is also called frontloading.

Three types of data and knowledge sourcesare identified; see also Figure 13. The first isautomated data collection from sensors and otherequipment in the plant. The second is non-automated data collection such as from inspectionreports. Inspection reports are directly entered asdata in a computer, such as lap-top or a hand-helddevice. These two sources of data are still raw andneed to be processed before they become useful.

Figure 13 shows that this is a two stageprocess. Raw data may be analyzed and diagnosedfor operational usage. Not all of that data is usefulfor other purposes. The filtered data are stored forlong term data analysis, such as for tacticalpurposes (maintenance planning and scheduling)and strategic purposes (continuous improvement).

Tactical analysis can be supported by aknowledge system, using rule based inference,while tactical analysis can be supported by datamining technology.

The third source is explicit knowledgerecording, such as in the form of idea’s andsuggestions for improvement.

The various kinds of data and knowledgeassociated with design modules are stored in a

Fig. 11. Northern part of The Netherlands withthe towns Groningen and Delfzijl. The light greyarea is the Groningen gas-field. Locations of gasproduction units (socalled clusters) are marked

with dots.

Fig. 12. Aerial photo of one gas productioncluster near a canal. The wells surface at the

light-grey rectangular area. The gas is treated ina plant before it is supplied to the international

gas distribution network.

Gielingh.pmd 10. 06. 08, 17:06395

396

�� !"��"#$

%��'

common data warehouse and managed with supportof a PDM system.

Apart from the PDM system, a large numberof other computer applications are used in thisproject, ranging from a variety of CA-applications,ERP, maintenance management and operationsmanagement systems. The logical knowledgestructure as described in this paper affects most –if not all – applications in use.

Practical limitations made it howeverunviable to change all applications according tothe principles outlined in this paper. The reasonwas that software changes had to be done in a fullyoperational environment, which would disrupt theongoing work too much. Therefore the principleswere applied as a working practice within theorganization, supported by the PDM/WFMsoftware.

Despite this restriction, the principles appearto be highly beneficial for the structuring andorganization of knowledge and supportedcontinuous improvement of the design,construction and servicing processes. Benefitsresult from cost reductions and higher end-uservalue. Lifecycle costs are estimated to be reducedbetween 25 and 30%, while the system also

contributes to other performance factors such ashigher availability, better reliability and increasedsafety [6].

Fig. 13. The GLT project uses three principal sources of data collection (top) that are analyzed andprocessed for operational, tactical and strategic usage

Fig. 14. The creation of a learning andinnovative organization by attaching knowledgecreated in all phases of the product lifecycle to

design objects. This results in discipline specificknowledge pools and an integrated knowledge

pool that is available for designers in newprojects.

Improvement Planning

AutomatedData Acquistion

Inspection Reporting

Operational Analysis

& Diagnostics

Tactical Analysis

Strategic Analysis& KnowledgeAcquisition

Knowledge Recording

raw dataraw data

Condition & Performance Data

Operation support data

Improvement Plans

Operation

Maintenance Planning & Scheduling

Design & MaintenanceImprovementPlanning data mining

rule based inference

rule based inference

& improvement concepts

Short Term Data Collection

Long Term Data Collection

Data Filtering

lessons learned

new inference rules

Facility 1

Facility 2

Facility n

KnowledgeFeedback

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

KnowledgeFeedback

discipline specific knowledge poolsfor process improvement

integratedknowledge pool for

product and process improvement

Front L oading

Front L oading

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

Facility 1

Facility 2

Facility n

KnowledgeFeedback

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

KnowledgeFeedback

discipline specific knowledge poolsfor process improvement

integratedknowledge pool for

product and process improvement

Front L oading

Front L oading

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

DesignandPlan

BuildOperate

andMaintain

Transferor

Dispose

Gielingh.pmd 10. 06. 08, 17:07396

397

�� !"��"#$

��

7 CONCLUSIONS ANDRECOMMENDATIONS

The theory and methodology that isdescribed here aims at the reuse and improvementof design and engineering solutions. It supportsContinuous Improvement (CI) as a task that is fullyintegrated with regular business processes. Genericdesign objects that are available in the design andplanning systems give designers and engineersaccess to actual performance data of these objectsin earlier projects. This enables them to learn fromthe past, even if the designers were not personallyinvolved in these projects.

The improvement process does not only relyon the creation of idea’s by people involved in eachbusiness process, but makes also use of the analysis ofdata that are collected automatically via sensors or viainspection reports, or of knowledge records.

Early implementations were based onparametric technology and led to substantialefficiency gains in business processes. Parametrictechnology offers the possibility to re-use designknowledge even in the context of entirely newspecifications. More details about theseapplications can be found in [7], [4] and [6].

A more recent application based on a PDMbased data warehouse closed the cognitive cycle andsupports a process of continuous improvement in alarge construction project for the oil and gas sector.

Parts of the theory have been published and/or used for standards for the exchange and sharingof product data, but are in these contexts notpresented as a solution for cognitive processes. Itcould be of interest for users and application vendorsto explore this aspect of the presented theory further.

Generic software that supports theory andmethodology via parametric technology did existin the past but was not maintained. Futureapplications would benefit from redevelopment ofthis software.

8 REFERENCES

[1] AIAG. Engineering change management -business case. AIAG CollaborativeEngineering Steering Committee, 2005.

[2] Bessant, J., Caffyn, S. High-involvementinnovation through continuous improvement,Int. J. Technology Management, vol. 14(1997), no. 1, p. 7-28.

[3] Gielingh, W.F. General AEC reference model(version 4); Document N329 of ISO TC184/SC4/WG1, Oct. 1988. Published as TNOReport BI-88-150.

[4] Gielingh, W., Suhm, A. IMPPACT referencemodel for integrated product and processmodelling. Springer-Verlag, ISBN 3-540-56150-1 / 0-387-56150-1, 1993.

[5] Gielingh, W., Los, R., Luijten, B., Putten,J.van, Velten, V. The PISA project, a surveyon STEP. Aachen: Shaker Verlag, ISBN 3-8265-1118-2, 1996.

[6] Gielingh, W.F. Improving the performance ofconstruction by the acquisition, organizationand use of knowledge. Delft, p. 372, ISBN 90-810001-1-X, 2005.

[7] Gielingh, W.F. A theory and methodology forthe modelling of complex systems. Submittedfor publication in the Journal for IT inConstruction, 2008.

[8] Lindberg, P., Berger, A. Continuousimprovement: design, organization andmanagement. Int. J. Technology Management,vol. 14 (1997), no. 1, p. 86-101.

[9] Morris, P.W.G., Hough, G.H. The anatomy ofmajor projects. John Wiley & Sons, ISBN 0-471-91551-3, 1987.

[10] Neisser, U. Cognition and reality - principlesand implications of cognitive psychology,New York, ISBN 0-7167-0477-3, 1976.

[11] Nonaka, I., Byosiere, P., Borucki, C.C., Konno,N. Organizational knowledge creation theory.International Business Review, 3 (1994).

[12] Nonaka, I., Toyama, R., Konno, N. SECI, Baand leadership: a unified model of dynamicknowledge creation. Long Range Planning,vol. 33 (2000), no. 1.

[13] Rocha, L. M. Complex systems modeling:using metaphors from nature in simulation andscientific models. BITS: Computer andCommunications News, Los Alamos NationalLaboratory, November 1999.

[14] Shimizu, H. Ba principle: new logic for thereal-time emergence of information. Holonics,5(1), 1995

[15] Smith, E.E. Concepts and thought. Thepsychology of human thought. Cambridge:Cambridge University press, ISBN 0-521-32229-4, 1988.

[16] Womack, J., Jones, D., Roos, D. The machinethat changed the world, New York, 1991.

Gielingh.pmd 10. 06. 08, 17:07397

Documents

COGNITIVE PRODUCT DEVELOPMENT: A METHOD FOR CONTINUOUS IMPROVEMENT OF PRODUCTS AND PROCESSES