A proposal to integrate TRIZ and CAD (Computer Aided TRIZ-based Design)

First presented at TRIZCON2001, The Altshuller Institute, March 2001.

Dr. Noel León-Rovira
Instituto Tecnológico y de Estudios Superiores de Monterrey,
Ave. Eugenio Garza Sada # 2501, Col. Tecnológico,
CP 6409, Monterrey, NL, Mexico. (email: nleon@campus.mty.itesm.mx)


This paper continues a series about the research work that is being undertaken at the Center of Design and Product Innovation at the Monterrey Institute of Technology (Mexico), looking for the integration of different design tools and methodologies to increase design effectiveness and productivity.

This time some theoretic reflections about the integration of TRIZ and CAD are presented with the objective to contribute to make this integration possible in a near future.

Computer Aided Design tools have evolved over the time and have become very useful for modeling the geometry but CAD software does not provides designers help in finding solutions to the increased amount of problems that arise when implementing embodiment and detailed design. It is analyzed how TRIZ tools as Inventive Principles, Substance Field diagrams, SUH diagrams and Trimming Modules may be integrated in a CAD environment to help designers to find solutions to conflicts arising during the embodiment and detailed design as they use CAD systems.

1. Introduction

This paper continues a series about the research work that is being undertaken at the Center of Design and Product Innovation at the Monterrey Institute of Technology (Mexico), looking for the integration of different design tools and methodologies to increase design effectiveness and productivity.

An integrated model of the Conceptual Design Process was presented at TRIZCON’99, which is based on integrating QFD, Functional Analysis and TRIZ [1]. Now some theoretic reflections about the integration of TRIZ and CAD are presented with the objective to contribute to make this integration possible in a near future.

It is intended to contribute to a reduction in product development time and to an improvement in quality and performance by creating the groundwork for integrating interactions between product development tools and methods, thereby allowing the exploration of alternatives.

Especially the integration between 3D Modeling CAD packages and TRIZ based Computer Aided Inventing Software could enhance the creativity needed for developing and improving products.

2. Background and state of the art

The product development process may be defined as the complex system of activities that produces the information required for bringing products to manufacture.

Therefore the design process is an information generating process, which starts with an abstract and often uncertain and confuse description of a new needed product performance. Based on this description a design team first creates the concepts and then the embodiment and detailed design required for manufacturing the product intended to satisfy the new need.

Very often during the design process information remains uncertain until design decisions are taken which allow going further and gathering new information derived from analysis or experiment done on the newly created models. This causes the design process to be iterative and recursive: that means going forward with uncertain information until the uncertainty may be overcome with information obtained through analysis of the recently created models. The designer may then continue or go back to the point where the uncertainty arise and redefine the decision about shapes, dimensions, physical principles, parameters, tolerances, etc. This traditional trial and error approach is expensive and time consuming.

Efficiently and effectively supporting the design process with computational tools and methods is limited due to the lack of integration of the methods and tools used.

Different software packages are commonly used to work on the required design tasks along the whole design process, as for example text processors; search engines; equations solvers and spread sheets; QFD software; drawing, drafting and painting tools; 3D-CAD packages; PDM and database managing tools; CAE software. Computer Aided Inventing tools based on TRIZ, are computer tools that have been used more often in the last years.

Although each of them is very useful, its lack of integration increases cycle time, while the knowledge burden on the designer keeps augmenting as more materials and more options become available.

After the Conceptual Design Process follows the Embodiment Design Process and a much bigger amount of decisions have to be taken at this stage. For example, Eppinger [2] has indicated that a copier redesign requires 400 people, 125 subassemblies, 2,000 engineering drawings, and over a million decisions. He has conducted a project named Information-Based Product Development and, from his point of view, product development information exchange today is more like a batch than a continuous process, as individuals or teams generate blocks of information which, only when complete, are passed along to those who need them. He says this leads to substantial delays.

Another statement supporting this view states that: “Complex products can contain millions of dimensions and characteristics (voltages, forces, etc.) each of which impact the performance of the product. However, only a small few of the millions of features, the Key Characteristics of the product, will significantly affect the final quality, performance, and cost of the product.” [3]

The annual Software Tools Symposium organized and facilitated by Finch [4] provided a forum for the sharing of both existing software tools and concepts for potential tools that arise from research. People from MIT, industry, and software developing companies came together to discuss how to adapt innovative ideas and tools from academic research into marketable products that American industry can use.

Chang [5] developed a new input method for conceptual design of mechanical assemblies that enables users to easily explore design concepts and design function by function not component by component. Unlike input methods in current CAD systems, his method enables users to input partial geometry and group it according to its functionality.

Flowers et. al. [6] conducted the project “Visualizing relationships in large information databases”, that was aimed to represent primarily textual information in a spatial context to take advantage of human visual processing skills. The databases used as case studies in this work include the patent database, the science citation index, and product development databases such as those used in tolerance buildup or identification of key characteristics.

Whitney et. al. [7], conducted the project “Information flow mapping to aid design of complex products”, looking for a better scale up of knowledge-based engineering (KBE) workstations, for products like cars and planes, which are complex and require coordination of many people and exchange of a lot of information. They are intended to be able to improve the effectiveness of product development projects by understanding the influence of information flow patterns, inherent in the product itself, on the way work is done.

Seering et. al. [8] are working on the concurrent application of tools, in order that these tools might be applied to the same product development program. Their objective is to develop an integrated demonstration of multiple tools applied to a single design problem that enables to identify differences and conflicts in the underlying assumptions of individual approaches.

The authors expect through this project to create a better understanding of the synergies and conflicts among tools and methodologies when applied to a single product development program.

Schön [9] expresses that during the design process a person needs be able to easily create a visual representation, even for abstract and verbal ideas, and then respond to it perceptually to discover new arrangements and shapes representing new ideas. The new concepts emerge from the visual representation.

The research question is how interactive systems can aid users in quickly creating and manipulating visual representations and whether they can support the discovery of new relationships, structures, and meanings in the materials.

Mitchell [10] points out that "... design is not description of what is, it is exploration of what might be. Drawings are valuable precisely because they are rich in suggestions of what might be.” He argues that designers frequently recognize emergent subshapes and subsequently structure their understanding of the design and their reasoning about it in terms of emergent entities and relationships -ones that they never explicitly input.

The whole purpose of the act of drawing might be seen to be in order to look at the result. In Gombrich's [11] discussion of Leonardo's creative process he suggested that: "in searching for a new solution Leonardo projected new meanings into the forms he saw in his old discarded sketches."

It is helpful to view the action process of drawing as quite independent of the perception process of looking. Reflection upon a design problem may lead to a drawing activity based upon a given structure that, through perception, generates emergent shapes that offer alternative structures to reflect upon.

The question may be: Is it possible to perform this activity without “looking”? Is it possible to perform automatically a drawing activity based upon a given structure, that generates new shapes which lead to alternative structures, which in turn lead to new structures and so on? The key question may be, how to “know” when the new structure matches the desired constrains?

The entity of a drawing might be thought of as a visual image together with an associated verbal description that imposes structure upon it. From this perspective, an emergent shape occurs when a revised description, or structure is discovered. The drawing looks the same but the verbal or structural description of it is new. In this case the new “structural description” is based on the recognition of “new” properties that had not been recognized yet.

Tann [12] tackles emergence, in the context of the use of line drawings in design, by formulating higher-level line descriptors, in terms of construction lines. The workshop addressed the issue of emergent shapes: the perception of shapes in a drawing that do not directly correspond to entities used to construct that drawing.

Stiny [13] describes four functions that can produce emergent shapes rather in the manner of a generative grammar. The basic strategy is, then, to take a drawing as described in its construction process, and to transform that description into an intermediate representation that enables the generation of emergent shapes. It was reported by Scrivener et al [14], that drawings considered in their bitmap, or pixel array, representation rather than the standard structural one were amenable to poststructuring. In other words, techniques were illustrated that could be used to impose structures on such a drawing that differed from the structures used to construct it. From this work, a considerable body of work has been undertaken in the development of a computational model of certain perceptual activities [15]. More recently attention has been given to how one might interact with drawings in which emergent shapes are perceived. The question here is how might the user economically indicate an emergent shape to the computer in order to manipulate it.

AID'96 was looking for papers on representations of visual information and reasoning processes on such representations that have played a very significant role in design problem solving and designers' reasoning. Compared to other types of problem solving activities, design is unique in the extent of use of visual representations. However, CAD tools merely make the creation of visual representations easy, but do not provide leverage on issues such as design reasoning.

It may be concluded that, from the CAD point of view, the research work has to uncover the semantics of models for design representations and how can such semantics may be represented in the computer and used in ways to intelligently assist the designer.

Following questions has to be addressed:

3. TRIZ and Product design

TRIZ has proved to be a very strong tool in helping to solve difficult technical problems that requires inventive thinking; that means problems where one or more technical contradictions are involved and which do not have known ways or means of solution.

TRIZ is not originally a tool that belongs to the classical product design methodologies and its place in the product design process has yet to be better identified in order to increase its efficiency. Some work has been already undertaken in this direction by Savransky [16] and Arciszewski and Zlotin [17].

Although not yet a comprehensive approach for the integration has been established and further work is being undertaken, several opportunities of synergy and need of improvement have been recognized between TRIZ and the embodiment design process.

It is known that useful ideas may be derived from the use of the contradiction matrix, during the conceptual design stage but less experience has been achieved in using the Contradiction Matrix during the embodiment design process. One possible way may be implementing some inventive principles more related to product design as segmentation, local quality, asymmetry, joining, nesting, counterweight, previous action, mediator combined with parameters of geometric nature as length of moving object, length of stationary object, area of moving object, area of stationary object, volume of moving object, volume of stationary object, weight of moving object, weight of stationary object to be used during the embodiment design process.

Furthermore the Matrix may be enhanced with solution principles that are often used to solve design problems but are not yet included in the matrix, as for example increasing the inertial moment of structural sections to solve the technical contradiction between strength and weight.

Other authors [18, 19] have recognized the need to enhance the Contradiction Matrix with new parameters and inventive principles that improve the success rate in using this tool. In our Center, further work is being developed in this direction.

It is known that SUH diagrams are widely used during the conceptual design stage, because they allow an extensive analysis of the possible solutions in order to increase Ideality.

Several modules in TechOptimizer (TO) are aimed to improve the product design as the Product Analysis Module that is mainly based on the functional decomposition and analysis of the products to be improved, to clarify what may be improved. The Trimming and Feature Transfer Modules help to complete the Product Analysis module by eliminating components or features of a product but maintaining its useful functions unaffected or improved or eliminating harmful side effects. The feature transfer module helps in transferring functions from one component or feature to another that requires improvement.

The product improvement process in TechOptimizer is supported with an extensive physical effects database, that helps in finding alternative physical ways of performing needed functions. However TO does not have an interface to CAD systems and therefore its recommendations has to be translated from designers in the CAD packages.


4. Ways and methods

Our approach involves identifying the relationships and role of textual and graphic information, during the product development process in an integrated context. The goal is to take advantage of both types of information representations. It is known that human visual processing skills are used to uncover relationships that arise when shapes and dimensions are combined into new design solutions. On the other side the rapidly developing computer searching and combining capabilities based on textual information may be used in semantic processors for creating new design concepts.

The new information generated during the design problem solving process may occur through finding, creation or a combination of both. Generation of new design information based on finding takes place during the adaptation of solutions gathered together based on search systems.

Generation of new design information based on creation takes place for example during the refinement of embodiment and detailing design, as new shapes and concepts representing new ideas emerge as response to perceptions and discovering, based on the human visual processing and abstraction skills. This is the case when computer aided modeling tools as (CAD) and Simulation software (CAE) are used to represent shapes and/or to analyze behaviors. Based on the graphical representation of the results, human designers are able to create, identify or infer new solutions.

Another case of design information generated through creation takes place during automated design or optimization processes. In these cases the new information created, is mostly a new combination of parameters based on invariant design concepts.

Generation of new design information based on the combination of finding and creation is the most complex type. This takes place for example when new qualitative solutions suggested through TRIZ tools are combined with the information generated based on emerging shapes and concepts during the embodiment and detailed design process.

The aim of our new approach is finding a way of integrating TRIZ based tools with 3D-CAD packages to provide designers help during the design process. This means providing this help just at the time designers are interacting with the user interface of 3D-CAD systems trying to find a design solution for a performance requirement without having to change their design environment.

4.1. Solution requirements

As design models are mainly of graphic nature and TRIZ tools concepts are mainly formulated in textual form, the integrated tools and methods should enable designers to interact with the graphic design interface of the CAD package being processed and when needed, finding TRIZ-based recommendations as directions of solutions.

Ideally the use of TRIZ based tools should be directly applied to the product model that is being developed. This means that ways should be found and/or developed to convert graphical and text information easily into each other and to identify relations between both types of information representation.

4.2 The object oriented 3D-CAD tree-structured graph.

In Figs. 1 and 2 is shown how objects modeled in 3D-CAD packages are related to the object oriented tree-structured graph (TSG) of its components. These TSGs contain the information about geometric shape, dimensions and topology of the objects, from which the parts and assemblies are modeled, therefore each component of a 3D parametric model may be accessed through textual information for edition or deletion.

For example for individual parts or components following items may be accessed through textual information of its TSG: the shape of its parametric profiles, its parameters, the relative position of the construction planes and the boolean and unit operations (join, cut, intersect, fillet, chamfer, etc) used to model each the part.

In case of assemblies or subassemblies the relative position and the way components or subassemblies are joined together is also accessible through the TSGs.

4.3. The Object-Action-Object-Result-Diagrams

Diagrams are commonly used problem solving tools, as they allow focusing the attention on the key information required when analyzing and solving problems. Specialized diagrams, as SF- and SUH diagrams and bond graphs are examples of how diagrams help in conceptualizing and visualizing the required information.

As objects in 3D-CAD models are represented in TSGs it is possible to add to its TSGs information about the way each component interact with each other (actions) and about the results obtained through this actions. Adding actions and results information to the TSGs a new type of Diagarams, the Object-action-object-result diagrams (O2-A-R-Diagrams), are obtained.

In Fig.3 an example of an O2-A-R-Diagram is shown, where the constituting objects O1-O6 may be seen with their geometric or topological relationships (GTi-j) and the functional actions among them as Ai-j arrows. The Results objects are linked with Ri arrows outgoing from the objects. A detailed description of the O2-A-R -Diagrams is contained in [20].

Fig. 1 3D-CAD Model of a gear reducer assembly composed of single
parts and its object oriented tree-structured graph. (TSG)

Fig. 2 3D-CAD Model of a gear reducer housing parts and its object oriented tree-structured graph. (TSG)

Fig. 3 Example of an O2-A-R-Diagram

O2-A-R-Diagrams are the main way in which product model information may be related to TRIZ based analysis tools for helping in finding solutions to design problems.

In O2-A-R-Diagrams an object is any physical component in a technical system, independently of its level of complexity in the system. That means that any technical system may be decomposed in objects, which may also again be decomposed in more simple objects, until a level is reached where the decomposition is no further possible or necessary. The decomposition of a technical system in objects is a logical process. This means that not only physically separable components may be decomposed, but also components which are physically inseparable, may also be logically decomposed in its building features.

Based on this a car model may be decomposed in several first level subsystems of the car (tier one), which constitute objects (assemblies) in a CAD-System. These are i.e. the steel body, the engine, the axles, and the transmission, among others. Each of these complex assemblies may be handled as a simple object which my again be decomposed into subassemblies. For example the engine may be divided in the starter system, the fuel injection, the cylinder head with valves, the crankshaft-connecting rod-piston subsystem and others. Each of these subassemblies may be further decomposed in parts and each part as i.e. the connecting rod may again be decomposed in its several constituting features from which it is modeled. The features are constituted of profiles, planes, dimensions, tolerances, etc. which, are also handled as objects in CAD systems.

O2-A-R-Diagrams are compatible with the TSGs of most important commercial 3D-CAD packages and are the basis for a software based user interface being developed which starts from the 3D-CAD TSG and allows relevant functional information (actions) to be added between each interacting object par (O2) for providing results (R). This approach resembles the way CAE software works, where additional actions as forces, velocities, accelerations are added to the existing objects in order to obtain simulation results as stress maps, velocity diagrams, etc.

The difference is that in this case no simulation is performed, as the aim of this approach is not to simulate but to look for new design solutions. The O2-A-R -Diagrams are appropriate for being processed with semantic processors to derive recommendations of solution concepts for each kind of insufficient or harmful actions or results contained in the diagram. As new design solutions are commonly searched through variation of shape, dimensions, topology, position, number of elements or its mechanical properties, directions of solutions should be presented graphically at the CAD display in order to help designers to overcome psychological inertia.


4.4. Further work with O2-A-R -Diagrams

Automatically formulating design solutions with O2-A-R -Diagrams requires yet following tasks to be completed:

Establishing a set of rules to be applied to typical situations of insufficient or harmful actions or results contained in O2-A-R -Diagrams as a subset of the inventive principles and/or standard solutions.

Building the interface for modifying the TSG in CAD systems in order that recommended directions of solution automatically appear as modifications to the 3D-CAD model to allow designers visually identify emergent shapes that reflect alternative structures to reflect upon.

As only few of the inventive principles and standard solutions are of geometrical, dimensional or topological nature, further rules derived from the best practices in embodiment design have to be implemented to enrich the rule set.

4.5. Some possible cases

In table II are shown several examples of the kind of recommendations that may be formulated by the semantic processing Design Advisor:

A first set of recommendations may be derived from those inventive principles that contain more geometric nature as segmentation, local quality, asymmetry, joining, nesting, counterweight, previous action, mediator, equipontentiality

A second set of recommendations may be obtained from the morphological matrix, in terms of changing position, shape, number and type of movement and also derived from DFMA practices in terms of merging parts, eliminating bolts, etc.

A third set or recommendations are of the kind of the recommendations derived from SUH diagrams at IWB or from the Product Analysis, Feature Transfer or Trimming modules in TO.

Table II Basic recommendation examples
A. Examples of recommendations derived from inventive principles

  1. Provide a mediator between Oi and Oj, that enhances [the] Action Ai,j and improves result Rj
  2. Analyze if the asymmetric shape of Oi provides an enhancement of action Ai,j and improves result Rj
    (Symmetric shape of objects is avoided and a non-symmetric alternative is presented)
  3. Changing the local quality of feature Oom in part Oi for enhancing action Ai,j
    (Marking feature(s) Oom-n in part Oi for stimulating designer to invent a local quality effect on it)
  4. Joining part Oi and part Oj, to improve actions Ai,j and Aj,j+1 for obtaining improved results Rj and Rj+1
    (parts Oi and Oj, are merged together automatically)
  5. Nesting part Oi in Oj to enhance action Ai,j
    (Marking parts Oi and OJ for stimulating designer to invent a nesting solution)

B. Examples of recommendations derived from the morphological matrix.

  1. Analyze if following alternative shapes of [the] (Object Oi), enhances [the] (Action Aij),
    (Several new shapes derived from the original shape (parametric profiles) Oi appear in the CAD display)
  2. Analyze if following alternativ positions GTi-j of (Oi) with respect to [the] (Oj) enhance [the] (Result Rj)
    (Several new positions of Oi with respect to Oj appear in CAD display)
  3. Increase the number of objects Oi performing the action Aij for enhancing result Rj
    (Instead of 1, several objects Oi are arranged to Oj performing the Action Aij)
  4. Alternative arrangement of Object Oi to Oj perfoming action Aij for enhancing result Rj
    (For example circular concentric insted of matrix array of Oi with respect to Oj)

C. Examples of recommendations derived from SUH diagrams or from the Trimming Module

  1. Find an alternative action (A*i-j) that does not cause [the] undesired (Result Rj)
    (not of geometric nature)
  2. Find a way to enhance [the] (Action Ai-j).
  3. Find a new action (A*i-j) that may be performed by object Oi on Oj to enhance [the] (Result Rj).
    (Not of geometric nature)

6. Conclusions

Synergies may be found among TRIZ tools and 3D-CAD-Systems, which allow improving the structure of the embodiment design process where inventive thinking is needed.

Significant conceptual advances in the way this integration may be performed have been achieved as results of our research work that permits implementing a prototype in the near future.

7. References

[1] León, N and Aguayo, H, A new Model of the Conceptual Design Process using QFD/FA/TRIZ, Paper Presented at the QFD-Symposium, Novi Michigan.

[2] S. Eppinger: Information-Based Product Development, http://me.mit.edu/groups/cipd/research.html, eppinger@mit.edu

[3] Key Characteristics Symposium organized and facilitated by Professor Anna Thornton,http://web.mit.edu/afs/athena.mit.edu/org/c/consortia/keychar/acthornt/www/papers.html

[4] Organized and facilitated by Dr. William Finch: http://web.mit.edu/wfinch/www/research/index.html

[5] Minho Chang, minho@mit.edu, Thesis Input methods for conceptual design of mechanical assemblies, Advisor: David Gossard.

[6] Woodie Flowers et. al. http://me.mit.edu/people/flowers.html, http://me.mit.edu/groups/cipd/projects/pelaez.html, Visualizing relationships in large information databases Mechanical Engineering, School of Engineering, MIT.

[7] Daniel Whitney et. al. Information flow mapping to aid design of complex products, School of Management, http://me.mit.edu/groups/cipd/projects/dong.html

Engineering, School of Engineering, MIT

[8] Warren P. Seering et. al. Concurrent application of thrust 2 tools


[9] Schön, D.A. The reflective practitioner. Maurice Temple Smith, London (1983).

[10] Mitchell, W. J. A computational view of design creativity. Preprints: Modelling Creativity and Knowledge-Based Creative Design. Dept of Arch. and Design Science, Univ. of Sydney. (1989) pp. 263-285

[11] Gombrich, E. H. Norm and Form: Studies in the art of the renaissance. Phaidon Press, Oxford (1966).

[12] Tann, M. Saying what it is by what it is like - describing shapes using line relationships. McCullough, Mitchell & Purcell (eds) The Electronic Design Studio. MIT Press, Cambridge, MA, (1990) pp 201-214.

[13] Stiny, G. Pictorial and formal aspects of shape and shape grammars. BirkhÖuser Verlag (1975).

[14] Scrivener, S. A. R., Edmonds, E. A. & Thomas, L. A. Improving image generation and structuring using raster graphics. Proc. CAD78. IPC Science and Technology Press. (1978) pp 223-229.

[15] Soufi, B. and Edmonds, E.A. 'The cognitive basis of emergence: implication for design support'. Design Studies (1996).

[16] S.D. Savransky, http://www.trizexperts.net

[17] Arciszewski T, and Zlotin B: IDEATION/TRIZ: Innovation Key to competitive advantage and growth (http://www.ideationtriz.com/report.html)

[18] Williams, T: Reversability of the 40 Principles of Problem Solving, http://triz-journal.com, May Issue, No. 1

[19] Savransky, S.D A few words about the Altshuller's contradiction matrix , http://triz-journal.com, August 1997

[20] León, Noel, Diseño Asistido por Computadora y TRIZ: Diagramas Objeto-Acción-Resultado, (Unpublished) Monterrey, Mexico, 1998.