Framework for a Structure Oriented Exchange of CAD Data

Contents

Preface

Acknowledgements

I. Introduction
1. Problem analysis and requirements definition
2. Review state of the art of exchange strategies
3. Analysis of modelling capabilities regarding feature modelling and structure representation
4. Needs-identification
5. Requirement definitions
6. Outline of the thesis

II. Approaches to the Exchange of CAD Data
1. Outline
2. General Classification of Exchange Approaches
2.1. Direct and Standard-based exchange Approaches
2.2. Horizontal and Vertical Exchange Approaches
2.3. Classification of exchange strategies
3. Capabilities of Scientific Exchange Strategies
3.1. On the Implicit Exchange of Feature-Based Product Model Data
3.2. Exchange of CAD Part Models Based on the Macro-Parametric Approach
3.3. The Unified Approach to Modelling Multidisciplinary Interaction
3.4. An exchange Approach in Conceptual-Based Design System
3.5. Open CAD Environment
4. Capabilities of Exchange Strategies in Commercial Areas
4.1. Vendor Exchange Method
4.2. Native Exchange Method
4.3. Neutral Exchange Method
4.3.1. IGES
4.3.2. VDA-FS
4.3.3. SET
4.3.4. XBF
4.3.5. PDDI
4.3.6. Projects Related to Neutral Exchange Methods
4.3.7. STEP (ISO-10303)
5. Further technologies
5.1. Parts Library
5.2. CAD – Services
5.3. ProSTEP
6. Conclusion

III. Analysis of today’s CAD systems – concepts and capabilities
1. Outline
2. Feature modelling definitions
2.1. Feature definitions
2.2. Feature model definition
2.3. Feature modelling definition
2.4. Feature attributes definition
2.5. Types of features
2.6. Feature properties
2.7. Feature properties classification
2.8. Generic and composite features
2.9. Feature taxonomies
2.10. Feature mapping
3. Analysis of feature modeling methods
3.1. Basic feature creation methods
3.2. Analysis of specific features modeling methods
3.2.1. Interactive feature modeling
3.2.2. Automatic (recognition) feature recognition
3.2.3. Design by feature modeling
3.3. Analysis of structural representation regarding feature modeling methods
3.3.1. Automatic feature recognition
3.3.2. Design by feature modelling
4. Major benefits of feature modelling methods
4.1. Design Intent
4.2. Multilevel structure
4.3. Feature-Kernel
5. Deficiencies as a result of feature data exchange
6. Analysis of commercial systems
6.1. Scope and process of analysis
6.1.1. Analysis of basic feature creation methods
6.1.2. Analysis of modelling kernel
6.1.3. Scope and process of model representation
6.2. Pro/Engineer’s analysis
6.3. I-DEAS’s analysis
6.4. Unigraphic’s analysis
7. Conclusions

IV. Exchange strategy – concept
1. Problem discussion
2. Needs identification on a Feature Model-Tree exchange methodology
3. Requirements definition - sequence of needs identification
3.1. Openness - neutral type of exchange data
3.2. Modularity – structure oriented data
3.3. Optimal Completeness – optimal density of exchange data
4. Requirements list
5. Established and new exchange strategies
6. Concept details
6.1. Fundamental ideas
6.2. General framework
6.3. Detail of the framework
6.4. Translation process

V. Evaluation and outlook References

Preface

Engineering Product Data Exchange (EPDE) is still in the Research and Development stage. It has been necessary to exchange data which describe a product – often, its geometric shape – since computers were first used for design and analysis in 1950s. That necessity still exists today. The data must to be available throughout the life cycle of the product and among all of the different enterprises.

This thesis will introduce the development of a framework for a structure- oriented exchange of CAD model data (CAD model data is a sub-data of Engineering Product Data).

The major tasks of this thesis are:

1. The a nalysis of CAD modelling capabilities regarding feature modelling and structural representation.
2. A survey of existing scientific and commercial strategies for an exchange of CAD model data.
3. The specification of an overall framework solving central technical problems.
4. Illustration of the framework’s functionality in a use case.

Acknowledgements

I especially thank Prof. Dr. F.L. Krause, head of Industrial Information Technology Department at Production Technology Centre, Germany and Dipl. Ing. R. Baumann, head engineer of Information technology department at Fraunhofer Institute, Germany, The writing of this thesis would not have been possible without their help, support and supervision.

I would also like to extend my thanks to Dr. S. Drehea for his support during my studies.

I am also grateful to the TU – Berlin for the opportunity to participate in the “Global Production Engineering” - Master’s program.

In addition to others not mentioned here personally, I especially thank my family for the their love and devotion to me through all the time.

I. Introduction

1. Problem analysis and requirements definition

Integration of a CAx (Computer Aided x) system throughout the product life cycle and among different enterprises is a major issue for industrial competitiveness and collaboration. One of the main successful factors for CAx system integration is efficient methodology for EPDE (Engineering Product Data Exchange). Data exchange is the totality of establishing the approach for and the successful achievement of the transfer of data between two distinct CAx systems.

Problem Statement:

- Why does an exchanged CAD (Computer Aided Design) model lose some modelling properties? – Especially losses such as model tree (design intent) and features.
- What reasons influence that phenomenon?
- How can these losses be minimized?

2. Review state of the art of exchange strategies

The review of exchange strategies is focused on which existing approaches are in use today, which capabilities are supported by them, which deficiencies they have, an understanding of state of the art is a precondition for beginning to deal with of the problem statement.

3. Analysis of modelling capabilities regarding feature modelling and structure representation

The analysis begins with a short review of existing feature modelling techniques, which will build up a framework for the analysis process. Three CAD systems are analysed – Pro/Engineering, I-DEAS and UniGraphics. Typical models, with the frequently occurring features, are reviewed depending on the feature modelling method and structural representation.

4. Needs-identification

The results of the analysis of modelling capabilities lead to the improvement of new methods and techniques. This defines the essential basis for the building of a concept framework.

5. Requirement definitions

- How can the model-tree to be exchanged?
- How will the exchanged model-tree act?

6. Outline of the thesis

The material is organized in three major sections. The first one, the state of the art, examines the fundamentals of exchange approaches, the current state of scientific and commercial exchange approaches and further related technologies.

The second one, presents the current state of feature modelling techniques and analyses of three commercial CAD systems according to feature modelling capabilities and structural representation.

The next section, the concept framework, designs a concept framework fitting the requirement definitions.

II. Approaches to the Exchange of CAD Data

1. Outline

The state of the art is discussed in this section according to existing approaches to the exchange of CAD data. Engineering Product Data Exchange (EPDE) refers to the task of expressing and transferring information about a given engineering product in digital format. This section will first outline the general classification of exchange approaches, such as direct and standard-based also horizontal-based and vertical-based. Furthermore, the classification of exchange strategies as a scientific and commercial. Review of capabilities on both approaches. Further technologies review additional exchange techniques. All these will be elaborated later in the next sections of the thesis.

2. General Classification of Exchange Approaches

2.1. Direct and Standard-based Exchange Approaches

The information generated during the design phase - the definition of the product - is the first precondition for data exchange. Basically, there are two types of CAD-product definition. One contains the data of a drafting and a solid model. The first one is composed of the vectors data for lines as solid lines, dotted lines, centre lines and extension lines together with annotation data for the dimension values, notes and symbols in the drafting. The second type of product definition data consists of a solid model with some annotation data.

Different CAD system store design results in their own system-specific structure for their product definition data, so it is necessary to convert the product definition data of one system to fit the structure of the other in order to transfer the data. A second translator is needed to transfer the data in the opposite direction between the same systems. Hence, for each pair of the systems are require two translators. Figure 1(a) shows these translators between each specific pair of systems, called direct translators. If we have n different systems, we have to provide n(n – 1)/2 pairs of systems. Example: There are 90 existing translators for 10 systems. And if n rises to n + 1, then additional translators are required to write 2n.

However, the data exchange can also be carried out by introducing a neutral data base structure, called a neutral file, or standard-based approach, which is independent of the number of existing n CAD systems.

illustration not visible in this excerpt

Figure 1. Exchanging data between different systems – two methods.

Figure 1(b) presents communication midpoint for different CAD systems. Each system has its own processors to transfer data to and from this neutral file format. The translator that transfers data from the database format of a given system to the neutral format is called a pre-processor and the translator in the opposite direction is known as a post-processor, as shown in Figure 2, hence a total number of processors required for n systems is 2n, and if n rises to n + 1 then the number of processors is 2n + 2.

illustration not visible in this excerpt

Figure 2. Data exchange using a neutral file.

Table 1 accounts for the main advantages and disadvantages of the direct and standard-based exchange approaches.

illustration not visible in this excerpt

Table 1. Comparison of Direct and Standard-based exchange approaches.

Conclusion:

Two general possible methods for EPDE are possible between different CAx systems are possible, direct or standard-based. The first one provides better quality of data exchange but less integration of new participating systems. The standard-based approach provides system openness (system-independent format) but lower quality.

2.2. Horizontal and Vertical Exchange Approaches

The second classification of general exchange approaches deals with so-called Horizontal and Vertical methods. Horizontal exchange methods transfer similar product data to the other method as other methods previously discussed. For example, designed product data in one CAD System is exchanged to another CAD System. Exchanges from CAM (Computer Aided Manufacturing) to CAM or CAPP (Computer Process Planning) to CAPP also appear in the horizontal exchange approach.

The following material emphasizes the vertical exchange method.

The definition of the Vertical exchange approach lies in the definition of the product life cycle. To understand its role, it is necessary to examine the activities and functions that must be accomplished in the design and manufacture of a product. With other words reviewing the product life cycle definition will generate the definition of vertical exchange approach.

The product life cycle, described by Zeid (199) and used by L. Kunwoo, is composed of two main processes: the design process and the manufacturing process. The design process originates from customer’s demands, which are identified through marketing activities and ends with a complete description of the product. The manufacturing starts with design specifications and ends with the shipping of the actual products.

The activities involved in the design process can be classified into two types: synthesis and analysis. The initial design activities, such as the identification of the design need, the formulation of design specifications, a feasibility study to collect relevant design information, and design conceptualisation, are part of the synthesis sub-process. The result of this is a conceptual design. Once the last step has been developed, the analysis sub-process begins with the analysis and optimisation of the design. Typical of the analysis are stress analysis to verify the strength of the design, interference checking to detect collision between components while they are being assembled and cinematic analysis to check whether the desired machine will be obtained from these activities. Once redesign has been completed, after optimisation, the design evaluation phase begins. Prototypes may be built for this purpose. Popularly used technology is rapid prototyping. If the design evaluation of the prototype model indicates that the design is unsatisfactory, the process is repeated with a new design, if the outcome is satisfactory, design documentation is prepared.

The completion of the last phase is a positive signal to begin the manufacturing process with process planning. Process planning is a function that establishes which process is to be used and what the proper parameter for that process. It also selects the machines. The outcome of process planning prompts a production plan, a material order, and machine programming.

The flow of information flow between the elements of the product cycle creates a framework of Vertical exchange approach. Each phase of the product life cycle is supported with respective software and together with human activities, such as human knowledge and qualitative decisions, precise information is produced as a result of this symbiosis. An example for well-supported product life cycle phase is a design. The software support this phase is knows as CAD system, which produces data in the proper format. The outcome information is input for the next phase CAM.

The vertical approach requires future research work and is still not full implemented. Completeness of that method is a key technology for carrying out Computer-Integrated-Manufacturing (CIM). CIM is often said to be more of a business philosophy than a computer system. In this thesis CIM is not dealt with further.

How the horizontal exchange approach could be implemented is shown in the Figure 3.

illustration not visible in this excerpt

Figure 3. Data Exchange between CAx systems

It shows how the CAD, CAM and CAE (Computer-Aided-Engineering) activities are integrated through the common database. CAD, CAM and CAM are all CAx systems, part of CIM. The common database of a product life cycle supports a corresponding model data for each CAx system which is generated through the development of a product model. It contains two general types of information: physical product design information represented by the design model data and process information represented by process model data. An additional type is factory model information, which represents the dynamic state of the manufacturing system.

Conclusion:

Creating a common base for product model information makes the Vertical exchange approach possible.

2.3. Classification of exchange strategies

- Scientific exchange strategies are identified as methodologies existing only in research area, and they are still under going construction and further development.

- Commercial exchange strategies are so-called technologies which have already developed and are available for engineering use.

In most cases scientific strategies are prototypes of commercial ones and the feedback to the first type of strategies gives additional requirements for the optimisation of the second type or prompts the development of a new strategy.

3. Capabilities of Scientific Exchange Strategies

3.1. On the Implicit Exchange of Feature-Based Product Model Data

The concept is being funded by the German Research Foundation with participation of H. Tönshoff; F. Krause; R. Baumann and P. Woelk [N].

The concept of an implicit feature model description assumes the use of features such as elementary objects of the product model. A model description consists of two substantial parts: a generic description of the available features and methods (“feature library”) and a structural description of how to create a model out of the features, specified in the feature library (“model history file”). Both the feature library and the model history file define the model. The feature library, which contains the generic feature description, must be made available to any system that takes part in the information exchange. The syntax for the definition of the feature library is based on EXPRESS. The language allows the definition of the design guidelines and constraints between features and parametric relations. The rules and methods for the transformation of design features into manufacturing features (“feature mapping”) allow systems supporting later stages of the product development process to participate in the information exchange (“vertical information exchange”). The feature library contains a description of geometric features, non-geometric features and auxiliary features. Figure 4 shows the contents of the feature library.

illustration not visible in this excerpt

Figure 4. The contents of the feature library.

illustration not visible in this excerpt

Figure 5. Regeneration of geometry out of the implicit model description.

The generic description offers a library of predefined features (user-defined features). The user may extend the set of features in the library. Due to the use of a formal language for the feature creation methods, User Defined Features (UDF) can be transferred from one CAx system to another.

The feature-based model history file is the second essential component of implicit data exchange. It contains the parameter values with which the features are created. Every entity corresponds to an instance of a feature specified in the feature library. The model structure corresponds to the order in which the entities occur in the model history file. A complete implicit description of the model is reached by the combination of the generic feature library and the description of the model history file (Figure 5).

3.2. Exchange of CAD Part Models Based on the Macro-Parametric Approach

The concept is being funded with participation of C. Guk-Heon; M. Duhwan; H. Soonhung in South Korea.

The concept of a macro-parametric approach provides capabilities to transfer parametric information including design intents. In this approach, CAD models are exchanged in the form of macro files. A macro file contains the history of user commands which are used in the modelling phase.

The macro-parametric approach is another kind of history-based parametric method. To transfer parametric information including the design history, a set of standard commands is defined and used, a neutral format. A macro file that records the modelling command sequence or user’s modelling history is exchanged. The history of user command, which defines a high-level dynamic interface, is recorded in a macro file, and that macro file is used for the static model exchange. The idea comes from the database recovery process where the transaction log file is used to restore the database after a crash. A set of user commands issued by a CAD designer during the design task is recorded as the modelling history, which implicitly includes the CAD designer’s intent. This is another method of product data exchange because the exchanged macro file regenerates the same model inside the receiving CAD system. Following Bill Anderson’s definition, authors define the term design intent, as “some functional requirements provided by customers, that is, a set of geometric and functional rules which the final product have to satisfy”. Therefore constraints, parametric features and design history represent the design intent. By translating the command sequence, design intent can also be implicitly translated. Figure 6 shows the data exchange model used in the macro-parametric approach.

illustration not visible in this excerpt

Figure 6. Architecture of the macro-parametric approach.

One is the schema mapping between the user commands set of a CAD system and the standard command set. The other is the actual data translation between the macro file of the commercial CAD system and the standard macro file. To translate data models between CAD systems, the macro file generated by a commercial CAD system is translated into a standard macro file which is again translated into a macro file of the receiving CAD system. To standardize the user modelling commands of commercial CAD systems, authors surveyed the modelling command set in commercial CAD system, as CATIA, Pro/Engineering, Solid Works and Solid Edge (Figure 7).

illustration not visible in this excerpt

Figure 7. Classification of standard modelling commands.

The standard commands set is a common set of modelling commands that are used in part modelling modules of major commercial CAD systems.

3.3. The Unified Approach to Modelling Multidisciplinary Interaction

NASA Langley Research Centre developed the concept with participation of J. Samareh and The Boeing Company with participation of K. Bhatia.

The Unified Approach has two essential steps. First, the data transfer process between two disciplines is modelled by a transformation matrix. Second, the CAD model is used to reduce the number of interactions. The interaction between two disciplines is modelled mathematically as

{F2} = [T21]{F1},

Where matrices {F1} and {F2} contain the information on discipline grids 1 and 2, respectively, and matrix [T21] is a transformation matrix. For example, {F1} could be an aerodynamic loads vector defined on a Computational Fluid Dynamics (CFD) grid and transferred to the Computational Structural Mechanics (CSM) grid as {F2}. Generally the transformation matrices are large and sparse. Only the non-zero elements need to be stored. If the transformation matrix [T21] is independent of the shape changes, then [T21] can be calculated once and used as long as there is no change in the grid connectivity. The concept of a transformation matrix simplifies integrated analysis such as aero-elastic calculation. Aero-elastic calculation has four distinct steps. First, the aerodynamic loads are calculated on the CFD grid. Second, the loads are transferred to the CSM grid. Third, the aero-elastic deflections are calculated on the CSM grid. Fourth, the deflections are transferred to the aerodynamic grid to recalculate the aerodynamic loads. This iterative process can be expressed as

illustration not visible in this excerpt

The first equation represents the aerodynamic load calculation. The term GF represents the CFD grid, and dF is the aero-elastic deflection on the CFD grid. The aerodynamics forces, FF, are transferred to the CSM grid as FS. The matrix [K] is the CSM stiffness matrix, and dS is the aero-elastic deflection on the CSM grid. The use of the transformation matrix simplifies the above set of equations to

illustration not visible in this excerpt

For a linear structure without rigid body degrees of freedom, the above equation set can be simplified to

illustration not visible in this excerpt

where

illustration not visible in this excerpt

If the transformation matrix is independent of the shape changes, then the formulation is especially beneficial.

The second ingredient of the proposed unified approach helps to reduce the number of transformation matrices. For a multidisciplinary application that involves n disciplines, the traditional process may require (n2 – n) transformation matrices. Figure 8 shows all possible interactions among eight disciplines.

illustration not visible in this excerpt

Figure 8. Multidisciplinaire interaction Figure 9. Unified multidisciplinary interaction

The new element of the CAD-based approach was the use of a CAD geometry representation to reduce the number of transformations from (n2 – n) to 2n. The reduction is accomplished by transferring the data to a CAD geometry model that serves as a common bridge or a data bus for sharing information among various discipline, as shown in Figure 9. This approach is of obvious benefit for multidisciplinary application when more than three disciplines are involved. The intermediate CAD grid isolates each discipline model from changes to all other discipline models. The overall approach has a strong potential for robust automation of all transformations, and ensures consistency among different renderings of proper configuration. First, the data is transferred from the individual source discipline to the CAD model as

{F_C} = [T_C1]{F₁}.

Then the data is transferred from the CAD to the target disciplines

{F₂} = [T_2C]{F_C}.

The advantage of this approach is that only [T_C1] and [T_2C] transformation matrices have to be calculated: 2n matrices instead of (n² – n)

The concept framework is the best-fit example for overcoming deficiencies throughout the vertical-based exchange approach.

3.4. An Exchange Approach in Conceptual-Based Design System

The following material was discussed by J. Shah and M. Mäntylä, and gives short fundamental definitions of Conceptual-based Design Systems and their approach to data exchange [N].

The top-down design approach is preferred by most designers for conceptual design, since the design of assemblies starts at a high level of abstraction. Assembly design does not always require detailed design of constituent parts and subassemblies. Hence the design can be carried out in terms of an abstract concept, and this helps the designer in validating some of the design concepts. Ideally a conceptual design should support transitions from high-level conceptual assembly models stressing the function of the assembly to detailed models of the individual components. A number of requirements have been identified (Mäntylä 1990). Here are some of them:

- The model should provide support for representing the assembly information on several levels of abstraction (abstract structure, overall geometry, detailed geometry);
- The model structure should represent design intent by preserving the “reason” for the various model entities.

illustration not visible in this excerpt

Figure 10. P-graph structure of lamp assembly.

Schubert (1979) introduced Part graph (P-graph, Figure 10) for assembly modeling. P-graphs are graph structures of hierarchical assembly models in which the nodes represent the objects or subassemblies, and the edges represent the nature of constraints and relationships

The Edinburgh Design System (EDS), developed by Popplestone (1980), allows the user to define two types of modules, concrete modules and interface modules; these in turn may have sub-modules or parts. Concrete modules contain parameters and variables whose values need to be explicitly specified at the time of design. These modules can be very abstract and are composed of sub-models in a hierarchical conceptual structure. Interface models specify constraints between these parameters and variables. For instance features may be defined as concrete modules and their spatial relationships can be encoded as interface modules. An important part for data exchange questions is that the existence of a conceptual-based design as a part of today’s CAD systems can generate a conceptual structure, the representing design intent of the product. The ability to exchange this structure will overcome some deficiencies CAD data exchange. Details of this are discussed further in the concept framework of this thesis.

3.5. Open CAD Environment

The Fraunhofer Institute is funding the concept for Computer Graphics in Darmstadt, Germany with participation of J. Encarnação and J. Ovtcharova.

This material presents an overview of the Open CAD Environment in the context of the CAD reference model.

The CAD Reference Model was founded a joint project of the BMFT (The Federal Ministry for Research and Technology in Germany) program “Work and Technology” founded as project BMFT (DLR – Aut, Bonn-FKY 01 HP 434 2). The goal of this project is to develop an adaptable user-friendly CAD environment, as requested by German industry, which is built around the needs of users and application and not CAD system technology. Reviewing CAD’s state of the art reveals that many systems do not contain all the functionalities desired by the users. Especially worth mentioning are the abilities to configure the user interface, integrate with other applications, and supply guidance for the user. Although the ISO 10303(STEP) standard for product data representation and exchange, covering the entire product life cycle, is available and under further development, today’s CAD systems rarely support conversion to the standard. Also, adequate knowledge representation in CAD system is still an unsolved problem. Figure 11 presents the architecture of the CAD Reference Model and shows how CAD systems should appear in the future and can be used as an evaluation tool for currently existing systems.

illustration not visible in this excerpt

Figure 11. Architecture of the CAD Reference Model.

One of the main advantages of the CAD Reference Model is the possibility of configuring the CAD system to different applications. Here, the goal of application dependent system configuration is to dynamically configure the CAD system at runtime depending upon the given design tasks or application areas. The basis of the configuration system is the open system architecture of the CAD Reference Model, which makes it possible to integrate different functionalities on different platforms. Other advantages of the CAD Reference Model stem from a new user interface system. This system addresses the main problems related to user interface; determining the order in which to perform actions and minimizing confusion over the complex functionality of many CAD systems. The CAD Reference Model requires user-controlled interactions, with a higher degree of flexibility than has been provided by previous CAD systems.

4. Capabilities of Exchange Strategies in Commercial Areas

4.1. Vendor Exchange Method

Vendor formats are a halfway house between direct data exchange and standard-based data exchange. The vendor publishes what he calls a neutral format but which is, in reality, closely aligned to his CAD system format. Any person wishing to transfer to and from that vendor’s system has just two processors to write and if the format becomes widely used the situation is in all respects similar to that of using a public neutral format. All the formats in this category convey minimal information. The best examples are ISIF by Intergraph, CATIA exchange format, and DXF by AutoCAD. These formats deal in basic geometrical elements and some annotation. Thus vendor formats are a way for a particular vendor to make an anaesthetized version of his internal data structure widely available and thus be able to show that access to his system from any other is easy.

The main advantage of that approach is the ability for direct exchange and the main disadvantages is that the number of information losses are equivalent to the standard-based approach and quality of translators increases with new CAD system’s participants.

4.2. Native Exchange Method

Generally the native exchange method is based on the direct exchange approach but it is only a sub-case.

The native exchange method is an excellent example for successful and complete CAD data exchange between different enterprises, but only in the case that the same CAD system is used. This approach provides excellent quality data exchange but less flexibility with enterprises with other CAD systems. The only problem with that approach occurs when the enterprises use different versions of their CAD system. The newer version reads the information coming from older version completely, but errors are generated when the older version tries to read the information from the newer.

[...]