The State of The Art

Design Using Solid Models

A Solid Model contains more information that any other geometric representation. This leads to very powerful software support, but may also create difficulties. Solid Modelers are often very slow, large, and create problems for data transfer. Despite these problems, the approach is often used for the benefit of representation. When using solids as a basis for representation more information may be associated with the solids. For example a solids model may have attributes like material, volume, geometry, positional and geometrical tolerances, and manufacturing operations. These attributes may be difficult, or impossible to include with a surface, or wireframe model.

In order to associate attributes to solid features, the standard graphics only interface must be augmented with a product structure interface. Some commercial packages do this with interactive command and text windows, but this may be awkward, and confusing. We have provided

a visual display of a tree of CSG operations, in which the nodes were represented as parts, instead of operations, which is more intuitive to the parts designer (Figures 3 and 4 show examples of this). This approach resulted in an easy to use system for visual construction of a product. This system was later modified to include tolerance form data for various primitives, based upon the ANSI Y14.5M Geometric Tolerancing Standard. This type of modeler is also discussed by Shah et. al. [1990], but his treatment of tolerance information was much more rigorous and involved the inclusion of datum information.

A majority of the Solids Modeler front ends attempt to mask the Solids Modeling with features. This has been well explored by Hoda ElMaraghy [1991] in her Intelligent Product Design and Manufacture (IPDM) system. The system uses a complete set of features for modeling geometry and tolerances. After modeling, the geometry may be recorded in a Product Description Language (PDL) file for storage and transmission. The PDL files are used by other modules which parse them, and attach manufacturing and analysis information to the Product.

Solids Modeling Systems for Concurrent Design

The University of Illinois the Knowledge-Based Engineering Systems Research Laboratory has been developing a concurrent engineering system [KBESRL, 1991]. At the front end the solids model is described with a LISP based approach which allows parts to be defined with primary (structural), and secondary (detailed) features [KBESRL, 1991, pp. 33-40]. The approach ties a generative CAPP system closely to the CAD system. Unlike our approach, their CAPP system must choose operations (their CAPP system is discussed in the next section). This causes a loss of intent of the designer which was available when the design was enacted. A related project [KBESRL, 1991, pp. 65-72] discusses using a network of CAD systems which share design knowledge, which seems to be largely geometrical at this point. They have an intricate set of problems to overcome because their system is distributed, (whereas the our approach in this paper allows centralized client/server relationships). Some of the discussions in this paper involve the conflicts which may arise when multiple agents access the same design simultaneously. It is interesting to note their approach uses a roll back mechanism.

Another project within their system focuses on communication issues [KBESRL, 1991, pp. 25-32]. Concurrent design is enabled by keeping common records of specifications and communications. If a conflict arises because of design changes, then users are informed. This work is important to mention because it will eventually be incorporated into their Solid Modeler, to provide complete concurrent engineering support.

Computer Aided Process Planning (CAPP) and The Solids Modeler

Process Planning is the vital link between Design and Production. Many interesting efforts have been made to develop tools to aid the Process Planner in his job. In particular computerized tools have been appearing in great numbers. At present, variant technologies are available to the commercial buyer, that allow planning using variations on existing process plans. Generative planners are able to develop process plans from the ground up, but these are still in research stages.

An interesting description of the Process Planning Problem is given by Kusiak [1991], who breaks the problem into eight sequential phases.

1. Interpretation of part design data (KB)

2. Selection of machining processes (KB)

3. Selection of machine tools, and fixtures (KB)

4. Machining optimization (OPT/KB)

5. Decomposition of machinable volumes (KB)

6. Selection of machinable volumes (OPT/KB)

7. Generation of precedence constraints (KB)

8. Sequencing machinable volumes (OPT/KB)

Simulation as a Verification Tool

It is clear that simulation plays an important role in new product design when attempting to determine producability and manufacturability. For example, there are a number of commercial systems which will simulate NC programs so that the tool path may be verified manually (for example CIMStation by Silma). There is still a shortage of methods for automatically determining validity of processes. One such algorithm has been published by Su et. al. [1991]. He discusses an approach for verifying design work by checking to see if a part is machinable on a 3 axis milling machine. This work is in its early phases, but would be a powerful feedback for designers.

Production simulation is a popular industrial tool for verifying production changes. As a result there are a number of vendors which supply software capable of performing various scheduling, breakdown, and analysis functions. (e.g. Hocus by P-E Inbucon, and ProModelPC by Production Modeling Corporation). The only shortfall to these systems is the lack of a good application interface for direct patching of production simulation into the software of the CIM system.

A Design System for Concurrent Engineering

A CIM Architecture has been developed which will support concurrent engineering from design to production. To do this, manufacturing knowledge has been introduced at the design stage so that the design will incorporate standard items like tolerances, operations, costs, and times. Using this data, a simple yet fully generative CAPP system can convert the designs into process plans. The resultant process plans can then be verified and checked using process and production simulators. The use of simulation allows fast feedback to the designers, so that a design may be altered in the early stages to overcome unforeseen problems with the choice of operations, or with the design itself.

A novel feature of this system is its support of a simultaneous design environment. In particular, designers will have simultaneous access to one design. Instead of a single designer who considers all design stages concurrently, a team of designers may cooperate on a multi-seat CAD package, applying their own specialties. This is made possible by having a main Design Engine (DE) which deals with storage and update of geometry, tolerance data, and manufacturing data. The users’ stations can request updates to the design, and display the results. There is only one copy of the design, but each user will have his own view.

The Product Design data serves as a framework for enabling analysis and simulation. At present the Product Design will be passed through the CAPP module to produce a Manufacturing Bill of Materials. The operations may then be simulated using the UWO Workcell Controller. This controller is equally capable of applying this to control the NC Milling machine and Robot, or appropriate simulations. A link is being established to a production simulation package.

This paper is organized to first discuss the work which has been done. This is followed with a description of previous research, and how it relates. The first section is a discussion of the Design Engine based Computer Aided Design for Manufacturing (CADM) system. The next section discusses how process plans are generated from the Product Designs, and how process planning faults are handled. Following this there is a discussion of process simulation for process plan verification, including the workcell controller. The final application section describes a production simulation module used to statistically determine the effects of one process plan when interacting with all other process plans in a factory. The application sections are followed by a discussion of related work by other researchers, and finally by conclusions.

Computer Aided Design for Manufacturing (CADM) with a Design Engine (DE)

The Computer Aided Design for Manufacturing (CADM) module has two objectives. It allows multiple designers to work on a Product Design concurrently. To do this the Design Engine (DE) was developed. The DE is an individual process (that may be running on a multi-process computer) which is the only holder of Product Design information. It stores all geometry, tolerance, and manufacturing information. Various views of this Design Model may be received by any number of remote designers, each using their own graphics programs (i.e., CADM). Figure 1 shows the hardware connections of these processes.

 

Figure 1 Networked Hardware Connection of Design Engine (the hardware used for this system was all Sun Workstations, running specialized software)

The diagram in Figure 5 may be deceptive: the DE is not responsible for production of graphics, it simply acts as a common data server for the Product Design, and will make updates to the Product Data. Figure 2 below shows how a CADM station would update a design after a user request.

 

Figure 2: Interaction of the Design Engine with a CADM Station

If the network was removed from Figure 2 this would be analogous to many CAD packages. The advantage of the present approach is that there may be multiple CADM stations changing a design concurrently. The only issue unresolved with this approach is an impasse which may occur when multiple CADM stations make conflicting changes. There are four basic methods which may be applied to this approach.

Do nothing and allow users to resolve conflicts with verbal agreement.

Allow a roll back mechanism which undoes conflicts, and requires resolution among designers.

Allow a locking mechanism for various design sub-structures. A user may lock part of a structure so that other users may not change it.

Use an activation token which is passed between CADM stations, and when a station has a token, it may accept user input.

Use tokens for sub-structures which limit changes to the token holder.

Each of these methods has merits, but at present there is no mechanism implemented (i.e., the first item on the list). This problem will be discussed more after the Product Design data structures have been presented.

The Product Design data structure is hierarchical and is based on a CSG approach. The data structure is not “parcelled-out”, so the structure remains one source of knowledge when reasoning and designing. The basic structure is seen in Figure 3 for a keytag with customized initials.

 

Figure 3: A Sample Product Structure For Manufacturing a Keytag

The product structure shown above, represents geometries and associations. Each node in the tree represents a physical feature, part, or assembly, and node each has associated attributes. The leaves of the tree are geometric primitives. On the diagram the ‘+’ and ‘-’ signs indicate CSG operations, and the ‘A’ indicates an assembly. The reader should note that the CSG operations on the tree may indicate manufacturing operations such as addition, joining, or removal of parts, assemblies and features. Thus, the tree structure may be equated with the Product Design Topology, and the Attributes may be equated with the Details of Design. A list of the various attributes that have been implemented are given below.

Part Attributes: Assembly Attributes:

Cost - Cost

Mass - Assembly Time

Centre of Mass - Mass

Volume - Centre of Mass

Material - Volume

Special Instructions - Orientation

Orientation - Position

Position

Manufactured/Purchased

Feature Attributes: Operation Attributes:

Cost - Cost

Geometry Type - Setup Time

Nominal Geometry Values - Operation Time

Tolerance Type - Operation Type

Tolerance Zone - Operation Variance

Datum - Tools Required

Orientation - Special Instructions

Position

Material

Material Properties

One of the benefits of this Product Design Structure is made obvious when its impact on CAPP is considered. A process plan can be partially described with the Product Structure, and the associated attributes. In this case the process planner simply converts the process plan to a Bill of Material, and assigns an order to the operations, then produces a Manufacturing Bill of Materials (or Routing Sheets). Note: the operations must be ordered because they are only stored in the form of a precedence graph.

A second benefit of this structure is that the DE may resolve multiple CADM user change conflicts by exploiting the structure of the product design tree. If the tree structure is locked at some node, then all subnodes down to the leaves cannot be added or deleted. This ensures that a structure will not become invalid. Even though a branch has been locked, the attributes of nodes and operations on the branch may be changed, and easily reversed if a conflict occurs.

The design system described in this paper has been developed at UWO using a CSG modeler and interface called TAGS, and some networking utilities for coarse grain concurrent processes [Jack et. al., 1992.]. A typical CADM window is shown in Figure 4, including the model, the tree structure, and some attributes which can be modelled. This package also includes tolerance analysis software which is not shown in Figure 4. Eventually, this software will be adapted to use ACIS (software by Spatial Technology), an object oriented solids modeler, capable of interpreting CSG and B-Rep definitions, as well as distributed ray tracing which permits photorealistic visualization.

 

Figure 4: A Screen Dump of the CADM Package (Based on TAGS)

Computer Aided Process Planning

Although the part design resulting from the CADM system has processes associated with features, the internal representation lacks a few details of a true Process Plan.

There are no machines assigned with the operations.

The operations require an order for production.

The TAGS tree must be converted to a Bill of Materials (BOM) and to Routing Sheets.

The operations must be verified for producability.

NC code must be produced for automated machinery.

In this case the CAPP system is simply doing routine chores, and in the event of error, it may directly inform the DE which in turn will make the information available to all CADM stations.

The tasks performed by the CAPP system are very routine, because most of the reasoned information is entered directly during the Product Design stage. Therefore the CAPP system must perform the following functions.

1. Map geometric operations to a database of available resources.

2. Determine operation precedence and assign an order.

3. Produce NC code from the Product Design model, and process descriptions.

4. Verify Process Plan Producability with a process simulation.

5. Verify Process Plan Manufacturability by doing a production simulation.

6. Reassign a plan order, or resource in the event of a scheduling failure. If not possible, then report failure to the DE.

These functions can be drawn out in a block diagram, as seen in Figure 5 below.

 

Figure 5: A Block Diagram of the CAPP System in Terms of the Complete System

This CAPP structure is very easy to construct, see Ruf et. al., [1990] for an example. A crude implementation of this module has been done in the present work, which verifies the connection between the DE and a BOM. A sample of the output may be seen in Figure 6; this was done with only one milling machine. This system requires a more complex product, and manufacturing facility for proper analysis of this system.

The Knowledge Base in Figure 5 will play a powerful role in the future CAPP system, by recognizing past failures, and making rules to avoid future situations. The Knowledge Base also holds information about sources of NC code for the design. When a design requires NC code generation, the Knowledge Base may suggest a program or subroutine which will generate the code. Existing software makes it possible to perform the feedback seen in the loops above [ElMaraghy, 1992]. The next section of interest is the process simulation.

BOM Sheets for Product ---------------------------

Product Name: Angle_Part

BOM Sheet for Part -------------------------------

Part Name: Angle_Part

 

Operation: Name: time: Cost: Source:

Block 0.00 0.00

Milling Peg 75.00 0.97

 

Totals:

Total Part Cost: $ 0.97

Total Part Time: 75.00s

BOM Sheet for Part -------------------------------

Part Name: Block

 

Operation: Name: time: Cost: Source:

Box 360.00 2.24

Milling Angle 3.57 1.17

 

Totals:

Total Part Cost: $ 3.41

Total Part Time: 363.57s

BOM Sheet for Part -------------------------------

Part Name: Box

 

Operation: Name: time: Cost: Source:

 

Totals:

Total Part Cost: $ 0.00

Total Part Time: 0.00s

Production Information ---------------------------

Total Product Cost: $ 4.38

Total Product Time: 438.57s

End of BOMS -------------------------------------

Figure 6: A Sample Bill of Materials From CAPP (for the part shown in Figure 4)

Process Simulation for Verification of Process Plans

While it is possible to select operations to produce a process plans, there is no guarantee that the designer has chosen operations which will succeed. But, it is not possible to verify this in the Design stage with only the operation (as specified in the DE). The simulation must occur after CAPP has assigned a particular machine to an operation.

An NC simulator, and robot simulator have been developed for an existing workcell at The University of Western Ontario. Screen dumps of the simulators may be seen in Figures 7 and 8. Each simulator runs directly in place of the actual NC Milling machine driver, and the robot driver, under the control software shown in Figure 9. This allows the complete generation of NC and robot programs, and then testing in a realistic environment, including manufacturing control software.

 

Figure 7: An NC Code Simulator for a DynaMite 2400 (This is the NC Code to Cut the keytag described earlier in the paper)

The NC Milling machine simulator will load the NC code destined for the Dyna 2400 Milling machine, and use the program to generate a graphical representation of the tool path. In Figure 7 the user will note that there are polygons, and lines. The lines indicate tool paths above the cutting surface, while the polygons represent the tool paths beneath the cutting surface. Here we can see the features described earlier in Figure 3, which had a hole, symbol, and initials (in this case SME). This is not a true tool path analysis, but for graphical purposes, the user may quickly determine if the results are satisfactory.

This program may be controlled through its user interface, or via commands passed through the network [Jack et. al., 1992], whereas the milling machine driver is only available over the network. The program to generate the NC code only communicates through the network, and is specific to these keytags. All of these applications were developed to be stand alone processes that run in the background of a heterogenous network.

 

Figure 8: A Robot Simulator for a CRS Plus Robot

The robot simulator is similar in setup to the NC simulator. The CRS Plus robot driver runs from the network only, whereas the robot simulator will accept commands from the network or from it’s user interface. The robot will track motion commands which would normally be sent to the real robot. This tool is excellent for predicting collisions, and operation sequences. The obviously poor resolution makes it impossible to determine exact positioning of the end effector. In a better simulation the complete manipulator would be modelled, along with other objects in the workspace.

 

Figure 9: Workcell Control Interface (This interface directly controls the robot and NC milling machine, but the workcell control architecture allows the real devices to be replaced by simulators, or run concurrently for verification).

The workcell controller in Figure 9 has its user interface to interpret commands from a user (left window), or a program (right window), and send them over the network to the other programs. The networking scheme used [Jack et. al., 1992] eliminates the need for the Controller to know which other processes are on the network. As a result, replacing the device drivers with simulations is trivial, and makes the simulation more realistic. It is also possible to run the simulation on a separate machine without affecting production.

Thus, it is easy to conclude that although these simulations are crude and require visual confirmation, they do allow a thorough verification of Process Plans. Eventually these simulators will be replaced by a commercial simulator called CIMStation by SILMA. The need for a powerful simulator will become even more important when a complex example is found which has many more devices and products.

Production Simulation for Verification of Process Plans

It is possible to develop process plans which are producible on available resources, and then find that the resources are not available. A manufacturer may waste valuable time if a bottle neck is discovered after process planning. By simulating the interaction of one process plan with all others, the loads may be more evenly balanced between available resources during the process planning stages. Also, long-term bottlenecks may be identified early enough to make capital investments. Even more importantly, since the simulations are closely linked to CAPP, which is closely linked to the DE, it is possible for a designer to try a few “what-if” scenarios with ease.

There are many excellent simulation packages in existence for various manufacturing problems. In general these are all discrete event simulators, capable of making various types of widgets. The main disadvantage of these packages is that they have all been developed as stand-alone packages with user interfaces. For this application it would be best if a simulator could be a background process. Also important is the lack of source code for these packages. Without source code the only source of reports is parsed text files. Needless to say this makes the system awkward and bulky. In response, an initiative has been made to develop a simple simulator which can run process plans, determine problems and feed them back to the CAPP system.

A discrete event production simulator, called FactorySIM, has been developed using color petri net subroutines developed at UWO. This simulator will take factory layout files, routing files, order files, process plan files, and inventory files then simulate manufacturing in a random manner. This is not efficient, and not the way that a proper scheduling system would work, but it does give a reasonable estimate of machine overloading. The main window from the simulator is seen in Figure 10 below. The various object shapes represent resources in the Resource file, and a small moving robot represents a workpart between operations.

 

Figure 10: The main simulation screen from FactorySIM (The factory may be viewed from two perspectives, and all operations are animated)

The files which drive this simulation are seen in Figures 11 to 15. These files contain cross referenced information, and may be uploaded during the simulation. This is a critical point for simulating the effects of introducing a new product during steady state production.

#

# Shop floor configuration file

#

 

start machine Press

size 1.0 1.0

pos 1.0 4.0

cycle 20.0

downtime 0.05

buffer 1 5

end

 

start machine Mill

size 1.0 1.0

pos 3.0 4.0

cycle 25.0

downtime 0.02

buffer 3 5

end

 

start machine Lathe

size 1.0 1.0

pos 5.0 4.0

cycle 18.0

downtime 0.07

buffer 5 5

end

 

end_of_file

 

Figure 11: An Example Resource Definition File for FactorySIM

#

# WIP routing between stations

#

 

# From inventory to machines

 

route Stores Press

step 0 0.0 -0.50 30.0

step 0 -0.50 0.0 30.0

step 0 -0.50 0.0 30.0

step 0 -0.50 0.0 30.0

step 0 -0.50 0.0 30.0

step 0 0.0 -0.50 30.0

end_route

 

route Stores Mill

step 0 0.0 -0.50 30.0

step 0 0.0 -0.50 30.0

end_route

 

route Stores Lathe

step 0 0.0 -0.50 30.0

step 0 0.50 0.0 30.0

step 0 0.50 0.0 30.0

step 0 0.50 0.0 30.0

step 0 0.50 0.0 30.0

step 0 0.0 -0.50 30.0

end_route

 

end_of_file

 

Figure 12: An Example Layout Definition File for FactorySIM

#

# A Process Plan

#

 

plan Test_Process_1

 

start operation Get_Material

uses Stores

produces 1 Raw_Part

precedence 1

end

 

start operation Cut_Hole

uses machine Mill

consumes 1 Raw_Part

produces 1 Widget

time 15.0

precedence 2

end

 

start operation Get_Other_Material

uses Stores

produces 1 Wrapper

precedence 1

end

 

start operation Widget_Cut

uses Press

consumes 1 Wrapper

produces 1 Box

time 10.0

precedence 2

end

 

start operation Assemble

uses Manual

consumes 1 Widget

consumes 1 Box

produces 1 Product

time 20.0

precedence 3

end

 

start operation Shipping

uses Shipping

consumes 3 Product

produces 1 Spiffy_box

precedence 4

end

 

end_of_plan

end_of_file

 

Figure 13: A Process Plan Definition File for FactorySIM

#

# Order File

#

 

order Big_order

part Spiffy_box

quantity 20

priority 1

process_plan Test_Process_1

end_order

 

order Small_order

part Spiffy_box

quantity 5

priority 0

process_plan Test_Process_1

end_order

 

end_of_file

 

Figure 14: An Order Definition File for FactorySIM

#

# Inventory Updata File

#

 

inventory Raw_Part

location Stores

quantity 20

end_inventory

 

inventory Wrapper

location Stores

quantity 25

end_inventory

 

end_of_file

 

Figure 15: An Inventory file for FactorySIM

This simulation is being developed at The University of Western Ontario, and will be used as a feedback for the CAPP system. The simulator will also use the network [Jack et. al, 1992] to allow it to fit into a heterogenous computing environment. For example, at present this package will currently run on either Silicon Graphics or Sun computers with no changes to the source code.

The Future; Discussion and Conclusion:

This paper discusses the successful implementation of a Concurrent CAD system which is capable of supporting multiple users, simultaneously working on various stages of the Product Design. The use of Manufacturing data in the CADM modules allows the CAPP system to be simplified. While the implementation of CAPP now is fairly crude, it will be fully developed in the near future. The ready access of simulation modules helps reduce the guess work which the design engineer must do. The simulation modules open up many opportunities for ‘trial and error’ sessions with new Product Designs. There are two types of simulations in this report, the first is a process simulator, which is fully functional. The second is a Production simulator is based on colored petri nets, that is partially functional.

Future work in this area will involve further testing and evaluation of the concurrent engineering paradigm. As the reader has seen, there is a test case using keytags. While this test case is simple, and manual transference of some data was done, this system has excellent promise. Work will also be performed for system development, such as model transfer to other systems using formats such as the PDES/STEP standard.

Concurrent Engineering principles make it possible to develop strong Design Engineering systems which will boost the performance of the modern manufacturer. Thus, this research will continue with the hopes of developing better support for the manufacturer of the future.

References:

6.1 KBESRL, 1991, Knowledge-Based Engineering Systems Research Laboratory Annual Report, 1991, Department of Mechanical and Industrial Engineering, University of Illinois.

6.2 ElMaraghy, H. A., 1991, “Intelligent Product Design and Manufacture”, appears in Artificial Intelligence in Design, edited by T.Pham, Addison-Welsly, pp. 147-167.

6.3 ElMaraghy, W. H., 1992, “Integrating Assembly Planning and Scheduling: CAPP Related Issues”, Submitted for the Annals of the CIRP 1992.

6.4 Jack, H. and ElMaraghy, W. H., “Interprocess Communication with MPS (Message Passing System)”, submitted for publication March, 1992.

6.5 Kusiak, A., 1991, “Process Planning: A Knowledge-Based and Optimization Perspective”, IEEE Transactions on Robotics and Automation, Vol.7, No. 3, June, pp. 257-266.

6.6 Ruf, T. and Jablonski, S., 1990, “Flexible and Reactive Integration of CIM Systems: A Feature Based Approach”, CSME Mechanical Engineering Forum.

6.7 Shah, J. J. and Miller, D. W., 1990, “A Structure for Supporting Geometric Tolerances in Product Definition Systems for CIM”, Manufacturing Review, Vol. 3, No. 1, March 1990, pp. 23-31.

6.8 Su, C-J. and Mukerjee, A., 1991, “Automated Machinability Checking for CAD/CAM”, IEEE Transactions on Robotics and Automation, Vol. 7, No. 5, Oct., pp. 691-699.

6.9 Wheeler, K., 1990, “Open Architecture in CIM: A Systems Integrator View”, Manufacturing Systems, Feb., 1990, pp. 22-25.

Additional Reading

The following list is divided into areas, an it includes references which include keynote, and some examples of research.

General

6.10 Izuchukuw, J., 1991, “The Design Team Approach to Class ‘A’ Engineering Is Working For Digital”, Industrial Engineering, Jan., 1991, pp. 37-39.

6.11 Koska, D. L. and Romano, J. D., 1988, “Countdown to the Future: The Manufacturing Engineering in the 21st Century”, a Profile 21 Publication, The Society of Manufacturing Engineers.

6.12 Lange, E., 1990, “Open Architecture in CIM: A Software Vendors View”, Manufacturing Systems, Feb., 1990, pp. 20-22.

6.13 SME, “Manufacturing 21 Report; The Future of Japanese Manufacturing”, Society of Manufacturing Engineers.

6.14 Whitney, D. E., 1988, “Manufacturing Design”, Harvard Business Review, July-Aug. 1988, pp. 83-91.

Moldeling

6.15 ElMaraghy, H. A., ElMaraghy, W. H. and Knoll, L., 19??a, “Design Specifications of Parts Dimensional Tolerance for Robotic Assembly”, source unknown.

6.16 Kjellberg, T. et. al., 1990, “Product Modelling: A Tool for Design and Manufacturing Systems Design”, source unknown (CIRP possibly)

6.17 Kroszynski, U. I., Palstroem, B., Trostmann, E. and Schlechtendahl, E. G., 1989, “Geometric Data Transfer Between CAD Systems: Solid Models”, IEEE Computer Graphics and Applications, Sept., 1989, pp. 57-71.

6.18 Takeuchi, Y., Sakemoto, M., Abe, Y. and Orita, R., 1989, “Development of a Personal CAD/CAM Systems for Mold Manufacture Based on Solid Modeling Technology”, Annals of the CIRP, Vol. 38, No. 1, pp. 429-432.

Petri Nets

6.19 Peterson, J. L., 1981, Petri Net Theory and the Modelling of Systems, Prentice-Hall, Inc., NJ.

6.20 Reisig, W., 1985, Petri Nets; An Introduction, Springer-Verlag

Process Planning

6.21 Alting, L. and Zhang, H., 1989, “Computer Aided Process Planning: the state-of-the-art-survey”, The International Journal of Production Research, Vol. 27, No.4, pp. 553-585.

6.22 Ham, I., 1988, “Computer Aided Process Planning: The Present and The Future”, Annals of the CIRP, Vol. 37, Jan., pp. 591-602.

6.23 Tsai, M. D., Takata, S., Inui, M., Kimura, F. and Sata, T., 1991, “Operation Planning Based on Cutting Process Models”, Annals of the CIRP, pp. 95-98.

6.24 Willis, D., Donaldson, I. A., Ranage, A. D., Murray, J. L. and Williams, M. H., 1989, “A Knowledge-based system for process planning based on a solid modeller”, Computer Aided Engineering Journal, Feb., pp. 21-26.

Software

6.25 P-E Inbucon, 19??, Hocus Manual, Hampstead, MD, USA.

6.26 Production Modeling Corporation, 1991, Promodel PC; User Manual, Orem, UT, USA.

6.27 Spatial Technology, 1991, “ACIS; Interface Guide”, Spatial Technology, Boulder, Colorado.