Elsevier

Computers in Industry

Volume 47, Issue 1, January 2002, Pages 39-53
Computers in Industry

Modelling and optimisation of Rapid Prototyping

https://doi.org/10.1016/S0166-3615(01)00140-3Get rights and content

Abstract

This paper proposes a Virtual Reality (VR) system for modelling and optimisation of Rapid Prototyping (RP) processes. The system aims to reduce the manufacturing risks of prototypes early in a product development cycle, and hence, reduces the number of costly design-build-test cycles. It involves modelling and simulation of RP in a virtual system, which facilitates visualisation and testing the effects of process parameters on the part quality. Modelling of RP is based on quantifying the measures of part quality, which includes accuracy, build-time and efficiency with orientation, layer thickness and hatch distance. A mathematical model has been developed to estimate the build-time of the Selective Laser Sintering (SLS) process. The model incorporates various process parameters like layer thickness, hatch space, bed temperatures, laser power and sinter factor, etc. It has been integrated with the virtual simulation system to provide a test-bed to optimise the process parameters.

Introduction

Rapid Prototyping (RP) or Layer Manufacturing (LM) refers to fabrication of parts layer-by-layer. It involves adding raw material successively, in layers, to create a solid of a predefined shape. These parts are used in the various stages of a product development cycle. Wholers [1] conducted a survey, and found that around 23.4% of RP parts are used as visual aids, whereas 27.5% of them are used as master patterns for secondary manufacturing processes and for direct tooling. Industries use 15.6% of them for fit and assembly tests, 16.1% for functional tests and the rest for quoting, proposal, ergonomic evaluation, etc.

Fig. 1 shows the flow of a typical RP process. The first step is to validate the 3D CAD model of a part, i.e. to ensure it is a solid, which must be repaired otherwise. The valid model is then oriented with respect to the build chamber, by considering the build-time and the surface quality. A few models may either be merged into a one-build assembly or nested for efficient utilisation of the machine and the material. Based on the process requirement, support structures may be added to the model, if necessary. It is then sliced with a set of horizontal planes. Each horizontal plane yields a planar slice contour, which is cross-hatched to determine the laser trajectories to control the sintering/solidification process. By scanning one layer over another, the part grows incrementally to its final shape. Thus, the main steps for process planning include orientation, support structure generation if necessary, slicing and selection of process parameters.

Process planning is performed to select the process parameters and to generate the control instructions to fabricate a part. In general, the designer carries out process planning by studying the part and quality requirements, which is indeed very time-consuming. Therefore, there is a need to automate the process. This can be achieved by linking the designer’s understanding and decision making with the physical process to create parts of the desired quality. Automation of process planning is also one of the fundamental aims of RP [2], which are

  • to build arbitrarily complex 3D shapes;

  • to use a generic fabrication machine which does not require part-specific fixturing or tooling;

  • to generate a process plan automatically, based on a CAD model;

  • to minimise human intervention.

RP facilitates fulfilment of the first two aims mentioned above. However, it requires a significant amount of human intervention to produce an optimal part. The optimality depends on the functional requirements, which include accuracy, build-time, strength and efficiency. The quality requirements, however, vary from visual aids to master patterns for secondary processes. Hence, a significant degree of expertise is required to produce parts of consistent quality. The process is, therefore, very costly and of a trial-and-error basis. The objectives of this paper are to quantify the requirements for optimisation of RP and to simulate the fabrication of prototypes for visualisation in Virtual Reality (VR).

VR is an advanced human-computer interface that simulates a realistic environment and allows a designer to interact with it. The essence of VR is immersion and interactivity, which differentiates it from CAD systems. Immersion means to block out distractions and to focus on selective information with which the designer wants to work. Interactivity implies the ability that humans interact with events in the virtual world [3]. Applications of VR have recently gained considerable momentum in industries. Resseler [4] presented a summary of its applications in manufacturing. Dai and Gobel [5] discussed the advantages of Virtual Prototyping (VP) over physical prototyping. They considered VP as the integration of VR with product design and simulation, and listed the constraints of computer graphics that have to be solved for effective implementation of VP. They suggested the use of electronic prototyping as an alternative to physical prototyping. However, this is not yet feasible with the present technology due to the weakness of the link between CAD systems and CAE components like FEA, kinematics and dynamics systems.

Few researchers have combined the advantages of VP with RP technologies. Gibson et al. [6] investigated the contributions of VR and RP towards a more efficient product development in ergonomic, aesthetic and functional aspects of design. They suggested the use of VR as a complementary technology to RP with an interface accommodated through a CAD system. Fadel and co-workers [7], [8] linked VR with RP to visualise the support structures of a part. They coupled RP with VR by developing the Interactive Virtual Environment for Correction of Stereolithography Tessellated List Files System (IVECS). It is a tool to perform minute surgeries on faulty tessellated models. Indeed, correcting a faulty Stereolithography Tessellated List (STL) file is tedious.

Automation of control code generation for the desired requirements is an emerging research aspect of RP. Diane et al. [9] classified RP process parameters into nuisance, constant and control parameters. Nuisance parameters include age of the laser, beam position accuracy, humidity and temperature, which are not controlled in the experimental analysis but may have some effect on a part. Constant parameters normally include beam diameter, laser focus and material properties, etc. The control parameters will affect the output of the process and are controllable in a run. These include layer thickness, hatch space, scan pattern, part orientation, shrinkage of the material and beamwidth compensation, etc. Diane et al. [9] concluded that layer thickness, hatch space, part orientation and depth of cure are the most vital among the control parameters. They conducted experiments with hatch space, orientation, layer thickness and overcure depth, and confirmed their influence on Stereolithography (SLA) parts quality is significant.

Zhou and Hersovici [10] presented accuracy problems in SLA process, and established that layer thickness, hatch space, overcure, gap and the position on the build-plane of the SLA process are control factors of accuracy. They employed Taguchi method to find the functional relationships between different combinations of control factors and part quality for standard surface features. However, extrapolation of these results to complex RP part surfaces is very difficult.

Thomson and Crawford [11] chose build-time, surface finish and part strength for manufacturing requirements and developed numerical methods to quantify the requirements with respect to the part orientation for the Selective Laser Sintering (SLS) process. A genetic algorithm was developed by Woodzaik et al. [13] to automatically place multiple parts in a workspace to reduce build-time, and thereby, increase efficiency. The parts are enclosed in rectangular boxes and are rotated 90° about the z-axis to aid part packing, without considering the surface accuracy and the support structure requirements. Anne et al. [14] presented an integrated software system for process planning for LM under development at University of Michigan. Ablani and Bagchi [15] developed a software system to find preferred orientations. It rotates a part in increments about the designer-supplied axes and slices the part to evaluate the errors due to the stair-step effect. However, they considered rotations of the part in the range from 0 to 360°, which is not entirely necessary for the estimation of surface accuracy.

Due to the complexity involved, most research work has been focused on the optimisation of a single requirement or parameter. For example, in the case of packing parts in the workspace, consideration is given only to minimising the build-time. Such algorithms are useful for solving a single requirement or optimisation of a parameter. However, they lack the flexibility for multiple requirements or tuning a few parameters according to the desired quality.

This paper proposes a VR system for modelling and optimisation of RP. It provides a test-bed for optimising various requirements of an RP process. Fig. 2 shows the flowchart of the proposed approach. The designer starts with building a 3D model of a part, and subsequently performs VR simulation of the RP process to optimise the control parameters. When the desired requirements are met, they can be used for physical fabrication. If the fabrication is to be carried out by a service bureau or another department, the designer can perform virtual simulation with the control parameters. The virtual part may subsequently be sent by conventional means or through Internet to the designer/customer for assessment, as represented in dotted lines. This facilitates feedback for design improvement. Once the desired quality is obtained through virtual fabrication, physical fabrication of the part may follow. This approach will consequently reduce the time required to communicate, as well as the number of iterations. The designer may build and break as many parts as required more quickly at a relatively low cost. Thus, virtual simulation of RP parts will help reduce, if not eliminate, the number of physical prototypes required to produce a part. The designer may conveniently realise and validate the intended part before committing to manufacture.

Section snippets

The virtual simulation system for Rapid Prototyping

The architecture of the proposed system is shown in Fig. 3. As an initial implementation, the system is first targeted at the SLS process. It may be subsequently enhanced to incorporate the characteristics of other RP processes. The input to the system is a CAD model in STL format, which is the de-facto standard in RP industry. The VR generator WorldToolKit (WTK) creates a virtual world describing the environment of the system. It also manages the geometry and rendering optimisation, and

Implementation of the system

A system that simulates the layer building process of a part to estimate the surface accuracy and build-time has been developed. It reads a product model in STL format and adopts uniform slicing to process layers for simulation of the RP fabrication. Though some new developments such as adaptive slicing and direct slicing have been reported, most RP machines still support only uniform slicing of STL files. The present system consists of the following modules, which include: (1) model display;

Case study—fabrication simulation of a part

To demonstrate the application of the system, fabrication of a turbine fan on a Sinterstation 2000 SLS machine with nylon was simulated, and the result is shown in Fig. 8. The laser diameter was 0.04 mm, while the hatch space and the layer thickness were both set as 0.1 mm. The turbine fan has different features like a hollow cylinder and several freeform surfaces around it. Depending on the distribution of the facets, the surface accuracy, build-time and efficiency could be different for

Limitations and further development

A limitation of the present system is that there is a decrease in the speed when it processes layer contours for simulation of relatively large and complex models. To improve this, effort is now devoted to developing a methodology to simplify the contour hierarchy of complex layers.

Another possible improvement of the system is the incorporation of shrinkage and warpage effects to enhance accuracy estimation. Effort is now focused on the development of a dexel-based RP modelling technique. A

Conclusions

This paper proposes a virtual system to aid a designer to select process parameters for RP. The system adopts surface accuracy, build-time and orientation efficiency as the key manufacturing requirements. Part orientation, layer thickness and hatch space were identified as the key control parameters that influence the requirements significantly. A mathematical model was developed to relate the requirements with key control parameters. Preliminary implementation of the system has been completed.

Acknowledgements

The authors would like to thank the Research Grants Council of the Hong Kong SAR Government and the CRCG of the University of Hong Kong for their financial support for this project. Thanks are also due to Dr. I. Gibson and his colleagues for providing access to the Sinterstation 2000 SLS machine.

S.H. Choi is associate professor in the IMSE Department at the University of Hong Kong. He obtained both his BSc and PhD degrees at the University of Birmingham. He worked in computer industry as CADCAM consultant before joining the University of Hong Kong. His current research interests include CADCAM, advanced manufacturing systems and VP technology.

References (19)

  • T. Wholers, Rapid Prototyping State of The Industry—1999 World Wide Progress Report, RPA-SME Publication,...
  • D.L. Bourell, J.J Beaman, H.L. Marcus, J.W. Barlow, Solid freeform fabrication: an advanced manufacturing approach, in:...
  • S.L Springer et al.

    State of the art Virtual Reality hardware for computer-aided design

    Journal of Manufacturing Systems

    (1996)
  • S. Resseler, Applying Virtual Environments to Manufacturing, NISTIR, p....
  • F Dai et al.

    Virtual Prototyping—an approach using VR techniques

    Computer in Engineering

    (1994)
  • I. Gibson, D. Brown, S. Cobb, R. Eastgate, Virtual Reality and Rapid Prototyping, in: Proceedings of IEE workshop on...
  • G.M. Fadel, D. Crane, L. Dooley, R. Geist, Support structure visualisation in a Virtual Reality environment, in:...
  • S.M. Morvan, G.M. Fadel, IVCES: interactively correcting STL files in a virtual environment, in: Proceedings of the...
  • A.S Diane et al.

    Optimising Stereolithography throughput

    Journal of Manufacturing Systems

    (1997)
There are more references available in the full text version of this article.

Cited by (0)

S.H. Choi is associate professor in the IMSE Department at the University of Hong Kong. He obtained both his BSc and PhD degrees at the University of Birmingham. He worked in computer industry as CADCAM consultant before joining the University of Hong Kong. His current research interests include CADCAM, advanced manufacturing systems and VP technology.

S. Samavedam got his degree in mechanical engineering from the Nagarjuna University, India, in 1994. He is now a research student in the IMSE Department at the University of Hong Kong, and his research interest is in the development of VR for RP.

View full text