DOI 10.1007/s00170-004-2388-9 ORIGINAL ARTICLE Int J Adv Manuf Technol (2005) Y.-M. Deng · G.A. Britton · Y.C. Lam Towards automatic shape modiﬁcation in injection-moulded-plastic-part design Received: 6 May 2004 / Accepted: 23 August 2004 / Published online: 18 May 2005  Springer-Verlag London Limited 2005 Abstract Injection-moulded-plastic-part design must ensure that the part can be manufactured to the desired quality level by the injection moulding process. Simulation software has been widely used in industry to assess mouldability and measures of quality. However, it cannot be used to improve a design directly. Design modiﬁcations must be performed by the designer after evaluation of the simulation results. Based on the authors’ pre- vious work on injection moulding CAD-CAE integration, this paper explores the strategies and methods for automatic-part- shape-modiﬁcation to attain a desired part quality. An enhanced CAD-CAE integration model is developed. This model is used to specify the shape-modiﬁcation variables, as well as the mould- ability and other quality measuring criteria. The shape modiﬁca- tion variables include positional and sizing parameters of each individual feature, as well as those associated with the part, such as part thickness. With this information, an iterative process of part-shape modiﬁcation and execution of simulation subroutines is carried out automatically, and the results are veriﬁed and eval- uated. Optimal shape, according to the speciﬁed criteria can thus be derived from the evaluation results. A software prototype has been developed. A design case study is presented to illustrate and demonstrate the usefulness of the proposed strategies and methods. Keywords Injection moulding · Plastic part design · Shape modiﬁcation Y.-M. Deng (✉) · Y.C. Lam Singapore-MIT Alliance (SMA), N2-B2C-15, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore E-mail: mymdeng@ntu.edu.sg Tel: +65-67904273 Fax: +65-68627215 G.A. Britton CAD/CAM Lab, School of Mechanical and Production Engineering, Nanyang Technological University, Nanyang Avenue, Singapore 639798, Singapore 1 Introduction Injection moulding is a major manufacturing method for the fab- rication of plastic parts. Part design is one of the most important design tasks in injection moulding. During the design process, designers must take into account functional requirements of the part as well as its mouldability. The quality of a part is inﬂu- enced by many factors, such as selection of material, moulding machine, and injection mould, as well as processing conditions. Concurrent engineering techniques are recognised as facilitating strategies in ensuring successful part design [1–4]. One way to apply concurrent engineering strategies in part design is to get part designers to work with both the customers and marketing personnel on the front end, as well as the mould designers, material suppliers, and process engineers on the back end [1]. Another important factor in the success of concurrent engineering is the use of design support tools that can provide relevant information and feedback with various degrees of de- tail [5]. Various computer-based tools have been developed to assist designers in assessing the mouldability, quality, and cost of a design. Among these tools, simulation packages are the most popular. For example, Moldﬂow is widely used in industry to simulate the moulding process before part design and moulding- process design are ﬁnalised and actual tooling is started. However, existing computer-based tools only provide infor- mation relating to the mouldability and performance of a design; that is, they can only act as an evaluation tool. They are not capable of directly improving design. To address this problem, a number of optimisation algorithms have been proposed based on the utilisation of the simulation software, such as gate loca- tion optimisation [6, 7], feed system optimisation [3], moulding condition optimisation [8], cavity balancing [9], and part thick- ness optimisation [10]. These optimisation algorithms require iterative executions of simulation routines. However, for each it- eration, modiﬁcation is only made to the CAE analysis model (mesh model and other input parameters). For example, in gate location optimisation, the gate location is changed, which can be implemented by setting a different mesh node of the analy-