A hybrid design support system for die casting scheme

Abstract

The increasing demand for casting quality, rapid production, process complexity and low cost requires continuous improvement of the die casting process. However, the complexity of the casting process rules out that an optimal solution can be solved purely by an empirical and analytical approach. Development and application of a computerized system is an important step to automate the die design process and compensate for the lack of design experts. This thesis employs the knowledge and feature-based approach, fuzzy judgment technique and computer-aided method to present a hybrid design support system (HDSS) for a die casting scheme. The design of a die casting scheme is a critical phase in die design. Die casting schemes often determine the successfulness of a die casting die. The design of the die casting scheme includes three parts, i.e., parting scheme design, gating system configuration and selection of filling parameters. In this thesis, the development of a hybrid design support system is described. The choice of a suitable parting scheme is a crucial step in determining the layout of a die and gating system. A parting scheme is mainly influenced by concave features in die casting. A concave feature may prevent the removal of a die casting part from a die cavity and influence the use of side cores. Geometrical feature analysis involves detection of concave features and determination of their release directions. A ray detection method is proposed for recognizing and extracting concave features from a die casting. With this ray detection method, each of the concave feature is assigned a block factor. A set of concave regions and their hulls can be built by examining the block factors, The points in a concave hull carry geometrical information required for the determination of normal vector and release direction. The release direction denotes a direction or a set of directions along which the concave region can be released. Each concave feature can correspond to a specific release direction or a set of release directions. A graphical method, release direction (RD) map, is used to study release direction. Such graphical methods can represent clear contours of the release directions. Based on the translation characteristic of a rigid object in mechanics, the release direction of die east objects is deduced. The minimal release direction principle and vector plane approach are proposed to solve for the release direction of concave regions. Parting scheme is a geometrical feature dependent in nature. In the present study, the parting scheme design is divided into three stages, parting scheme generation, parting scheme evaluation and parting scheme optimization. In the stage of parting scheme generation, the direction-shifting approach is used to generate the parting schemes. Each possible parting direction will match one or more planes as possible parting schemes based on heuristic knowledge or design rules. All possible parting schemes can be generated by shifting all of the possible parting directions. The system allows designers to evaluate the performance of parting schemes. The performance of a parting scheme is identified by multiple design rules. In order to evaluate the parting schemes, a series of evaluation models is developed by formulating design rules as satisfaction or test functions. Then the parting scheme is evaluated by the function values, each of which reflects an aspect of the performance of a parting scheme. Thus a quantitative measurement on parting scheme performance can be realized. A multi-criteria weighted optimization model is used for the determination of an optimal parting scheme from candidate parting schemes. Analytical formulae are used to model the parting scheme of casting dies. Before a gating system is designed, the value of the parameters should be selected. A fuzzy synthetic evaluation technique is proposed for the selection of process parameters. The fuzzy synthetic evaluation divides factors and value domain of parameter into discrete quantities to create a fuzzy factor set and an alternative set. By fuzzy transform, the contribution of each discrete level and each factor can be synthetically considered. The fuzzy synthetic evaluation can give reasonable and consistent result if the factor set and alternative set are provided. The proposed approach overcomes the shortcoming of the conventional method in the selection of die casting parameters. In development of gating design system, a knowledge-based technique is applied. The case-based approach and feature-based CAD approach are combined to generate a gating configuration. The case-based approach is used to match the successful cases of gating configuration by case reasoning. With reference to the matched case, a new gating system can be designed. The feature-based CAD approach is used to design a gating system under an interactive CAD environment based on casting feature and gating knowledge. A knowledge representation scheme based on the knowledge object is developed for gating configuration. The hybrid knowledge representation can integrate the feature representation and inference in a knowledge object. The dimension dependency network is developed to define gating dimensions. A gating configuration is produced either by case-based approach or by feature-based CAD approach. The gating position, gating type and gating dimensions can be recommended. In the thesis, the proposed methodology was tested based on real castings from industry and the recommended designs were found to be capable of producing sound and acceptable castings. The above exploratory work links the knowledge of different disciplines to develop a hybrid design support system for a die casting scheme. The hybrid approach proposed in this thesis is one of the first of its kind and contributes to the automation of CAD system in die casting die design.