Fault and cause diagnosis of casting defects: A case study

Sweeting, F. B. J.; Thorpe, W. R.; Pettitt, A. N.

doi:10.1002/qre.4680080203

Quality and Reliability Engineering International

Explore this journal >

Explore this journal >

Article

Fault and cause diagnosis of casting defects: A case study

Authors

F. B. J. Sweeting,
1. School of Information Technology, University of Southern Queensland, Darling Heights, Toowoomba, Australia 4350
Search for more papers by this author
- Jeff Sweeting, prior to joining the staff of the University of Southern Queensland, was a senior partner in the international management consultancy of Urwick Orr & Partners, now Price Waterhouse Urwick. He has been involved in quality and productivity assignments in a wide range of industries and technologies. A graduate of the City and Guilds College of the Imperial College of Science and Technology, London he has served as a ford Foundation Advisor in their management aid programme to develop Asian consultants. A mechanical and industrial engineer, he is currently researching the application of information technology and statistical techniques to the cost-effective diagnosis of the cause of intermittent or erratic nonconformity in grey iron castings.
W. R. Thorpe,
1. Department of Mining and Metallurgical Engineering, The University of Queensland, St. Lucia, Queensland, Australia 4067
Search for more papers by this author
- Warren Thorpe received his Ph.D. in Physical Metallurgy from the University of Queensland in 1976. After a year at Griffith University working on computer modelling, he was awarded an AINSE Postdoctoral Fellowship at the University of Queensland to carry out research on high temperature mechanical properties of materials. Since joining the CSIRO division of Manufacturing Technology in 1979, he has worked on a variety of materials projects, specializing in computer applications. In 1989 he transferred to a joint appointment at the University of Queensland, Department of Mining and Metallurgical Engineering, where he is a Reader in Materials Engineering, and Officer in Charge of the Brisbane Laboratory of the CSIRO Division of Manufacturing Technology. His current research interests cover both the computer modelling and physical metallurgy of solidification as well as the statistical analysis of industrial data.
A. N. Pettitt
1. School of Mathematics, Faculty of Science, Queensland University of Technology, Brisbane, Australia 4001
Search for more papers by this author
- Tony Pettitt is currently Professor and Head, School of Mathematics, Queensland University of Technology, Brisbane, Australia. He obtained his Ph.D. in mathematical statistics from the University of Nottingham, U.K., and lectured for several years at Loughborough University of Technology, U.K. Recent positions in Australia have included Principal Research Scientist with CSIRO in the Division of Mathematics & Statistics and Biometrics Unit. A statistician by profession, current interests include applied statistics aspects of quality improvement, statistical modelling, biometrics and statistical consulting.

First published: 1992Full publication history
DOI: 10.1002/qre.4680080203 View/save citation
Cited by (CrossRef): 2 articles Check for updates
- Citing literature

Abstract

The product selected for detailed study was a cored grey iron disc, cast in a five cavity mould, subject to rejection at a highly variable rate. The behaviour of process variables in the sub-systems of green sand, core making, metal pouring and metallurgical composition was analysed to detect any statistically significant correlations with the incidence of blisters, which was the predominant cause of rejection.

To supplement the recorded data and to separate causal factors from correlations, a series of direct experimental trials was conducted to trace a faulty casting to the cavity, mould box, pouring conditions, green sand, cores and metallurgical composition pertaining at the time of actual production.

From a progressively accumulated database of some 100 hours of the specific product runs, built up over several months, the production variables and responses were analysed using regression, ANOVA, correlation and factorial experimentation techniques. Three major interventions involving a change in bentonite supply and two modifications to the runner system were accompanied by a reduction of some 60 per cent in the mean level of blister rejects with an improved dispersion.

Continue reading full article