Modelling of Charpy V test rejection probability S. Tamminen*, I. Juutilainen and J. Ro ¨ning The purpose of this study was to develop a product design model for estimating the impact toughness of low alloy steel plates. The rejection probability in a Charpy V test is predicted with process variables and chemical composition. Joint modelling of the mean and deviation was used in order to improve the results. The proposed method is suitable for the whole production line, including all grades of steel in production and it is not restricted to a few test temperatures. Using the proposed model the product design group could have recognised most of the rejections before production. Next, the developed model will be implemented into a graphical simulation tool that is in daily use in the product planning department and already contains some other mechanical property models. The model will guide designers in predicting the related risk of rejection and in producing desired properties in the product at lower cost. Keywords: Impact toughness, Charpy V test, Product design, Heteroscedastic regression, MLP Introduction Manufacturing industries share a growing interest in quality control. Rejections in qualification tests are very expensive for a company, as one rejected product can easily cost thousands of Euros. As a result, the motiva- tion to reduce the number of rejected plates has aroused interest to develop models that provide help for process planning. Typically, one steel plant can have hundreds of products, and yet, the customers still make enquiries about new ones. The product design department will benefit from a model that has good interpolation capabilities when responding to the customers’ quality requirements. If the whole process is included to modelling, the transfer of information between products is possible. It is not an easy task, however, but powerful data mining methods can help to achieve this goal. Qualification of the impact toughness requirements of a steel plate is verified with a Charpy V test (CVT), which is a cost effective material testing procedure. The test is performed on three different samples from every steel plate, and the plate is accepted if the average of the measurements is higher than the requirement and none of the measurements is 30% below the requirement. When a product is designed, the risk of rejection in the CVT should be minimised. A model for rejection probability prediction would help in achieving this goal. Transition behaviour is typical for ferritic steel qualities. However, the impact toughness of these qualities can be affected by chemical composition and thermomechanical treatments. The effect of carbon concentration on transition behaviour is illustrated in Fig. 1. Steel is ductile at higher temperatures (the upper shelf) and it gets brittle at low temperatures (the lower shelf). The transition temperature is determined from the average of the upper and lower shelves. When the carbon concentration is low, the transition region from ductile to brittle is narrow, the upper shelf is high, and the slope between the shelves is steep. An increase in the carbon concentration lowers the upper shelf and also widens the transition region. 1 The effect of other alloying elements and process parameters on transition behaviour is similar (or reversed, if the parameter has a positive effect on impact toughness). The complicated interactions between these factors bring a challenge to modelling, as elements that are harmful alone can produce a desirable effect together with another component (e.g. nitrogen and aluminium). Steel’s behaviour in the transition region brings uncertainty to modelling, because in this area the force required to break the test bar can vary dramatically. Factors that raise the transition temperature (e.g. carbon, nitrogen, grain size) have a negative effect on impact toughness, as well. Furthermore, if the grain size is not uniform in the product, the transition temperature is affected by the biggest grain size instead of the average grain size. At room temperature steel can perform well in the impact toughness test, but when the temperature falls, its performance weakens. In this study the most demanding steel qualities are tested at temperatures as low as 2100uC. When modelling impact toughness the assumption of constant variance should be considered critically. In this article it is shown that rejection probability modelling can benefit from the use of an additional variance model together with a mean model in prediction. Department of Electrical and Information Engineering, University of Oulu, PO box 4500, FIN-90014, Oulu, Finland *Corresponding author, email satu@ee.oulu.fi ß 2010 Institute of Materials, Minerals and Mining Published by Maney on behalf of the Institute Received 31 October 2008; accepted 15 August 2009 DOI 10.1179/030192309X12506804200780 Ironmaking and Steelmaking 2010 VOL 37 NO 1 35