Statistical Process Control (SPC) based on the Shewhart’s type control charts, is widely used in contemporary manufacturing industry, including many foundries. The main steps include process monitoring, detection the out-of-control signals, identification and removal of their causes. Finding the root causes of the process faults is often a difficult task and can be supported by various tools, including datadriven mathematical models. In the present paper a novel approach to statistical control of ductile iron melting process is proposed. It is aimed at development of methodologies suitable for effective finding the causes of the out-of-control signals in the process outputs, defined as ultimate tensile strength (Rm) and elongation (A5), based mainly on chemical composition of the alloy. The methodologies are tested and presented using several real foundry data sets. First, correlations between standard abnormal output patterns (i.e. out-of-control signals) and corresponding inputs patterns are found, basing on the detection of similar patterns and similar shapes of the run charts of the chemical elements contents. It was found that in a significant number of cases there was no clear indication of the correlation, which can be attributed either to the complex, simultaneous action of several chemical elements or to the causes related to other process variables, including melting, inoculation, spheroidization and pouring parameters as well as the human errors. A conception of the methodology based on simulation of the process using advanced input - output regression modelling is presented. The preliminary tests have showed that it can be a useful tool in the process control and is worth further development. The results obtained in the present study may not only be applied to the ductile iron process but they can be also utilized in statistical quality control of a wide range of different discrete processes.
The paper presents an application of advanced data-driven (soft) models in finding the most probable particular causes of missed ductile iron melts. The proposed methodology was tested using real foundry data set containing 1020 records with contents of 9 chemical elements in the iron as the process input variables and the ductile iron grade as the output. This dependent variable was of discrete (nominal) type with four possible values: ‘400/18’, ‘500/07’, ‘500/07 special’ and ‘non-classified’, i.e. the missed melt. Several types of classification models were built and tested: MLP-type Artificial Neural Network, Support Vector Machine and two versions of Classification Trees. The best accuracy of predictions was achieved by one of the Classification Tree model, which was then used in the simulations leading to conversion of the missed melts to the expected grades. Two strategies of changing the input values (chemical composition) were tried: content of a single element at a time and simultaneous changes of a selected pair of elements. It was found that in the vast majority of the missed melts the changes of single elements concentrations have led to the change from the non-classified iron to its expected grade. In the case of the three remaining melts the simultaneous changes of pairs of the elements’ concentrations appeared to be successful and that those cases were in agreement with foundry staff expertise. It is concluded that utilizing an advanced data-driven process model can significantly facilitate diagnosis of defective products and out-of-control foundry processes.