Improvement of sampling and analysis procedures for clean and high purity steel

The ultimate objective is to improve steel making capability such that steels can be made to closer composition tolerances and with greater consistency from cast to cast. This require improvements to sampling and analysisprocedures. In order to fulfil this main objective, four major activities are performed : - adaptation of new developments in the field of analytical methods (PDA-OES and FGA) to improve the control of steel making process. Most of the developments concern the PDA-OES technique with the objective of improving the detection limits and the analytical precision and providing faster information on cleanness of steel products, - rapid description of the possible heterogeneities (defects or segregations) of the metal at different stages of the process (mapping system), - improvement of liquid steel sampling to be sure that informations can be used for predicting the final cleanness and purity of the metal. These improvements should result from fundamental design studies with quantification of their effectiveness by analytical techniques, - adaptation of the techniques and procedures to on site analysis. Some of the developments will be tested in industrial plants to control their effectiveness. The potential of the PDA-OES technique has been studied in term of significant of the emitted signal. The influence of the applied energy on the extracted matter and consequently on the possible signal quantification has permit to quantify particularly Al on oxide form as well as oxygen content. The possibilities in term of size distribution is defined and propose a low limit of detection which is around 1,5μm inclusion size diameter. This project has examined the sample-taking process, and tested methods to improve sample quality and to make more effective use of data from poor samples. Consideration has also been given to measurement of segregation and cleanness. The quality of samples from different sampler types and with different deoxidants were compared; many samples were found to contain cavities, which could be the cause of erroneous analysis. Samples taken at the mould have been found to be of inferior quality to tundish samples, but, even for the latter, sample quality was unsatisfactory without a deoxidant. Those containing aluminium or titanium deoxidant were found to give better samples than those without, although a minimum of 1mm of material had to be removed from the surface to ensure homogeneity. This needed to be balanced against the likelihood of meeting cavities at greater depths. For grades where neither of these deoxidants may be used zirconium was tested as an alternative, and gave better results than samples where no deoxidant was added. Samplers containing these plates were found to give significantly improved results on low carbon steels (<0.5%), but this was not replicated at higher levels. The potential to improve results by the use of pulse discriminated optical emission data has been demonstrated. This information was used to identify poor excitation periods by measuring the amount of variation between individual sparks in a burn. PDA techniques have also been used to map and measure segregation of carbon and other elements in slab samples, and to detect inclusions.