Revamping of the automation and measurement systems of a steel mill and uncertainty evaluation

The AOD (Argon Oxygen Decarburization) system is the part of a steel mill, found upstream in the manufacturing process, that modifies the steel composition to give it its main characteristics. The AOD is controlled by a PLC (Programmable Logic Controller) system and a weight measuring instrument is used to choose the quantity of material charged on a hopper. The hopper is then transported to the melted steel so that it can dump its content to change the composition. The weight indication given by the scale has to have a maximum admitted error that is under the threshold of 1% of the weight, so that the characteristics of the final material do not have appreciable variations and, as a consequence, expensive additions do not have to be made afterwards to reach the target composition. During the thesis activity, a revamping of the system used to control the process has been done to update the older PLC system composed by two obsolete CPU (Central Processing Unit) models with up to date components that have new functionalities. The study of the indication uncertainty has been done by analysing the measuring system while consulting the modules data sheets and by studying the uncertainty propagation through the measuring chain. The PLC update is done by creating a program that can work in the new hardware configuration, which has been implemented by consulting its electrical schemes. The weight uncertainty is evaluated using the probabilistic method and the PLC program is written using the TIA (Totally Integrated automation) portal software. Blocks already used in the current system have been updated to fit in the new program and new functions have been written to handle the movement of the hopper. One of the main outcome of the measuring system investigation was that the most important uncertainty contribution is related to a not proper calibration of the PLC weighing module, the Siwarex, while the contributions related to the loading cells, the junction box and the analog-to-digital converter negligibly affect the weight measurement. For this reason, particular attention has been paid towards the correct implementation of calibration procedures for these types of weighing systems. The PLC program shows that code written in a up to date environment presents several advantages over code written in older systems and this makes possible maintenance interventions easier to be performed; the update of older control systems seems like a good direction to take for future development of other parts of the steel manufacturing process. The weight uncertainty study has shown that, after a proper scale calibration, the expanded uncertainty (confidence level of 95%) is of about 0.15 % of the net weight. A program that handles the revamped system and which can operate the hopper in four different ways has been developed and successfully simulated.