Advanced predictions tools, foundry process control and knowledge management for irons castings

It is well known that the large number of variables that interact during the foundry process bring about many drawbacks. These difficulties are even bigger when we try to predict the behaviour of the process since it is extremely complicated to establish correspondence relationships among the most critical variables on the basis of relating data. The simulation tools, the control devices and the process management systems used, are very useful but they do not satisfy the stringent demands of the foundry industry. The above mentioned tools work independently and they do not take into account the relationships existing among them. This research takes into consideration certain IT generic tools that, once adapted to the foundry process and implemented on the basis of specific knowledge, are capable of processing and interrelating a huge amount of data in such a way that they can predict the final quality of the castings, maintaining at the same time the process under controlled conditions. These tools manage the information coming directly from the foundry plant, what allows to strengthen the process and make a continuous progress by a constant information feedback, helping to improve the own rejection rate levels, even shown in ppm, etc. The fact of developing tools capable of managing all these concepts has been considered a utopia in the foundry process for a long time. The analytical process used is based on the selection of concrete incidences, parameters and defects. The system assigns them the potential causes considered more probable by the classic knowledge and later on, they are selected and given priority according to objective criteria. The conclusions reached are based on applications and verifications carried out on different foundries. They have allowed us not only to validate the correct functioning of the system but also to verify its efficiency according to the success rate. It is possible to ‘master the process’, reduce the variability rate, minimise incidences and manage efficiently the own knowledge by using the data existing in each foundry and by integrating the different prediction and control tools.