Statistical analysis of dross data for hydro aluminium casthouses

Abstract

Reducing the formation of dross is important for a sound economic result in aluminium casthouses. In order to reduce the amount of dross the main drivers affecting the dross creation need to be identified. The first step towards identifying these drivers is to measure the dross amount on a charge basis. With a sufficiently large data set it is possible to apply statistical methods to correlate different process variables and the dross amounts. It is also possible to rank the different variables and identify those that are the most important for dross formation. In this paper multivariate statistical analysis is used to correlate the various input variables and dross formation on a charge basis and to identify the most important drivers for dross formation. Examples from two remelt extrusion ingot casthouses and a primary extrusion ingot casthouse are given and discussed.