Why data-density method is better than step-cooling method to identify Liquidus temperature

摘要

This paper advances an alternative data-density method, which provides a more accurately and reliably identify on-line liquidus temperature (Tl) of molten aluminium electrolyte. An in-house computer program of data-density algorithm was developed for a complete investigates and analysis the relationship between data-density and temperature curve of molten aluminium electrolyte. Traditional step-cooling algorithm detects Tl largely depending only on temperature curve geometry, such as inflection point is obvious or obscure and imperceptible. When there is not inflection point on temperature curve, in such case, traditional step-cooling algorithm is unable to firm clarify there exists Tl or not, not mention to identify Tl. Experiments and tests show that even slight or nearly no nuance of inflection point presented on the temperature curve, but there exists a little different data-density on the temperature curve, which can be captured and identified by data-density algorithm. Direct measurements in industrial cells by data-density algorithm is clearly outperforms conventional step-cooling first derivatives and step-cooling second derivatives methods for determining the Tl of molten aluminium electrolyte.