Thermal conductivity of the side ledge in aluminium electrolysis cells: compounds as a function of temperature and grain size

TL;DR

This study develops temperature- and grain size-dependent models for thermal conductivity of materials in aluminium electrolysis cell side ledges, combining experimental data, physical models, and ab initio calculations to improve numerical simulations.

Contribution

It provides a comprehensive formulation of thermal conductivity for various phases in the side ledge, integrating experimental data, physical modeling, and first-principles calculations.

Findings

Model predictions agree well with experimental data.

Thermal conductivity varies with temperature and grain size.

New formulations enable better numerical modeling of the side ledge.

Abstract

In aluminium electrolysis cells, a ledge of frozen electrolyte is formed, attached to the sides of the cell. The control of the side ledge thickness is essential in ensuring a reasonable lifetime for the cells. Numerical modelling of the side ledge thickness requires an accurate knowledge of the thermal transport properties as a function of temperature. Unfortunately, there is a considerable lack of experimental data for the large majority of the phases constituting the side ledge. The aim of this work is to provide, for each phase possibly present in the side ledge, a formulation of the thermal conductivity as a function of both temperature and size. To achieve this, we consider reliable physical models linking the density of the lattice vibration energy and the phonon mean free path to key parameters: the high temperature limit of the Debye temperature and the Gruneisen constant. These model parameters can be obtained by simultaneous fitting of (i) the heat capacity, (ii) the thermal expansion tensor coefficient and (iii) the adiabatic elastic constants, on relevant physical models. Where data is missing, first principles (ab initio) calculations are used to determine directly the model parameters. For compounds for which data is available, the model's predictions are found to be in very good agreement with the reported experimental data.