Abstract:
1. OBJECTIVE OF STUDY
While considerable advances have been made in aluminium smelting to reduce energy consumption and minimize excess materials uses, the literature is devoid of technical information relating to relevant properties and contributions made by the cover material. The cover material acts as an insulator, absorbent for emissions from the cells, and a barrier to secondary reactions associated with air burn, and therefore can play an important role.
The broad objective of this Thesis is to address this omission by investigating the status of cover material, characterize its properties, and enable more precise predictions to be made for formulation changes. The study has been split into several parts, these being:
• survey of cover practices in smelters around the world
• sampling and characterizing cover materials from operating cells
• measuring thermal conductivities of constituent materials that are present in the cover (for different size fractions)
• modelling and analyzing the data
2. SURVEY OF TOP COVER
A survey of top cover in the smelters across the world has been conducted. The survey questionnaire has included cell design, operating conditions and the top cover and covering practice questions. The survey result has shown that the top cover is being required to play an increasing role in energy balance, material balance and anode protection. The typical cover materials have shifted from alumina to the optimized mixtures of crushed recycled anode cover with alumina. The management of top cover materials and the cover handling process has become rather challenging. The material stream needs to be maintained at a potline material balance, and the cover composition needs to be adjusted. In the cover handling process, the cover particulate size needs to be controlled, and the impurities needs to be removed.
Based on the survey, most of the smelters have updated one or more than one items of their cell technologies. 78% of the potlines have side-by-side cell arrangement, and about 60% of potlines have turned up the line amperage. As a result of original design or point feed conversion, almost all of the prebaked cell potlines are being operated at central point feeding mode. The side channel has become narrower, consequently larger anodes has been used to reach high cell productivity. The cell operating conditions in most of smelters have been improved with an optimized bath composition and temperature window.
3. CHARACTERISTICS OF COVER MATERIALS IN HALL-HEROULT CELLS
The samples from Hall-Heroult cells have been taken for XRD analysis, particle size distribution, phase analysis, and for thermal conductivity measurement in laboratory. These samples have included alumina, course crushed bath (CCB), large crushed bath (LCB), medium crushed bath (MCB), fine crushed bath (FCB) and the mixtures of CCB with alumina at the percentages of 20%, 40%, 60% and 80%.
SEM images have shown that the bath particles are irregular in shape, neither plate-like nor needle-like nor sphere-like, and a single bath looks like a pore-free and dense block. XRD analysis has shown that crushed bath contains Na3AlF6, Na5Al3F14, and NaF1.5CaF2.AlF3 as well as other minor components.
The powder packing density has been found to be sensitive to particle size distribution. CCB with full size range has the highest packing density, and other powders with discrete size distributions sieved from CCB have lower packing densities.
4. THE THERMAL CONDUCTIVITY OF GRANULAR MATERIALS
An experimental apparatus has been designed for radial heat flow to measure the thermal conductivity of granular materials from Hall-Heroult cells. The granular materials were held between two concentric cylinders to achieve and maintain a sample shape like a cylinder. The measurement was conducted under steady-state condition. Experimental design and arrangement have had the following advantages:
• The inner combined heating element has been deliberately designed to have a properly large diameter and low curvature surface so that a good contact is kept between the granular materials and the element surface, to produce a very homogeneous heat source so that a very homogeneous temperature is distributed on the element surface.
• The thermocouples have been embedded into the cylindrical surfaces in an axial direction so that the temperature gradients can be accurately measured with accurate locations and the packed granular material layer is to minimum extent interfered with the thermocouples.
• Metal surfaces have a homogenization function for temperature distribution and are favorable to measure temperature gradients for inhomogeneous powder phase like alumina or its mixtures with CCB with relatively large size. It means that the calculation of temperature gradients through the packed material layer is based on two averaged temperatures over two cylindrical surfaces rather than a limited numbers of point temperatures within the material layer, which exactly represents what is required by Fourier's equation given in integral form.
• End effect has been minimized by deliberately designing a compact apparatus and its end cover connections. In addition, end effect could be also measured by mounting two thin-film heat flux sensors onto end covers.
The thermal conductivities have been measured as a function of temperature in powder systems. Linear relationships between thermal conductivity and temperature have been found for all the samples in given temperature ranges with the following regressed equations:
Powder Equation Temperature Range Void Fraction
Al2O3 0.143+0.393x10-3T 140°C~800°C 0.685
80% Al2O3+CCB 0.160+0.450x10-3T 160°C~600°C 0.666
60% Al2O3+CCB 0.155+0.616x10-3T 160°C~600°C 0.626
40% Al2O3+CCB 0.206+0.653x10-3T 160°C~600°C 0.495
20% Al2O3+CCB 0.277+0.751x10-3T 160°C~600°C 0.378
CCB 0.270+0.913x10-3T 160°C~600°C 0.348
LCB 0.126+0.115x10-2T 160°C~600°C 0.478
MCB 0.224+0.857x10-3T 160°C~600°C 0.377
FCB 0.140+0.718x10-3T 160°C~600°C 0.447
The experimental reproducibility is within 1.4% and the maximum relative error in thermal conductivity measurement is less than 20%.
5. MODELING THERMAL CONDUCTIVITY
Theoretical analysis has been made on thermal conductivity of a powder system, and two types of series-parallel models based on Ohm's law and heat transfer have been discussed. As thermal conductivity in a powder system is strongly related to some macroscopic characterizations of particle packing, particle packing behaviors and packing models have been further discussed.
The effect of radiation on thermal conductivity has been observed from the LCB and the alumina-CCB mixtures at higher temperature and larger void fraction rather than larger particle as explained in the literatures. This effect will be disappeared when the void fraction becomes less than 0.4.
Kunii-Smith model and Deissler-Eian model have been chosen out of various models to predict thermal conductivity in the current powder systems and compared with the experimental results to test their validity. It is noted that there is a very good agreement at lower temperature but there is a larger difference at high temperature between predictions and measurements on thermal conductivity, as the radiant contribution at high temperature should be considered but it is not included in the simplified Kunii-Smith model or Deisser-Eian model, which further verifies the effect of radiation heat transfer on thermal conductivity in a real powder system.
6. HEAT LOSS AND ENERGY BALANCE IN INDUSTRIAL HALL-HEROULT CELLS
Anode yokes and stubs provide a parallel mechanism for heat flow from the upper surface of the anode carbon. It has generally been assumed that, as the anode ages, and the butt thickness is reduced, the heat flux via this mechanism will increase.
The measurement of heat flux through several types of top covers has been conducted in different prebaked cells. Temperature measurements have been made on a selection of anode assemblies of different design, including yokes, stubs and rods. These results have been also included in order to complete the picture for heat flow through both the cover material and the top surface of the anode.
As a result of heat flux measurements around the cell surfaces in four prebaked cells, energy balances have been calculated and heat losses in different zones have been evaluated. Finally a heat assembly model has been proposed for Hall-Heroult cells to evaluate a thermal state in the concerned zones when energy balance changes from one state to another state.