Modelling the electrical resistivity of green carbon anodes for aluminium industry
|Advisor:||Darvishi Alamdari, Houshang; Mashreghi, Javad|
|Abstract:||Primary aluminium produced nowadays is obtained by electrolysis of alumina at 960 °C, following the process proposed by Charles Martin Hall, and Paul Louis Toussain Héroult, in 1886. This process, named Hall-Héroult, due to the name of its creators, consists in applying an electrical current trough a mix of molten cryolithe in which is dissolved alumina. The current flows between the carbon anodes and the cathode. Anodes are mostly composed of calcined petroleum coke aggregates and remnant particles of used anodes, butt particles. Those carbonaceous particles are tied together using coal tar pitch. Once the anode bloc is formed, it is backed to gain mechanical strength. To allow an efficient yield during the electrolysis process, anodes, through which the current flows, shall have the lowest electrical resistivity. Furthermore, those consumable anodes, are also required to bring the carbon as reactant for the electrolytic reaction, consequently, their chemical composition must be pure enough not to diminish the produced aluminium quality. The research project presented focuses on the study and understanding of electrical behaviour of the anode prior to its backing, for quality control during manufacturing process. To determine the electrical resistivity of a composite material, Nielsen’s model appears as an interesting tool, reliable for multiple electrical modelling applications. Using this model implies knowing several properties of the different phases present in the anode. In the present case, anode material is restricted to a specific size fraction of coke particles and binder matrix, made of pitch and fine particles. The properties to know are, for the coke particles, their intrinsic electrical resistivity, aspect ratio, and the maximal packing fraction that can be reached. Only electrical resistivity is required for the binder matrix. In aluminium industry, only the electrical resistivities of beds of particles are measured to follow the evolution of this parameter depending the batch used. To implement the electrical resistivity of coke particles in Nielsen model, its intrinsic electrical resistivity is required. To obtain this value, measurement were performed using four probes setting, which allows measuring the electrical resistivity of material presenting low resistivity. Then, inter-particles void present in the measurement volume shall be removed in order to only take account of the volume occupied by the material investigated. Finally, the contacts between particles are estimated by numerical calculation, using discrete element method. The number, average surface and disposition of the contacts were assessed to implement them in a mathematical model allowing calculating the electrical resistivity of coke. Several measures lead to the proposition of a consistent and reliable value for electrical resistivity of coke. To measure the electrical resistivity of binder matrix, as well as the anode’s, an accurate method is required. The standardised method used in the industry reveals strong standard deviation and rather not often reproducible values. The method of measurement of electrical resistivity of continuous phases proposed by Van der Pauw allows retrieving highly reproducible results, with a much lower standard deviation. This technique can be useful to measure the electrical resistivity of slices cut out of anode cores or cast binder matrix. Several measurements, performed on anode like samples with different composition revealed that the electrical resistivity measured in laboratory does not fit with the one calculated using Nielsen’s model, using the physical parameters of the characterized materials. To explain the difference, between the model and laboratory measurements, an optical microscopic analysis was performed. It was disclosed that the binder matrix layer between particles remain too thick to allow the creation of an electrical path from particle to particle within the material. This implies that the electrical resistivity of the binder matrix mostly dictates the one of the composite material, in this case.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||21 August 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.