| dc.description.abstract | Carbon required for the reduction of alumina by electrolysis in the Hall-Héroult process is supplied by the anode. Anodes are produced from a mixture of petroleum coke with coal tar pitch as the binder. The minimum theoretical carbon consumption is 0.334 kg C/kg Al. But due to current efficiency and oxidation losses the consumption is typically 400 kg C/kg Al.
Petroleum coke is a by-product from crude oil refineries, but as its value represents only about 2 per cent of the overall production it has a limited interest to the producers. Mechanical and physical properties of coke are influenced by the crude oil, processes within the refineries and calcining of the coke. Continuous high demand for calcined coke by aluminium smelters has created a difficult situation with respect to quality and availability, leading to the use of lower quality coke in aluminium smelters and potential disturbances in the production.
Understanding the consequences of varying calcined coke qualities is crucial for to possibly compensate and adjust process parameters in the subsequent use of the coke, in order to obtain economical production of aluminium.
In the present work, the subject of study was four different petroleum coke types, where three were derived from a single source (SSA, SSB and SSC) and one was from a blend of different cokes. All of the cokes had different chemical and physical properties. Single source cokes have homogenous chemical and physical composition in the whole size range while the blended coke had heterogeneous composition due to the mixture of different coke types used. The blended coke consisted of several different coke types (25) in order to meet the required specification.
This work also describes new characterisation/measurement methods which can contribute to a better understanding of variations in material properties due to physical and mechanical changes in the calcined coke feed materials.
Each step in the production of pilot scale anodes is described. A new method was developed for the characterisation of the mechanical strength of calcined coke. This method, the drop test, originates from the study of materials for road construction for the determination of the grain stability of rock materials. The method determines the coke grain strength in the entire size distribution. On the other hand, established measuring methods for calcined coke, HGI (ASTM D5003-95) and grain stability (ISO 10142), determine only the mechanical strength in a specific size range, +0.6-1.18 mm and +4-8 mm, respectively. In other words, both methods give limited information about coke mechanical strength distribution throughout the whole size range, which is especially important when blended cokes are used. Additionally, the results from the drop test are expressed in volume reduction, size reduction, impact force and amount of fines below 148 μm generated during the test. A good correlation between ISO 10142 and the drop test was found. The results show good correlation between the size and volume reduction and the specific grinding energy. The method can also determine the grain strength of a specific size fraction which might be critical to further processing during anode production.
Totally, about 8 tons of four calcined coke types were processed during this work. The majority of this material was ground in the air swept ball mill, for production of the finest fraction, dust. The pilot scale air swept ball mill circuit is identical to the equipment used in industry today. This investigation showed the importance of process optimisation for the stable production of dust with the required specification (particle size and Blaine). Each coke exhibited different behaviour in the grinding circuit, due to different mechanical strengths, which is reported as a specific grinding energy. The mechanically weakest coke produced dust with the lowest specific grinding energy. The particle size variation in the coke used in the mill produced a significant disturbance in the product quality. It was found that the sweeping speed through the mill influences the particle size distribution in terms of the proportion of the finest particles. It also affected the particle roundness of the produced dust, which can influence the flowability of the coke when mixed with pitch. The ratio of ball sizes and the rotational speed of the mill influenced the particle size distribution, Blaine and specific grinding energy.
On-line particle size control was utilised during the dust production. This system allowed continuous control of the product size by regulation of the air classifier rotor speed. Three different dust sizes were produced in the air swept ball mill, which were all below about 200 μm but contained different proportions of material below 63 μm.
Sieved coke fractions together with dust were weighed according to a specified recipe and blended in a sigma blade mixer together with coal tar pitch. A vibration compactor was used for the production of green anodes under optimal conditions. The anodes were produced with three different pitch contents (15, 18 and 20 wt%) and three different dust sizes (45, 63 and 94 %, -63 μm).
Core samples of green anodes were investigated in an improved dilatometer for the determination of thermal expansion and shrinkage during baking. Improvements made to the dilatometer during this work contributed to better reproducibility, increased precision and a healthier working environment. An increase in the heating rate influenced the initial expansion, shrinkage and baking loss of the anode. The heterogeneous pitch distribution throughout the anode due to the forming force, friction between coke particles and between the paste and mould showed the importance of selecting a consistent sampling position. The results showed that the bottom samples have greater expansion due to a thicker pitch bridge layer while the top samples have reduced expansion. The sample packing material contributed to a higher initial expansion due to the reduced free surface area for binder volatiles release, and the shrinkage was reduced because of the radial thrusts from the packing material. Thermal dilation results showed differences between samples with varying pitch, dust size and coke type, which can be used to indicate the final properties of baked anodes. A good correlation was found between the thermal dilation after the pitch expansion and the dust content.
Green anodes were baked under controlled conditions and then characterised. The baked apparent density showed a good correlation with the pitch content and dust size. The results showed that 18 % pitch content and 63 % -63 μm produced the highest baked densities. The variation in coke type exhibited differences in baked density, specific electrical resistance, air permeability, CO2 and air reactivity and modulus of elasticity. | nb_NO |
