Chapter 2 Literature Review
In this chapter a literature review is presented to provide an overview of the knowledge about process control strategies in aluminium smelting in general and an understanding of the aluminium smelting process in particular. Sections 2.1.1 to 2.1.4 of this chapter focus on investigations undertaken in the area of alumina feed control, mass and energy balance control, discrete events within a continuous process and liquid electrolyte mass. Sections 2.2.1 to 2.2.7 discuss control strategies and statistical methods applied in the chemical and metallurgical industries. In the last part of this chapter, Sections 2.3.1 and 2.3.2 review new developments in measurements and process control to improve the ability to control the aluminium smelting cells. The constraints in existing control strategies regarding the ability to deal with continuous changing conditions like line current, power modulation and raw materials used are discussed.
Aluminium Smelting Process
Alumina is the single largest raw material input into a smelting cell. Due to increased line current the mass of alumina fed to a cell per day is now almost equal to the amount of liquid mass available for dissolution (Martin et al., 2008). Effective feeding and dissolution is critical to maintaining the target alumina concentration band and electrolysing it efficiently. It is also pivotal in the control of heat balance, as well as avoidance of anode effects and sludge formation (Taylor et al., 1990). The essential requirements during alumina feeding are effective transport into the electrolyte, rapid dispersion in the electrolyte, rapid dissolution as small alumina grains and finally uniform mixing of the dissolved alumina in the electrolyte (Graduate Certificate, 2008). The saturation solubility in industrial electrolytes is typically in the range of 5-6 wt%. Electrolyte additives and impurities generally lower the solubility of alumina in the bath (Section 1.3.3).
The design of feeders with respect to electrolyte flow characteristics, coupled with a stable anode cover and transport of alumina along the surface of electrolyte in the feeding channel are essential for good dissolution (Purdie, 1995). The dispersion of alumina in the electrolyte during feeding due to a combination of electrolyte movement (Section 1.3.5) and gas bubble flow characteristics is another determinant of good dissolution (Moxnes et al., 1998). In addition, transition alumina with a high internal surface and structural hydroxyl content will essentially blow apart when added to the electrolyte, if the grains are initially well dispersed (Metson et al., 2005). The gas release due to this explosion will actually aid dispersion and supports rapid dissolution of alumina. This is if agglomeration of the alumina grains with frozen electrolyte is prevented in the seconds after alumina addition. A high alpha content (no hydroxyl groups in its structure) has the reverse effect because of the absence of gas evolution.
Beside moisture and alpha content, there are several other alumina properties that are important for its dissolution in the electrolyte (Taylor, A., 2004). In terms of particle size distribution, coarser fractions are favoured as there are problems with the dissolution of finer particles linked to freezing and aggregation. Furthermore, the morphology of alumina particles is a function of the Bayer process and affects the attrition index of the particles as they are handled. An increased variability in the quality of alumina used results in a decrease in the controllability of the target alumina concentration band in the smelting process. More knowledge is required to understand the development of structural features during calcination of Bayer Gibbsite to produce smelter grade alumina (Perander et al., 2008).
Together with the ability to dissolve alumina in the electrolyte, the dissolution rate is also important for the reduction operation. What limits the alumina dissolution rate is the amount of sensible heat that is available in the electrolyte before it freezes onto the alumina (Wai-Poi et al., 1994). Without natural mixing as described previously, freezing will retard alumina dissolution significantly, because the total energy required for preheat, dissolution and volatile evolution is substantial (up to 2000 kJ/kg) (Taylor et al., 1990). Studies show also a relatively long duration for the dissolution process to take place, including its distribution under all anodes within a cell (Thonstad et al., 2001). In order to ensure alumina dissolution is continuously maintained, the smelting cell must also have sufficient superheat (typically 8-10°C) to allow the liquid electrolyte mass transfer of the sensible heat. In this respect power modulation results in large deficiencies in energy input with a corresponding decrease of the sensible heat available for dissolution.
The initial endothermic heat supply to maintain the chemical dissolution rate within poorly dispersing alumina additions may not be supported by interfacial convection from the melt and localized freezing will occur – resulting in a smaller overall temperature decrease of the bulk electrolyte (Taylor et al., 1986). The main consequence of incomplete alumina dissolution is sludge formation on the cathode. The composition of sludge is different from liquid electrolyte and consists of 40-70% alumina, 0-5% AlF3, and cryolite saturated with dissolved alumina (Taylor et al., 1990). The measured solidification temperature of sludge is almost invariant at 948 – 954°C. The dissolution of alumina from sludge into the electrolyte is likely to be controlled by the access of fresh electrolyte to the sludge (Equation 2.1 and Figure 2.1) (Grjotheim and Kvande, 1993; Whitfield et al., 2004).
Where, kAl2O3 is the mass transfer coefficient, Abath-slduge is the effective interfacial electrolyte/ sludge area, [Al2O3]sat is the saturation concentration of alumina and [Al2O3] is the actual alumina concentration in the bulk electrolyte. The saturation concentration is a function of the electrolyte temperature and composition.
Most alumina feeding control philosophies are based on the molten electrolyte’s resistance dependency on the alumina concentration (Welch, 1965). The decomposition potential (Erev), anode polarisation (ηAC), bubble resistance and electrolyte resistance (Relectrolyte) vary with alumina concentration in the bulk electrolyte and are sensitive to the anode current density (Equation 1.5). Alumina feed control is normally based on resistance or pseudo-resistance (Rpseudo) in order to filter out current fluctuations (Equation 2.2). The pseudo-resistance is derived from the cell voltage and voltage intercept (Vextrapolated) of cell voltage versus current extrapolated from a small change in line current to zero current (Kvande and Haupin, 2000). Although not technically correct, the voltage intercept is often referred to the Bemf which is made up of the decomposition potential and overvoltages.
Demand feed or an underfeed/overfeed strategy for alumina feed control is used in modern point feed cells (Bearne, 1998). In both cases alumina is fed to the cell at a reduced feed rate leading to a decline in alumina concentration and therefore an increase of the pseudo-resistance. A period of underfeed ends at a fixed alumina concentration and a period of overfeed starts. Demand feed is controlled by a fixed resistance change (∆R), whereas underfeed/overfeed strategy switches at a given slope (dR/dt) and/or the second derivative of the cell pseudo-resistance (d2R/dt2). In the latter case the alumina concentration is subject to increased variability at increased line current due to a rise in the alumina depletion rate.
The main assumption is that changes in alumina concentration due to the applied feeding strategy have a significantly higher effect on the resistance than changes in anode-cathode distance (ACD) or external resistances. This sets a lower limit for feeding rate, otherwise other effects like variations in metal production, anode consumption or even thermal effects dominate changes in the resistance. The upper feeding rate limit is linked to alumina dissolution and determines the maximum concentration band. The shape of the curve of cell pseudo-resistance (R) versus alumina concentration varies with anode-cathode distance (ACD), alumina concentration ([Al2O3]), temperature (T), aluminium fluoride concentration ([AlF3]), anodic current density (acd) and other factors. This is conceptually described as a chain differentiation of partial derivatives in Equation 2.3. Therefore a deterioration of the controllability of the alumina feeding is introduced by the employment of power modulation due to a varying ACD, temperature, bath chemistry and current density over time.
The characteristic variation of pseudo-resistance allows the alumina feed strategy to target an average alumina concentration, although the actual slope, ∂R/∂[Al2O3], is unique for every resistance curve as is shown in Figure 2.2 (Grjotheim and Kvande, 1993). Curves c, a, and b represent the curvature at increasing ACD. It is hypothesised that the presence of an alumina layer at electrolyte-metal interface leads to significant differences between theoretical and measured pseudo-resistance curves (Figure 2.1). There is evidence for this layer is also dependent on the electrolyte temperature (Whitfield et al., 2004).
The mass and duration of overfeeding and underfeeding are specified as control settings. An increase of the underfeed rate will lead to a longer time in underfeed given the same endpoint in alumina concentration. If the overfeed rate is increased, the range of the alumina concentration in the cycle will increase, average concentration will shift towards higher value and a longer time in underfeed will be found. Broadening the regulation bandwidth of the cell voltage will lead to a decrease in the average alumina concentration due to a higher ACD operation (=less beam down movements), but will not change over/underfeed duration.
The lowest alumina concentration is set by proximity to the anode effect concentration. When the bulk concentration of alumina in the electrolyte falls below a critical value, typically below 1.5%, an anode effect will occur (Taylor, Welch et al., 1984). The anode effect is characterized by an abrupt sudden and rapid increase in cell voltage. The wettability of the bottom anode surfaces decreases with decreasing alumina content in the electrolyte resulting in combination of de-wetting and increased coverage of the anode surface by gas bubbles. The anode effect occurs when the local current density exceeds the critical limiting current density resulting in the evolution of greenhouse gases like CF4 and C2F6.
The highest alumina concentration is influenced by the alumina dissolution effectiveness along with the cell pseudo-resistance characteristic. As demonstrated in Figure 2.2 control of the alumina concentration is lost in case of a positive relationship between the alumina concentration and pseudo-resistance, which is represented by the region on the right side of the minimum in the resistance curve. By their design, both a demand feed as well as underfeed/overfeed control strategies initiate extra feeding of alumina under these conditions resulting in an acceleration of sludge formation (Bearne, 1998). Furthermore, the rate of alumina dissolution is reduced according to Equation 2.1, which also leads to more sludge formation
Mass and Energy Balance
The mass and energy balance are defined by Equations 1.8 and 1.9 respectively. Several control parameters can be used to optimize the steady state mass and energy balance like target settings for the cell voltage, temperature and chemistry (combined as superheat), electrolyte and metal level, anode cover and suction rate. However, the smelting process is rarely in a steady state equilibrium due to the necessity for frequent batch-wise operations like alumina feeding, metal tapping, anode changing and anode covering (Section 1.3.6). The frequency of these operational activities and hence the associated impact to the mass and energy balance increases at increased line current. Moreover, power modulation results in large deficiencies in the energy input to a cell as demonstrated in previous work (Stam and Schaafsma, 2007)
Chapter 1 Introduction
1.1 Motivation and Background
1.3 Electrolytic Reduction of Alumina
1.4 Outline of the Thesis
Chapter 2 Literature Review
2.1 Aluminium Smelting Process
2.2 Process Control and Statistical Methods
2.3 Developments in Measurements and Control in Aluminium Smelting Cells
Chapter 3 Theoretical Development
3.1 Process Mechanisms
3.2 Control Theory Development
3.3 Formulation of Control Model
3.4 Architecture Principles
Chapter 4 Experimental Design
4.1 Methodology for Industrial Experiments
4.2 Conditions for Industrial Experiments
Chapter 5 Results and Discussion
5.1 Main Results
Chapter 6 Conclusion and Further Development
6.2 Further Developments
GET THE COMPLETE PROJECT