Thread Content
Abstract Through industrial experimental studies, the relationship curve between the apparent resistance of aluminum electrolyzers and alumina concentration was determined. On this basis, by analyzing the experimental data, a fuzzy control model for alumina concentration was established, enabling fed-dosing in aluminum electrolysis cells as needed and significantly reducing the anode effect coefficient. Models for slot condition self-diagnosis, pole pitch adjustment, set voltage self-correction, and anode effect prediction were developed simultaneously, and they were successfully applied in a new type of aluminum electrolysis cell control box based on PLC control. Introduction: Many institutions at home and abroad have been working on the development and research of automatic control technologies for aluminum electrolysis. Starting from early methods such as constant-regulation of cell voltage or apparent cell resistance, as well as timed feeding controls, to more recent adaptive control techniques used abroad, and now to fuzzy control of alumina concentration in China [1], the software capabilities have gradually improved. On the hardware side, a two-level distributed control system based on custom-made industrial computers or microcontrollers has been continuously used. Such control systems have low reliability, often experiencing issues such as crashes and malfunctioning. Previously, PLCs were also used for control, but it was mostly centralized control. With the development of large-scale prebaked cells, their poor real-time performance, as well as slow data processing and transmission speeds, made it impossible to make timely control decisions for aluminum electrolysis cells. To address the above issues, we adopted the MicroPLC from GE90 series in the United States as the main controller to implement one-to-one distributed control, with one PLC assigned to each cell control box. By combining expertise from those familiar with aluminum electrolysis, an expert fuzzy control model was used to develop a two-level distributed intelligent fuzzy system for aluminum electrolysis cells, thereby achieving a perfect integration of the system’s software and hardware. The control system adopts a two-level distributed control scheme of DDC (direct digital control) and SCC (process monitoring), and its hardware composition is shown in Figure 1. Each cell control box controls one electrolyzer, and each of these boxes is equipped with an independent PLC as the main controller. This PLC is responsible for collecting data on series current, cell voltage, and various digital input signals, performing data analysis and fuzzy reasoning to determine the alumina concentration in the cell. By adjusting the feeding interval and pole distance, it is possible to achieve material balance and heat balance within the electrolyzer. The process monitoring level communicates with the PLC in the slot control box via an RS422 communication bus to obtain relevant information. By processing this information, it enables diagnosis of the slot conditions, thereby allowing for the modification and optimization of parameters at the DDC level. It also provides a reliable basis for manual operations and maintenance decisions. Multiple regional SCCs can be connected to form a local area network, expanding the existing secondary level by an additional MIS (Management Information System) level and enabling connectivity with the entire plant’s computer network, thereby facilitating decision-making by management. 2 Software Introduction The software structure block diagram of the control system is shown in Figure 2. Due to the large number of software models, it is not possible to introduce them all here; instead, we will focus on discussing the following few models. 2.1 Experimental testing of the R-C curve To develop a correct and reliable model for controlling alumina concentration, we conducted extensive experimental research. After carrying out multiple tests on the 140kA electrolyzers at the **large-scale aluminum electrolysis test facility in Qinyang, we determined the relationship between the cell resistance R and the alumina concentration C, resulting in the dynamic R-C curve. Its characteristic curve is shown in Figure 3. The area near point A is a low-concentration zone, where the probability of anodic effects is higher ； The area near point B is a high-concentration zone where precipitation is likely to occur. As the curve moves from point B to point A, the resistance in the groove increases gradually, the slope becomes steeper, and the concentration of alumina in the groove decreases. Although the current efficiency is high, anode effects occur easily, leading to increased power consumption ； If the curve moves in the reverse direction, from point A to point B, as the concentration of Al2O3 increases, the cell resistance gradually decreases, and the slope also diminishes; the curve becomes flatter. Near the lowest point, there is a region of little sensitivity to changes. In this area, precipitation is likely to occur within the cell, which leads to an increase in the cell temperature and a rise in the voltage drop at the bottom of the cell. This, in turn, increases power consumption and reduces current efficiency. The curve measured under these conditions can drift due to the effects of the manufacturing process, especially the movement of the anode, which causes the curve to shift. Therefore, during the curve testing process, we try to avoid abnormal operations such as lifting the busbar, performing edge processing, or replacing the anode, in order to ensure the accuracy of the curve. 2.2 Development of the fuzzy control model for alumina concentration: Since electrolyzers are nonlinear, time-varying systems with large time lags, and contain high-temperature, highly corrosive electrolyte melts, it has been impossible to measure many parameters within electrolyzers online over long periods of time, both domestically and internationally. Therefore, we employed three different algorithms to analyze the cell resistance, the rate of change of cell resistance, and the R-C curve. By combining expertise from electrolysis specialists with the experience of field operators, we developed a comprehensive fuzzy control model for alumina concentration. This model takes the material discharge rate as the output variable, with the fuzzy linguistic values of this rate defined as five levels: \"large shortage, shortage, normal, excess, and large excess\". Fuzzy control rules are generated based on expert experience and measured process knowledge. By controlling the feeding rate of alumina, the alumina concentration in the tank is kept to vary between low, high, and low levels, thereby preventing the occurrence of sudden anode effects and precipitation and maintaining material balance within the tank. The main functions of this model are reflected in the following aspects: (1) It determines the alumina concentration in the tank, as well as its trend and rate of change, based on R, ΔR, and (indirect) factors. (2) Fuzzy inference is used to determine the slopes of the switching points for various feeding states. (3) Prediction of the anode effect and its handling during the transition from high concentration to low concentration. (4) It reduced operation at high alumina concentrations, preventing precipitation. 2.3 Effect Prediction and Treatment Model A set of effect prediction and treatment models was developed through experimental studies on 140kA aluminum electrolysis cells at the **large-scale aluminum smelting industrial test base in Qinyang. The model provides accurate and reliable predictions; it can forecast 20–40 minutes in advance, with a 100% success rate ； At the same time, this model can nip the “sudden anode effect” in the bud. The effect prediction curve is shown in Figure 4. This curve was measured and calculated during the wait time for the anode effect. 2.4 Alumina concentration tracking control model: After the electrolyzer has been in operation for a long period of time (24 hours or more), it is necessary to determine the alumina concentration within the cell again. To do this, feeding to the cell must be stopped, the pole distance must be kept constant, and the cell resistance as well as its rate of change must be analyzed in order to determine the current alumina concentration in the electrolyzer. This information is then used to decide on the subsequent feeding rate and the mode of feeding to be used. This is very useful for consuming the excess alumina in the electrolyzer. 2.5 Polarity Control Model The structure of this control system is shown in Figure 5. The purpose of adjusting the pole pitch is to maintain the thermal balance of the electrolyzer. This pole pitch adjustment model provides pole pitch adjustments for 3 main scenarios: In the first scenario, the slot resistance is not within the normal set range, and the amount of pole pitch adjustment is determined based on the degree of deviation. The second type is a specialized pole pitch adjustment model for special operating conditions, including aluminum pouring, anode replacement, busbar lifting, and other abnormal operations, which will not be discussed in detail here. The third method is pole pitch adjustment under special slot conditions, which involves making corresponding adjustments to the anode pole pitch depending on the specific slot conditions. 3 System Features (1) The alumina concentration control model and anode effect prediction model, developed using fuzzy control algorithms based on expert experience and control principles, have been successfully applied to 140kA prebaked cells at a **large-scale aluminum electrolysis testing facility, achieving excellent technical and economic results. (2) The main controller PLC in the slot control box is a general-purpose device that combines the advantages of computers and industrial process control systems, and can be widely used in various process control applications; it is capable of operating safely and reliably in any harsh environment. This approach of using a single PLC to control a single electrolyzer, combined with an upper-level monitoring system to form a two-level distributed control system, is the first of its kind in China’s aluminum electrolysis industry. (3) The control software can be modified and transmitted online via the human-machine interface on SCC-grade computers, and the operation status of the tank control box software can be monitored through the SCC-grade computer interface, which is unmatched by other tank control boxes. (4) The SCC-grade computer human-machine interface allows for online modification of the operating parameters of the slot control box (nearly 90), enabling it to adapt to various slot types. (5) The layout inside the slot control box is reasonable and compact; it is a small wall-mounted cabinet structure. 4 Control Effect Through tests conducted on 140kA aluminum electrolyzers at the **large-scale aluminum smelting test facility in Qinyang, adverse factors such as long service life of the electrolyzers, poor operating conditions, and unstable electrolyzer performance were overcome, yielding satisfactory results. The bottom of the test tank was relatively clean; the electrolyzer shifted from unstable operation to stable operation, with the effect coefficient dropping to around 0.25 on average. The anode effect could be prevented for up to 10 days, and the alumina concentration was maintained between 1.0% and 3.5%. Just in terms of energy consumption due to these effects, each unit achieved a savings of 160 kWh/t compared to the previous control system, resulting in good economic and social benefits. It therefore has high potential for widespread application.