Improving Anode Quality Based on Process Parameters in Aluminum Industry

Abstract

Abstract : Carbon anode production is crucial in aluminum production and accounts for a high percentage of production costs. Therefore, anode quality improvement is essential for cost reduction and annual production increase. Raw material properties and process parameters highly influence carbon anode quality. The problem to be investigated in this research study is the unavailability of a well-known mathematical relationship or model to control the quality of anodes produced. The main objectives of the study are to develop a mathematical model of anode quality based on its process parameters, obtain the significant process parameters determining the anode quality, investigate the main causes of variation in anode quality, and define the relationship between anode quality and process parameters. The objectives of the study were achieved by using a multivariate statistical (latent variable modeling) method to build a mathematical model between anode quality properties and process parameters. The model built is used to predict a new dataset to validate prediction ability. Different interpretation methods are used to obtain significant process parameters and investigate the main causes of variation in anode quality. The data collected to achieve these objectives are actual anode quality properties, process parameters, and raw material properties data extracted from Carbon Plant 2 of Aluminium Bahrain (ALBA). Two projection to latent structure (PLS) models have been built using SIMCA software to improve anode physical properties, and chemical and reactivity properties. The research provides good anode quality models with 64.0% and 64.1% variances for the physical properties model and chemical and reactivity model, respectively. The model's prediction ability is 51.1% and 57.2% variances for the first and second models, respectively. Also, the primary source of variation in anode quality was mainly the variability of raw material properties, precisely calcined petroleum coke (CPC) properties. Moreover, the research illustrates that the most influential variable for the first model was green anode density (GAD), and for the second model, was CPC iron content and paste plant process parameters. Furthermore, the most important variables were green anode density and fire soaking time for the first and second models, respectively. The study estimated the net carbon consumption (NCC) reduction using the model built and found it to be 1.65 kg per ton of aluminum. Accordingly, the estimated annual cost reduction is $825,000.