Predictive model of gas consumption and air emissions of a lime kiln in a kraft process using the ABC/MARS-based technique

Abstract

The kraft manufacturing process is the main pulping process in the paper industry. The kraft chemical recovery process is an efficient technology that enables the recycling of the pulping chemicals and the generation of electrical power. However, this process presents substantial issues related to energy consumption and environmental emissions. One of the main fundamental elements of the kraft process is the lime kiln. Lime kiln gas consumption, SO2, and NOx air emissions are key factors from the energy saving point of view (i.e., energy efficiency) and environmental pollution in this industrial process, respectively. Knowledge of the process variables involved in a lime kiln and how these are related to gas consumption and air emissions is essential to predict the kiln’s behavior and minimize its environmental effects. The aim of this research study is to build a regression model for each one of the three prime variables (gas consumption, SO2, and NOx emissions) of a lime kiln employed in the paper manufacturing process using the multivariate adaptive regression splines (MARS) method in combination with the artificial bee colony (ABC) technique. These two statistical learning techniques were combined, thereby obtaining an easy-to-interpret mathematical model with a high goodness-of-fit. A coefficient of determination greater than 0.9 is obtained for all the modeled variables. Moreover, the particular contribution or importance of the input variables in each model is also calculated. The results thus obtained are a useful instrument to gain a better understanding of the dynamics of the lime kiln and the involvement of the process variables in gas consumption and gas emissions.