Application of ANN in Six Sigma for CO modelling and energy efficiency of blast furnace: a case study of an Indian pig iron manufacturing organisation. (2015)

• Record Type: Journal Article
• Authors: Sen, Parag
• Is Part Of: International journal of six sigma and competitive advantage. Volume 9:Number 2/3/4(2015)
• Page Start: 109
• Page End: 125
• Publication Date: 2015
• Subjects: carbon monoxide modelling -- six sigma -- ANNs -- artificial neural networks -- process capability -- blast furnaces -- pig iron manufacturing -- energy efficiency -- case study -- India -- process parameters -- carbon emissions -- define measure analyse improve control -- DMAIC -- coke consumption -- air pollution; Six sigma (Quality control standard) -- Periodicals; Quality control -- Periodicals; 658.4013
• Languages: English
• ISSNs: 1479-2494

Abstract

The current paper presents a case study of an Indian pig iron manufacturing organisation to model the CO emission from the blast furnace by applying Six Sigma. The problem for the current organisation is the variation of CO from blast furnace which often exceeds the prescribed limit. Hence, the main objective of the paper is to identify significant process parameters responsible for CO emission by applying define-measure-analyse-improve-control (DMAIC) methodology of Six Sigma because of its structured approach to problem solving. However, this paper also shows a new technique by using artificial neural network (ANN) as a tool in Six Sigma. Results suggest that coke consumption is the most important parameter to influence CO emission from the perspective of cost. Frequent high concentration of CO implies that heat is leaving the furnace in the form of coke consumption, which needs to be improved using best available technologies (BAT). After implementing BAT, the process capability improves from 0.80 to 1.18, resulting in satisfactory reduction of CO concentration and energy consumption. Application of Six Sigma may help the organisation to save around Rs. 5 crore (50 million) annually.