Detection of Coke Oven Through-Wall Leakage using Data-driven Techniques

28 Sep 2011

M. Saiepour, P. Griffith; Tata Steel RD&T, Swinden Technology Centre, United Kingdom
G. Yi, J. Zhang, J. Morris; Newcastle University, United Kingdom
P. Warren; Tata Steel Long Products, Scunthorpe, United Kingdom

Within a European collaborative research project aimed at reducing emissions, maximising energy
efficiency and improving the performance of coke oven heating, rules-based and multivariate statistical process monitoring techniques have been applied for the on-line detection of through-wall leakage. Coke oven heating process, in steelworks, is a significant contributor to environmental emissions.
Legislation has made it essential to manage and reduce these emissions. Through-wall leakage of
material from oven chamber to heating flues is a major cause of incomplete combustion leading to
excessive emissions of dust, visible as black smoke at the battery stack. Hence, detection and location of leakage at an early stage is beneficial. This paper describes research undertaken at a Tata Steel coke oven plant (UK) to develop a diagnostic system to detect and locate through-wall leakage.
Historical coke oven process and waste gas data were analysed and regions selected when no leakage was suspected. These were used as normal condition data to build a principal component analysis (PCA) model with 4 principal components that represented about 80% of the data variations. The warning (95%) and the control (99%) limits for the two monitoring statistics, the square prediction error (SPE) and Hotelling’s T2, were then determined from the PCA model and the normal process operational data. In this study, an abnormal operating condition is detected when the SPE or T2 statistics of a consecutive number of data samples exceed the control limits. In order to determine if this is a through-wall leakage problem, data related to oven charging and combustion are examined. Results have demonstrated that the proposed method is effective in detecting through-wall leakage. A real-time diagnostic and advisory system is currently being developed.

The research leading to these results has received funding from the European Union Research
Programme of the Research Fund for Coal and Steel (Contract number – RFCR-CT-2008-00007).

Paper presented at Clean Technologies in the Steel Industry, Budapest, Hungary, September 26-28, 2011