Prediction of coke AMS through data mining: a practical approach

The appropriate specification of size analysis of blast furnace coke remains debatable even today. But operators prefer consistent quality and the right size of coke for the smooth operation of the blast furnace. The arithmetic mean size (AMS) of coke and its distribution inside the blast furnace ar...

Full description

Saved in:

Bibliographic Details
Published in:	International journal of coal preparation and utilization Vol. 42; no. 8; pp. 2366 - 2383
Main Authors:	Kumar, Deepak, Saxena, V. K., Tiwari, H. P., Aleti, B. T., Tiwary, V. K.
Format:	Journal Article
Language:	English
Published:	Abingdon Taylor & Francis 03-08-2022 Taylor & Francis Ltd
Subjects:	Algorithms Blast furnace practice Coal blend Coke coke AMS Coke fired furnaces Data mining operating parameters Permeability Prediction models Process parameters recovery stamp charge cokemaking Regression analysis
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	The appropriate specification of size analysis of blast furnace coke remains debatable even today. But operators prefer consistent quality and the right size of coke for the smooth operation of the blast furnace. The arithmetic mean size (AMS) of coke and its distribution inside the blast furnace are essential for maintaining the efficient operation of the blast furnace. The coke AMS is significantly affecting the blast furnace permeability, not a small extent. Therefore, the proper sizing of coke inside the blast furnace can contribute to the increase in production of a blast furnace with the optimal coke rates. However, limited methods that incorporate process parameters and blend properties in the prediction of coke AMS exists. The present work focuses on the assessing of coke AMS using an algorithm on plant process data like parent coal characteristics, carbonization, and operational conditions that can influence coke properties like the mean size of the coke. This work using classification and regression tree (CART) and random forest algorithms provide the basis to futuristic prediction model using process and coal blend parameters.
ISSN:	1939-2699 1939-2702
DOI:	10.1080/19392699.2020.1845662