Prediction of CRI and CSR of cokes by two-step correction models for stamp-charged coals - Statistical analysis

Abstract

A new approach for the prediction of Coke Reactivity Index (CRI) and Coke Strength after Reaction with CO2 (CSR) was proposed. For this purpose, 41 coking coals (40 USCB and 1 Appalachian coals) and 36 coal blends were studied. Cokes were carbonized using a pilot-scale coke oven by a stamp charging. Coals and coal blends were characterized by 14 properties and cokes by CRI and CSR indexes. Experimentally determined CRI and CSR indexes of cokes (type II) did not fit well with their calculated values from regression equations based on different coal blend properties (R < ca. 0.370). The similar discrepancy was also observed if experimentally determined coke indexes were compared with weighted averages of C<(R)over tilde>I and C (S) over barR (R < ca. 0.390). With the view of low reliability, herein, the prediction of CRI and CSR values of 36 cokes (type II) was based on 2 newly proposed two-step models. In model 1, the first step consists of preliminary calculation (estimation) of CRI/CSR indexes using modified regression equations originally calculated for the prediction of cokes (type I) carbonized from unblended coking coals; in model 2 calculation of CRI/CSR values (type II) in the first step is carried out using calculated weighted averages of C<(R)over tilde>I and C (S) over barR from type I cokes. In both models, estimated CRI/CSR values were corrected using calculated correction coefficients from regression equations based on coal blend properties. If compared with a traditional one-step procedure, the results document that this novel (two-step) approach provides much better prediction of the coke characteristics.