Use of vitrinite reflectance categories on coke strength prediction of bi-modal and multi-modal coal blends

This project aimed to explore the potential use of vitrinite distribution categories (VDCs) to explain the in-situ coking behaviour and improve the prediction of CSR for coal blends. VDCs, previously developed capture the overall shape and modality of the vitrinite reflectance distribution in coal blends. In this project, eight blends with bi-modal and continuous (overlapping) V-step distributions were prepared by deliberate blending of coals at a wide range of coal ranks and their performance was compared with two target single coals.

This project encompassed two key components; in-situ investigation of the coking behaviour of coal blends and development of regression models to assess the use of VDCs in the prediction of coke quality.

The 4kg coke oven and the permeability/dilatation test facilities at UON was used to evaluate the influence of blend modality and distribution of V-steps on in-situ coking performance and thermoplastic properties and to draw correlations with coke quality. The results showed that the distribution of V-steps and the VDC category of the blends did not influence thermoplasticity and the internal gas pressure (IGP) during coking. Thermoplasticity and IGP depended strongly on blend components and decreased when high inertinite coal was added to the blends.

Coke quality analysis results showed that the blends performed better than the sum of the contribution from the blends, regardless of the modality of the blends. The addition of an appropriate “bridging” coal to bi-modal blends significantly improved coke CSR. This was attributed to the enhanced overlap between the plastic phases of blend components, leading to improved bonding necessary for the formation of a strong coke. Even at a bi-modal V-step distribution, the inclusion of a high fluidity coal improved the CSR, suggesting that a strong coke can be produced from bi-modal blends of Australian coals with an appropriate selection of blend components.

Regression models were developed to evaluate the potential of the VDC parameter in improving the CSR prediction from blends. The limited number of observations - 5 single coals and 8 blends - presented some challenges in identifying the benefits of the VDC parameter which resulted in MBI, MBI2 and VMdaf alone explaining 91% of the variability in the dataset. When calculated plastics (LogMF or Total Dilatation) were used, the addition of the categorical VDC parameter which represents the breadth and modality of the vitrinite reflectance distribution was unnecessary. When sumV6to10 (as the rank parameter) was used in combination with measured Total Dilatation, the addition of the VDC variable notably improved the goodness of model fit.

Alternative parameters were developed to VDC, viz. WtSDVsteps measuring the breadth and modality of the V-step distribution and the “IsBlend” categorical variable as a facile distinction between single coals and blends. The modelling results showed a greater benefit of IsBlend or WtSDVsteps terms to base models using measured plastics for CSR prediction.