For just over three decades, Digipet, our automated digital coal petrographic system has been used daily, in the Fingerprinting of coals in North America. In a process we call reflectance profiling, a high-resolution digital monochrome camera, collects 88.8 million reflectance measurements from 339 images of the sample. This is how we massively oversample the reflectance of a coal, and produce  histograms, unique for any coal, that we call Fingerprints or reflectance profiles. These are ideal for monitoring variation in single seam production, compositional changes of blends, and vessel or train cargo consistency. Our Fingerprints are based on leading-edge technology, that has been developed in-house. Our custom software can analyze, filter and display single-seams, and blends, and for several decades, has been used to determine the proportions of coals in blends at SunCoke Energy’s heat-recovery coke batteries.

DIGIPET – ICCP Accreditation results

Digipet participated in the 2000 & 2002 ICCP’s Single-Coal Accreditation exercise, measuring Random Vitrinite Reflectance, and Percentage Vitrinite content of six coals. Digipet’s results, shown along side, were obtained by numerical modelling of the Fingerprints, and are superior to the results of the coal petrographers who participated using manual methods. The ID’s of the coals have been blanked out, but the comments from the then Convener of the exercise are included. For Vitrinite Random reflectance, this table shows that the system is accurate to better than 1/2 of a standard deviation for the set of six coals, equivalent to about 0.015 reflectance. For measurement of the Vitrinite content, the system is accurate  to better than 0.28 of a standard deviation, equivalent to about 1% Vitrinite content, for these samples. Because of this demonstrated level of accuracy over manual methods, we call the technique “Fingerprinting your coals”, but in 2007, Digipet was suspended from further participation in the program!

In the table of Digipet’s results, the closeness of six different coals to established mean values of reflectance, and vitrinite content, are shown. The same statistical technique has been modified to quantify product consistency among samples, including blends, that are allegedly the same, using Fingerprints. By computing mean cell frequencies, a mean Fingerprint for a coal recipe is established, and by applying a modified Pearson Chi Square test, a value can be determined for the amount that another sample of the coal, or blend, departs from the mean recipe Fingerprint. This statistic can be determined for single-seam mine products, for coke battery blends, and for power station feeds, to quantify blend consistency; and a software macro can create Shewhart run charts of the parameter.

Our online Fingerprint Library demonstrates the uniqueness of these histograms.

Single-sourced coals are characterized by one population of vitrinites, and a single vitrinite peak in Fingerprints. As demonstrated in the figure, this pattern remains unchanged even when tens of millions of measurements are made.