Technical Market Support » Thermal Coal
During pulverised fuel combustion, coal particles are rapidly pyrolysed to yield char particles which are then burnt. The conversion of coal particles to char is the rate determining step of the coal combustion process. For a single coal, the individual coal particles can vary significantly in organic composition, pore structure, mineral matter and petrography. Hence coal particles are able to produce char particles having significant variations in char physical structure. These structural variations are seen in char particles from different maceral concentrates and they affect combustion conditions.
A representation of char morphotypes is typically captured under an optical microscope, with the resulting char images being a two-dimensional representation of the groups to which they belong. Optical analysis of coal and char particles is done on polished blocks to provide images of coal macerals and the internal structures of chars. The char classification systems have already been established, but a new approach to automatic char classification is required to adequately describe particle morphotypes that clearly relate to different maceral types and their distribution in the original coal. There is an incentive to automate the char classification to significantly increase the number of individual particles which are analysed, avoid the need of a skilled petrographer for the analysis, and to provide size detail on the individual particles.
The goal of this study was to link unique Coal Grain Analysis (CGA) information for parent coal particles to information acquired by using image analysis methods for daughter char particles to quantify the transformation of different coal grain types (ie pure components, composite particles) to specific char types for a better understanding of maceral influence on char morphology and combustion performance.
For the purpose of automation of char classification, CSIRO's semi-automated imaging CGA technique was adapted for char analysis. CGA is based on coal petrography methods, which collects a large number of high resolution, calibrated, contiguous images of coal and char samples in incident white light at 500x total magnification with a digital pixel resolution of 0.13μm x 0.13μm. A feature of this system is that colour images, collected with an air objective, are mosaicked together to enable detailed (micron level) information to be obtained on particle sizes from 1 micron up to 4mm, thus providing new insights into understanding the performance of thermal coals. The mosaics of these images are than analysed using CSIRO's ParIS (Particle Imaging System) software to obtain detailed information on individual parent coal particles and daughter char particles.
Each individual coal particle was classified as a single component, maceral dominant or maceral rich composite grain. A standard char classification method developed by the Combustion Working Group of the International Committee for Coal and Organic Petrology led by Ed Lester is commonly used to manually classify char particles into 9 char types. CGA char analysis recognised 4 char categories, including minerals. The decisive parameter for sorting into individual categories was the porosity of each individual particle (ratio of pores to the whole char body area). The process firstly sorted out mineral particles, which had more than 50% of mineral content into the category MINERAL. Secondly, particles with the porosity greater than 65% were categorised as THIN WALLED, particles with the porosity less than 40% were categorised as SOLIDS, all the others fell into the THICK WALLED category.
It was found that there was a strong correlation between the vitrinite content of particles and the formation of the Thin Walled char particles: the more vitrinite, the more thin walled particles. Furthermore, CGA coal analysis provides the abundance of pure vitrite in parent coal. It was observed that vitrite content has an even stronger influence on forming the Thin Walled char particles than does total vitrinite content of the coal. An additional correlation was determined between the inertinite content of the parent coal and solid char particles, the more inertinite in the parent coal, the greater the proportion of solid char particles.
As the CGA method provides size detail (particle width, length and area) it is possible to determine size attributes for both the char particle classes and the parent grain types which formed these chars. This information was used to determine whether specific types of char particles exhibited swelling, erosion or remained unchanged during the char formation process from the parent coal. A comparison was made of the mean maximum diameter of coal particles with the mean maximum diameter of chars' Thin Walled particles, the outcome clearly showed that all Australian coals showed swelling after transformation to char.
To be able to compare the CGA results against a simplified standard char classification, manual char analysis was also done. Outcomes of this project show that in terms of the individual char morphologies, results obtained from the CGA char classification system varied from the manual classification by a minimum of 0% and maximum of 39%. In contrast to the CGA char classification, the traditional manual method of char classification requires an expert petrographer to use multiple structural parameters, such as material connectivity and wall thickness within each particle, to classify each particle. For CGA to be fully comparable and better than manual char classification, it is necessary to take into consideration more of the typical char morphology features (char wall thickness, voidage and fused/unfused material).
This project has demonstrated that there is great potential for semi-automation of char image analysis. CGA proved to be a suitable platform, although there is still space for improvement of the system. A trial of a modified CGA system was tested during this project with promising results: compared to manual char analysis, there was only 0.5% variance from an equivalent manual char analysis result.