Prediction of macerals contents of Indian coals from proximate and ultimate analyses using artificial neural networks Manoj Khandelwal a, * , T.N. Singh b a Department of Mining Engineering, College of Technology and Engineering, Maharana Pratap University of Agriculture and Technology, Udaipur 313 001, India b Department of Earth Sciences, Indian Institute of Technology Bombay, Powai, Mumbai 400 076, India article info Article history: Received 21 October 2008 Received in revised form 11 November 2009 Accepted 18 November 2009 Available online 6 December 2009 Keywords: Macerals Ultimate analysis Proximate analysis Multi-variate regression analysis (MVRA) Artificial neural network (ANN) abstract Coal, a prime source of energy needs in-depth study of its various parameters, such as proximate analysis, ultimate analysis, and its biological constituents (macerals). These properties manage the rank and cal- orific value of various coal varieties. Determination of the macerals in coal requires sophisticated micro- scopic instrumentation and expertise, unlike the other two properties mentioned above. In the present paper, an attempt has been made to predict the concentration of macerals of Indian coals using artificial neural network (ANN) by incorporating the proximate and ultimate analysis of coal. To investigate the appropriateness of this approach, the predictions by ANN are also compared with conventional multi- variate regression analysis (MVRA). For the prediction of macerals concentration, data sets have been taken from different coalfields of India for training and testing of the network. Network is trained by 149 datasets with 700 epochs, and tested and validated by 18 datasets. It was found that coefficient of determination between measured and predicted macerals by ANN was quite higher as well as mean absolute percentage error was very marginal as compared to MVRA prediction. Ó 2009 Elsevier Ltd. All rights reserved. 1. Introduction The development of any country is directly related to per capita consumption of energy. Coal is one of the prime sources of energy in India, and accounts for nearly 70% of the total commercial en- ergy produced by the country [18]. Coal is mainly of two types, coking and non-coking. Indian coal belongs to two principal geo- logical periods, the lower Gondwana coals of Permo-carboniferous age, and tertiary coals of Eocene to Miocene age [19]. Majority of the Indian coal is of the non-coking type and is available in many states, like Jharkhand, Chattisgrah, Orissa, Madhya Pradesh, Maharashtra, Andhra Pradesh, etc. Tertiary Lignite deposits are available in Tamil Nadu, Kashmir, Rajasthan, Gujarat, Assam and Jammu-Kashmir. Coal is an extremely complex heterogeneous material that is difficult to characterize. Coal may be defined as an organic rock composed of an assembly of macerals, minerals and inorganic ele- ments held molecularly by the organic matter. The elementary composition of coal is very simple, carbon (C), hydrogen (H) and oxygen (O) being the principal constituents, along with small amounts of nitrogen (N) and sulfur (S). Chemically, coal consists of a mixture of complex organic compounds along with small amounts of inorganic mineral matter and moisture. Physical char- acteristics of coal vary with the rank of the coal. Chemical compo- sition of the coal is defined in terms of its proximate and ultimate (elemental) analysis [9]. Coal can be broadly classified into coking and non-coking variety based on the degree of coalification pro- cess. Coal is composed of a number of different organic entities called macerals. These macerals are particularly of great impor- tance to estimate the degree of maturity or rank, coke quality and carbon matter, etc. [6]. Takahashi and Sasaki [23] proposed the automatic analysis for the identification of macerals by measuring the reflectance at some fixed intervals from the distribution pattern. Pearson [15] devel- oped a method of probability analysis from vitrinite reflectance data to evaluate mixing-technology and to monitor blend consis- tency to improve the plant efficiencies for blended coking coal. Vasconcelos [24] investigated the spatial distribution of macerals group analyses of world coal, and proposed the boundaries be- tween the different categories based on VLI – data for coal all around the world. By combining vitrinite reflectance (VR) and fluo- rescence alteration of multiple macerals (FAMM) analyses, Kalk- reuth et al. [2] have proposed a technique, which provides insights into the organic chemical nature of vitrinites (i.e., perhy- drous vs. orthohydrous vs. subhydrous compositions) in the Perm- ian coal of the Parana Basin, Brazil. Parikh et al. [14] calculated the higher heating value of coal from proximate analysis. Balan and Gumrah [1] assessed the shrinkage-swelling properties in coal seams using rank dependent physical coal properties. Ravi and Reddy (1999) proposed ranking of coking and non- coking coals of India for industrial use, using fuzzy multi-attribute 0016-2361/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2009.11.028 * Corresponding author. Tel.: +91 294 2471 379; fax: +91 294 2471 056. E-mail address: mkhandelwal@mpuat.ac.in (M. Khandelwal). Fuel 89 (2010) 1101–1109 Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel