Application of Genetic Algorithm for the Assessment of Spontaneous Heating Susceptibility of Coal

Abstract

When coal is exposed to air, oxygen is absorbed and some portions of the coal substance with production of some gases (mainly CO, CO2), water vapour with the evolution of some amount of heat. This oxidation takes place even at ambient temperatures and humidity. It is a slow process and the heat evolved is carried away by air. This process of self heating of coal or other carbonaceous material resulting eventually in its ignition is termed as “spontaneous heating” or “auto oxidation”. Mine fire is major hazard in coal mining industry worldwide. The first step to take proper preventive measure is the assessment of liability of coal seams towards self heating. Various expert systems are being adopted these days to predict the self heating properties of coal. In the present work a new soft computing approach, viz. genetic algorithm has been applied for categorization of coal samples, depending upon their spontaneous heating susceptibility.
EXPERIMENTAL INVESTIGATION
Eleven samples were collected from South Eastern Coalfield Limited, six samples from Mahanadi Coalfield Limited, three samples from Eastern coal field Limited, three samples from BCCL, two samples from Central Coal field Limited and one each from Western coal field Limited and NCL. The samples were collected following channel sampling procedure. Some of the samples were well known for their high and low susceptibility in Indian coalfields. The proximate constituents were determined in laboratory using the standard experimental procedure. The crossing point temperature of these coal samples was determined at a heating rate of 10C/ minute with an air flow rate of 80 cc/min. The onset temperature from DTA thermograms of these coal samples was determined by intersection between two tangents drawn at the inflexion point of the endothermic region and another tangent was drawn at rising portion of curve of stage III. The wet oxidation potential of the coal samples were determined using KMnO4 solution.
Table 1 Proximate constituents, CPT, onset temperatures and wet oxidation potential difference of coal samples
Sample no. Moisture (%) VM (%) Ash (%) CPT (0C) Tc (0C) WOPD (mV)
1 4.974 31.949 23.913 159 171.07 112
2 9.955 30.395 12.6 171 129.49 112
3 5.785 27.35 18.6 163 123.4 114
4 5.894 30.706 26.25 167 143.41 119
5 7.669 30.921 10.8 162 138.36 101
6 9.164 26.586 16.241 185 123.12 113
7 6.488 28.262 10.395 165 162.56 109
8 3.96 26.134 23.3 173 163.63 80
9 2.932 30.883 12.037 163 186.8 99.9
10 1.948 33.052 17.527 156 184.15 108
11 2.345 29.641 15.725 165 189.56 103
12 4.48 25.01 44.02 169 155.93 102
13 2.4 23.27 52.3 174 147.29 104
14 14.5 31.97 12.35 155 158.46 51.2
15 8.97 29.49 22.88 147 153.34 47.5
16 9.59 28.93 7.68 149 124.82 54.1
17 11.13 25.19 38.46 149 171.87 99.3
18 1 17.36 15.73 180 188.98 26.6
19 1.9 33.08 16 155 128.38 63.9
20 0.6 22.32 10.73 160 169.66 41.8
21 8.43 24.43 9.6 152 162.3 87.8
22 1.8 36.13 14.27 150 129.33 106.6
23 10 32.27 18 144 132.94 143.1
24 10.52 29.47 11.91 152.5 136.87 121.8
25 7.67 29.83 18.88 155 145.33 123.8
26 14.39 29.31 12.76 150 128 161
27 9.68 29.8 18.12 138 122.67 206

APPLICATION OF GENETIC ALGORITHM
A genetic algorithm (GA) inspired by Darwin's theory of evolution is a search technique used in computing to find exact or approximate solutions to optimization and search problems. Genetic algorithms are a particular class of evolutionary algorithms (EA) that use techniques inspired by evolutionary biology such as inheritance, mutation, selection, and crossover. Genetic algorithms are. In the present work coal seams are classified by using genetic algorithm to determine their spontaneous heating susceptibility, taking Euclidean distance as fitness function. For this purpose a program was written in C++. The program was run using four variables:
1. Moisture
2. Volatile matter
3. Ash
4. One of the following susceptibility indices at a time.
• Crossing point temperatures
• Onset temperatures
• Wet oxidation potential values
It was observed that the results were consistent, when four clusters were obtained, so four clusters were obtained in all cases.
DISCUSSION AND CONCLUSION
The genetic algorithm takes into account the intrinsic parameters determined by proximate analysis, which is regular affair in the field as it is required to determine the grade of coals. It was observed that the categorization matched with the field results fairly accurately. It is hoped that the output of the work will benefit the practicing mining engineers and researchers to a greater extent in categorizing coal samples, depending upon their spontaneous heating susceptibility and accordingly they can plan the mining activities and adopt advance precautionary measures to deal with fire problems in mines.

REFERENCES
• Ramlu, M.A.; 1991, ” Mine Disasters and Mine Rescue”, Chapter-1, Mine fires, Oxford and IBH publishing Co.Pvt.Ltd, p: 10-20

• Banerjee, S.C.; 1998,” Prevention and Combating Mine fires”, Allied publisher, p: 30-66

• Goldberg D.E.,1999,”Genetic Algorithms”,Chapter-1, A gentle introduction to genetic algorithms, Addision wesley longman publication, 1st edition, page:1-25