Comprehensive Case Study of Coals from Fărcăşeşti Area

 

Elena Maria PICĂ

Technical University of Cluj-Napoca, Romania, empica@yahoo.com

 

 

Abstract

      A statistical study of analysis results was made for lignite from the Fărcăşeşti area (Gorj County, Romania), exemplified for the eight characteristic properties, as moisture content (imbibitions and hygroscopic), volatile, density, sulfur, ash softening content, higher heating value and seam. Previously, the properties dependencies were investigate in pairs of two. In present study the properties was investigated using an automat processing routine for multivariate regression, available at address:

http://vl.academicdirect.org/applied_statistics/linear_regression/multiple/v1.5/

      The program is capable to identify multiple dependencies between given properties. Few significant results were obtained, that make possible to simplify analysis procedure of coals by reducing number of determinations and/or measured properties. All equations are made for predicting heating value Qsi from other measured properties excluding fixed carbon content Cfi. Present article is focused on identifying dependencies between Wii, Whi, Vi, Sti, Qsi, ti, ro and seam (see text). Application of the model among others at prospecting new coalfields and coal conversion, can contribute to the reduction of drilling and analysis costs.

 

      Keywords: Coal analysis, Regression models, Software implementation study.

 

 

 

                        Introduction

 

            In field of statistical data processing it exist a large set of software to compute and fit the regressions, but few are free. Even for free software, another problem it appear, operating system license and portability of the software. Other questions require an answer: We want a server-based application or client based application? We want a server side application or a client side application? As example, a client side application can have disadvantage of execution on client, and dependence of processing speed by power of client machine. If we prefer this variant, a java script or visual basic script is our programming language. Under Apache, we have the possibility to execute programs already compiled in C, Fortran and Java, under Unix machines we can directly execute Perl programs, and, most important, under all operating system platforms we can execute PHP programs if we previously install PHP language and module binaries.

            The advantage of PHP programs consist in his portability under most of operating system platforms and internal compilation feature that do not necessity the compilation “by hand” from the user. Putting PHP programs on a web server into a data folder and executes by them using PHP module. The output of the PHP program is in HTML style and can be viewed by any web client (Microsoft Internet Explorer [1], Mozilla [2], Opera [3], Netscape [4], Konkueror [5]).

 

 

                        Model Implementation

 

            Many statistical procedures for processing data are available [6]. Most of them offer a voluble set of possibilities and variants, but which one to consider them? That is not a easy question and the frequent answer is: that is choice of analyst [7, 8].

            Data mining technology offer in this area of knowledge some answers, but not a complete answer [9]. By other hand, to interpret experiment results, data need to be well processed [10]. Modeling of structure is benefit to property predictions [11, 12]. Nonstandard statistical evaluation procedures then are helpful [13]. The design of statistical processing program is presented in another paper [14].

 

                        Data Mining

 

            To characterize a coal seam the results of proximate analyses as function of depth as related to the initial sample (i), the sample for analysis (a), or anhydrous sample (anh) can be considered [15,16].

A set of measured data from Farcăseşti area was taken into statistical analysis. The probes for analysis were taken from the 64040.15 platform at different seams. The analysis results are given in table 1.

            All measured data from table 1 refer to the initial sample “i” and are expressed in percents (excepting the ash softening temperature, the density and the number of seam). The imbibitions moisture content is Wii, the hygroscopic moisture content is Whi, the volatile content is Vi, the fixed carbon content is Cfi, the content of total sulfur is Sti, the higher heating value is Qsi, the ash softening temperature is ti, the density is ro, and the seam is represented by a number.

 

Table 1. Data values and Qsi predicted values

Wii

(x0,A)

Whi

(x1,B)

Vi

(x2,C)

Cfi

(x3,D)

Sfi

(x4,E)

Qsi

(x5,F)

ti

(x6,G)

ro

(x7,H)

Seam

(x8,I)

34.4

8.1

23.4

14.7

1.03

2225

1130

1.23

16

24.7

9.9

25.3

17.5

0.69

2529

1100

1.19

14

27.1

7.9

20.1

12.1

0.64

1802

1250

1.39

13

25.0

10.5

26.2

19.2

0.90

2700

1250

1.22

12

29.0

8.4

22.1

14.6

0.95

2117

1150

1.11

10

33.0

9.6

26.0

17.9

1.25

2641

1105

1.10

10

28.5

9.1

25.5

18.2

1.32

2590

1120

1.25

10

32.2

8.9

27.2

18.6

0.97

2816

1130

1.12

10

33.3

9.6

25.0

18.6

0.93

2647

1115

1.10

10

25.5

9.4

30.5

20.9

2.10

3043

1100

1.22

10

30.0

8.3

21.2

14.0

1.67

2025

1105

1.13

8

34.7

9.8

26.6

20.9

0.88

2919

1115

1.08

8

26.9

10.7

28.5

19.8

1.63

2983

1100

1.02

8

33.1

9.5

27.1

20.6

1.07

2949

1085

1.20

7

34.4

9.1

26.3

18.7

1.97

2692

1125

1.28

6

25.0

8.8

29.5

15.7

0.82

2650

1110

1.20

5.9

27.5

10.2

26.9

19.0

1.69

2737

1120

1.13

5.1

25.4

10.7

27.5

19.1

2.23

2741

1115

1.20

5

 

            The used functions for Qsi prediction are:

f(x0,x1,x2) = x0∙34.2+x1∙168+x2∙102-2620;                                                                 r = 0.978; s = 0.127

f(x0,x1,x2,x4) = x0∙33.9+x1∙170+x2∙103-x4∙36.2-2630;                                                             r = 0.978; s = 0.124

f(x0,x1,x2,x8) = x0∙34.3+x1∙171+x2∙107+x8∙13.1-2920;                                                            r = 0.977; s = 0.127

f(x0,x1,x2,x4,x8) = x0∙34.3+x1∙171+x2∙107-x4∙2+x8∙13-2920;                                 r = 0.977; s = 0.127

f(x0,x1,x2,x6,x7,x8) = x0∙45+x1∙175+x2∙124+ x6∙2+x7∙282+x8∙5.47-6260;          r = 0.952; s = 0.185

f(x2,x3) = x2∙74.8+x3∙80.5 -770;                                                                                                      r = 0.992; s = 0.098

f(x0,x2,x3) = x0∙13.1+x2∙75.1+x3∙69.4-9750;                                                                               r = 0.994; s = 0.066

f(x2,x3,x4) = x2∙71.8+x3∙82.9-x4∙47.7-6740;                                                                                r = 0.994; s = 0.079

f(x0,x1,x2,x3,x4) = x0∙16.3+x1∙65.6+x2∙76.2+x3∙51.3-x4∙37.2-1330;                      r = 0.996; s = 0.054

                The  Qsi functions are used for prediction. There are the predicted values:

 

Table 2. Qsi predicted values

f(Wii,Whi,

Vi)

f(Wii,Whi,

Vi, Sti)

f(Wii,Whi,

Vi,seam)

f(Wii,Whi,

Vi,Sti,

seam)

f(Wii,Whi,

Vi,ti,

ro,seam)

f(Vi,

Cfi)

f(Wii,

Vi, Cfi)

f(Vi,

Cfi,Sti)

f(Wii,Whi,

Vi,Cfi,Sti)

2304

2286

2358

2355

2347

2164

2263

2176

2261

2469

2471

2511

2508

2377

2531

2473

2560

2522

1684

1679

1681

1679

1847

1708

1737

1742

1759

2671

2669

2694

2691

2911

2735

2663

2756

2714

2037

2023

2007

2004

1969

2058

2087

2078

2091

2773

2753

2767

2763

2748

2616

2663

2617

2691

2485

2462

2473

2469

2469

2603

2587

2603

2559

2751

2741

2748

2745

2795

2762

2791

2775

2769

2682

2672

2670

2667

2658

2597

2640

2619

2667

2942

2898

2957

2951

2957

3194

3112

3148

3020

1963

1921

1902

1897

1788

1943

1990

1929

1975

2926

2920

2897

2894

2938

2902

2938

2927

2945

3005

2977

2987

2983

2932

2956

2903

2936

2937

2872

2860

2831

2828

2842

2915

2934

2928

2915

2768

2721

2709

2704

2830

2703

2759

2671

2718

2722

2722

2666

2663

2694

2700

2669

2707

2678

2778

2746

2713

2709

2729

2772

2735

2752

2749

2851

2802

2789

2784

2806

2825

2760

2778

2778

 

           

 

 

                        Results and Discussion

 

            The fig. 1 presents the regressions between Qsi and a function that cumulate the contributions of imbibitions and hygroscopic moisture and volatile contents by a regular linear equation. The fig. 2 presents same dependency by an origin forced regression equation. The r squared values about 0.95 indicate a very good correlation.   

 

Figure 1. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

Figure 2. Dependencies of Qsi by other measured data from table 1 (see table 2 also)

Figure 3. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

            The fig. 3 presents the regressions between Qsi and a function that cumulate the contributions of imbibitions and hygroscopic moisture and volatile contents and also the content of total sulfur by a regular linear equation. The fig. 4 presents same dependency by an origin forced regression equation. The correlation analysis shows that total sulfur content adding do not increase de accuracy of predicted higher heating value.

 

Figure 4. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

Figure 5. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

            The fig. 5 presents the regressions between Qsi and a function that cumulate the contributions of imbibitions and hygroscopic moisture and volatile contents and also the seam identification number by a regular linear equation. The fig. 6 presents same dependency by an origin forced regression equation. The correlation analysis shows that we consider the seam number do not increase significant the accuracy of predicted higher heating value.

 

Figure 6. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

Figure 7. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

 

Figure 8. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

            The fig. 7 presents the regressions between Qsi and a function that cumulate the contributions of imbibitions and hygroscopic moisture, volatile, total sulfur contents and also the seam identification number by a regular linear equation. The fig. 8 presents same dependency by an origin forced regression equation. The correlation analysis shows prove the observations from figs. 3-6, thus the best accuracy of predicted higher heating value is obtained from contributions of imbibitions and hygroscopic moisture and volatile contents.

 

Figure 9. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

            In the fig. 9 and 10 is presented a tries to include in higher heating value descriptor equations the ash softening temperature and the density of coal. The r value proves that the trying is uselessly.

 

Figure 10. Dependencies of Qsi by other measured data from table 1 (see also table 2)

 

            The dependency plotted in fig. 11 considers a function that cumulates the contributions of fixed carbon and volatile contents. Apparently, the correlation analysis does not show a better correlation than previous results. But, looking at fig. 12, where are considerate also the contribution of hygroscopic moisture content, it can observe that the predictor function is comparable in power of estimation with predictor from fig. 1.

 

Figure 11. Dependencies of Qsi by a Cfi based equation

 

 

Figure 12. Dependencies of Qsi by a Cfi based equation

            Another prediction tries is depicted in fig. 13, where the contributions of volatiles, fixed carbon and total sulfur are considered. The r value proves that the trying was successful, the predictor is better than previous ones.

 

Figure 13. Dependencies of Qsi by a Cfi based equation

 

            Last figure (fig. 14) presents also a very good correlation and is based on both fig. 1 and fig. 13 observations. The dependency includes imbibitions and hygroscopic moisture, volatile, fixed carbon and sulfur contents.

 

Figure 14. Dependencies of Qsi by a Cfi based equation

                        Conclusions

            The study shows the possibility of reducing number of analysis for physical and chemical parameters of coals without reducing the quality of information. The study shows the possibility of prediction of higher heating value from another measured data. From fig. 1 it result that if are available only imbibitions and hygroscopic moisture and volatile contents is proper to predict higher heating value from them. From fig. 13 it result that if are available only fixed carbon, volatile and sulfur content, is proper to predict higher heating value from them. Finally, from fig. 14 it result that if are available imbibitions and hygroscopic moisture, volatile, fixed carbon and sulfur contents, the best choice to predict the higher heating value is to consider all of them.

 

 

                        References

 

1.      http://microsoft.com

2.      http://www.mozilla.org

3.      http://www.opera.com

4.      http://www.netscape.com

5.      http://www.konqueror.org

6.      H. Nascu, L. Jäntschi, T. Hodisan, C. Cimpoiu, G. Câmpan, Some Applications of Statistics in Analytical Chemistry, Reviews in Analytical Chemistry, Freud Publishing House, XVIII(6), 1999, 409-456.

7.      L. Jäntschi, G. Katona, M. Diudea, Modeling Molecular Properties by Cluj Indices, Commun. Math. Comput. Chem. (MATCH), 2000, 41, 151-188, ISSN 0340-6253.

8.      L. Jäntschi, H. Nascu, Numerical Description of Titration (MathCad application), International Conference on Quality Control, Automation and Robotics, May 23-25 2002, Cluj-Napoca, Romania, work published in volume "AQTR Theta 13" (ISBN 973-9357-10-3) at p. 259-262.

9.      M. Diudea, L. Jäntschi, A. Graovac, Cluj Fragmental Indices, Math/Chem/Comp'98 (The Thirteenth Dubrovnik International Course & Conference on the Interfaces among Mathematics, Chemistry and Computer Sciences), June 22-27, 1998.

10.  L. Jäntschi, M. Unguresan, Parallel processing of data. C++ Applications, Oradea University Annals, Mathematics Fascicle, 2001, VIII, 105-112, ISSN 1221-1265.

11.  L. Jäntschi, M. Unguresan, MathLab. Statistical Applications, International Conference on Quality Control, Automation and Robotics, May 23-25 2002, Cluj-Napoca, Romania, work published in volume "AQTR Theta 13" (ISBN 973-9357-10-3) at p. 194-199.

12.  L. Marta, I. G. Deac, I. Fruth, M. Zaharescu, L. Jäntschi, Superconducting Materials: Comparision Between Coprecipitation and Solid State Phase Preparation, International Conference on Science of Materials, August 20-22, 2000, Cluj-Napoca, Romania, work published in volume 2 (ISBN 973-686-058-2) at p. 627-632.

13.  C. Sârbu, L. Jäntschi, Statistic Validation and Evaluation of Analytical Methods by Comparative Studies. I. Validation of Analytical Methods using Regression Analysis (in Romanian), Revista de Chimie, Bucuresti, 49(1), 1998, 19-24, ISSN 0034-7752.

14.  Lorentz JÄNTSCHI, Automat Server Side Processing of Statistical Data, UNITECH'02 International Scientific Conference, November 21-22 2002, Gabrovo, Bulgaria, work published in volume "ISC UNITECH'02 GABROVO Proceedings" (ISBN 954-683-167-0) at p. 185-189.

15.  Blum I., Barca Fr., Chemistry and preparation of solid combustors (in Romanian), Ed. Did. Ped., Bucharest, 1966.

16.  Enache C-tin., Focşa I., Pică Elena Maria, Niac G., The Study of Lignite from the Fărcăşeşti Area, Acta Technica Napocensis, Section: Construction Machines and Materials, 36 (1993), pp.79-86.