Professional Documents
Culture Documents
Variogram Modeling of Lime Saturation
Variogram Modeling of Lime Saturation
net/publication/343599084
CITATIONS READS
2 216
6 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Irfan Marwanza on 04 September 2020.
Index Terms: limestone, Lime Saturation Factor, geostatistical method, Chi-Square, variogram, Nugget Effect, Sill, Range
—————————— ◆ ——————————
2. Median
___________________________________________ Median is the center value of a group of data sorted
sequentially.
• The author's name is a lecturer in the Mining Engineering
Department, Universitas Trisakti, Jakarta PH: +62-815-8619-1888.
E-mail: irfanm@trisakti.ac.id ……………............….…. (3)
Information:
Lo = lower limit of the median class
1679
IJSTR©2020
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 03, MARCH 2020 ISSN 2277-8616
c = width of interval class descriptive statistics used to describe the lowest observation
n = data amount size, first quartile (Q1), median (Q2), third quartile (Q3), and
∑ft = the frequency of all classes lower than the highest observation (Fig.2.1). It is also used to describe data
median class distribution and outliers, which are from extreme values and
f_med= the frequency of the median class far from others thereby, affecting the amount of variance.
3. Mode
The mode is the value of most frequent data
……………..........…. (4)
Information:
Lo = lower limit of mode class
c = width of interval class
b1 = the difference between the mode class frequency
and one class before the mode
b2 = the difference between the mode class frequency
and one class after the mode
Fig. 2.1 Boxplot [16]
4. Range
The range is the highest data reduced by the lowest after it is 2.3 Geostatistics
sorted by value. Geostatistics aims to present quantitative descriptive of natural
Range = highest data – lowest data...…..............…... (5) variables that are distributed to quantify the spatial uncertainty
data. [17],[18]
5. Standard Deviation
Standard Deviation shows the dispersion of data set relative to 2.3.1 Variogram
its mean. The variogram is a geostatistical analytical method that
quantifies the degree of similarity and variability between
separate data at certain distances. Data close to the estimated
……...………..............................…. point are similar to those farther apart [19]
(6)
Information: ……........................(9)
σx = standard deviation Information:
x̅ = mean (h) = Variogram value at interval h
x_i = data number - i Xi = Data number - i
n = data amount N = Pairs of data
7. Coefficient of Variation
The coefficient of variation is a measure of data distribution,
otherwise known as relative standard deviation.
……..................................................(8)
Information:
CoV = coefficient of variation
σx = standard deviation
x̅ = mean
Fig. 2.2 Variogram Parameter [21]
8. Histogram
A histogram is the representation of data using bar charts. Its
The range is the variogram distance of a stable period which
use shows the continuous frequency of data.
affects Nuggets at minimal distances, while sill is the
saturation point where the samples obtained do not correlate
9. Boxplot
[22], [23].
Boxplot, known as chart box-and-whisker, is a type of
1680
IJSTR©2020
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 03, MARCH 2020 ISSN 2277-8616
A good variogram model occurs when the original value of Range 504.977
data and its estimate approaches a straight line of 45 degrees.
Besides, it can also be seen through the calculation of the The results obtained from the statistical analysis shows that
error between the actual value and the estimated value. large data variations were detected, with varying values of
calculated average (Mean) and median, indicating that the data
3 METHODOLOGY is not normally distributed. The characteristics of the boxplot
The research is located in Kedungbunder Village, Gempol LSF were studied which showed deviations and abnormalities in
District, Cirebon Regency, West Java Province with 108°24'15 the statistical analysis above and below the whisker. The LSF
'' - 108 24'30"E and 6°43'35" - 6°43'55 "S and a total of 346 Boxplot graph is seen below (Fig 4.1).
hectares wide. The cement mill is located 20 km west of
Cirebon City, while the quarry is situated towards the south of
the cement mill at a distance of 2.5 km, precisely in the
Kromong Mountains. Furthermore, the study was conducted
with a quantitative descriptive approach by collecting actual
information using field conditions, which are represented with
images and graphs. The presentation of information is based
on quantitative equations of geostatistical methods. [22].
Primary data are used with a total of 35 respondents using the
following research procedures:
1. Sampling
The sampling location was determined based on the mining
plan. The samples used were obtained from drilling, and some
were collected using the cone quartering method.
2. Samples Analysis
The physical properties of the cement are more important than Fig. 4.1. Boxplot LSF
the chemical properties[25]. Samples were analyzed using the
laboratory process, which consists of various stages such as 4.2 Fitting Test Chi-Square Method
crushing, grinding, making compact rings, and analysis of the Fitting Test using the Chi-Square method is used to test data in
chemical composition using X-Ray Fluorescence Spectrometry the frequency distribution table [26]. Before analyzing the fitting
(XRF) machines. The LSF value is determined for each test, presenting descriptive statistics using histograms need to
sample from the chemical composition. be conducted. The results of the histogram plotting show an
abnormal distribution, as shown in Fig. 4.2.
3. Statistics Test
LSF values of 35 samples were carried out via statistical tests
using Matlab to produce data distribution and boxplot analysis
to detect outliers.
4. Variogram Modeling
Using the LSF data pair, a variogram model was generated,
and an optimal model produced using Surfer software.
1681
IJSTR©2020
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 03, MARCH 2020 ISSN 2277-8616
TABLE 4.3
ERROR VARIOGRAM GAUSSIAN MODEL
Sill
Range
Fig. 4.3. Data Distribution 30000 35000 40000 45000 50000
150 18.33% 18.3% 18.41% 18.53% 18.63%
Visually, the log type is normal, and the exponential distribution
175 19.86% 19% 18.39% 17.8% 17.31%
has a shape that follows the histogram value of the LSF data.
However, the Chi-Square calculation needs to be conducted to 200 23.8% 23.47% 23.1% 22.67% 22.26%
ascertain the type of LSF distribution. [26] The calculation
225 23% 23.27% 23.46% 23.58% 23.7%
begins with determining the midpoint value of each histogram
bar and comparing the middle values of all data distributed. Chi- 250 18.92% 19.58% 20.14% 16.63% 21.03%
Square Method Fitting Test Results are seen in table 4.2 below.
TABLE 4.4
TABLE 4.2 ERROR VARIOGRAM EXPONENTIAL MODEL
FITTING TEST CHI-SQUARE METHOD RESULTS
Distribution Type Sill
Value Range
Normal Lognormal Exponential Gamma 30000 35000 40000 45000 50000
Mean 129.76 131.26 129.76 129.76 150 13.66% 13.65% 13.64% 13.63% 13.63%
Variance 16367.5 21713.1 16837.1 11982.6 175 13% 12.99% 12.99% 12.97% 12.99%
Std.Dev 127.94 147.35 129.76 109.47 200 12.75% 12.74% 12.74% 12.73% 12.74%
CoV 0.98 1.12 1 0.84 225 12.65% 12.64% 12.64% 12.62% 12.64%
Skewness 1.24 1.11 1.22 1.45 250 12.58% 12.56% 12.56% 12.55% 12.56%
1682
IJSTR©2020
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 03, MARCH 2020 ISSN 2277-8616
TABLE 4.5
ERROR VARIOGRAM SPHERICAL MODEL
Sill
Range
30000 35000 40000 45000 50000
150 51.95% 51.92% 51.91% 51.9% 51.88%
175 47.01% 46.98% 46.96% 46.94% 46.98%
200 43.71% 43.68% 43.68% 43.68% 43.68%
225 40.01% 39.98% 39.98% 39.98% 39.98% Fig. 4.5. Variogram cross-validation
250 37.09% 37.07% 37.07% 37.07% 37.07%
Visible data points evenly spread on the centerline with the
distribution of points above and below the line.
• The error calculation is based on the absolute value of
the difference in the estimated variogram and LSF
recorded in percents. From the three models analyzed, 5 CONCLUSION
the smallest error was 12.55% in the Exponential In conclusion, LSF contains outliers obtained from the boxplot
model using the variogram parameters and is stated as and an exponential distribution of the Chi-Square analysis.
follows: Sill: 45000 The most optimal variogram model with the smallest error size
• Range: 250 meter is the Exponential variogram model with parameters Sill
• Nugget Effect: 100 45000, Range 250, Nugget Effect 100, and a total error of
12.55%.
Fig. 4.4 shows the experimental graph of the LSF variogram 6 ACKNOWLEDGMENT
that has been modeled with the Exponential Model, therefore, This research was supported by the Mining Engineering
the sill, range, and nugget effect is obtained. Department, Faculty of Earth and Energy Technology,
Universitas Trisakti, Jakarta, Indonesia. Thanks are expressed
to PT Indocement Tunggal Prakarsa Tbk. Palimanan unit,
Cirebon, and all those that contributed to the success of this
research for their support and assistance. Thanks also to the
STEEM (Science, Technology, Engineering, Economics,
Education, and Mathematics) Conference.
7 REFERENCES
[1]. Aldidamouni Hamdi Ahmed, D. Ahmed Saleh Mohammed
Taha (2011) book: Chemistry and Technology cement.
[2]. Walter, H. D (1985). Cement Data Book. Weisbaden Berun:
Bauverlag GmBH.
[3]. Rikoto I.I, Nuhu, S (2019). Effect of Free Lime and Lime
Fig. 4.4. Variogram Model Exponential Saturation Factor on Grindability of Cement Clinker.
International Journal of Engineering Research and Reviews
Variogram assessment requires the expertise of geostatistics ISSN 2348-697X (Online) Vol. 7, Issue 1, pp: (61-66)
experts, and visually, the model chosen is based on the number [4]. Anonymous, (2003), "Company Overview", PT. Indocement
of pairs that intersect with the model. However, due to Tunggal Initiative Tbk, Cirebon
limitations in the study and outlier data from boxplot analysis, [5]. N. H. Mtarfi, Z. Rais, M. Taleb (2017). Effect of clinker free
the variogram was chosen based on the smallest error. lime and cement fineness on the cement physicochemical
Subsequently, cross-validation was conducted to satisfy the properties. Journal of Materials and Environmental
rules of BLUE (Best Linear Unbiased Estimator) in the analysis. Sciences. ISSN: 2028-2508. JMES, 2017 Volume 8, Issue
[30]. Cross plot showed the alignment of the distribution of 7, Page 2541-2548.
points and the values of the observed and estimated data. [6]. Boughanmi, S., Iabidi, I., Tiss, H., and Megriche, A (2016)
Cross-validation of the model is shown in Fig. 4.5 for the LSF The Effect of Marl and Clay Compositions on the Portland
because the RMSE is closer to zero, fitted Exponential models Cement Quality, Journal of the Tunisian Chemical Society,
yielded better results, and was adopted in this study. 2016, vol. 18, pp. 43–51
[7]. Lefond. J. Stanley (1983). Industrial Minerals and Rocks
(Nonmetallics other than Fuels). Society of Mining
Engineers. New York.
[8]. Rakesh Rana, Richa Singhal (2015). Chi‑square Test and
its Application in Hypothesis Testing. Journal of the Practice
of Cardiovascular Sciences . Volume 1. Issue 1
[9]. N. H. Mtarfi, Z. Rais, M. Taleb (2017). Effect of clinker free
lime and cement fineness on the cement physicochemical
1683
IJSTR©2020
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 9, ISSUE 03, MARCH 2020 ISSN 2277-8616
properties. Journal of JMES, 2017 Volume 8, Issue 7, Page And Kufa. Journal of Applicable Chemistry, 3 (2): 812-822.
2541-2548 ISSN: 2278-1862
[10].Hendrick G van Oss (2012) Cement, Mineral Commodity [26].Zhanyu Ma, Arne Leijon, and W. Bastiaan Kleijn (2013).
Summaries, U.S Geological Survey, pp 38-39. Vector Quantization of LSF Parameters With a Mixture of
[11].A. C. Ozdemir, A. Dag, and T. Ibrikci (2018). A Comparative Dirichlet Distributions. IEEE Transactions On Audio,
Assessment on Cement Raw Material Quarry Quality Speech, And Language Processing, Vol. 21, NO. 9,
Distribution via 3-D Identification. Journal of Mining September 2013. DOI: 10.1109/TASL.2013.2238732
Science, Volume 54, Issue 4, pp 609–616. [27].Chiles, J.-P., & Delfiner, P. (2012). Geostatistics Modeling
[12].Soumaya Ibrahimi, Néjib Ben Jamaa, Mohamed Bagané, Spatial Uncertainty: Second Edition. New Jersey: John
Mekki Ben Ammar, André Lecomte, Cécile Diliberto (2015). Wiley & Sons, Inc.
The Effect Of Raw Material’s Fineness And Lime Saturation [28].Deutsch, J.L., Szymanski, J. and Deutsch, C.V (2014).
Factor On Clinker’s Grindability And Energy Efficiency In Checks and measures of performance for kriging estimates,
The Gabes Cement Industry. Journal of Multidisciplinary Journal of the Southern African Institute of Mining and
Engineering Science and Technology (JMEST)ISSN: 3159- Metallurgy, 114:3, 223-230.
0040. Vol.2, Issue 11. [29].Deutsch, J.L., Palmer, K., Deutsch, C.V., Szymanski, J. and
[13].V.C. Johansen, L.M. Hills, F.M. Miller and R.W. Stevenson Etsell, T.H (2015) Spatial Modeling of Geometallurgical
(2002). The Importance of Cement Raw Mix Homogeneity, Properties: Techniques and a Case Study, Natural
International Cement, Chicago, USA, online on America's Resources Research, 25:2, 161-181.
Cement (2003) [30].Bora Oz and Clayton V. Deutsch (2000). Cross-validation
[14].Rao D.S, Vijayakumar T.V., Prabhakar S, Bhaskar Raju G for Selection of Variogram Model and Kriging Type:
(2011). Geochemical Assessment Of A Siliceous Limestone Application to IP Data from West Virginia.
Sample For Cement Making. Chin.J.Geochem Chinese
Journal of Geochemistry. Volume 30, Issue 1, pp 33–
39.DOI: 10.1007/s11631-011-0484-8. Springer Link
[15].Supardi, U. (2013). Statistics Application in Research.
Jakarta: Change Publication.
[16].Junaidi. (2004). Data Description Through Boxplot. Faculty
of Economics and Business, University of Jambi.
[17].Marwanza, I., Nas. C, Azizi. A. M, Dahani. W, Subandrio
(2019). Geostatistic Application In Determining The Spread
Of Claystone Rock Mass Characteristics In Coal Mining.
IJSTR Volume 8 - Issue 10, October 2019 Edition - ISSN
2277-8616
[18].Isaaks, E.H., and Srivastaya, R.M (1989). An
Introduction To Applied Geostatistics, Oxford
University Press: NewYork. 561p
[19].Marwanza, I., Nas, C., Badaruddin, S.P., Azizi, M.A (2018).
Copper Resource Estimation in PT X Batu Hijau, Regency
of West Sumbawa, West Nusa Tenggara Province using
geostatistical method. IOP Conference Series: Earth and
Environmental Science
[20].Amine Soufi, Lahcen Bahi and Latifa Ouadif (2018).
Contribution Of Geostatistical Analysis For The Assessment
Of Rmr And Geomechanical Parameters. VOL. 13, NO. 24,
ISSN 1819-6608. ARPN Journal of Engineering and Applied
Sciences ©2006-2018 Asian Research Publishing Network
(ARPN). All rights reserved.
[21].Bohling, G. (2005). Introduction To Geostatistics And
Variogram Analysis, Earth, 1-20.
[22].Zhanglin Lia, Xialin Zhanga, Keith C. Clarkec, Gang Liua,
Rui Zhuc. (2018). An automatic variogram modeling
method with high-reliability fitness and Estimates. Journal of
Computers & Geosciences, 120. DOI: 10.1016/j.cageo. ©
2018 Elsevier Ltd. All rights reserved.
[23].Han, C., Wang, J., Zheng, M., Wang, E., Xia, J., Li, G.,
Choe, S. (2016). New Variogram Modeling Method Using
MGGP And SVR. Earth Sci. India.
https://doi.org/10.1007/s12145-016-0251-9.
[24].Rossi, M.E. and Deutsch, C.V. (2013). Mineral Resource
Estimation, Springer, Dordrecht. 332 P.
[25].Suaad Abdel-Mahdi Abd- Nour and Fatima Allawi Abdul
sejad (2014). Measure And Compare The Chemical And
Physical Properties For Cement In The Laboratory Of Najaf
1684
IJSTR©2020
www.ijstr.org
View publication stats