International Journal of Scientific Research in Knowledge, 2(3), pp. 143-150, 2014 Available online at http://www.ijsrpub.

com/ijsrk ISSN: 2322-4541; 2014 IJSRPUB http://dx.doi.org/10.12983/ijsrk-2014-p0143-0150

Full Length Research Paper Probability Distribution of Engineering Properties of Soil at Telecomunication Sites in Indonesia
Marelianda Al Dianty*, Ahmad Shukri Yahaya, Fauziah Ahmad
School of Civil Engineering, Univeristy Sains Malaysia, 14300 Nibong Tebal, Penang, Malaysi *Corresponding Author : mareliandaaldianty@gmail.com
Received 15 December 2013; Accepted 20 February 2014

Abstract. Engineering properties of soils, collected from different sites designated for the erection of telecommunication towers, were obtained and analyzed. Probability distribution functions were fitted to this properties. Probability distribution function were fitted to this properties to predict the incidences of their occurance.. Correlation analysis also carried out to check for the relationships if any between the properties. A linear relationship was observed between plastic index and liquid limit, saturated unit weight and angle of internal friction, saturated unit weight and dry unit weight, degree of saturation and water content,saturated unit weight and cohesion, saturated unit weight and dry unit weight. The void ratio,bulk unit weight and dry unit weight were found to follow normal distribution; water content, liquid limit and plasticity index followed the gamma distribution; gravity specification, porosity, saturation, angle of internal friction and cohesion fitted weibull distribution and saturated unit weight fitted log-normal distribution. The study also found that porosity and void ratio have high probability of occurance at their threshold levels. This indicates an unfavorable condition for the soil at this sites Keywords: Index properties, Plasticity properties, Strength properties, Probability distribution, Correlation.

1. INTRODUCTION The choice of foundation type and its geometry depends to a large extent on the engineering properties of the soil base. Engineering properties of soil are fundamental aspect that, used in the analysis of site condition and foundation of the structures of tower telecommunication. A geotechnical investigation must be conducted in order to obtain data about the site. Geotechnical investigation includes field and laboratory testing of both disturbed and undisturbed soil samples. Because of the volume of work and the accompanying cost involved in determining all the engineering properties required before taking decision on foundation design parameters, to minimize these, it may be necessary to analyze probabilistically the distribution of the engineering properties to help in speedy judgment of the soil condition at a particular site telecommunication before construction begin. A distribution is selected based on how well it represents a sample data set from the population, the judgments of experts, the characteristics of the underlying population, or some combination of these factors (Frey and Rhodes, 1999). Popescu et al. (1998) observed that the distribution of soil strength in shallow layers were prevalently positively skewed,

while for deeper soils the corresponding distributions tended to follow more symmetric distributions. Corotis et al. (1975) carried out an investigation on whether a number of properties of three groups of soils could be described by normal or lognormal distribution. Yahaya et al. (2007) choosing North Penang in Malaysia as study cases, investigated the distribution of SPT N-Values on residual soils. The statistical distributions chosen were normal, lognormal, weibull, rayleigh, exponential and gamma. Weibull and gamma distributions were found to fit well at some depths. Bower et al. (2012) carried out probability distribution to get better reflect river flow for wet and dry season. The result found that lognormal distribution better than power law for wet season. While, dry season was better if use lognormal distribution and power law. Suchomel et al. (2011) have used probabilistic distribution to solve geotechnical problem , strip footing. Some soil parameters followed normal, whereas other followed lognormal distribution. This paper attempts to determine the probability distribution of index properties (e, Gs, Sr, n, Wc, t, sat, d), plasticity properties (LL, PL, PI) and strength properties (c and ) at telecommunication network tower sites. The geotechnical investigation was carried out to

143

Al Dianty et al. Probability Distribution of Engineering Properties of Soil at Telecomunication Sites in Indonesia

determine the physico-mechanical properties of the soils. 2. MATERIALS AND METHODS 2.1. Soil sampling process

Properties" and "Engineering Properties". The testing were conducted such as specific gravity test, natural moisture content, bulk density, dry density, saturation index, atterberg limit, sieve and hydrometer, direct shear, triaxial test and consolidation test. 2.2. Analysis Method

The data were obtained from 73 telecommunication sites at East Java province. East Java Province has a total area of 46,428.57 km. It is the largest area in Java Island with a population of more than 37,476,011 (2010 Census), making East Java provine the second most populous in Indonesia. East Java Province lies between longitudes 111 and 114E and latitudes 7 and 8S. There are three step of soil sampling processes at telecommunication sites. (1) site survey, (2) field investigation, (3) laboratory testing. The purpose of site survey is to find closet candidates to nominal point at random sites. The survey must bring a few tools such as GPS compass,gauge and digital camera. GPS used to determine position of latitude and longitude; compass used to determine direction; gauge used to determine dimension of site; Digital camera used to take pictures around rented sites. Before go to the sites, it should be make sure that bring radio network planning nominal point created by software for the candidate sites. Next step is choose the shortest distance from the provided coordinate. It this process, must be consider the environment of site, like main road elevation, access road elevation, land to be rented elevation. It will involve the additional cost and drainage plan. Next step is field investigation process by hand boring test. Boreholes shall be made to obtain information about the subsoil profile, its nature and strength and to collect soil samples for strata identification and for conduction laboratory tests. The soil test result is important to site design team and it will influence the progress of site construction. One point-boring were conducted to obtain disturbed and undisturbed soil samples. Undisturb soil sampling conducted at a depth of -1.00; -3.00 and - 5.00 meters above the local ground by using the "Thin Walled Tube Sampler" diameter 65 mm and length 50 cm. Tests conducted on the collected disturbed samples includes index properties test, while on collected undisturbed samples, strength test was carried out. Lastly, Laboratory Testing were conducted to obtain accurancy the soil data. There are two (2) main job at laboratory testing, which is to determine "Physical

The methodology adopted in analyzing the results of the laboratory test of the soil properties, involved the use of bar-chart plots of the number of occurrences of given parameters within a given values. To determine the linear relationship between two parameters, correlation analysis was employed. Four types of distributions were used; normal, lognormal, weibull and gamma. The distribution models are shown in Tabel 1. The fitness of the distribution models were assessed using Goodness-of-Fit test, such as the KolmogorovSmirnov (K-S) and Chi-Square tests (Ang and Tang, 1975; Baecher and Christian, 2003; Fenton and Griffiths, 2008; Kim et al, 2012). Some of the properties wete then predicted using the best-fit of the selected distributions. Finally, the distributions for selected parameters were the obtained. In order to determine the best distribution, four performance indicators were used as given by Eqs. (1) to (5). The performance indicators comprised of (a) error measures NAE and RMSE and (b) accuracy measures IA, PA and R. The formulas were given below: NAE = (1) RMSE = IA PA = 1= (2) (3) (4)

(5)

Where, and are predicted and observed values respectively; and are predicted and observes mean values respectively; and are standard deviations of predicted and observed values respectively, and N is the sample size.

144

International Journal of Scientific Research in Knowledge, 2(3), pp. 143-150, 2014

Table 1: Model of probability distributions used

No Distribution Model

Normal

Lognormal

Gamma

Weibull

3. RESULTS AND DISCUSSIONS The histogram of engineering properties were presented in figure 1, 2 and 3. The distribution of plasticity parameters shown in Figure 1. Those With outlier on the right side, the histograms showed that the distributions are not symetric. The mean of values of LL and PL are below median, implying a left skew.The plasticity index shows normal curve or symmetric pattern. Histograms of the other index properties presented in figure 2, shows the values of

porosity and Gs around a the central or median value with no bias to the left or right, while the histograms for values degree of saturation and saturated unit weight shows two peaks values. Histograms of the values for porosity, unit weight and unit density spread to the right. Only histograms of values of water content showed normal distribution trend. The histograms for the values shear stregth parameter in figure 3, shows non normal distribution, the shape of curves are skewed to the right side.

Histogram of all.depth.tfm$LL
7

Histogram of all.depth.tfm$PL
12

Histogram of all.depth.tfm$PI

10

Frequency

Frequency

Frequency
10 15 20 all.depth.tfm$PL 25 30

10

20

30

40

50

60

0
0

10

10

20

30

40

50

60

all.depth.tfm$LL

all.depth.tfm$PI

Fig. 1: Histogram of plasticity properties

145

Al Dianty et al. Probability Distribution of Engineering Properties of Soil at Telecomunication Sites in Indonesia

Histogram of all.depth.tfm$e
10 7

Histogram of all.depth.tfm$Gs
14

Histogram of all.depth.tfm$Sr
15 Frequency

Histogram of all.depth.tfm$Wc

10

Frequency

Frequency

Frequency

0.5

1.0 all.depth.tfm$e

1.5

10

20

30

40

30 40 50 60 70 80 90 100 all.depth.tfm$Sr

10

10

20

30

40

50

60

70

all.depth.tfm$Gs

all.depth.tfm$Wc

Histogram of all.depth.tfm$n
10 15 20 25 30 25

Histogram of all.depth.tfm$yt
25

Histogram of all.depth.tfm$ysat

Histogram of all.depth.tfm$yd

Frequency

20

Frequency

Frequency

20

Frequency

15

15

10

10

10

20

30

40

50

60

0.6 0.8 1.0 1.2 1.4 1.6 1.8 all.depth.tfm$yt

0.5

1.0

1.5

2.0

2.5

10

15

0.4 0.6 0.8 1.0 1.2 1.4 1.6 all.depth.tfm$yd

all.depth.tfm$n

all.depth.tfm$ysat

Fig. 2: Histogram of index properties

Histogram of all.depth.tfm$O

Histogram of all.depth.tfm$C

12

10

Frequency

Frequency
0 5 10 15 all.depth.tfm$O 20 25 30

0
0.0

10

20

30

40

0.5

1.0 all.depth.tfm$C

1.5

2.0

Fig. 3: Histogram of shear strength properties

The correlation matrices between soil engineering properties are shown in tabel 2. Strong relationship between plasticity index and liquid limit, measuring 0.8223, was observed, whereas sat and recorded R2 value of 0.8003 R2 value of 0.73499 was recorded between sat and d.; while Sr and Wc , sat and c and sat and d, recorded 0.744, 0.6717 and 0.6064 R2 values respectively. These results are to some extent, in agreement with findings by Rthti (1988) and

uzielli et al (2007). R2 value of 0.5574 for correlation between cohesion and angle of internal friction indicates moderate result. Result of normality test by kolmogorov-smirnov is shown in table 3. Three engineering properties (void ratio, bulk unit weight, dry unit weight) were considered for the test. The test recorded p-value of 0.200 ( > 0.05) for the three engineering properties.

146

International Journal of Scientific Research in Knowledge, 2(3), pp. 143-150, 2014

Tabel 2: Correlation between parameters of engineering properties LL LL PL PI e GS Sr n Wc t sat d c 1 PL 0,4611 1 PI 0,822 0,208 1 e 0 -0,1 -0,1 1 GS 0,0908 -0,033 0,1462 0,4898 1 Sr 0,0444 0,1592 -0,06 0,2502 -0,071 1 n 0,0417 -0,122 -0,028 0,9235 0,6242 0,3012 1 Wc 0,2438 0,1546 0,2049 0,2427 -0,17 0,744 0,1658 1 t 0,2355 0,1859 0,2114 -0,507 -0,264 0,3953 -0,589 0,0748 1 sat -0,03 0,2278 0,2583 -0,692 -0,281 -0,289 -0,708 -0,078 0,414 1 d 0,055 0,201 0,156 -0,834 -0,185 -0,216 -0,829 -0,289 0,606 0,735 1 -0,0914 0,2129 0,046 -0,6625 -0,5293 -0,199 -0,7273 0,0759 0,3661 0,8001 0,535 1 c -0,14 0,0865 -0,055 -0,346 -0,128 0,0405 -0,386 -0,002 0,2117 0,6718 0,3955 0,5574 1

Tabel 3: Kolmogorov-Smirnov (KS) result

Engineering properties Void ratio (e) Porosity (n) Specific Gravity (Gs) Saturation (Sr) Bulk unit weight (t) Saturated unit weight (sat) Kolmogorov-Smirnov Statistic Df Sig. .049 197 .200* .087 197 .001 .282 197 .000 .147 167 .000 .059 193 .200* .239 196 .000 Engineering properties Dry unit weight (d) Liquid Limit (LL) Plastic Limit (PL) Plasticity Index (PI) Angle of shearing resistance () Cohesion (c) Kolmogorov-Smirnov Statistic Df Sig. .055 197 .200* .102 182 .000 .105 179 .000 .092 179 .001 .235 193 .000 .392 190 .000

From table 4, it was observed that Wc, LL and PI followed gamma distribution, Gs, n, Sr, and c followed Weibull distribution, while sat and PI followed log-normal distribution. From the best distribution test, it was observed that the probability of the LL exceeding the threshold (LL > 50) value, is 0.5435. The probability that PI values exceeds the threshold (PI > 45) is 0.8575, while the probability of Sr exceeding the threshold (Sr > 80) is 0.5792, of n exceeding the threshold (n > 45) is 0.8575, of e exceeding the threshold (e > 0.8) is 0.6424 and of d exceeding the threshold (d > 2.2) is 0.1668. In terms of fundamental part in determining engineering behavior of the soil, Observation of the results for plasticity index, saturation, void ratio, porosity and dry density, showed that the distributions of data fits well to the distributions models considered. The values of coefficient of determination

of the correlation between the engineering properties of the soils were found to be between 0.1681 and 0.9999. It should be noted that these observation depends on the soil type, as stated by Lacasse and Nadim (1996). 4. CONCLUSION The study has proposed a set of probablility distributions that can be used to predict soil conditions and their charactheristics. This study found that porosity (n) and void ratio (e) of the soils in the studied sites, has high probability of occurance at their threshold values (probability greater 0.5), while d and Sr has a medium probability of occurance at their threshold values. This indicates that the soil conditions at the investigated sites are unfavorable in their natural state.

147

Al Dianty et al. Probability Distribution of Engineering Properties of Soil at Telecomunication Sites in Indonesia

Table 4: Probability distribution of engineering properties of the soils

Liquid Limit (LL) Distribution Weibull Log-Normal Gamma NAE 0.5775 0.0343 0.0309 RMSE 29.981 24.306 0.732 IA 0.4198 0.9875 0.9905 PA 0.9756 0.9802 0.9826 R 0.9414 0.9504 0.9549 Distribution Weibull LogNormal Gamma NAE 0.0164 0.0337 0.0321 GS RMSE 0.0676 0.1080 0.1025 IA 0.9395 0.8887 0.8961 PA 0.9147 0.8059 0.8148 R 0.8281 0.6430 0.6572

Plasticity Limit (PL) Distribution Weibull Log-Normal Gamma NAE 0.0612 0.0243 0.0262 RMSE 19.143 0.7841 0.8375 IA 0.9569 0.9910 0.9896 PA 0.9472 0.9825 0.9796 R 0.8871 0.9546 0.9489 Distribution Weibull LogNormal Gamma NAE 0.1716 0.1716 0.0544

Water Content RMSE 106.453 106.453 78.121 Porosity PA 0.9675 0.9830 0.9849 R 0.9255 0.9556 0.9592 Distribution Weibull LogNormal Gamma NAE 0.0153 0.0503 0.0427 RMSE 13.456 4.098 33.069 IA 0.9923 0.9431 0.9603 PA 0.9862 0.9111 0.9316 R 0.9628 0.8218 0.8592 IA 0.9367 0.9367 0.9592 PA 0.8813 0.8813 0.9392 R 0.7689 0.7689 0.8731

Plasticity Index (PI) Distribution Weibull Log-Normal Gamma NAE 0.0786 0.0786 0.0549 RMSE 26.356 28.862 17.287 IA 0.9791 0.9777 0.9905

Angle of internal friction () Distribution Weibull Log-Normal Gamma NAE 0.1931 0.1754 0.1908 RMSE 23.908 62.388 25.240 IA 0.9762 0.8793 0.9721 PA 0.9544 0.8215 0.9493 R 0.9013 0.6678 0.8917 Distribution Weibull LogNormal Gamma

Degree of Saturation NAE 0.0505 0.0503 0.0491 RMSE 55.136 64.014 58.380 IA 0.9608 0.9481 0.9559 PA 0.9608 0.9046 0.9172 R 0.9260 0.8086 0.8321

Cohesion (c) Distribution Weibull Log-Normal Gamma NAE 0.5237 0.4066 1..3855 RMSE 0.4149 0.4196 0.4941 IA 0.8425 0.8387 0.8766 PA 0.8843 0.8814 0.8530 R 0.7737 0.7687 0.7031 Distribution Weibull LogNormal Gamma

Saturated Unit weight NAE 0.0873 0.0422 0.0458 RMSE 0.2136 0.1051 0.1107 IA 0.8439 0.9381 0.9331 PA 0.7970 0.8855 0.8749 R 0.6287 0.7761 0.7577

REFERENCES Ang AHS, Tang WH (1975). Probability concepts in engineering planning and design. New York: John Wiley & sons. Anon (1979). Classification of rocks and soils for engineering geological mapping. Part I: Rock and soil material. Buletin International Association Engineering Geology, 19: 364-71. Ayyub BM (1998). Uncertainty modeling and analysis. United States of America: CRC Press . Baecher GB, Christian JT (2003). Reliability and statistical in geotechnical engineering. John Wiley & Sons Ltd., Chichester, pp. 177203. Bell FG (2000). Engineering properties of soils and rocks. fouth edition. Blackwell Science Ltd. Bowers MC, Tung WW, Gao JB (2012). On the distributions of seasonal river flows : Lognornomal or power law?. Water Research Journal, 48: W05536.

Bowles JE (1982). Engineering Properties of Soil. Vol.2. New York: McGraw-Hill. Corotis RB, Azzouz AS, Krizek RJ (1975). Statistical evaluation of soil index properties and constrained modulus. In E. Schultze (ed.) Proceedings of the 2nd International Conference on Applications of Statistics and Probability in Soil and Structural Engineering, Aachen, September 1518, 1975: 273294. Essen: Deutsche Gesellschaft fr Erd- und Grundbau. Fauziah A, Ahmad SY, Farooqi MA (2007). Distribution of SPT N-values on residual soil for North Penang Pluton, Malaysia. World Engineering Congress Fenton GA, Griffiths DV (2008). Risk assessment in geotechnical engineering. John Wiley & Sons, Inc., Hoboken, New Jersey, pp. 161201. Frey and Rhodes (1999). Quantitative analysis of variability and uncertainty in environmental data and model. Water Resources and

148

International Journal of Scientific Research in Knowledge, 2(3), pp. 143-150, 2014

Environmental Engineering Program Department of Civil Engineering. Vol. 1. Hahn GJ, Shapiro SS (1967). Statistical Models in Engineering. John Wiley and Sons: New York. Kim D, Kim KS, Ko S, Choi Y, Lee W (2012). Assesment of geotechnical variability of Songdo silty clay. Engineering Geology, pp.133134. Kuncicky DC (2004). Matlab programming. United States of America: Pearson Prentice Hall. Lacasse S, Nadim F (1996). Uncertainties in characterising soil properties. In C.D. Shackleford, P.P. Nelson and M.J.S. Roth (eds.), Uncertainty in the Geologic Environment: From Theory to Practice, Geotechnical Special Publication ASCE, 58(2): 4975. Rthti L (1988). Probabilistic solutions in geotechnics. NewYork: Elsevier

Sofyan Y, Heri K (2009). SPSS complete: Teknik analysis statistik terlengkap. Jakarta: Salemba Infotek. Suchomel R, And masin D (2001). Probabilistic analyses of a strip footing on horizontally stratified sandy deposit using advanced constitutive model. Computers and Geotechnics, pp 363374. Uzielli M, Lacasse S, Nadim F, Phoon KK (2006). Soil variability analysis for geotechnical practice. In Tan TS, Phoon KK, Hight DW, Leroueil S (2006) eds., Proceedings of the 2nd InternationalWorkshop on Characterisation and Engineering Properties of Natural Soils, Singapore, November 29December 1, 2006. The Netherlands:Taylor & Francis. Yahaya AS, Ahmed A, Gabda D, Su C (2007). Problem and solution in statistics for engineers and scientist. Prentice Hall. Pearson Malaysia Sdn. Bhd.

149

Al Dianty et al. Probability Distribution of Engineering Properties of Soil at Telecomunication Sites in Indonesia

Marelianda Al Dianty is a Ph.D candidate in geotechnical engineering at Universiti Sains Malaysia. She obtained her Master degree in Science and Technology from Universiti Kebangsaan Malaysia. She received her degree in Civil Engineering from Universitas Tarumanegara, Jakarta in Indonesia. She has been been a part time lecturer of statistics for two years. Her professional experience started as a Project Manager with Korean IT Company at Jakarta and she has completed her internship from Petronas during her Master degree. Her last career was as a Project Quality Manager around 3 years at a telecommunication company, Huawei.

Asscociate Professor Ahmad Shukri Yahaya obtained his first degree from University of Nottingham, England in Mathematics. He pursued his Master degree in Industrial Mathematics at University of Aston in Birmingham, England. He has published numerous books and papers at national and international levels. He has also conducted consultancy works and seminars at national and international level.

Professor Dr Fauziah Ahmad is a geotechnical professor at Univeristi Sains Malaysia. She obtained her Bachelor in Civil Engineering at University of Strathclyde, Glasgow, UK in 1985. She continued her doctorate degree without pursuing the Master degree at the same university. She is a member of the geotechnical council and an editorial board member. She has received many awards and medals at various exhibitions in the world especially for slope protection and soil erosion. Currently, Prof Fauziah is deputy dean in IPS (Institute Postgraduate Studies) of Universiti Sains Malaysia.

150