Jurnal JCB aku.pdf

7/25/2019 Jurnal JCB aku.pdf

1/11

Ground penetrating radar wave attenuation models for estimation

of moisture and chloride content in concrete slab

S.F. Senin a,b, R. Hamid b,

a Faculty of Civil Engineering, Universiti Teknologi MARA Pulau Pinang, 13500 Permatang Pauh, Penang, Malaysiab Department of Civil and Structural Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia

h i g h l i g h t s

GPR wave attenuation is measured on a concrete slab saturated with water and chloride.

The moisture content (MC) shows a strong linear correlation with the wave attenuation.

The chloride content (CC) attenuates radar waves at a higher rate than the MC.

Two models to estimate MC and CC in concrete slab are proposed.

The extent of these corrosion agents are useful in condition rating decisions.

a r t i c l e i n f o

Article history:

Received 20 March 2015

Received in revised form30 November 2015

Accepted 22 December 2015

Keywords:Ground penetrating radar

Moisture

Chloride

Amplitude attenuation

a b s t r a c t

The detection of moisture and chloride ingress through concrete cover is important for estimating the

extent of corrosion in reinforced concrete (RC) components. The concentrations of the substances are

monitored by observing the ground penetrating radar (GPR) amplitude attenuation in water and chloride

saturated concrete slab samples. The amplitude attenuation significantly correlates with the amount of

both substances. Two multiple nonlinear regression models were developed. The proposed models

demonstrate a strongcorrelation with the radar amplitude attenuation data as both substances are varied

in the concrete cover. The developed models can be employed to estimate the moisture and free chloride

content in concrete cover for improved quantification of corrosion level.

2015 Elsevier Ltd. All rights reserved.

1. Introduction

The deterioration of reinforced concrete structures due to corro-

sion has received substantial attention from civil engineers due to

its common occurrence and high cost of repair. This impairment is

accelerated if concrete possesses high permeability due to impro-

per mixing and compaction during the concrete casting process

[1]. A high watercement ratio in concrete facilitates the ingressof water and chloride ions through the capillary pores in its cover

prior to the destruction of the rebar in subsequent stages.

The existence of moisture and chloride ions in a concrete cover

is a form of conditioning that can initiate the corrosion mechanism.

Chloride may also exist in a concrete surface via cracks in concrete

or the presence of contaminated sand or aggregates. Once the

chloride ion reaches the rebar level and exceeds the chloride

concentration threshold value, the protective thin passive layer

surrounding the rebar will gradually be destroyed, which will

cause the development of anodic and cathodic sites on the rebar.

Due to generations of electrical potential differences at anodic

and cathodic sites on the rebar, iron is oxidized at the anode and

dissolved in the concrete pore solution as ferrous ions (Fe2+),

whereas electrons that move towards cathode and hydroxyl ions

(OH) will be released. With sufficient oxygen at the anodic sites,

a rust component Fe(OH)2 is formed when the hydroxyl ionsreacted with the ferrous ion. Further oxidization of Fe(OH)2 into

different corrosion products will produce a high volume of

corrosion products that comprise six times the original volume

[2]. With this volume increase, the interface between concrete

and steel will experience high tensile stresses that can cause

damage initiation (cracking, delamination and concrete spalling).

The rebar corrosion-damage process in concrete structures is a

time-consuming process that exhibits symptoms once the electro-

chemical reactions at the anode and cathode sites produce

corrosion-damage to concrete due to expanding rust on the

concrete-rebar interface.

http://dx.doi.org/10.1016/j.conbuildmat.2015.12.156

0950-0618/ 2015 Elsevier Ltd. All rights reserved.

Corresponding author.

E-mail address: [email protected](R. Hamid).

Construction and Building Materials 106 (2016) 659669

Contents lists available at ScienceDirect

Construction and Building Materials

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o n b u i l d m a t
http://dx.doi.org/10.1016/j.conbuildmat.2015.12.156mailto:[email protected]://dx.doi.org/10.1016/j.conbuildmat.2015.12.156http://www.sciencedirect.com/science/journal/09500618http://www.elsevier.com/locate/conbuildmathttp://www.elsevier.com/locate/conbuildmathttp://www.sciencedirect.com/science/journal/09500618http://dx.doi.org/10.1016/j.conbuildmat.2015.12.156mailto:[email protected]://dx.doi.org/10.1016/j.conbuildmat.2015.12.156http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://crossmark.crossref.org/dialog/?doi=10.1016/j.conbuildmat.2015.12.156&domain=pdfhttp://-/?-


2/11

The presence of moisture can caused several forms of attack on

concrete which leads to chemical or physical deteoriation. The

forms of attack on concrete can be caused by internal or external

source of moisture. One of the common form of attack in concrete

structures is due to the sulphate attack (SA). This traditional attack

is initiated by the chemical interaction of sulphate-rich soil or

water with high sulphate content within the cement paste. The

destructive nature of SA causes cement paste to disintegrate,

which eventually leads to a weaker cement matrix [3]. Several

attempts were done on how to assess concrete resistance to SA,

however, the experiments required a long term observation that

will take years as it involves the sulphate ingression by diffusion

mechanism in concrete, thus. An accelerated test has been

proposed to access the sulphate resistance on concrete made with

ordinary Portland cement and slag cement [4].

Water expansion in concrete pores during winter season is

another phenomena that can cause damage to the concrete in long

term duration. When water freezes, it will expands approximately

15% from its original volume [5]. As the water in moist concrete

freezes, it produces pressure inside the capillaries and concrete

pores. Once the tensile strength of concrete is exceeded, the pores

will dilate and rupture. Further successive freezethaw cycle will

caused significant expansion, cracking and crumbling of the

concrete. In view of this situation, few methods has been applied

on concrete to improve its freezethaw durability. The use of

air-entraining agent and the hydraulically pressed concrete were

the possible ways and its performance of this method can be found

elsewhere from other references[6,7].

Alkali-silika reaction (ASR), has been identified as one of the

deteorioration phenomena in concrete facilitated by the presence

of moisture. In this phenomena, some reactive aggregates react

with the akali hydroxide in concrete to produce gels; causing

expansion and cracking over a period of years. The typical indicator

of this reaction is the formation of random map cracking on areas

with frequent supply of moisture, i.e. near joints and close to the

waterlines. Moisture allows migration of alkali ions to the reaction

sites in concrete and the resulting gel absorb moisture, leading toexpansion on concrete. The expansive reaction can occur in

concrete having a relative humidity more than 80% [8].

Therefore, an assessment of the chloride and moisture content

in concrete cover during the early stage of corrosion is essential

for planning the maintenance of deteriorated concrete structure

with a relatively lower cost compared with maintenance after

severe damage over a certain period of time.

To assess the state of moisture and chloride ion content in a

concrete cover, a number of techniques have been developed to

measure both contaminants in concrete. The most prominent and

direct method for measuring moisture content in concrete is the

gravimetric method[9]. This method can be used to measure the

moisture content by observing the change in mass; however, it is

not a practical technique for quantifying moisture in large concretestructures. The neutron hydroprobe is a non-invasive method that

relates the scattering of neutrons with the moisture content of con-

crete [10]. Unfortunately, this technique is only applicable in a lab-

oratory and may pose health risks to an operator. For quantifying

the chloride content in concrete, concrete dust samples are

obtained at different depths and titration method using silver

nitrate is used to determine the chloride content. However, this

method is time consuming to be applied to a large area of concrete

surface. Ground penetrating radar (GPR) has great potential in

enabling operators to measure the moisture and chloride content

using electromagnetic waves. This nondestructive method facili-

tates rapid data collection in large concrete areas, and only

requires a single-sided surface of concrete for inspection work.

Previous GPR researches have provided GPR wave attenuationmodels based on one parameter: the effect of the water content

[1113] or the effect of the chloride content [14]. The effect of

water content on GPR amplitude with 1.5 GHz antenna was stud-

ied using concrete sampels of 7 cm thickness [11]. They provide

a numerical simulation by modelling the amplitude variation with

degree of saturation and the results are correctly simulated with

the experimental values, except in the range of saturation laying

between 0% and 20%. The quantification of volumetric water con-

tent in 12 cm thickness concrete samples was conducted by Klysz

& Balayssac, 2007[12]. They analysed the direct waves of 1.5 GHz

antenna and successfully model the relationship between the nor-

malized amplitude with the saturation degree. The relationship

between the volumetric water content in fresh concrete mix and

its relative dielectric constant was obtained using GPR wave [13].

Microwave non-destructive testing has been used to evaluate the

separate effect of moisture and chloride contamination in concrete

and its relationship with relative dielectric constant and loss factor

were determined [14]. However, this current study will focussed

on studying and modelling the effect of both moisture and free

chloride content in concrete cover to the ground penetrating radar

amplitude; which has been identified to be the objective and the

novel aspect in this work. The first scope of the work is the collec-

tion of radar signals on concrete cover that was saturated with

varying moisture and free chloride content. The following scope

of work is to develop the attenuation model of radar amplitude

due to moisture and free chloride content in the concrete cover.

1.1. Ground penetrating radar

GPR is a method that is common for infrastructure inspection

work involving concrete structures. GPR can be used for subsurface

condition assessment and to monitor concrete infrastructures,

such as bridge decks [1517]and building components [18]. It is

a nondestructive method that has been employed in geological

studies to map the embedded geological features [19], however,

with the advancement of new high-frequency GPR antenna and

better data processing software, the application of the method

has evolved from geological field applications to civil engineering

applications, such as the estimation of pavement thickness [20]

and the detection of defects in pavement [21].

In principle, this nondestructive method is dependent on mea-

suring the reflection of the reflected the electromagnetic wave that

propagates in certain lossy dielectric mediums, such as concrete,

after it impinges any embedded layer that possesses different elec-

trical properties of the propagated medium, i.e., the dielectric con-

stant and the conductivity. The degree of this wave reflection is

quantified by the reflection coefficient R, which is computed as

the ratio of the incident wave amplitude at an interface or object

to the reflected wave amplitude on the targeted layer or object,

as shown in Eq. (1)[22]:

R AAp

1

whereA is the reflection wave amplitude from the top surface of the

concrete structure andApis the wave reflection amplitude from the

metal plate placed at the concrete soffit.

The reflection of the wave in concrete is influenced by two main

electrical properties of the medium, i.e., the dielectric constant and

the conductivity, as concrete is a lossy material. The dielectric con-

stant describes a materials ability to store and release electromag-

netic wave energy by electrical charge displacement and a

polarization process when an electrical field from the radar

antenna is applied [23]. The original energy of waves is subse-

quently converted to heat energy during the displacement and

polarization process, which causes amplitude attenuation to theoriginal wave amplitude. The moisture content in concrete has

660 S.F. Senin, R. Hamid / Construction and Building Materials 106 (2016) 659669


3/11

been described in certain studies[24,25]as a substance that influ-

ences the dielectric constant in concrete; thus, it can characterize

the moisture content based on the radar signal changes due to

the dielectric variation in a concrete cover. The ability of concrete

to pass free electric charges under the influence of an applied elec-

tric or electrical conductivity of concrete also produced changes in

electromagnetic wave reflection. Dissolved cations such as chlorideions (Cl) in concrete pores will rapidly accelerate and randomly

collide with each other, which produces energy loss when exposed

to an electrical field from the radar. This situation causes

amplitude attenuation as the energy is converted to heat energy.

The degree of moisture and chloride content in concrete can be

estimated by analysing the direct and reflected waves in the time

domain. These two radar waves can be easily recognized and

identified by locating the wave peaks, as depicted in Fig. 1.

Direct waves (DW), which are usually termed as direct-

coupling, comprise the first radar wave energy to reach the

antenna receiver after propagating in an air medium and is

reflected by the top of the concrete. A direct wave is not a surface

wave, such as a Rayleigh wave, but is part of the radiating energy

that directly propagates along the airconcrete interface [26].

Depending on the radar frequency, this wave is only reflected by

the top surface of concrete to a few centimetres of its depth[27].

The remaining radar wave, which has less energy, propagates

through the concrete and is reflected by the concreteair or con-

cretemetal interface with a contrast value in the electrical proper-

ties. According to reference [27], the depth of the object must be

greater than the antenna offset between the transmitter to the

receiver to prevent the inevitable fuzzy zone and an incorrect

propagation distance introduced by a very shallow object depth.

In this study, the proposed thickness of sample, D, is selected as

0.07 m (7 cm) to simulate the thin concrete cover dimension and

set to be greater than antenna radar offset distance (0.058 m) as

proposed by Klysz et al. [27]. Previous succesfull works had used

sample thickness ranging from 7 cm[28], 1 cm to 8 cm[29] and

8 cm [30] on relating the moisture content and chloride contentin concrete by GPR wave and they did not observed and reported

the overlapping of direct and reflected waves. Thus, based on these

research works, the minimum sample depth was selected as thin as

7 cm to ensure that correct and better signals can be captured

before analyses are conducted.

Both direct waves and reflected waves are displayed in a GPR

system as a two-dimensional plot of the image or waveform in real

time, which can be analysed to examine its amplitude and arrival

time. The time delay between the reflection and the direct wave

can be used to compute the dielectric constant, e of the concrete

and the wave velocity.

2. Materials and methods

2.1. Materials

Thirty-one unreinforced concrete slab samples with the dimensions of

0.25 m 0.25 m 0.07 m were prepared with a very low water-to-cement ratio

of 0.7 to simulate a high permeable concrete (i.e., deteriorated concrete). A mini-

mum depth of 70 mm is selected for the sample as this depth exceeds the antenna

offset of 58 mm. The mix design compositions, which are shown in Table 1, were

prepared based on reference [31]. The fine aggregates is obtained fromnatural river

sand and the coarse aggregate is crushed limestone (density of 2.7 g/cm3) with sin-

gle size of 10 mm. Clean tap water was used for the samples mixture.

2.2. Sample preparation

In this study, seven samples were prepared to evaluate the effect of moisture

content on the radar amplitude, and twenty-four samples were prepared to inves-tigate the effect of chloride content on radar amplitudes. All samples were initially

cured in the curing tank for 28 days. At the end of the curing period, the saturated

sample weights were recorded as wsat.

The effect of moisture content on the radar signals was evaluated using seven

samples. All samples, with the exception of the saturated sample, were dried in

an oven at 80 C and weighted until a constant weight at the targeted degree of

moisture content,wdrywas achieved. The percentage of water content in the sam-

plesx1 is determined using Eq.(2):

x1 wsatwdrywsat

100 2

In order to distribute homogenous moisture content across the slab cross sec-

tion, all samples were sealed by thin aluminium foils and dried in an oven at

80 C for two months as shown inFig. 2. The weight of each sample was checked

on a weekly basis to ensure no weight loss prior to the radar measurement.

The effect of free chloride content in concrete on radar signals was evaluated

using twenty-four samples. As with the previous samples, this sample was ovendried until a constant sample weight was achieved to evaporate the pore water

inside the concrete pores. The samples were directly immersed in the distilled

water, which contained different chloride concentrations of 10 g/L, 20 g/L, 30 g/L,

40 g/L and 50 g/L for one month duration to ensure that the entire concrete cover

is filled with chloride ions. All samples were partially dried at 80 C to achieve cer-

tain targeted degree of moisture content. In order to obtain homogenous distribu-

tion of moisture in each samples, all sampels were sealed by thin aluminium foils

and oven-dried for two months. The weight of each sample was regularly checked

to ensure no weight loss during this process.

2.3. Ground penetrating radar (GPR) signal measurement

The GPR signal acquisition was performed on the top surface of all samples

using SIR-3000 system with ground-couple 1.6 GHz, monostatic antenna. All sam-

ples were tested at a marked point in the middle of the samples with a metal plate

inserted beneath each sample. One hundred measurements of radar signals were

performed, and the average of each signal amplitude was used to represent themeasured amplitude of the radar to minimize the signal-to-noise ratio as the num-

Fig. 1. Identification of direct and reflected waves from the GPR scan on concrete.

Table 1

Sample mix composition of the slab.

Materials Mix composition

weight (kgm3)

Ordinary Portland Cement 402

Coarse aggregates (10 mm size, crushed limestone) 939

Fine aggregates (river sand, density 2.7 g/cm3) 1058

Water 286

S.F. Senin, R. Hamid / Construction and Building Materials 106 (2016) 659669 661
http://-/?-http://-/?-


4/11

ber of measurements was increased. The radar reflection on the metal plate can be

deemed as the real radar reflection from the top surface of the rebar embedded in

theconcrete structures. Each signal was sampled at a sampling frequency of 85 GHz

to prevent an aliasing effect on the digitized signals [32]. In this study, the ampli-

tude is digitized using a 16 bit format as integer (32, 768 to +32, 768). The noises

in the signals were excluded from the original signal by a signal filtering process

that was performed prior to the analysis.

The plot of radar signals produced by both conditions was performed using

MATLAB. The arrival of the direct waves (DW) can be easily identified as the first

peak positive amplitude of both situations. As a metal plate has infinite values of

dielectric, it is capable of reversing the signal amplitude polarity; the arrival of

reflected waves (RW) is identified by locating the phase reversal of the signal pro-

duced by the sample with the metal plate [33]. The peak-to-peak amplitude of DW,

whichare denoted as DWPPandRWPP, correspond to the peak-to-peak amplitude of

reflected waves, were determined fromwaveforms and normalizedto peak-to-peak

amplitude of directwaves or reflected wave signalpropagates in the control sample

App to obtain the radar wave attenuation per meter, a as shown in Eq. (3):

a 20D

log10

AcApp

3

where A c is either the peak-to-peak amplitude of the direct wave (DWpp) or the

peak-to-peak amplitude of the reflected wave (RWpp) for certain moisture and

immersed chloride content and D is the sample thickness. As concrete is a low-

loss material[34], the velocity of the radar wave, vcan be computed using Eq. (4)

and the sample dielectric constant, e can be computed using Eq. (5):

v 2dt3 t1 4

v cffiffiffie

p 5

Fig. 2. Samples are sealed by thin aluminium foils.

DWarrivalme , t1

RWarrivalme , t3

Mulple GPR reflecon

Fig. 3. Waveforms of GPR amplitudes with various moisture content.

Table 2

Peak-to-peak amplitude and amplitude attenuation.

Moisture

content (%)

Ad (mV) Ar (mV) aDW(dB/m) aRW(dB/m) Remark

10.1 3958.50 6816.30 99.543 130.424 Saturated

9.4 4304.70 7028.00 89.140 126.629

7.1 4501.10 7968.10 83.604 111.051

5 4661.60 9804.10 79.256 85.321

3.2 5411.30 12394.50 60.751 56.230 1.7 5922.10 14792.70 49.559 34.281

0 8829.40 19499.90 0.0000 0.0000 Dry

Fig. 4. Waveforms of GPR amplitudes with 3.2% of moisture content.

Fig. 5. DW amplitude attenuation at various moisture content.



5/11

wheret3 and t1 are the arrival time of the peak positive amplitude of the reflected

wave and the arrival time of the peak positive amplitude of the direct wave, respec-

tively;cis the speed of light in free space (0.3 m/ns) ande is the sample dielectric

value; d is the actualpropagation distance(0.07577m) andeis the sample dielectric

value.

3. Results and discussion

Six samples and one control sample were employed in this

study to observe the effect of moisture content on the peak-to-

peak radar amplitude; the remaining twenty-four samples were

analysed to determine the effect of free chloride on the peak-to-

peak radar amplitude.

3.1. Effect of moisture content on amplitude attenuation

3.1.1. Control sample

A dry control slab sample with no chloride contamination was

prepared in this study to serve as the reference sample. The control

sample in this study is a sample that exhibits GPR wave attenua-

tion as electromagnetic waves travel through the sample without

moisture and without free chloride content. The measured peak-

to-peak amplitudes of direct and reflected radar waves by the

GPR antenna is employed as the reference radar amplitude App to

be normalized to the measured peak-to-peak amplitude of the

remaining samples with varying moisture content and chloride

content.

3.1.2. Radar amplitude and peak-to-peak amplitude attenuation

The amplitudes of the radar signals of the samples with varying

moisture content were depicted as a series of waveforms inFig. 3.

These waveform amplitudes comprised the average amplitude of

100 radar scans, which were conducted at the middle point of

the top slab surface. The control sample amplitude measurement

is represented by a black line waveform in Fig. 3. The peak-to-

peak amplitudes of the direct waves Ad and the reflected waves

Ar of each waveform and the amplitude attenuations are listed in

Table 2.

To compute the values ofAdandArinTable 2, a waveform with

a moisture content of 3.2% is employed; the calculation is shown in

Fig. 4. As shown in Fig. 3, the peak-to-peak amplitudes of the direct

waves and reflected waves are attenuated by 55.2% and 65%,respectively, from the amplitude of its control sample as the mois-

ture content has increased from 0% (dry state) to 10.1% (saturated

state). This amplitude attenuation can be explained by the electro-

magnetic wave energy loss of the radiated waved from the trans-

mitter antenna. The moisture or free water that exists in

concrete pores absorb the energy in the waves and convert it to

heat energy.

The attenuation of the amplitude by moisture is analysed by

computing the radar wave attenuation using Eq. (3), as shown in

Figs. 5 and 6. Both plots demonstrate an acceptable linear relation-

ship (R2 = 0.82 and 0.97) with respect to the water content of the

samples; these findings are consistent with the findings obtained

by reference [35]. The direct wave amplitude attenuations were

Fig. 6. RW amplitude attenuation at various moisture content.

Fig. 7. Dielectric constant of the current work and others researchers.

Table 3

Wave velocity for different values of moisture content.

Moisture (%) Wave velocity (m/ns)

10.1 0.095

9.4 0.095

7.1 0.101

5 0.108

3.2 0.118

1.7 0.128

0 0.150

0

1

2

3

4

5

6

7

8

9

10

1 1 3 5 7 9 11 13

Amplitudeaenuaon(dB),y

Water content (%), x

DW

Sbartai

Linear (DW)

Linear (Sbartai)

y= 0.4306x + 4.5633

y= 0.556x + 1.7196

Fig. 8. Comparison of current DW amplitude attenuation with[24].

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


6/11

decreased by 7.94 dB/m, whereas the reflected wave amplitude

attenuations were decreased by 12.70 dB/m. These reductions in

the amplitude attenuation with the increased moisture content

causes an increase in the dielectric constant, which affects the

GPR energy absorption attenuation, as shown in Fig. 7. The radar

wave propagates through the samples with a different wave veloc-

ity, vas the moisture content varies from a dry state to a saturated

state; the results are tabulated inTable 3. The wave velocity rangeis consistent with the findings in reference[36]. When the electro-

magnetic wave travel is transmitted by the radars antenna

through the sample without any water content, it travels at a

higher speed with a certain degree of moisture content. This result

is attributed to the increasing dielectric value of the samples as the

degree of polarization increases as the moisture content is

increased. The experimental dielectric values ranged between the

values obtained by Zhangs equation[25]and the values obtained

by Chens equation[13], as depicted inFig. 7, which are consistent

with findings by other researchers. The moisture variation in the

samples influences the wave velocity, the moisture content and

the amplitude attenuation.

To compare this study with the work of other researchers, the

existing amplitude attenuation per meter results are multiplied

y = 0.8887x + 0.8057

y = 0.6467x + 0.1871

1

1

3

5

7

9

11

1 1 3 5 7 9 11Amp

litudeaenuaon(dB),y

Moisture content (%) , x

RW

Sbartai

Linear (RW)

Fig. 9. Comparison of current RW with reference [18].

Table 4

Peak-to-peak amplitude and amplitude attenuation.

Chloride content (g/L) Moisture content (%) Ad (mV) Ar (mV) aDW (dB/m) aRW(dB/m) Remark

0 0 8829.40 19499.90 0.00 0.00 Dry

10 10.1 3958.50 6816.30 99.54 130.42 Saturated

9.4 4304.70 7028.00 89.14 126.63

7.1 4501.10 7968.10 83.60 111.05

5 4661.60 9804.10 79.26 85.32

3.2 5411.30 12394.50 60.75 56.23

1.7 5922.10 14792.70 49.56 34.28

20 9.3 4961.50 5897.00 71.52 148.40 Saturated

4.9 5256.70 6345.00 64.35 139.30

4.7 5295.10 6581.90 63.45 134.77

4.4 5335.30 6652.30 62.51 133.45

3.2 6778.10 8838.10 32.81 98.19

0 8477.70 17078.10 5.04 98.19 Dry

30 12.1 3833.40 1941.60 101.92 286.25 Saturated7.6 5135.60 3362.60 67.24 218.10

6.4 5508.80 4088.90 58.54 193.84

5.5 5114.45 5443.10 67.75 158.34

3.9 6516.30 7780.90 37.69 158.34

0 8985.20 20339.30 2.17 5.23 Dry

40 9 4077.90 1040.50 95.86 363.65 Saturated

5.7 4987.20 1598.80 70.88 310.35

5.2 5406.20 2300.20 60.87 265.22

4.4 6822.30 6860.90 32.00 129.62

2.9 6707.20 6439.20 34.11 137.49

0 8923.30 19066.10 1.31 2.79 Dry

50 10.2 3666.50 775.50 109.66 400.13 Saturated

5.9 4983.70 2198.40 70.97 270.84

10.2 3738.00 542.10 106.66 444.56

7.9 4132.90 497.70 94.19 455.16

7.4 4424.70 704.40 85.73 412.06

5.5 5180.80 1844.90 66.15 292.59

= 0.406x 87.275

R = 0.314

107

102

97

92

87

82

77

720 10 20 30 40 50

Amp

litudeaentuaon

(dB/m)

Free chloride content (g/L), x

Fig. 10. DW amplitude attenuation at various immersed free chloride content.



7/11

by the thickness of the sample as shown in Figs. 8 and 9. The exist-

ing DW and RW amplitude attenuation is similar to the findings in

reference[35]. The values of the DW amplitude attenuation in this

study are lower than the results in [35] with average of 66%,

whereas the RW results are higher than the values in [35] with

average of 41%. The lower value of amplitude attenuation pre-

dicted by the current linear relationship may be attributed to thelower water content on the top surface of the concrete as com-

puted by Eq. (2). The lower water content on this surface will

absorb less radar energy and produce less DW attenuation. In the

case of the RW with a 1.6 GHz antenna, the absorption of radar

energy will exceed the absorption noted in reference [35], in which

a 1.5 GHz antenna was employed.

3.2. Effect of free chloride on amplitude attenuation

3.2.1. Radar amplitude and peak-to-peak amplitude

The amplitude values of the radar signals for the samples of var-

ious chloride content are listed in Table 4.Figs. 10 and 11 display a

plot of the amplitude attenuation of direct and reflected waves as

the immersed chloride content is varied from 0 g/L to 50 g/L for the

saturated samples. Both plots present linear relationships with sat-

isfactory determination coefficients for the reflected wave (R2 of

0.95), whereas the direct wave exhibited a very weak linear rela-

tionship (R2 of 0.31). This weak linear relationship may be attribu-

ted to the varying surface roughness conditions on the sample

surface which results in altering the GPR absorption rate character-

istics as compared with the very smooth surface sample condition.The direct wave and reflected wave amplitude attenuation were

linearly decreased by 4.06 dB/m and 68.67 dB/m, respectively. This

wave attenuation can be explained by the increased mobility of the

free chloride ions in the moisture inside concrete pores, which

increases the ionic conduction and polarization. These findings

concluded that the attenuation of direct and reflected waves

increases as the chloride increases, which is consistent with the

findings by[37].

Figs. 12 and 13show the comparison between the current DW

and RW amplitude attenuation with reference [35]after the cur-

rent amplitude attenuation values were multiplied by the thick-

ness of the sample D. The results indicate that both the DW and

the RW follow the same trend demonstrated by the moisture con-

tent case: the current DW amplitude attenuations were lower than

the values in[35]with average of 70% and the current RW ampli-

tude attenuations were higher than the values provided by [35]

with average of 18%. However, the slope value of the RW equation

(8.6735) is higher than the slope provided by [35]. Researchers

have hypothesized that the use of a higher antenna frequency will

contribute to higher radar energy absorption and produce a higher

slope value for RW.

3.3. Comparison of amplitude wave attenuation between water and

free chloride content

The comparison of the GPR amplitude wave in the previous sec-

tion between water content and free chloride content showed sig-

nificant amplitude attenuation. A comparison of the GPR

y = 0.0284x + 6.1092

y = 0.0533x + 10.13

0

2

4

6

8

10

12

14

0 10 20 30 40 50


Free Chloride content (g/L),x

DW

Sbartai

Linear (DW)

Fig. 11. Comparison of current DW results with reference [24].

= -6.8665x - 123.91

R = 0.9507

-445

-395

-345

-295

-245

-195

-1450 10 20 30 40 50

Amplitudeaenuaon

(dB/m)

Free Chloride Content (g/L), x

Fig. 12. RW amplitude attenuation at various immersed free chloride content.

y = 0.4807x + 8.6735

y = 0.3437x + 7.7396

0

5

10

15

20

25

30

35

0 10 20 30 40 50


Free Chloride content (g/L), x

RW

Sbartai

Linear (RW)

Linear (Sbartai)

Fig. 13. Comparison of current RW with reference [24].

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


8/11

amplitude attenuation by both corrosive agents is shown in Fig. 14.

The comparison of the effect of both corrosion agents on RW indi-

cated that the chloride content attenuates the GPR amplitude at a

higher rate than moisture content. The results differ for the DW as

the moisture content attenuates the GPR amplitude at a higher rate

than chloride content. As the linear relationship between the direct

wave and the chloride content is very weak, i.e., R2 is 0.314, the

GPR attenuation in this case is not considered in the comparison

with the RW case. The chloride content in the sample attenuates

the GPR amplitude at a higher rate than the moisture content.

3.4. Development of models

3.4.1. Modelling equations for the attenuation of direct and reflected

waves

The thirty one samples were analysed and the radar amplitude

attenuation for direct wave and reflected waves were computed

based on Eq. (2). The amplitude attenuation of the radar signal

can be directly linked to the energy loss from the energy absorp-

tion by water molecules and chloride ions in the concrete pores.

UsingTable 4, the plots of the direct wave and the reflected wave

attenuation, respectively, as the moisture and immersed free chlo-

ride content was varied are denoted by red points in Figs. 15 and

16. A multiple nonlinear regression equation algorithm was used

80 60 40 20 0

DW

RW

Wave aenuaon (dB/m)

Chloride

Moisture

Fig. 14. Comparison of moisture and chloride on wave attenuation.

Fig. 15. DW attenuation model for varying moisture and free chloride content.

Fig. 16. RW attenuation model for varying moisture and free chloride content.



9/11

to correlate the amplitude attenuation per meter, a with the

immersed chloride contentx and the moisture in concretey basedon the general form of the following relationship in Eq. (6):

a y0 ax by cx2 dy2 exy 6wherey0,a,b, c, d ande are the fitted coefficients of the nonlinear

multiple regression using an iterative least squares algorithm and

is the error term. The multiple regression equations plot of both

waves are shown inFigs. 15 and 16.

Reference [35] revealed a linear relationship if the amplitude

attenuation is related to either moisture or free chloride content;

however, the radar amplitude attenuations in this study were

observed to behave nonlinearly with variations in the moisture

and free chloride content in concrete. The coefficient of multiple

determination of the multiple nonlinear regression equation R2

indicates close agreement between the computed amplitude atten-

uation and the predicted value. The direct wave multiple regres-

sion model and reflected wave multiple regression models

exhibit coefficient of determination of 0.917 and 0.948 respectively

3.4.2. Accessing the quality of the modelling equations

The quality of the modelling equations can be evaluated by

studying randomness of errors, the trend of errors and the nature

of the errors distribution. The randomness of the errors is quanti-

fied by the average variability of errors, whereas the trend pattern

of the residual show the direction of each deviation about the

mean value.

3.4.2.1. Constant variance and trend pattern test on residuals on themodelling equations. To access the quality of the developed mod-

elling equations, plots of the errors in both models were analysed.

The error in the model, is defined as

ap am 7where, ap and am represent the predicted wave attenuation per

meter by the developed model and the measured wave attenuation

per meter from this study, respectively.

The scatter plots of the errors in direct wave attenuation model

and reflected wave attenuation model are shown in Figs. 17 and 18,

respectively. Both error terms fluctuated around zero values and

do not reveal a significant trend, which indicates that the errors

were not independent and random. The developed model satisfied

the assumptions of normality and equal variance and may beemployed for the prediction of radar wave attenuation[38].

3.4.2.2. Normal probability tests on residuals of the modelling

equations. The normal probability plots of residuals of direct waveand reflected waves models were shown inFigs. 19 and 20respec-

tively. Based on the normal probability plots, it appears that the

actual residual data points from both models lies almost to the the-

oretical normal probability lines. It was observed approximately

29% (9 points) and 13% (4 points) of the residuals plotted for direct

and reflected waves were not laying on the theoretical normal

probability lines. This is consistent with the coefficient of determi-

nation,R2, from the previous section as direct wave models.

As a conclusion, the residuals data of direct wave and reflected

waves models behave randomly and satisfies the normally distri-

bution assumption.

3.4.3. The potential application of models on estimating moisture and

free chloride content in concrete cover

The potential assessment of the moisture and free chloride con-

tent in concrete cover can be estimated simultaneously using the

developed models shown in Figs. 15 and 16. In order to perform

the estimation of either moisture or free chloride content, the

developed modelling equation as shown in Fig. 15 will be

converted to contour plots as shown in Fig. 21. The radar signals

collection will firstly performed above the rebar location. The first

amplitude detected by the GPR antenna is analysed as the direct

Fig. 17. Residual plot of the DW residual. Fig. 18. Residuals plot of the RW residual.

Fig. 19. Normal probability plot of the direct wave residual.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


10/11

wave whereas the signal that is reflected by the top rebar surface is

considered as the reflected wave.

The peak-to-peak amplitude each direct and reflected waves is

computed and divided to the peak-to-peak amplitude when the

GPR radar signal is measured in air as the normalized value of

direct wave, aDW, and reflected wave, aRW. The moisture content

of the concrete cover, x can be measured using the moisturemeter

or the free chloride content,y based on dust analysis. The intersec-

tion point between the aDWor aRWwith eitherx ory value will be

marked on the contour plot and the estimate value of moisture or

chloride content can be determined by identifying the projected

value. The process on finding the estimate moisture or chloride

content is shown inFig. 21respectively. The same procedure can

be repeated on the direct wave to estimate the moisture or freechloride content in the cover concrete.

4. Conclusions

The main contribution of this study is to provide a nondestruc-

tive evaluation of the moisture and chloride content in a concrete

structure, which is based on GPR direct and reflected wave ampli-

tudes. The results of this study are useful for monitoring the state

of chloride contaminant and moisture content in concrete struc-

tures that have been exposed to a chloride environment, basedon the measured radar amplitude attenuation. We highlight the

following results of this study:

1. The amplitude of GPR is significantly influenced by the amount

of moisture in concrete. The attenuation of the radar wave

amplitude exhibits a linear relationship between the direct

waves and the reflected waves measured in the moisture-

immersed slab samples.Reasonable correlation coefficients with

R2 = 0.82 and 0.97 were obtained. The reflected waves attenuate

more than direct waves for the varying moisture content in the

concrete slab samples.

2. Chloride content in concrete influenced the radar amplitude

attenuation higher than the moisture content in the samples.

This result is due to the free chloride ions that will conduct elec-

tricity produced by the GPR antennas electrical field and

increase the degree of the attenuation of the radar amplitude.

Linear relationships between the chloride content and the

amplitude attenuation of the direct and reflected waves were

observed. Correlation coefficients with R2 = 0.31 and 0.95 were

obtained. A weak correlation with the direct wave attenuation

was observed and may be attributed to the surface textures of

the sample.

3. Two nonlinear models to estimate the moisture and chloride

content in concrete were developed. The attenuation residuals

of the models are not dependent on both the MC and CC and

are random in nature, which satisfies the assumptions of nor-

mality and equal variance. The proposed models can be used

to estimate the moisture and chloride content in concrete

structures.

4. This field of research has the significant potential or impact on

monitoring and diagnosing the quality of concrete cover at site

by characterizing the amount of moisture and free chloride con-

tent using GPR. As concrete cover protects the structural ele-

ments and the reinforcement, the monitoring and evaluation

of moisture and chloride by using this method serve as the indi-

cator of structural quality. This quality evaluation helps the

engineers and contractors to decide the replacement program

on the concrete cover within its life cycle.

Acknowledgment

The authors acknowledge Universiti Kebangsaan Malaysia fortheir financial support via the allocation grant underproject GUP-

2013-017 and DLP-2013-033.

References

[1] M.G. Richardson, Fundamentals of Durable Reinforced Concrete, Spon Press,

2002.

[2] R.R. Aveldano, N.F. Ortega, Characterization of concrete cracking due to

corrosion of reinforcements in different environments, Constr. Build. Mater.

2011 (25) (2011) 630637.

[3] F. Mittermayo, A. Baldermann, D. Klammer, A. Leis, J. Tritthart, M. Dietzel.

Sulfate attack-Reaction mechanism revealed by a multi proxy approach, Civil

and Environmental Research; 5: Special Issue for International Congress on

Materials and Structural Stability, 2730 Nov 2010: 96100.

[4] A. Leeman, R. Loser. Accelerated sulphate resistance test for concrete-

Chemical and microstructural aspects, Second International Conference on

Microstructural-related durability of cementitious, composites, Netherland,1113 April 2013: 18.

Fig. 20. Normal probability plot of the reflected wave residual.

Contour Plot of Direct Wave attenuation Model

Free chloride percentage,x

0 1 2 3 4 5

Moistu

recontentpercentage,y

0

2

4

6

8

10

12

-100

-100

-80

-80-80

-80

-60

-60 -60

-60

-40

-40-40

-40

-20

-20-20

-20

0

00

-100

Col 3

6% MC and -70

dB/m

The esmated free chloride content is 0.9 %

-120

Fig. 21. Contour plot of direct wave attenuation model.

http://refhub.elsevier.com/S0950-0618(15)30827-8/h0005http://refhub.elsevier.com/S0950-0618(15)30827-8/h0005http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0010http://refhub.elsevier.com/S0950-0618(15)30827-8/h0005http://refhub.elsevier.com/S0950-0618(15)30827-8/h0005http://refhub.elsevier.com/S0950-0618(15)30827-8/h0005


11/11

[5]G. Smoak, Guide to Concrete Repair, The Minerva Group Inc., 2002. 192 pages .

[6] H.-S. Shang, T.-H. Yi, Freeze-Thaw Durability of Air-Entrained Concrete, The

Scienties World Journal, 2013. 16.

[7]J.E. Hardy, P. Gerald, M. Ryall, Bridge management, CRC Press, 1990. 812 pages.

[8] D. Stark, The moisture condition of field concrete exhibiting alkali-silica

reactivity, Second International Conference SP-126, American Concrete

Institute, Farmington Hills, Michigan, 1991. 973987.

[9] G.D. Desmaraij, Derome, P. Fazio, Experimental Setup for the Study of Air

Leakages Patterns, Thermal Performance of the Exterior Enveloped of Buildings

VII, 99108, American Society of Heating, Refrigerating and Air-conditioning

Engineers, Inc., Atlanta, 1988.[10] J .S. Robert, E.C. Mary, Structural Materials Technology: A NDT Conference, CRC

Press, 1994.

[11] S. Laurens, J.P. Balayssac, J. Rhazi, G. Klysz, G. Arliguie, Non-destructive

evaluation of concrete moisture by GPR: experimental study and direct

modeling, Mater. Struct. 38 (2005) 827832.

[12] G. Klysz, J.-P. Balayssac, Determination of volumetric water content of

concrete using ground-penetrating radar, Cem. Concr. Res. 37 (2007) 1164

1171.

[13] W.Chen, P. Shen, Z. Shui, Determination of water content in freshconcrete mix

based on relative dielectric constant measurement, Constr. Build. Mater. 34

(2012) 306312.

[14] M. Jamil, M.K. Hassan, H.M.A. Al-Mattarneh, M.F.M. Zain, Concrete dielectric

properties investigation using microwave nondestructive techniques, Mater.

Struct. 46 (2013) 7787.

[15] A.M. Alani, M. Aboutalebi, G. Kilic, Applications of ground penetrating radar

(GPR) in bridge deck monitoring and assesment, J. Appl. Geophys. 97 (2013)

4554.

[16] R. Parrillo, R. Rober, A. Haggan. Bridge Deck Condition Assesment using

Ground Penetrating Radar, ECNDT Conference Proceedings 2006, Berlin,

Germany: 2529.

[17] M. Scott, A. Rezaizadeh, A. Delahaza, C.G. Santos, M. Moore, B. Graybeal, G.

Washer, A comparison of non-destructive evaluation methods for bridge deck

assessment, NDT and E Int. 36 (2003) 245255.

[18] V. Barrile, R. Pucionotti, Application of radar technology to reinforced concrete

structures: a case study, NDT and E Int. 38 (2005) 596604.

[19] J.H. Bungey, Sub-surface radar testing of concrete: a review, Constr. Build.

Mater. 2004 (18) (2004) 18.

[20] A. Loizos, C. Plati, Accuracy of pavement thickness estimation using different

ground penetrating radar analysis approches, NDT and E Int. 40 (2007) 147

157.

[21] T. Saarenketo, T. Scullion, Road evaluation with ground penetrating radar, J.

Appl. Geophys. 43 (2000) 119138.

[22] U.B. Habale, H.J. Siriwardane, S. Pyakurel, R. Kiriakidis, R. Ingram. Water level

detectionin siltymaterials using Ground Penetrating Radar, Proceedings of the

3rd Annual Review of Progress in Quantitative Non Destructive Evaluation,

2006; 26, Portland Eregon, July 30August 4; 2006.

[23] M.J. Harry, Ground Penetrating Radar: Theory and Applications, Elsevier B.V.,

2009.

[24] Z.-M. Sbartai, S. Laurens, D. Breysse. Concrete Moisture Assesment Using Radar

NDT Technique-Comparison between Time and Frequency Domain Analysis,

NDTCE09; Nantes, France, June 30July 3; 2009.

[25] J. Zhang, B. Han, J. Yang, Application of radar technology on concrete water

content testing, Adv. Mater. Res. 2011 (250253) (2011) 31033106.[26] G. Klysz, J.-P. Balayssac, S. Laurens, Spectral analysis of radar surface waves for

non-destructiveevaluationof cover concrete, NDTandE Int.37 (2004)221227.

[27] Geophysical Survey System Inc., GSSI handbook for RADAR inspection of

concrete; 2003.

[28] S. Laurens, M.E. Barrak, J.P. Balayssac, J. Rhazi, Aptitude of the near-field direct

wave of ground-coupled radar antennas for the characterisation of the

covercrete, Constr. Build. Mater. 2007 (21) (2007) 20722077.

[29] K. Viriyametanont, S. Laurens, G. Klysz, J.P. Balayssac, G. Arliguie, Radar survey

of concrete elements: effect of concrete properties on propagation velocity and

time zero, NDT and E Int. 2008 (41) (2008) 198207.

[30] A. Kalogeropoulos, J. Kruk, J. Hugenschmidt, S. Busch, K. Merz, Chlorides and

moisture assessment in concrete by GPR full waveform inversion, Near Surf.

Geophys. 2011 (9) (2011) 277285.

[31] Department of the Environment, Design of Normal Concrete Mixes, Building

Research Establishment, Watford, UK, 1988.

[32] B.P. Lathi, Principles of Linear Systems and Signals, 2nd Ed., Oxford University

Press, 2009.

[33] S.S. Hubbard, J. Zhang, P.J.M. Monteiro, J.E. Peterson, Y. Rubin, Experimental

detection of reinforcing bar corrosion using nondestructive geophysics

techniques, ACI Mater. J. 100 (6) (2003) 501510.

[34] C. Maierhofer, H.-W. Reinhardt, G. Dobmann, Non-Destructive Evaluation of

Reinforced Concrete Structures, Elsevier (2010).

[35] Z.M. Sbartai, S. Laurens, J.-P. Balayssac, G. Arliguie, G. Ballivy, Ability of the

direct wave of radar ground-coupled antenna for NDT of concrete structures,

NDT and E Int. 39 (2006) 400407.

[36] M. Loken. Use of Ground Penetrating Radar to Evaluate Minnesota Roads,

Minnesota Department of Transporation; Report No: MN/RC 200701; 2007.

[37] Z.M. Sbartai, S. Laurens, J. Rhazi, Balayssac, G. Arliguie, Correlation using

electrical resistivity, J. Appl. Geophys. 62 (2007) 361374.

[38] J.E. Matson, B.R. Huguenard, Evaluating aptness of regression model, J. Stat. Edu.

15(2)(2007). .

http://refhub.elsevier.com/S0950-0618(15)30827-8/h0025http://refhub.elsevier.com/S0950-0618(15)30827-8/h0025http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0035http://refhub.elsevier.com/S0950-0618(15)30827-8/h0035http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0050http://refhub.elsevier.com/S0950-0618(15)30827-8/h0050http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://refhub.elsevier.com/S0950-0618(15)30827-8/h0090http://refhub.elsevier.com/S0950-0618(15)30827-8/h0090http://refhub.elsevier.com/S0950-0618(15)30827-8/h0095http://refhub.elsevier.com/S0950-0618(15)30827-8/h0095http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://refhub.elsevier.com/S0950-0618(15)30827-8/h0105http://refhub.elsevier.com/S0950-0618(15)30827-8/h0105http://refhub.elsevier.com/S0950-0618(15)30827-8/h0105http://refhub.elsevier.com/S0950-0618(15)30827-8/h0115http://refhub.elsevier.com/S0950-0618(15)30827-8/h0115http://refhub.elsevier.com/S0950-0618(15)30827-8/h0125http://refhub.elsevier.com/S0950-0618(15)30827-8/h0125http://refhub.elsevier.com/S0950-0618(15)30827-8/h0125http://refhub.elsevier.com/S0950-0618(15)30827-8/h0130http://refhub.elsevier.com/S0950-0618(15)30827-8/h0130http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0155http://refhub.elsevier.com/S0950-0618(15)30827-8/h0155http://refhub.elsevier.com/S0950-0618(15)30827-8/h0160http://refhub.elsevier.com/S0950-0618(15)30827-8/h0160http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0170http://refhub.elsevier.com/S0950-0618(15)30827-8/h0170http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0185http://refhub.elsevier.com/S0950-0618(15)30827-8/h0185http://refhub.elsevier.com/S0950-0618(15)30827-8/h0185http://www.amstat.org/publications/jse/v15n2/datasets.matson.htmlhttp://www.amstat.org/publications/jse/v15n2/datasets.matson.htmlhttp://www.amstat.org/publications/jse/v15n2/datasets.matson.htmlhttp://www.amstat.org/publications/jse/v15n2/datasets.matson.htmlhttp://refhub.elsevier.com/S0950-0618(15)30827-8/h0185http://refhub.elsevier.com/S0950-0618(15)30827-8/h0185http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0175http://refhub.elsevier.com/S0950-0618(15)30827-8/h0170http://refhub.elsevier.com/S0950-0618(15)30827-8/h0170http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0165http://refhub.elsevier.com/S0950-0618(15)30827-8/h0160http://refhub.elsevier.com/S0950-0618(15)30827-8/h0160http://refhub.elsevier.com/S0950-0618(15)30827-8/h0160http://refhub.elsevier.com/S0950-0618(15)30827-8/h0155http://refhub.elsevier.com/S0950-0618(15)30827-8/h0155http://refhub.elsevier.com/S0950-0618(15)30827-8/h0155http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0150http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0145http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0140http://refhub.elsevier.com/S0950-0618(15)30827-8/h0130http://refhub.elsevier.com/S0950-0618(15)30827-8/h0130http://refhub.elsevier.com/S0950-0618(15)30827-8/h0125http://refhub.elsevier.com/S0950-0618(15)30827-8/h0125http://refhub.elsevier.com/S0950-0618(15)30827-8/h0115http://refhub.elsevier.com/S0950-0618(15)30827-8/h0115http://refhub.elsevier.com/S0950-0618(15)30827-8/h0115http://refhub.elsevier.com/S0950-0618(15)30827-8/h0105http://refhub.elsevier.com/S0950-0618(15)30827-8/h0105http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://refhub.elsevier.com/S0950-0618(15)30827-8/h0100http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0095http://refhub.elsevier.com/S0950-0618(15)30827-8/h0095http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0090http://refhub.elsevier.com/S0950-0618(15)30827-8/h0090http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://refhub.elsevier.com/S0950-0618(15)30827-8/h0085http://-/?-http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://refhub.elsevier.com/S0950-0618(15)30827-8/h0075http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://refhub.elsevier.com/S0950-0618(15)30827-8/h0070http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://refhub.elsevier.com/S0950-0618(15)30827-8/h0065http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://refhub.elsevier.com/S0950-0618(15)30827-8/h0060http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://refhub.elsevier.com/S0950-0618(15)30827-8/h0055http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0050http://refhub.elsevier.com/S0950-0618(15)30827-8/h0050http://refhub.elsevier.com/S0950-0618(15)30827-8/h0050http://-/?-http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0045http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0040http://refhub.elsevier.com/S0950-0618(15)30827-8/h0035http://refhub.elsevier.com/S0950-0618(15)30827-8/h0035http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0030http://refhub.elsevier.com/S0950-0618(15)30827-8/h0025http://refhub.elsevier.com/S0950-0618(15)30827-8/h0025

Jurnal JCB aku.pdf

Documents

Transcript of Jurnal JCB aku.pdf