info@auctorespublishing.com
+1-(302)-520-2644
+1-(302)-520-2644

Implementation of Nondestructive Test (NDT) to Model the Strength Properties of Asphalt Concrete

  1. Home
  2. Journals
  3. Advances in Architecture and Civil Engineering
  4. Current Issues
Submit Guidelines

Research Article | DOI: https://doi.org/10.31579/JACE.2019/001

Implementation of Nondestructive Test (NDT) to Model the Strength Properties of Asphalt Concrete

  • Saad Issa Sarsam 1*
  • Nazar Sajad Kadium 2

1 Professor, Department of Civil Engineering, College of Engineering, University of Baghdad, Iraq.
2 MSc. Student, Department of Civil Engineering, College of Engineering, University of Baghdad, Iraq 

*Corresponding Author: Professor, Department of Civil Engineering, College of Engineering, University of Baghdad, Iraq.

Citation: Saad I. Sarsam, Nazar S. Kadium (2019). Implementation of Nondestructive Test (NDT) to Model the Strength Properties of Asphalt Concrete. Advances in Architecture and Civil Engineering, 1(1): Doi: 10.31579/JACE.2019/001

Copyright: ©2019. Saad Issa Sarsam. This is an open-access article distributed under the termsof the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Received: 14 September 2019 | Accepted: 27 September 2019 | Published: 03 October 2019

Keywords: modeling; asphalt concrete; NDT; testing; pulse velocity; strength properties

Abstract

Asphalt concrete is a composite material, it consists of aggregates, mineral filler and asphalt cement. Its mechanical behavior is complex due to its susceptibility of temperature, moisture, loading frequency, and strain level. The quality control process during construction of asphalt concrete pavement layers is labor, cost and time consuming. Implementation of nondestructive testing NDT could possess a sustainable solution for such issue.   In this investigation, asphalt concrete specimens of wearing, binder and base courses have been prepared in the laboratory and subjected to Marshal and indirect tensile strength determination. Another group of specimens were subjected to temperature susceptibility evaluation. Specimens have practiced ultrasonic pulse velocity traversing the specimen using pundit instrument before conducting the strength test. The dynamic and seismic moduli have been calculated based on the ultrasonic pulse velocity. Test results were analyzed and modeled. A linear statistical model relating ultrasonic pulse velocity with the strength properties of asphalt concrete was obtained. It was concluded that such models can explain (75.4) % of the variation in the strength properties respectively under the testing techniques implemented.

Introduction

Implementation of nondestructive test technique NDT and modeling the physical properties of asphalt concrete are considered as a sustainable issue. The quality control process during construction of asphalt concrete pavement layers is labor, cost and time consuming. Implementation of nondestructive testing NDT could possess a sustainable solution for such issue. The ultrasonic pulse velocity method of testing is one of the most widespread wave-based methods in nondestructive testing techniques NDT as reported by Zhou and Chen, 2019. However, its potential for assessing the quality of asphalt concrete materials is limited because it is only based on measurement of wave velocity, Jiang, 2007. NDT was implemented for testing of concrete, while there has been ongoing research on the use of ultrasonic testing for asphalt concrete mixtures. However, no standard wave-based test protocol is available yet for quality assessment of asphalt concrete, Witczak et al, 2002. The ultrasonic testing of asphalt concrete is considered difficult because the material is viscoelastic and susceptible to temperature variation. The composite characteristics of the materials result in scattering of waves and limit the penetration of waves into the object being investigated, Sztukiewicz 1992. The quality of asphalt concrete materials is often characterized by material properties such as dynamic modulus determined based on the simple performance tests recommended by the Federal Highway Administration (FHWA) Bonaquist et al., 2003. The dynamic modulus test is the oldest test recently modified for use as a part of simple performance tests for the evaluation of the long-term performance of asphalt concrete mixture, Tigdemir et al, 2004. Sarsam and Kadim, 2018 investigated the strength and volumetric properties of asphalt concrete wearing course using pulse velocity test. Specimens were prepared in the laboratory at various asphalt percentages and tested for pulse velocity, then subjected to indirect tensile strength and punching shear strength determination. The impact of moisture damage and testing temperature on pulse velocity were also investigated. It was concluded that implementation of non-destructive testing with the aid of pulse velocity is feasible for predicting the quality of asphalt concrete within the limitations of the testing program implemented. The good correlation between the pulse velocity and the volumetric and strength properties demonstrates the potential benefit of using the wave parameters for condition assessment of asphalt concrete. Surface wave data acquisition has successfully been used to determine dynamic modulus and thickness of the top asphalt concrete layer in the field. The principle is that the velocity of wave propagation depends on the elastic modulus, density, and Poisson’s ratio of the medium concerned. In practice, however, the velocity-strength relationship is affected by the combined effect of aggregate size, air void, and moisture condition which can cause some degree of variability in the results, Arabani et al, 2009.
The aim of this investigation is to model the strength properties of asphalt concrete with the aid of NDT. Specimens will be prepared and tested for strength, and sensitivity to temperature variation after practicing the ultrasonic pulse velocity test. The dynamic and seismic moduli will be calculated based on the ultrasonic pulse velocity. A statistical model could be obtained relating the pulse velocity with the strength properties.

Materials And Methods

Asphalt Cement
Asphalt cement of penetration grade (40-50) obtained from Dora refinery was implemented in this investigation; the physical properties of asphalt cement are listed in Table 1.

                                                                  Table1. Physical Properties of Asphalt Cement

Coarse and Fine Aggregates
Crushed Coarse and fine aggregates have been obtained from Al-Nibaee quarry, Table 2 illustrates the physical properties of aggregates.

                                                         Table 2. Physical Properties of Al-Nibaee Coarse and fine Aggregates

Mineral Filler

The mineral filler used in this work is limestone dust and was obtained from Karbala plant. The physical properties of the filler are presented in Table 3.  

                                                               Table 3. Physical Properties of Filler (Limestone dust).

Selection of Asphalt Concrete Combined Gradation

The selected gradation in this work follows the SCRB, 2003 Specification for dense graded base, binder, and wearing courses, with (25, 19, and 12.5) mm nominal maximum size of aggregates. Figure 1 exhibits the implemented aggregate gradations.

Preparation of Hot Mix Asphalt Concrete

The aggregates were dried in an oven to a constant weight at 110 ºC, then sieved to different sizes, and stored separately. Coarse and fine aggregates were combined with mineral filler to meet the specified gradations of various asphalt concrete layers as per SCRB specifications shown in Figure 1. The combined aggregate mixture was heated to 150ºC before mixing with asphalt cement. The asphalt cement was heated to the same temperature of 150ºC, then it was added to the heated aggregate to achieve the desired amount and mixed thoroughly using mechanical mixer for two minutes until all aggregate particles were coated with thin film of asphalt cement.

Figure 1. The implemented aggregate gradation according to SCRB, 2003

Marshall Size specimens were prepared in accordance with ASTM D1559, 2013 using 75 blows of Marshall hammer on each face of the specimen for binder and wearing courses. On the other hand, 50 blows of Marshall hammer on each face of the specimen was implemented for base course. Specimens with optimum asphalt content and 0.5% of asphalt above and below the optimum have been prepared for each layer. Figure 2 shows part of the prepared specimens.

Figure 2. Part of the prepared specimens

Ultrasonic Pulse Velocity Measurement

The portable ultrasonic non-destructive digital indicating tester (Pundit) was implemented in this study. The device generates and receives ultrasonic waves and has a digital display of the results. A frequency of 54 kHz and accuracy of 0.1 was implemented through this study to measure the ultrasonic pulse velocity for the specimens. The direct transmission arrangement was used in this study. The pulser and receiver were placed on opposite specimen parallel surfaces according to ASTM C597, 2013. Calibration of the pundit was done before testing to check the accuracy of the transit time measurements. This is achieved by the calibration with the reference bar. A thin layer of Vaseline oil was applied on the surface of the tested points to act as a couplet between the transducer and the asphalt concrete specimen’s surface and to prevent dissipation of transmitted energy. Eight readings were performed and averaged for each specimen. The ultrasonic pulse velocity test setup is demonstrated in Figure 3. All the prepared asphalt specimens have practiced the NDT before they were subjected to strength test.

Figure 3. Ultrasonic pulse velocity determination in process

Indirect Tensile Strength Test And Temperature Susceptibility

The indirect tension stress test as specified by ASTM, 2013 was conducted. The test was performed on the cylindrical specimens, 102 mm in diameter and 63.5 mm in height. The prepared Marshall Size Specimens of the three layers were subjected to the indirect tensile stress test at (25 and 40) ºC. Specimens have been tested in triplicate, and the average value was considered for analysis.
Marshal Stability and Flow

Part of the prepared specimens have been subjected to Marshal properties determination according to ASTM D1559, 2013. Table 4 demonstrates the strength properties obtained of various asphalt concrete mixtures implemented (independent variables).

Statistical Analysis and Modeling

                                      Table 4. Strength Properties of Asphalt Concrete Mixtures (Independent Variables)

Statistical Analysis and Modeling
The statistical analysis was implemented with the aid of statistical package (SPSS V22) software, Basheer, 2003 for the development of model relating the dependent variables (ultrasonic pulse velocity) to the number of independent variables including (Indirect tensile strength ITS at (25 and 40) °C, Temperature susceptibility TS, Marshal stability and Flow, Seismic and Dynamic moduli). A mathematical relationship between dependent and independent variables was the goal in mind of measuring future values of those predictors and inserting them into the mathematical relationship to predict future values of the target variable. It is desirable to give some measure of uncertainty for the predictions, typically a prediction interval that has some assigned level of confidence like 95%. Another task in the process is model building. Thus, a significant level of 0.05 was chosen. Model selection, fitting and validation are the basic steps of the model building process.

Identification Of Dependent And Independent Variables Of The Developed Model

To achieve the requirements of modeling, several variables were used. These variables are listed in Table 5.

Table 5. Independent and Dependent Variables Considered in Regression Analysis.

Checking of the Sample Size

Sample size was checked as shown in Table 6 and calculated by using equation 1, kennedy and neville, 1986.

Sample Size = (Z- score) ² SD (1-SD) / (significant level) ²     .......(1)
Where:

Z-score: constant value corresponding to the confidence level (for confidence level of 95%, Z-score= 1.96.

SD: standard deviation

Significant level: 0.05

Table 6. Sample size for the model.

Checking for Outliers.

The experimental work include collected data, the distribution of data contain of group concentrated within limited margin calculated from the frequency of data, but sometimes due to mistakes or other abnormal condition, a data set is considered as extreme or outliers by checking those extreme values using Chauvinist’s criterion and absolute tabulated sample size value as in Table 7. It was noticed that all tabulated values are more than the test results, thus, there are no outliers. To eliminate the data for modeling purposes to increase the R2, the standardized residuals are removed from data matrix. The standardized residuals are an indication of how the residuals are large in the standard deviation units, and it is equal to the observed minus the estimated value and divided by the standard deviation. In the case of nonlinear regression, the residuals of high absolute value are removed resulted from observed minus estimated values, (Montgomery et al, 2009).

                                                                                    Table 7. Chauvinist’s test

Testing of Normality

Kolmogorov Smirnov (K-S) and Shapiro Wilk test was used to check the distribution of variables were used to developed model relate resilient modulus of asphalt mixture to numerals of variables. Herrin et al, 1995 stated that the K-S statistics D is based upon the maximum distance between F (y) and F n (y); that is equation 2. Table 8. show the result of normality checking of the model.

D=max. [F(y) - Fn(y)] …………. (2)

Where:

F(y) = Normal cumulative probabilities (From normal distribution table)

Fn (y) = Sample cumulative distribution function.

 

 

Ultrasonic Pulse Velocity (mm/microsecond)

N

244

Normal Parameters

Mean

3.7468

Std. Deviation

0.313

Most Extreme Differences

 

 

Absolute

.123

Positive

.123

Negative

-0.044

Test Statistic

0.071

Kolmogorov-Smirnov Z

4.999

 Sig. (2-tailed)

.004

 

Table 8. Result of One Sample K-S Test

Multicollinearity

SPSS software version (22) is employed to find the correlation between independent variables with one another using Multicollinearity (collinearity and intercorrelation). A statistical procedure using stepwise regression technique. The independent variables are eliminated according to the significant contribution of X variable on the produced model. The process is repeated with each variable until significant predictor variable remained and the insignificant ones are removed. This Would suggest that those variables with high inter correlation may be eliminated according to the decision to add or remove a variable to improve the model; at that Point, interactions among the variables are considered. A correlation matrix is produced to determine the correlation coefficients for the Variables. 

Stepwise Regression Model

The best and commonly method used to determine parameter of prediction model is stepwise method, Kennedy and neville, 1986. This method computes the simple regression model for each independent variable. The independent variable is with the largest F-statistic, in other words, the smallest p-value is chosen as the first entering variable. SPSS software uses the F-statistics and the standard is usually set at F=3.8, which is chosen because the significant level is about 5 %. The standard is called the F-to-enter. If at least one variable exceeds the standard, the procedure continues. It then considers whether the model would be improved by adding a second independent variable. It examines all such models to determine which is best and whether the F-statistic of the second variable (with the first variable already in the equation) is greater than F-to-enter. If two independent variables are highly correlated, only one of them will enter the equation. Once the first variable is included, the added explanatory power of the second variable will be minimal and its F-statistic will not be large enough to enter the model. Tables 9 show the summary of stepwise regression for the model while Table 10 exhibit the Coefficients.

Error analysis

                                          Table 9. Model Summary
                                           Table 10. Coefficients

Error analysis

For nonlinear model goodness, the homoscedasticity hypothesis assumes that the error is with constant variance, the standardized residual scatter plot as solve the problem, and by setting the estimated values of dependent variable on x axis and plotting the difference of observed value and theoretical value on y axis which is standardized residual, we can decide the goodness of the model. The pattern plots as demonstrated in Figure 4 is satisfactory and the residuals are randomly distributed.

Figure 4. Scatter plot of Standardized Residual

Variance Analyses Of Anova Test

To check the significant differences of the independent variable s mean, the ANOVA is applied by F-test. The F -test called the test of linearity which determines by a straight line the deviations of means. Table 11 shows the ANOVA test for the model. The results show statistically significant relation since the p-value is less than 0.05, thus the overall variables used in regression (independent variables) are effective on the dependent variable. And by comparing the F ratio value with critical value from relative frequency distribution we decide the significances of results.

 

Model

 

Sum of Squares

df

Mean Square

F

Sig.

1

Regression

 

20.943

12

1.745

59.962

.000b

Residual

 

7.160

246

.029

 

 

Total

 

28.103

258

 

 

 

 

 

Table 11 ANOVA for the Model

 

 a. Dependent Variable: UPV.mm/micro sec

 

 

 

b. Predictors: (Constant), dynamic Modulus (GPa), ITS @40 (KPa), ITS @ 25C (KPa), temp. susceptibility, flow (mm), Seismic Modulus (GPa)

         

 

Checking of R-critical.

A high correlation coefficient R value does not guarantee that the model fits the data well. The correlation between x and y is considered significant at the given probability level when the calculated R exceeds the tabulated R value. Table 12 demonstrates the calculated and tabulated R- value.

                                                                            Table 12. Tabulated R-values for the model

Model Limitation

The limitation of the data used to establish the model is presented in Table 13. The intention of the limitation is not to suggest that the modeling effort has not been successful. It merely serves to alert of the limitations of the data. Table 14 exhibits the developed model.

                                               Table 13. The Intention of the Limitation
                                                                   Table 14. The Developed Model

Where:

Dependent variable:

Ultrasonic pulse velocity = (mm/microseconds)

Independent variables:

F = Marshal Flow (mm)

DM = Dynamic Modulus (GPa)

TS = Temperature Susceptibility (kPa/ ̊ C)

SM = Seismic Modulus (GPa)

Validation of the Developed Model

The graphic plotting of observed and estimated data is a most useful method of evaluating the overall performance of a regression equation. If the point which result from the plot of estimated with observed data tend to stand nearby the line drawn at 45o, then the result model is considered satisfactory. This can be done by data splitting in to two sets. 70% of data are used to build the models and 30% of it is used for the validation process. Figure 5 shows the expected vs. observed data for dependent variable, the model is considered good because of the small variance of the points measured as compared with predicted values.

Conclusions

Figure 5. Scatter plot for expected and observed values

Based on the limitations of the testing program, the following conclusions may be drawn.

1- The developed linear statistical model can explain 75.4 % of the variations in ultrasonic pulse velocity among various strength properties.

2- The dynamic and seismic moduli of asphalt concrete exhibit significant influence on ultrasonic pulse velocity among other strength parameters.

3- Temperature susceptibility TS exhibit the lowest influence on ultrasonic pulse velocity, while Marshal Flow shows significant impact.

4- Such modeling is considered as a sustainable issue in predicting the strength properties of asphalt concrete, it can limit the testing requirements, time and cost of quality control.

References

  1. Zhou J., Chen X. (2019). Modeling the Linear Viscoelastic Behavior of Asphalt Concrete via Optimization Method. CICTP
    View at Publisher | View at Google Scholar
  2. Jiang Z. (2007) Innovative nondestructive testing (NDT) for condition assessment of longitudinal joints in asphalt pavements. MSc Thesis, University of Waterloo, Ontario, Canada.
    View at Publisher | View at Google Scholar
  3. Witczak, M.W., Kaloush, K., Pellinen, T., El-Basyouny, M., and Von Quintus, H., (2002), Simple Performance Test for Superpave Mix Design, NCHRP Report 465, Transport Research Board, Washington D. C.
    View at Publisher | View at Google Scholar
  4. Sztukiewicz R. J. (1992). Ultrasonic description of the condition of the surface layer of asphalt concrete pavement. NDT and E International, Vol. 25 No. 4-5, P. 227-228.
    View at Publisher | View at Google Scholar
  5. Bonaquist, R.F., Christensen, D.W., and Sump W., (2003), Simple Performance Tester for Superpave Mix Design: First-Article Development and Evaluation, NCHRP Report 513, Transport Research Board, Washington D. C.
    View at Publisher | View at Google Scholar
  6. Tigdemir M., Kalyoncuoglu F., Kalyoncuoglu Y. (2004). Application of ultrasonic method in asphalt concrete testing for fatigue life estimation. NDT and E International, Vol. 37 No. 8, P. 597-602.  
    View at Publisher | View at Google Scholar
  7. Sarsam S. I., Kadim N. S. (2018) Feasibility of Using Pulse Velocity to Evaluate Asphalt Concrete Properties. Journal of Advances in Civil Engineering and Construction Materials; 1(1): 51-63.
    View at Publisher | View at Google Scholar
  8. Arabani M., Kheiry P., & Ferdosi B. (2009).  Laboratory Evaluation of the Effect of HMA Mixt Parameters on Ultrasonic Pulse Wave Velocities, Road Materials and Pavement Design, 10:1, 223-232.
    View at Publisher | View at Google Scholar
  9. ASTM. American Society for Testing and Materials. (2013). Road and Paving Material, Vehicle-Pavement System, Annual Book of ASTM Standards, Vol.04.03.
    View at Publisher | View at Google Scholar
  10. SCRB, (2003), General Specification for Roads and Bridges, Section R/9 Hot-Mix Asphalt Concrete Pavement, Reverse Edition, State Corporation of Roads and Bridges, Ministry of Housing and Construction, republic of Iraq.
    View at Publisher | View at Google Scholar
  11. Basheer, S. Z. (2003). Guide to the SPSS statistical program, The Arab Institute for Training and researches in Statistics, Republic of Iraq, Baghdad.
    View at Publisher | View at Google Scholar
  12. Kennedy, j., Neville (1986). Basic statistical methods for engineers and scientists, JW &Son, third edition.
    View at Publisher | View at Google Scholar
  13. Montgomery, D. C., & Peck, E. A. (2009).  Introduction to Linear Regression Analysis, 2nd Edition, John Wiley & Sons, Inc., New York.
    View at Publisher | View at Google Scholar
  14. Herrin G. D., Scheaffer R., McClave J. T. (1995). Probability and Statistics for Engineers. The American statistician, Vol.49 No.2, P.238.
    View at Publisher | View at Google Scholar
a