Académique Documents
Professionnel Documents
Culture Documents
net/publication/323272762
CITATIONS READS
0 65
2 authors, including:
Gauhar Sabih
University of New Mexico
13 PUBLICATIONS 10 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Gauhar Sabih on 22 February 2018.
Transportation Geotechnics
journal homepage: www.elsevier.com/locate/trgeo
Technical Paper
a r t i c l e i n f o a b s t r a c t
Article history: The format of the design and performance prediction of rigid pavements was reformed with the advent of
Received 21 October 2017 Pavement mechanistic-empirical (ME) design procedure, which now serves as the state-of-the-art tool in
Revised 24 January 2018 pavement design. Various state agencies have either completed or in the process of calibration of distress
Accepted 11 February 2018
prediction models and characterization of concrete materials to provide accurate inputs required by
Available online 19 February 2018
Pavement ME design software. There are numerous concrete properties for which input data is required
in ME design software, but with previous research, it was found that the concrete strength and thermal
Keywords:
properties including elastic modulus, modulus of rupture and coefficient of thermal expansion (CTE) are
Concrete
Coefficient of thermal expansion
the most important ones that affect the design and performance of rigid pavements. Accurate rigid pave-
Modulus of rupture ment design is heavily dependent on accuracy of these material inputs. This study is part of a New Mexico
Characterization Department of Transportation (NMDOT) research project that focuses on the development of guidelines
International roughness index for characterizing Portland cement concrete (PCC) materials for paving mixes being used in New Mexico.
Elastic modulus Concrete mixes with 5 different coarse aggregates were tested for these pivotal concrete properties at the
curing age of 7, 14, 28 and 90 days, and for CTE at 28 days. The laboratory test results represent level 1
PCC material inputs. The data collected offered an excellent opportunity to validate and refine the ME
default level 2 models for estimating flexural strength and elastic modulus based on compressive
strength data. The data demonstrated a slight deviation from the nationally calibrated models. CTE values
of concrete based on aggregate type were established for these paving mixes. Further analysis verified the
benefit of using the level 1 inputs over the default level 3 inputs for accurate pavement design and per-
formance prediction. It was also highlighted that transverse cracking is the most significantly affected
performance parameter between the pavement designed with level 1 and level 3 material inputs.
Ó 2018 Elsevier Ltd. All rights reserved.
Introduction United States (U.S.) and there are about 18 different material prop-
erties of concrete, which are required as input for design of JPCP
A concrete pavement is a structure comprising of a layer of Port- according to mechanistic-empirical (ME) design procedure. After
land cement concrete usually supported by underlying layers on studying the sensitivity of various input factors on rigid pavement
the subgrade. Concrete pavements may be either unreinforced or performance, Schwartz et al. [1] identified the modulus of rupture
reinforced depending on the preference to control the cracking. (MOR), the modulus of elasticity, and the coefficient of thermal
The high strength and rigidity of concrete provides a concrete expansion (CTE) as input factors that had significant impact on
pavement with a reasonable degree of flexural strength which the concrete pavement’s performance with regards to transverse
leads to distribution of externally applied wheel loads to a wider cracking, joint faulting and pavement roughness. Tanesi et al. [2]
area. This results in minimizing the pressure applied to the sub- worked to determine the effect of the variability of the CTE on
layers. The load capacity of a rigid pavement is solely dependent the predicted pavement performance. They performed a sensitivity
on the concrete layer. This study found that jointed plain concrete analysis by varying the CTE values on a single jointed plain con-
pavement (JPCP) is the most widely used rigid pavement across the crete pavement design and found that with the increase in CTE
value of concrete, the percentage of slabs with transverse cracking
also increases. Hein [3] conducted his research and described that
⇑ Corresponding author. thermal expansion and contraction of a concrete pavement can
E-mail addresses: gsabih@unm.edu (G. Sabih), tarefder@unm.edu (R.A. Tarefder). have a significant effect on its performance. Thermal contraction
https://doi.org/10.1016/j.trgeo.2018.02.003
2214-3912/Ó 2018 Elsevier Ltd. All rights reserved.
G. Sabih, R.A. Tarefder / Transportation Geotechnics 15 (2018) 20–28 21
can result in transverse cracking of slabs depending on the joint version 2.3 to evaluate the impact of level 1 and level 3 material
spacing. Thermal effects also impact slab bending and curling inputs on JPCP performance predictions.
and when joints/edges are curled upwards, they do not have full
contact with the base and are subject to cracking under traffic Objectives
loading. This could be particularly significant for long, thin slabs
under heavy, frequent loading. Sabih and Tarefder [4–6] investi- Testing and evaluation of concrete for CTE, MOR and elastic
gated the effects of variability of mechanical and thermal proper- modulus to develop a database to be used with pavement ME
ties of concrete on performance predictions of concrete design for design of rigid pavements in New Mexico. Moreover,
pavements and found that concrete properties such as elastic mod- to develop level 2 interconversion correlations between com-
ulus, flexural strength and coefficient of thermal expansion signif- pressive strength, MOR and elastic modulus.
icantly affect the performance of concrete pavement over the To conduct the sensitivity analysis of these material input levels
pavement design life including transverse cracking, joint faulting on rigid pavement performance.
and pavement roughness.
The process of design and performance prediction of concrete
Casting of concrete specimens (cylinders and beams)
pavements was mostly empirical until the advent of Pavement
ME design procedure which was originally developed under the
This study consisted of laboratory testing of 5 concrete paving
NCHRP Project 1-37A. This procedure was eventually adopted by
mixes from different districts of New Mexico prepared with differ-
AASHTO as the standard for pavement design and AASHTO made
ent coarse aggregates. Concrete cylinder and beam specimens were
standard guidelines for agencies to implement the procedure and
casted from these mixes to determine compressive strength, elastic
perform local calibration [7]. According to the survey conducted
modulus, MOR and CTE. The test results provided pavement engi-
in 2014 by NCHRP to determine the implementation status of
neers with level 1 inputs to be used in rigid pavement design with
Pavement ME design, it was found that most of the states are using
Pavement ME design software for these specific mixes and level 2
the default input values as they are in the process of local calibra-
correlations were produced for interconversion of compressive
tion and implementation [8]. Pavement analysis and design can
strength into elastic modulus and MOR for New Mexico paving
currently be performed using the software program commonly
mixes. Forty cylindrical specimens and twelve beam specimens
referred to as Pavement ME design. It is based on mechanistic-
were prepared from each of the five mixes. These mixes are indi-
empirical design concepts, which means that the design procedure
cated as CA-ID-1 to CA-ID-5 for data composition and analysis pur-
calculates pavement responses such as stresses, strains, and deflec-
poses. The detail of coarse aggregate mineralogy of these mixes is
tions under axle loads and climatic conditions and then accumu-
listed in Table 1.
lates the damage over the design analysis period. The procedure
then empirically relates calculated damage over time to pavement
distresses and smoothness based on performance of actual projects Details of mix design
throughout the U.S. Pavement ME design performs a wide range of
analysis and calculations in a rapid, easy-to-use format [9]. The concrete mixes consisted of different coarse aggregates
According to the previous research, accurate determination of with different mix proportions and mix properties. The details of
concrete material properties enhances the accuracy of the designed all the tested mixes are tabulated in Table 2. The concrete speci-
pavement. The variability of CTE, MOR and elastic modulus mens were prepared in the field and transported to the pavement
impacts the performance indicators of JPCP such as faulting, rough- laboratory at UNM, after an initial setting, and placed in tempera-
ness and cracking. These key parameters can be determined for ture controlled curing tanks at 75 °F for final curing. All of the pav-
each paving mix through laboratory tests. They are used by the ing mixes considered in this study contains fly ash as a
structural response models and performance prediction models supplementary cementitious material, which contributes to the
for damage calculations and performance predictions. One of the properties of fresh and hardened concrete. Fly ash improves the
features of the Pavement ME Design is its ability to use default, workability, reduces the water demand and reduces the heat of
regional, or site-specific values for materials data inputs which hydration of plastic concrete. It also increases the ultimate
are regarded as level 3, 2, and 1 respectively. These levels of inputs strength and durability and reduces the permeability of hardened
define the accuracy of the output in the form of designed pave- concrete.
ment. Level 1 inputs comprise of the laboratory tested data of
the specific paving mix while level 2 inputs are the values based Compressive strength and elastic modulus testing
on the default interconversion models and the values obtained
from the average of the constituents of concrete. Level 3 has lowest Compressive strength testing and results
accuracy in design as it comprises of the software default values.
This study encompasses the testing of New Mexico paving mixes The compressive strength of concrete cylinders (4 8 in.) was
to generate level 1 input data of these parameters and refine the determined per ASTM C39-16 [10] and AASHTO T-22-14 [11] at
level 2 models correlating compressive strength, elastic modulus, the age of 7, 14, 28 and 90 days. This test method covers determi-
and MOR. Simulations will be conducted in Pavement ME design nation of compressive strength of cylindrical concrete specimens
Table 1
Summary of concrete mixes with coarse aggregate mineralogy.
Mix ID Coarse aggregate source Coarse aggregate mineralogy Fine aggregate Location
CA-ID-1 Dark Canyon Un known Grand Falls Hobbs
CA-ID-2 Steele pit Granite Steele pit Clovis
CA-ID-3 Tinaja Limestone Tinaja Grants
CA-ID-4 Placitas Quartzite Placitas Santa Fe
CA-ID-5 Avispa Limestone Dryer Vado
22 G. Sabih, R.A. Tarefder / Transportation Geotechnics 15 (2018) 20–28
such as molded cylinders. The testing was conducted on 3 replicate Elastic modulus testing and results
specimens for each age group for each paving mix. The comparison
of compressive strength results is shown in Fig. 1. The results are Concrete cylindrical specimens were tested per ASTM C469-14
consistent with a constantly increasing trend with concrete age. [13], which covers determination of chord modulus of elasticity
The 28-day compressive strength values range from 4422 to (Young’s modulus) of concrete cylinders. The cylindrical specimens
7694 psi whereas the strength at 14 days range from 3658 psi to of all the mixes were tested for elastic modulus at the age of 7, 14,
6482 psi. These values far exceed the minimum NMDOT specifica- 28 and 90 days. The testing was conducted on 3 replicate speci-
tion requirement of 3000 psi. The average 28-day PCC compressive mens for each age group for each paving mix. The comparison anal-
strength values in the long-term pavement performance (LTPP) ysis of these results is shown in Fig. 2. The 28-day elastic modulus
database range from 3034 to 7611 psi with an average value of values range from 4.29 to 6.49 106 psi, with an average of 5.4
5239 psi [12]. These NMDOT paving mixes have a higher than aver- 106 psi. These values can be considered high relative to the average
age compressive strength value compared to the mixes used in the value of 4.38 106 psi corresponding to LTPP sections used in the
sections included in the national calibration of the ME design national calibration of rigid pavement models for ME design.
models.
Table 2
Details of mix designs.
00:5
Ec ¼ q1:5 33 f c ð1Þ
00:5
Ec ¼ 57000 f c ð2Þ
With the help of tested data of five mixes i.e. CA-ID-1 to 5, two
inter-conversion models have been proposed for these paving
mixes. First model is 0.5 power model similar to ME design model,
designated as Model-1, and the second model is power model
shown as model-2. The regression analysis of these two models
is presented in Figs. 3 and 4 respectively.
The two models are presented as follows:
00:5
Ec ¼ 67290 f c Model-1
00:7464
Ec ¼ 7986:8 f c Model-2
Fig. 4. Power model for elastic modulus.
The proposed 0.5 power model has a R2 (coefficient of determi-
nation) value of 0.77 and the power model gives R2 value of 0.88,
which shows that both the models fits the data well but the power Modulus of rupture testing of concrete beams
model works better than 0.5 power model.
Testing methodology
Analysis of proposed models This test is conducted per the specifications of ASTM C78-15
[14] and AASHTO T 97-14 [15]. This test method covers the deter-
The analysis of the predicted elastic modulus values was carried mination of the flexural strength of concrete using a simple beam
out with the measured values and the results have been plotted as with third-point loading.
shown in Fig. 5, which clearly shows that the proposed models
gives better results for NMDOT mixes when compared to the ME
design default model. In comparison of the two proposed models, Test results and discussion
the power model works better as the predicted values from this
model are closer to the measured elastic modulus values as com- The concrete beams casted from all the mixes were tested for
pared to the 0.5 power model. flexural strength at the age of 7, 14, 28 and 90 days. The testing
The density of the paving mixes used in this study ranges was conducted on 3 replicate specimens for each age group for
between 137.1 to 144.7 lb/ft3 and the multiplying factor in the each paving mix. The comparison of average values of flexural
default model (Eq. (2)) for these mixes comes in the range of strength is plotted in Fig. 6. The results are consistent with a con-
52,900 to 57,400 with these values of densities. Whereas, the pro- stantly increasing trend and flexural strength increases with con-
posed 0.5 power model has the factor of 67,000. Thus, it is evident crete age. The 28-day flexural strengths are in the range of 665
that the default model will under predict the elastic modulus val- psi to 963 psi with an average value of 833 psi. These values are
ues of these mixes. characteristic of high strength mixes. A typical 28-day PCC flexural
The standard error of estimate (SEE) was determined for all of strength value used in rigid pavement design is 650 psi. The target
the interconversion models. The SEE values for power model, 0.5 flexural strength of the LTPP sections, which represent the newly
power model and default model were 0.35 106, 0.49 106 and constructed rigid pavement experiments nationwide, was 550
0.90 106 respectively. It is evident that the proposed power psi. The 28-day flexural strength values reported in the LTPP data-
model has the lowest value of SEE which again proves that it is base range from 489 to 1006 psi with an average of 735 psi [12]. It
the best model for the prediction of elastic modulus for these pav- is evident that these paving mixes have higher than average 28-
ing mixes. day flexural strengths.
24 G. Sabih, R.A. Tarefder / Transportation Geotechnics 15 (2018) 20–28
clearly shows that the proposed model gives better results for Coefficient of thermal expansion testing of concrete cylinders
NMDOT mixes when compared to the ME design default model.
The standard error of estimate (SEE) was determined for all the The concrete cylinders were tested for determination of CTE per
interconversion models. The SEE values for power model, 0.5 AASHTO T 336-11 [16]. This method determines the CTE of a cylin-
power model and default model were 48, 53 and 103 respectively. drical concrete specimen maintained in a saturated condition. The
It is evident that the proposed power model has the lowest value of testing was conducted on 2 replicate specimens for each paving
SEE which again proves that it is the best model for the prediction mix. The comparison of the results at the age of 28 days is shown
of MOR for these paving mixes. in Fig. 10. The CTE property in NMDOT mixes vary over a fairly
large range. The CTE values range from 3.71 to 5.39 106 in/in/
°F. The CTE of CA-ID-1, 2 and 4 are consistently higher than that
of CA-ID-3 and CA-ID-5 which are 100% limestone. This confirms
the findings from the literature that limestone has the lowest
CTE value as compared to other minerals. The impact of coarse
aggregate on the CTE values is evident with these results. For the
tested mixes the standard deviation values for the test data (same
lab, same mix design) are within 0.1 106 in/in/°F which shows
excellent repeatability.
The comparison of CTE test data for NMDOT mixes with the ME
design default data was conducted. The average CTE values for con-
crete with limestone, quartzite, and granite are different than the
Fig. 8. Power model for MOR prediction. average values of ME default data. The average value for concrete
ing between the results of level 1 and level 3 inputs. The change in spond to the difference in transverse cracking as IRI is dependent
transverse cracking with input levels is 39–54% which is highly sig- on transverse cracking parameter along with other factors. These
nificant. With these results, it is evident that the pavement must be results necessitate the importance of using level 1 inputs while
designed with the accurately tested level 1 inputs for the paving designing a concrete pavement. These results also confirm the
mix to be used so that the designed pavement can last for the requirement of generating a database of CTE values and the inter-
entire service life. The ME default CTE data will not produce accu- conversion models for compressive strength, elastic modulus and
rate design for NMDOT paving mixes. MOR for the paving mixes to be used in New Mexico.
is also confirmed with the experimental data. Furthermore, it is [4] Sabih G, Tarefder RA. Impact of variability of mechanical and thermal
properties of concrete on predicted performance of jointed plain concrete
confirmed that the default interconversion correlations need to
pavements. Int J Pavement Res Technol 2016:1–13.
be adjusted for locally used paving mixes for accurate rigid pave- [5] Sabih G, Tarefder RA. Effects of concrete stiffness on mechanistic-empirical
ment design. performance of unbonded jointed plain concrete overlay. In: Proc of the
international conference on highway pavements & airfield technology,
Philadelphia, PA, USA, Aug 27–30; 2017. p. 25–32.
Acknowledgements [6] Sabih G, Tarefder RA. Effect of CTE, MOR and Elastic Modulus on Performance
of Rigid Pavement by MEPDG Simulation. In: Proc of the 4th Geo-China
This study is part of a project funded by the NMDOT. The International Conference, July 25 to 27; 2016, Shandong, China. p. 86–93.
[7] AASHTO. Mechanistic – Empirical pavement design guide, A Manual of
authors would like to express their gratitude to Jeffrey Mann for Practice, American Association of State Highway and Transportation
being the Advocate of this project and David Hadwiger for being Officials; 2008.
Project Manager. Special thanks to Parveez Anwar, Sean Brady [8] NCHRP. Implementation of the AASHTO mechanistic-empirical pavement
design guide and sofware. National Cooperative Highway Research Program,
and Naomi Gaede for their help and support in execution of this Synthesis 457, Transportation Research Board, Washington, D.C.; 2014.
project. The help of the pavement research group at UNM in col- [9] Mallela J, Titus-glover L, Bhattacharya B, Darter M, Von Quintus H. Idaho
lecting concrete mixes and specimens’ preparation is greatly AASHTOWare Pavement ME Design User’s Guide; 2014.
[10] ASTM C39/C39M–12a. Standard test method for compressive strength of
appreciated. cylindrical concrete specimens. In: ASTM International, 100 Barr Harbor Drive,
PO Box C700, West Conshohocken, PA 19428–2959, United States; 2012.
Appendix A. Supplementary material [11] AASHTO T-22-14. Standard method of test for compressive strength of
concrete specimens. American Association of State Highway and
Transportation Officials, 444 North Capitol Street N.W., Suite 249,
Supplementary data associated with this article can be found, in Washington, D.C., 20001; 2014.
the online version, at https://doi.org/10.1016/j.trgeo.2018.02.003. [12] Rao C, Varner R, Barstis W. PCC material characterization in mississippi for
aashtoware pavement ME design. Annual meeting of Transportation Research
Board; 2016.
References [13] ASTM C469/C469M–10. Standard Test Method for Static Modulus of Elasticity
and Poisson’s Ratio of Concrete in Compression. ASTM International, 100 Barr
[1] Schwartz C, Li R, Kim SH, Ceylan H, Gopalakrishnan K. Sensitivity evaluation of Harbor Drive, PO Box C700, West Conshohocken, PA 19428–2959, United
MEPDG performance prediction. National cooperative highway research States; 2010.
program research results digest, RRD 372, Washington, D.C.: National [14] American Society for Testing and Materials. Standard test method for flexural
Research Council; 2011. strength of concrete (Using Simple Beam with Third-Point Loading) 1, vol.
[2] Tanesi J, Kutay ME, Abbas A, Meininger R. Effect of coefficient of thermal C78–02, No. C; 2010.
expansion test variability on concrete pavement performance as predicted by [15] AASHTO T 97–14. Standard method of test for flexural strength of concrete
mechanistic-empirical pavement design guide. Transp Res Rec specimens. American Association of State Highway and Transportation
2008;2020:40–4. Officials, 444 North Capitol Street N.W., Suite 249, Washington, D.C., 20001;
[3] Hein DK, Sullivan S. Concrete Coefficient of Thermal Expansion (CTE) and its 2014.
significance in mechanistic-empirical pavement design. Advances in Pavement [16] AASHTO T 336–11. Coefficient of thermal expansion of hydraulic cement
Evaluation and Instrumentation of the 2012 Annual Conference of the concrete. American Association of State Highway and Transportation Officials,
Transportation Association of Canada Fredericton, New Brunswick; 2012. 444 North Capitol Street N.W., Suite 249, Washington, D.C., 20001; 2011.