Performance Checks for Unbound Aggregate Base Permanent ...

Procedia Engineering Volume 143, 2016, Pages 979–990 Advances in Transportation Geotechnics 3 . The 3rd International Conference on Transportation Geotechnics (ICTG 2016)

Performance Checks for Unbound Aggregate Base Permanent Deformation Prediction Models under Dynamic Stress States Induced by Moving Wheel Loading Yuanjie Xiao1 and Erol Tutumluer2* 1

Department of Geotechnical Engineering, School of Civil Engineering, National Engineering Laboratory for High-speed Railway Construction, Ministry of Education Key Laboratory for Heavyhaul Railway Engineering Structures, Central South University, Changsha, Hunan, China 2 Department of Civil and Environmental Engineering, University of Illinois at UrbanaChampaign, Urbana, Illinois, U.S. [email protected], [email protected]

Abstract Permanent deformation (rutting) is the most critical load-associated distress that develops in unbound aggregate layers significantly affecting their performance. Despite past research work has focused on estimating rutting of unbound aggregates using a variety of prediction models, few of them can be applied with adequate confidence to actual field conditions, i.e., dynamic stress states due to moving wheel loads and varying aggregate source properties. This paper aimed to evaluate these rutting models by comparing computed permanent deformation to that measured in advanced laboratory repeated load triaxial (RLT) tests. The data source analyzed was from a series of laboratory RLT tests conducted to characterize two crushed aggregate materials for their permanent deformation trends when subjected to moving, complex gear loading of next generation aircraft. Both constant and variable confining pressure tests were performed using an advanced RLT testing device to clearly account for the effects of dynamic stress states, including stress ratio, stress magnitude, and stress path loading slope (representing rotating principal stress directions) on permanent deformation accumulation. The applicability of several commonly used unbound aggregate rutting models to dynamic stress states was evaluated using this comprehensive laboratory database. The comparison of model predictions against laboratory RLT test results indicated the need for further enhancement of these models. Finally, rational modifications were suggested to improve their predictive ability, which would provide an improved analysis and design of flexible pavements for the rutting distress. Keywords: Unbound aggregates, permanent deformation, rutting, repeated load triaxial tests, dynamic stress states, moving wheel loading *

Corresponding author. Tel.: +1 (217) 333-8637

Selection and peer-review under responsibility of the Scientific Programme Committee of ICTG 2016 c The Authors. Published by Elsevier B.V. 

doi:10.1016/j.proeng.2016.06.087

979

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

1 Introduction Rutting or accumulation of permanent deformation is the primary distress mechanism of unbound aggregate base/subbase layers in pavements. Accordingly, rutting resistance is a major performance measure for designing these pavement foundation layers. Field experiences, such as those from the AASHO road test, Minnesota Road Research test facility (Mn/ROAD), and the National Airport Pavement Test Facility (NAPTF) among many others, have clearly shown that granular base/subbase permanent deformation may contribute significantly to the overall flexible pavement surface ruts (Christopher et al., 2006; Dai et al., 2007; Kim and Tutumluer, 2006). To ensure satisfactory pavement performance and longevity, unbound aggregate layers should be properly designed to possess adequate rutting resistance, especially for unsurfaced or thinly surfaced pavements where nonlinear and inelastic behavior induced by dynamic wheel load stresses become quite crucial. Despite being a highly researched topic in the past few decades, realistic characterization and accurate prediction of rutting accumulation still remain challenging with great complexity arising from both traffic and environmental loading aspects. Better understanding of long-term rutting mechanism as well as more accurate and reliable rutting models become urgent needs in the design of new pavements, especially those built with locally available marginal quality and/or recycled materials possessing greater performance uncertainty than traditional “standard” materials with proven field performance (Xiao et al., 2012). Both experimental and field results indicated that permanent deformation response depends strongly on stress level and material state conditions (e.g., water content and density). Aggregate type, quality, and physical properties are long realized as individual factors to greatly influence the unbound pavement layer rutting performance, yet the underlying mechanisms of their interactions with in-situ stress states controlling field rutting development have not been fully explored. Mishra and Tutumluer (2012) studied a factorial combination of aggregate type, particle shape and surface texture, type and amount of fines, and moisture and density in relation to required compaction conditions for their effects on mechanistic pavement response and rutting performance through both laboratory repeated load triaxial (RLT) and field accelerated pavement testing (APT). They found that improved behavior in terms of both permanent deformation and resilient modulus could be achieved by maintaining a stable aggregate matrix (e.g., optimal fines contents); meanwhile, the stability of the aggregate matrix also depends on the interactions among individual aggregate physical properties such as moisture and plastic (cohesive) fines content, defined in this paper as the amount of material passing sieve No. 200 or 0.075 mm. Gabr and Cameron (2013) studied permanent strain behavior of recycled concrete aggregate for unbound pavement construction and related the accumulated permanent strain to moisture content and maximum dry density corresponded to the optimum, weighted plasticity index, and maximum shear stress ratio. Empirical tests such as California Bearing ratio (CBR) cannot truly assess rutting performance; instead, rutting resistance should be assessed from tests that realistically mimic actual traffic (moving wheel) loading encountered in the field. Although repeated load triaxial (RLT) test procedures, such as the AASHTO T307-99, have been established for determining resilient modulus of granular materials, there is currently no standard test procedure in the U.S. for evaluating permanent deformation or field rutting potential of unbound aggregate materials. The need to include permanent deformation aspects along with the resilient properties, when characterizing unbound granular materials, has recently been confirmed. High moduli materials could still undergo severe plastic strain or permanent deformation, indicating that relying on resilient properties alone may not prevent excessive deformation or failure. Further, stress path RLT testing procedures are in fact needed to properly account for moving wheel loading and its effects on permanent deformation in pavement granular layers. Tutumluer et al. (2001, 2005, 2007) considered in laboratory testing the dynamic nature of moving wheel loads. They found that greater rutting damage could occur especially in the loose base/subbase under the extensioncompression-extension type rotating stress states induced by a moving wheel pass. This kind of

980

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

laboratory testing result is consistent with previous field findings by Hornych et al. (2000) that permanent deformations observed under moving wheel load conditions can be up to three times higher than those obtained from plate loading tests. Gräbe and Clayton (2003, 2009, 2014) also concluded that principal stress rotation (PSR) caused by moving wheel loads has a significant and deleterious impact on both resilient moduli and permanent deformation of some types of road and track foundation materials, i.e., reduced resilient moduli and increased rate of permanent strain. Research using a hollow cylinder triaxial apparatus (Chan, 1990) has shown that permanent strains increase significantly when a rotation of the principal stresses is applied (for the same principal stress amplitude). Evident from the reasons outline above, further modeling of the plastic strain response of unbound granular materials (UGMs) with respect to complex physical conditions and dynamic stress states is indispensable for improvements with the mechanistic based design of conventional flexible pavements. Based on results from laboratory, either single-stage or multistage, repeated load tests and/or field APT testing, a wide range of models have been proposed to predict unbound aggregate permanent deformation accumulation due to repeated loading. A comprehensive survey of those rutting models was given by Bodin (2013). The majority of those rutting models is empirical in nature as they relate rut depth (or permanent strain) with the number of load applications, common material physical properties, and applied stress states. Their applicability is often limited to materials and test conditions similar to those for the model development because no fundamental material property or true rutting mechanism is included in the model. Therefore, there is still much need of a unified rutting prediction model in a sense that it can be regarded as truly universal and adapted with confidence for a much wider range of materials and conditions, as well as the need to establish the relationships between rutting accumulation and various strength theories.

2 Objective and Scope The primary objective of this paper is to evaluate the applicability of several commonly used unbound aggregate rutting models to field stress states induced by moving wheel loads and examine if they can be applied with greater confidence to a wider range of conditions, i.e., the original Tseng and Lytton (1989) model, the Mechanistic-Empirical Pavement Design Guide (MEPDG) rutting model (ARA, 2004), the Gidel et al. (2001) model, the Korkiala-Tanttu (2008) model, and the Xiao et al. (2015) model. To achieve this purpose, such model predicted permanent deformation results were compared against those measured in advanced laboratory repeated load triaxial (RLT) tests, respectively. The scope of the paper deals with utilizing a comprehensive aggregate mechanistic property database recently compiled at the University of Illinois and including resilient modulus, shear strength, and permanent deformation test data for two crushed aggregate materials used in airport pavement foundation layers, namely the P209 granular base and P154 subbase materials. The scope of the paper covers first the unbound aggregate mechanistic property database used in this study and discusses the deficiency of current MEPDG rutting prediction model for unbound granular materials. The applicability of selected permanent deformation prediction models for unbound aggregates to dynamic stress states is assessed next. Finally, the need to develop a stable permanent strain accumulation rate model is emphasized and its potential use is discussed for longterm rutting evaluation.

3 Unbound Aggregate Mechanistic Property Database The data source of unbound aggregate permanent deformation results of laboratory RLT tests was from the FAA’s National Airport Pavement Test Facility (NAPTF) research study (Kim and

981

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

Tutumluer, 2006). Two crushed aggregate materials, specified herein as the P209 granular base and P154 subbase materials, were tested for permanent deformation trends in the laboratory to study effects of moving wheel loads and the degree of compaction. The P209 material was a crushed limestone and the P154 material consisted of manufactured screenings specified by FAA’s Advisory Circular 150/5370-10B. The P209 aggregate material was classified as A-1-a according to AASHTO procedure and as GP-GM (poorly graded gravel with silt) according to ASTM procedure; whereas the P154 aggregate material was classified as A-1-b according to AASHTO procedure and as SW-SM (well graded sand with silt) according to ASTM procedure. The P209 has much less amount of fines fraction than the P154. Material properties, stress path RLT permanent deformation test programs, and corresponding laboratory test results for this study were detailed elsewhere (Kim, 2005). A pavement element constantly experiences a combination of varying magnitudes of static and dynamic stresses depending on the depth in the pavement layer and the radial offset from the wheel load. The principal stress rotation and the constantly rotating fields of stresses under moving wheel loads have been observed from field measurements. In the case of aggregate bases, the cyclic component of load imposes a change (increment) of stress state which is not co-axial with the stress state under the static (overburden) load. To mimic such stress states, both constant confining pressure (CCP) and variable confining pressure (VCP) tests were performed using an advanced repeated load triaxial testing device named UI-FastCell to clearly account for the effects of variable stress states, including stress ratio, stress magnitude, and stress path loading slope (representing rotating principal stress directions) on permanent deformation accumulation for the FAA P209 and P154 aggregate materials. As illustrated in Figure 1(a), in the CCP tests, vertical wheel load stresses were repeatedly applied on laboratory specimens and the effects of applied stress magnitudes and stress ratios (vertical to horizontal stress) on the permanent deformation accumulation were studied to adequately simulate the conditions realized directly under the wheel where the highest applied stress ratios are experienced in the field. In contrary to the CCP tests, the variable confining pressure (VCP) tests were performed by subjecting the specimens to various constant stress path loadings realized under actual traffic. Specifically, two dynamic stresses of prescribed magnitudes were pulsed simultaneously in both horizontal and vertical directions, in addition to the applied constant hydrostatic confining stress. Different combinations of the stress path slope (m) and its length (L) were therefore applied to simulate the various dynamic loading conditions experienced under the moving wheels. Shear strength properties of the two materials were also determined from rapid shear strength tests. The P209 aggregate had a friction angle of 61.7o with a cohesion intercept of 132 kPa and the P154 material had a friction angle of 44o with a cohesion intercept of 182 kPa. Figure 1(b) illustrates the actual stress path loading scheme applied for both P209 and P154 aggregate materials in the laboratory. The total number of load applications is 10,000 for P209 aggregate material and 40,000 for P154 aggregate material, respectively.

4 Evaluating Unbound Aggregate Rutting Models for Dynamic Stress States Under the application of moving wheel loads, unbound pavement layers are constantly subjected to dynamic stress states. That is, not only changes the magnitude of the load, but also the direction of the principal axes of the stresses rotates during the approaching and the departure of moving wheels. In this section, the original Tseng and Lytton model (1989) and the nationally calibrated MEPDG rutting model (ARA, 2004), together with the models by Gidel et al. (2001) and Korkiala-Tanttu (2008), were verified against laboratory RLT test results obtained using both CCP and VCP tests performed. The first two models correlate rutting in the unbound layers inversely with resilient modulus, resulting in the general misunderstanding that a high modulus is regarded to lead to high rutting resistance. The

982

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

universal constants of the MEPDG rutting model, as well as its general applicability to a wide range of materials under complex stress states, need to be validated further for it to become a truly universal model; otherwise, it cannot be expanded to unbound granular materials with confidence.

4.1 MEPDG Rutting Model The original Tseng and Lytton model (1989) is formulated as Equation 1 where the regression coefficients ε0, β, and ρ, regarded as material constants, were obtained from laboratory RLT tests by fitting permanent strains against the number of load applications N. Empirical relationships were also developed to estimate those material constants from water content, bulk stress and resilient modulus. The Tseng and Lytton model was subsequently modified and calibrated by El-Basyouny et al. (2005) for implementation into the current MEPDG (ARA, 2004). As shown in Equation 2, it normalizes the predicted permanent strain to the resilient strain and thus requires knowledge of resilient modulus. E H p  N H 0 e - U N

§H · log ¨ 0 ¸ © Hr ¹

0.80978 - 0.06626Zc - 0.003077V T  10-6 Er

R

2

R

0.60



log E

-0.9190  0.03105Zc  0.001806V T -1.5 u 10-6 Er

log U

-1.78667  1.45062Zc - 3.784 u 10-4 V T2 - 2.074 u 10-3 Zc2V T -1.05 u 10-5 Er

R

2

0.66

Hp N Hr

§ H0 · ¨ ¸ © Hr ¹ log E

2

0.74

(1)



§ H0 ¨ © Hr

§U·

· -¨© N ¸¹ ¸e ¹

E

§ § U ·E § e U E u a u E b1 ·  ¨ e¨© 109 ¸¹ u a u E b9 ¨ ¸ ¨ 1 9 © ¹ ¨ © 2

· ¸ ¸ ¸ ¹

§ § U ·E · § e U E u 0.15 ·  ¨ e¨© 109 ¸¹ u 20 ¸ ¨ ¸ ¨ ¸ © ¹ ¨ ¸ © ¹ 2

-0.61119 - 0.017638Zc

U

§ ¨ C0 109 u ¨ ¨ ª1- 109 ¨« ©¬

C0

§ a u E b1 ln ¨ 1 ¨ a u E b9 © 9

 · ¸¸ ¹

1

·E ¸ ¸ Eº ¸ »¼ ¸¹

(2) 1

§ ·E ¨ -4.89285 ¸ ¸ 109 u ¨ ¨ ª1- 109 E º ¸ ¨« »¼ ¸¹ ©¬



-4.89285

where ωc is gravimetrical moisture content (%); σd and σθ are deviator stress (psi) and bulk stress (psi), respectively; ε0, β, and ρ are material constants; εr and Er are resilient strain and resilient modulus (psi), respectively; and a1=0.15, b1=0, a9=20, and b9=0 are the fitting parameters finally selected by the Arizona State University (ASU) researchers after a considerable study to provide the best prediction for a wide range of unbound material types. However, Hashem and Zapata (2013) recently found that the use of a9=50 yielded better prediction results as compared to a 9=20 adopted in the current MEPDG model. They also developed new prediction equation for β by replacing resilient modulus with the shear stress to strength ratio.

983

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

Permanent deformation data from laboratory RLT tests, as described previously, were first analyzed to obtain the accumulated permanent strain due to different loading along multiple stress paths, respectively. Both permanent strain models were used to calculate the accumulated permanent strain after the corresponding number of load applications, respectively. The model parameters ε 0, β, and ρ, whose values were not shown herein due to space limitation, were solved for based on the least square error between the measured and calculated permanent strains. High R2 values close to 1 were obtained, thus indicating that the Tseng and Lytton model form can fit very well the permanent deformation curves obtained at each dynamic stress state. The regressed parameters ε 0/εr, β, and ρ were then compared with those estimated from the original Tseng and Lytton’s empirical relationships (see Equation 1) and with the universal constants of the MEPDG model (see Equation 2), respectively. Note that the original regression coefficients in Tseng and Lytton’s empirical relationships (Equation 1) provided poor estimates for model parameters ε0/εr, β, and ρ in terms of very low R2 values; therefore, they were re-calculated by regressing fitted parameters ε0/εr, β, and ρ against the same independent variables in the original relationships (Equation 1), i.e., moisture content, bulk stress and resilient modulus, respectively. The relationships between the measured and predicted values of ε0/εr, β, and ρ are plotted in Figures 2 and 3 for FAA P154 and P209 materials, respectively.

(a)

(b)

Figure 1: (a) Schematic representation of stress path loading; (b) stress path loading scheme of laboratory repeated load triaxial permanent deformation tests for FAA NAPTF aggregate materials

984

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

(a)

(b)

(c)

(d)

Figure 2: Relationship between laboratory measured and predicted values of (a) log(ε 0/εr), (b) logβ and (c) logρ by the Tseng & Lytton model form and (d) by the MEPDG model for P154 material

It can be seen from both Figure 2 and Figure 3 that low R2 values were overall observed for all three stress path slopes by using the Tseng and Lytton’s empirical relationships, even though the regression coefficients were re-calculated. This indicates that the model parameters of Tseng and Lytton’s permanent strain model vary significantly with applied stress states along different stress paths. As contrary to current practice, relying on resilient modulus and bulk stress alone may not be sufficient enough to provide accurate prediction of rutting accumulation, especially for in-situ moving wheel load conditions. Note that the current MEPDG permanent strain model, simplified from the original Tseng and Lytton’s model, eliminates the dependence of model parameters on stresses and resilient moduli; instead, they are only dependent on moisture content. This explains why the same model parameters were obtained regardless of applied stress states. As shown in Figure 2 and Figure 3, such simplification apparently results in poor permanent deformation prediction that contradicts laboratory RLT test results.

4.2 Rutting Models by Gidel et al. (2001), Korkiala-Tanttu (2008), and Xiao et al. (2015) The major motivation of developing a reliable rutting model is to predict permanent deformation accumulation with a high degree of confidence no matter what the applied stress states are in the unbound aggregate layer and whether or not these stresses are due to stationary or moving wheel loads. The aforementioned FAA database consists of permanent deformation results that were obtained at different dynamic stress states. In this sub-section, two commonly used rutting models incorporating stress terms were assessed first for their general applicability to dynamic stress states. The first model assessed was the one suggested by Gidel et al. (2001), as shown in Equation 3. It was

985

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

developed using repeated multi-stage triaxial test apparatus and uses the maximum deviator (qmax) and mean (pmax) stresses to compute the permanent strain. This model explicitly combines the effects of stresses and number of cycles on the axial permanent strain. The second model evaluated has a similar formulation as the first one and is proposed by Korkiala-Tanttu (2008). According to this model, the development of permanent deformation is directly related to the distance from the stress point to the Mohr-Coulomb failure line expressed in q-p space and it also relates the effect of stress on the permanent deformation through a hyperbolic function (see Equation 4). This model accounts for the effect of the number of cycles through the log-log approach suggested by Sweere (1990). This model is given by Equation 4. In addition to these two models, another one recently proposed by Xiao et al. (2015) as a unified rutting model (see Equation 5) was also evaluated that incorporated critical stress variables and essential material properties governing rutting behavior. It was clearly demonstrated that permanent deformation accumulation of unbound aggregate layers is dictated by the shear stress ratio and the working stress states to which pavement layers are subjected. The inclusion of the bulk stress term responsible for the volume change behavior is meant to represent the confinement level. As given by Equation 5, this eight-parameter Power-law model by Xiao et al. (2015) was denoted as “Proposed Model” in this study. Despite its applicability to varying aggregate source properties was verified against laboratory RLT test results of aggregate materials studied in Illinois Center for Transportation (ICT) research studies (Tutumluer et al., 2009; Mishra, 2012), the exploration of its applicability to complex dynamic stress states has not been pursued yet.

(a)

(b)

(c)

(d)

Figure 3: Relationship between laboratory measured and predicted values of (a) log(ε 0/εr), (b) logβ and (c) logρ by the Tseng & Lytton model form and (d) by the MEPDG model for P209 material

ª

§ N · ¸ © N0 ¹

H p  N E1H 0 «1  ¨ « ¬

986

B º

n

§L · § · q s » ¨ max ¸ ¨ m   max ¸ pmax pmax ¹ » © pa ¹ © ¼

1

(3)

Performance Checks for Unbound Aggregate Base Permanent ...

H p  N E1CN b

R A R

(4)

§ T · § Wf · d a¨ ¸ ¨ ¸ N © p0 ¹ © W max ¹ c

b

Hp

Xiao and Tutumluer

(5)

Wf ª T º ˜ EXP «e ˜ S  Sopt  f ˜ wPI  g ˜ wPI ˜  h ˜ S  Sopt ˜ » p0 ¼ W max ¬









2 2 where N 0 is the reference number of load applications (in this study=100), Lmax , pmax  qmax and n, B, m, s, and H 0 are material constants; C and b are material parameters, A is a parameter

independent of the material (A=1.05) and R is the deviatoric stress ratio given by m 6sin I  3  sin I

R

qmax qf

qmax s  m ˜ pmax

,

and s 6c ˜ cos I 3  sin I , I is the angle of internal friction, c is the cohesion of the

material, and qmax and pmax are the maximum deviatoric and hydrostatic stresses of the load application, respectively; H p is permanent axial strain (%), T is bulk stress,

Wf W max

is shear stress to

shear strength ratio, p0 is normalizing unit pressure (1 psi), N is the number of load applications, S is achieved saturation level in decimal, Sopt =optimum saturation level in decimal, wPI =weighted plasticity index, and a, b, c, d, e, f, g, and h=regression coefficients. The analysis was limited to permanent axial strain, as this is the most important for the estimation of rutting. Kim (2005) also developed CCP and VCP permanent deformation models using the same FAA database. According to Kim (2005), the R2 values of the developed VCP models for m = -1 (triaxial extension) were in general the lowest possibly due to the high noise and fluctuations in the recorded triaxial data; therefore, permanent deformation test results for m = -1 was not analyzed in this study. As an example, permanent deformation results of FAA P209 material at different CCP and VCP stress states were used for comparison. Note that the s and m values required for calculating R value in Equation 3 are s=19.46 psi and m=2.49 for P209 material, and s=54.36 psi and m=1.80 for P154 material according to the shear strength test results, respectively. Figure 4 plots the measured permanent strain values against those predicted by each of the three models. For all the permanent deformation curves combined, the R2 and Standard Error of Estimates (SEE) values for the model by Gidel et al. (2001) are 0.57 and 0.13, and 0.67 and 0.11 for the one by Korkiala-Tanttu (2008), respectively. Improved accuracy was obtained by the Proposed Model, as indicated by R 2=0.90 and SEE=0.06. It can also be observed from Figure 4 that the Proposed Model yielded predictions with all the permanent deformation curves combined that match the measured values more closely. However, it should be noted that the use of the classic power-law model form in the Proposed Model still limit its ability to fit well only the primary stage of the permanent deformation. To better fit the secondary stage of permanent deformation, the use of a more advanced two-stage model may become necessary, which is the goal of another separate paper forthcoming.

5 Summary and Conclusions A comprehensive laboratory repeated load triaxial permanent deformation database for unbound aggregates typically used in airfield pavement base/subbase layers was used to investigate the influence of in-situ dynamic stress states induced by moving, complex gear loading of next generation aircraft on the permanent deformation characteristics. The main objective of this study was to evaluate the feasibility of applying commonly used unbound aggregate rutting models to predict the permanent

987

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

deformation results of laboratory RLT tests that mimicked the in-situ dynamic stress states. The rutting models evaluated included the original Tseng and Lytton (1989) model, the MechanisticEmpirical Pavement Design Guide (MEPDG) rutting model (ARA, 2004), the Gidel et al. (2001) model, the Korkiala-Tanttu (2008) model, and the Xiao et al. (2015) model. Based on the comparison results of the original Tseng and Lytton model and the current MEPDG rutting model against laboratory RLT test results, the need for further enhancement of both models becomes obvious. To improve the predictive ability of both models, rational modifications were suggested that include the dependency of permanent deformation on the stress states and shear stress ratio. This would allow an improved analysis and design of flexible pavements against rutting distress. Among the three models that relate permanent strain to stress terms, the model by Xiao et al. (2015) appears to match laboratory test results more closely than the models by Gidel et al. (2001) and Korkiala-Tanttu (2008); however, all of them did not exhibit satisfactory accuracy in predicting permanent strains in the secondary stage. To this end, a more advanced two-stage model that incorporates essential rutting mechanisms and key variables representing dynamic stress states would need to be developed. Once well validated and calibrated, the application of this model in pavement design would allow the consequences of using either traditional aggregate materials or those not completely following material specifications to be quickly evaluated in terms of rutting distress, without performing timeconsuming, costly, and complicated RLT test procedures on a routine basis, especially for low-volume roads.

(a)

(b)

(c)

(d)

Figure 4: Measured cumulative permanent axial strain vs. predicted cumulative permanent axial strain by different models for P209 material at a loading slope of (a) m=3 and (b) m=1.5

988

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

Acknowledgements The database analyzed in this paper was collected from a previous study funded by FAA and conducted at the Center of Excellence for Airport Technology where the corresponding author served as the Principal Investigator (PI) and Dr. In Tai Kim worked as the primary research assistant then. The first author is also grateful to the financial support by the National Natural Science Foundation of China under Grant No. 51508577. The contents of this paper reflect the views of the authors who are responsible for the facts and the accuracy of the data presented herein. This paper does not constitute a standard, specification, or regulation.

References ARA. (2004). Guide for the mechanistic empirical design of new and rehabilitated pavement structure (Report 1-37A). Washington, DC: Transportation Research Board of the National Academies. Bodin, D. (2013). Feasibility study of using wheel-tracking tests and finite element modeling for pavement permanent deformation prediction. Austroads Publication No. AP-T228-13, Sydney, Australia, Austroads Ltd. Chan, F.W.K. (1990). Permanent deformation resistance of granular layers in pavements. PhD thesis, Dept. of Civ. Engrg., University of Nottingham, Nottingham, England. Christopher, B. R., C. Schwartz, and R. Boudreau. Geotechnical Aspects of Pavements. Publication NHI-05-037. National Highway Institute, FHWA, U.S. Department of Transportation, 2006. Dai, S., Boerner, D. and Isackson, C. (2007). Failure Analysis of Flexible Pavement Section on MnROAD. Transportation Research Board 86th Annual Meeting Compendium of Papers CD-ROM, Transportation Research Board, Washington D.C. El-Basyouny, M.M., Witczak, M., and Kaloush, K. (2005). “Development of the Permanent Deformation Models for the 2002 Design Guide.” In TRB 84th Annual Meeting Compendium of Papers CD-ROM, Transportation Research Board, Washington, D.C. Gabr, A.R. and Cameron, D.A. (2013). Permanent Strain Modeling of Recycled Concrete Aggregate for Unbound Pavement Construction. ASCE Journal of Materials in Civil Engineering, Vol. 25(10), pp. 1394-1402. Gidel, G., Breysse, D., Hornych, P., Chauvin, J.J., and Denis, A. (2001). A New Approach for Investigating the Permanent Deformation Behavior of Unbound Granular Material Using the Repeated Load Triaxial Apparatus. Bulletin des LCPC – 233 July – August. Gräbe, P.J. and Clayton, C.R.I. (2003). Permanent Deformation of Railway Foundations under Heavy Axle Loading. Proceedings of the International Heavy Haul Association, Specialist Technical Session, Dallas, Texas, USA. Gräbe , P.J. and Clayton, C.R.I. (2009). Effects of principal stress rotation on permanent deformation in rail track foundations. ASCE Journal of Geotechnical and Geoenvironmental Engineering, 135(4), pp. 555-565. Gräbe, P.J. and Clayton, C.R.I. (2014). Effects of Principal Stress Rotation on Resilient Behavior in Rail Track Foundations. J. Geotech. Geoenviron. Eng., 140(2), 04013010. Hashem, B. and Zapata, C.E. (2013). Enhancement of the Permanent Deformation Model for Unbound Materials Used by Darwin-ME. In CD-ROM of the 92nd Annual Meeting of TRB, Washington, D.C. Hornych, P., Kazai, A., and Quibel, A. (2000). Modeling A Full-Scale Experiment of Two Flexible Pavement Structures with Unbound Granular Bases. In Unbound Aggregates in Road Construction,

989

Performance Checks for Unbound Aggregate Base Permanent ...

Xiao and Tutumluer

Edited by A.R. Dawson, A.A. Balkema Publishers, Proceedings of the Unbound Aggregates in Roads (UNBAR5) Symposium, University of Nottingham, England, June 21-23, pp. 359-367. Kim, I.T. (2005). Permanent Deformation Behavior of Airport Flexible Pavement Base and Subbase Courses. Ph.D. Dissertation, Department of Civil and Environmental Engieering, University of Illinois, Urbana, IL. Kim, I.T. and E. Tutumluer. Unbound Aggregate Rutting Models for Stress Rotations and Effects of Moving Wheel Loads. Transportation Research Record: Journal of the Transportation Research Board, No. 1913, TRB, Washington, D.C., 2005, pp. 41-49. Kim, I.T. and E. Tutumluer. Field Validation of Airport Pavement Granular Layer Rutting Predictions. Transportation Research Record: Journal of the Transportation Research Board, No. 1952, TRB, Washington, D.C., 2006, pp. 48-57. Korkiala-Tanttu, L. (2008). Calculation Method for Permanent Deformation of Unbound Pavement Materials. VTT Publications 702, Espoo 2008. Mishra, D. (2012). Aggregate Characteristics Affecting Response and Performance of Unsurfaced Pavements on Weak Subgrades. Ph.D. Dissertation, Department of Civil and Environmental Engineering, University of Illinois, Urbana, IL. Mishra, D. and Tutumluer, E. (2012). Aggregate Physical Properties Affecting Modulus and Deformation Characteristics of Unsurfaced Pavements. ASCE Journal of Materials in Civil Engineering, 24(9), pp. 1144-1152. Sweere, G.T.H. (1990). Unbound Granular Bases for Roads. PhD Dissertation, University of Delft, Delft, The Netherlands. Tseng, K., and Lytton, R. (1989). Prediction of permanent deformation in flexible pavement materials. STP1016-EB, ASTM International, West Conshohocken, PA, pp. 154–172. Tutumluer, E., Ceylan, H., and Seyhan, Ü., (2001). Advanced Characterization of Granular Material Behavior Using Artificial Neural Networks. XVth International Conference on Soil Mechanics and Geotechnical Engineering, Istanbul, Turkey, August 27-31. Tutumluer, E., Mishra, D., and Butt, A. (2009). Characterization of Illinois Aggregates for Subgrade Replacement and Subbase. Final Report, Illinois Center for Transportation (ICT) R27-1 Project, University of Illinois Urbana-Champaign, Urbana, IL. Xiao, Y., Tutumluer, E., Qian, Y., and Siekmeier, J. (2012). Gradation Effects Influencing Mechanical Properties of Aggregate Base and Granular Subbase Materials in Minnesota. Transportation Research Record No. 2267, Washington, D.C., pp. 14-26. Xiao, Y., Tutumluer, E. and Mishra, D. (2015). “Performance Evaluations of Unbound Aggregate Permanent Deformation Models for Different Aggregate Physical Properties.” Transportation Research Record, In Press.

990