Introduction

Pavement structures are subjected to different levels of traffic loading with variable speeds and environmental changes that may lead to various distresses. Pavement performance mainly relies on material properties, traffic levels, climate conditions, and construction quality [1]. The mechanical properties of the unbound granular materials (UGMs) are affected by material type and its characteristics along with moisture content variations, which depend on the climate condition.

In recent years, the waste materials have received more attention from researchers and practitioners worldwide. The interest in these materials was necessitated by the population growth, which led to large consumption of natural resources. Thus, the use of recycled materials appeared promising from a wide variety of viewpoints, which increases the awareness to a greener environment. The waste materials include a wide range of excavated materials such as rock, soil, reclaimed asphalt pavement (RAP), bricks, concrete, plastic wastes, scrap tires, foundry sand, oil sand marble dust, and steel slag [2,3,4]. A great deal of research effort has been performed in this direction, and still there is ongoing research along with field studies for better understanding the behavior of these materials and increasing their utilization. Aside from benefits and encouraging research recommendations and practical results obtained from recycling waste materials as a substitution to natural materials, the lack of actual experience and some environmental issues have delayed the wide applications of such materials [4].

RAP is considered as one of the most recycled materials widely used in pavement construction. It is produced from the milling of the aged asphalt layers during rehabilitation, resurfacing, or pavement reconstruction projects, which increase every day. RAP has several applications, for example as a granular material in the base or subbase layers in paved and unpaved roadways, gravel roads rehabilitation, shoulders, bicycle paths, driveways, parking areas and as a fill material [5,6,7,8,9]. Extensive laboratory and field studies in the literature were conducted on the evaluation of RAP performance as unbound granular material either blended with virgin aggregate (VA) or alone. Table 1 summarizes different previous literature studies focused on characterizing properties of blends of RAP and virgin aggregates as base/subbase materials. Based on laboratory and/or field testing, these studies reported superior properties of the RAP/VA blends as compared to virgin aggregates. The studies also reported that RAP is a viable and cost-effective base/subbase material if blended with VA. Furthermore, some studies reported that California bearing ratio (CBR) decreases as the RAP amount increases [9, 11, 13, 17, 19, 22, 25]. However, the resilient modulus (Mr) has a opposite trend, which increases with the increase in RAP amount in the blend [5, 7, 9, 11, 12, 14,15,16,17, 20, 23, 24]. This is attributed to the binder in the RAP, which leads to better resilient behavior. The reasons for the different behaviors of CBR and Mr with the increase in RAP may be the high load applied during the CBR test, which is a failure test compared to the Mr test that facilitates the sliding of the asphalt-coated aggregates over each other leading to higher deformation. For the Mr, the addition of RAP to virgin aggregate mixes provides some kind of stabilization because of the presence of asphalt coating in the RAP aggregate, which makes the mixture stiffer at the low state of stress during the Mr test. Other studies discussed the performance of treated or untreated RAP/VA blend in terms of CBR, unconfined compressive strength (UCS) and Mr [26, 27].

Table 1 Summary of previous studies of RAP blends with virgin aggregate
Full size table

Few studies are available for the prediction of RAP performance by the Mechanistic Empirical Pavement Design Guide (MEPDG) software. Alam et al. [5] evaluated the effect of RAP amount as subbase (different percentages: 0%, 30%, 50%, 70%, and 100%) on pavement performance. The simulation results indicated that as RAP amount increased in the blend, the predicted fatigue cracking of the AC layer decreased. However, the predicted rutting of the subbase was slightly affected by the RAP content. The difference in rutting was found to be less than 0.05 in. (1.4 mm) between 0 and 100% RAP blends. In a similar study, Schwartz et al. [28] conducted sensitivity analyses of performance predictions to MEPDG design inputs under three traffic volume levels and five climatic conditions. The authors reported that the longitudinal and bottom-up fatigue cracks are very sensitive to the thickness of base material and Mr of both layers of subgrade and base.

Although the laboratory characterization of RAP as a pavement material was well investigated, limited studies are available on the prediction of pavement performance containing recycled products especially RAP as base material and the significance of RAP blends as base/subbase on rutting and fatigue performance. Furthermore, the license of MEPDG is too costly at least for developing countries like Egypt and it requires many inputs that may not be readily available at many agencies.

Research objectives

The main objective of this research is to build a framework for the prediction of field performance of RAP blends by incorporating the KENLAYER and quality-related specifications software (QRSS) along with the MEPDG sophisticated models which consider cost saving. The pavement response will be determined by designed Excel sheets that solve the sophisticated models of MEPDG in a simplified method.

Mechanistic Empirical Pavement Design Guide

The MEPDG is an advanced tool for the design and analysis of both new and rehabilitated flexible pavement structures [29]. It predicts the accumulated damage based on monthly/bimonthly changes of traffic, climate, and consequently construction material properties. After that, damage is converted into expected smoothness and pavement distresses using empirical transfer functions calibrated either globally or locally. For flexible pavements, performance is indicated in terms of rutting (total pavement rutting and individual layer rutting), top-down longitudinal cracking, bottom-up fatigue cracking and transverse (thermal) cracking [29].

MEPDG has three different hierarchical levels of inputs regarding traffic, materials, and environmental conditions. For material inputs, level 1 possesses the highest accuracy and reliability level and lowest error as the input values are obtained from direct laboratory measurements. In level 2, the input data are based on correlations with routine engineering tests to calculate material properties. Finally, default values based on experience are used as inputs for level 3 [29].

MEPDG distress prediction models

The most important pavement structural distresses are rutting, bottom-up fatigue cracking. In the following sections, MEPDG rutting and fatigue cracking models are detailed.

Rutting models

Total rutting is the summation of the permanent deformation of each layer of the pavement structure (i.e. hot mix asphalt, HMA layers, unbound layers, and subgrade soil). Table 2 displays the models (Eqs. 1–3) used in MEPDG for HMA rutting calculations [29]. Equations (4) and (5) depict the models for computing the rut depth in subgrade soil and unbound pavement layers (refer to Table 2).

Table 2 MEPDG rutting models [29]
Full size table

Alligator fatigue cracking model

Equation (6) is used for calculating Nf (allowable number of load repetitions) that results in fatigue failure (based on the tensile strain at the bottom of AC layer) as presented in Table 3 [29]. The damage ratio (DI) is used to represent the amount of damage. DI is the division of the axle loads actual number by the allowable number of axle loads to fatigue failure (Nf) as displayed in Eq. (8). Finally, the empirical transfer function is used to determine the area of fatigue cracking as displayed in Eq. (9).

Table 3 MEPDG alligator fatigue cracking model [29]
Full size table

Quality-related specifications software (QRSS)

The quality-related specifications software (QRSS) was developed under the NCHRP Project 9-22 [30] by Fugro Consultants, Inc. and Arizona State University (ASU). This software is a simplification of the MEPDG. The QRSS predicts the rutting of HMA layer and fatigue and thermal cracks for any pavement structure based on the dynamic modulus of the HMA which is a function of traffic speed, volumetric properties of HMA, and climatic data. The prediction of the HMA rutting and alligator fatigue cracking in the QRSS methodology is based on the effective dynamic modulus |E*| of the HMA as the main property characterizing the HMA stiffness. Even though there are many models available in literature for |E*| prediction of the HMA [31], the QRSS adopted two different models for |E*| prediction. These models are Witczak NCHRP 1-37A and Witczak NCHRP 1-40D [30]. In this research, the NCHRP (1-37A) model (Eq. 11, Table 4) was selected as it was reported by several studies to yield better predictions compared to the NCHRP (1-40D) model [31,32,33,34,35,36,37,38,39]. This is particularly true for the traditional mixes using Marshall mix design and the binder penetration grading system, which are still being used in Egypt.

Table 4 Models used in QRSS
Full size table

The QRSS relies on the effective E*, which is based on the effective temperature (Teff). Teff can be defined as the single temperature where an amount of distress would be equivalent to that occurs from the temperature fluctuation throughout the annual cycles of temperature. In QRSS, Teff for rutting prediction is determined by the methodology presented in NCHRP 2011 and shown in Eq. (12), while for fatigue cracking, Eq. (13) is used as provided in Table 4 [30, 40, 41].

The effective asphalt modulus |E*| is based on the effective temperature and effective frequency. The effective frequency for fatigue cracking depends on the effective depth, traffic speed, and layer modulus as shown in Eq. (14) [40, 41]. The effective depth for rutting prediction can be computed by Eq. (15), whereas for fatigue, it is located at the bottom of the AC layer.

Materials and methodology

The RAP material used in this research was sourced from a wearing AC surface layer of a major road under rehabilitation located in Port Said Governorate, Egypt. The road has a medium traffic levels and was constructed nearly 20 years ago, and the main pavement distress was the asphalt rutting. The RAP was obtained by a cold milling machine. The RAP properties were examined in the laboratory, and the average percentage of bitumen was 5.2% (by total weight of the mix) based on the extraction results [42]. The asphalt mixture aggregate was crushed dolomite with 19 mm nominal maximum size, and the binder was 60/70 penetration grade.

A crushed dolomite virgin aggregate (VA) was supplied from the Ataqa quarry located in Suez governorate, Egypt. This is a typical aggregate type usually used for the base layer construction in Egypt. Table 5 presents the engineering properties of the 100% RAP and virgin aggregate materials compared with the Egyptian specifications [49].

Table 5 Engineering properties of the virgin aggregate and RAP materials
Full size table

A control section (containing 0% RAP) was structurally designed according to the AASHTO 1993 guide [50] for comparing the effect of the base material properties containing different RAP percentages on the field performance over the service life. The laboratory resilient moduli of the base materials, which were used to predict the performance, are available in [24, 51]. The HMA dynamic modulus was predicted by the QRSS. Three different Egyptian climatic conditions of Alexandria, Cairo, and Aswan and two levels of design speeds of 10 and 50 mph (16 and 80 km/h) were chosen for the analysis. Climate data for the three different cities were taken from Elshaeb et al. [52].

The predicted dynamic moduli of the HMA layer |E*| along with a typical flexible pavement structure were used for the pavement structural analysis model using the Multi-Layer Elastic Analysis (MLEA) software (KENLAYER). The horizontal tensile strain at the bottom of the asphalt layer and the vertical resilient strain at the critical locations within the pavement structure were calculated.

The pavement performance in terms of total pavement rutting and HMA fatigue cracking was predicted using MLEA along with the MEPDG performance models and transfer functions. The same analysis methodology was followed previously by Arisha et al. [53]. Figure 1 outlines the performance analysis process. It should be emphasized that the predicted performance in this research is only valid for comparison purposes because of the simplifications made and also the lack of local calibration of the transfer functions.

Fig. 1
figure 1

Flowchart of performance analysis process

Full size image

Pavement analysis and performance prediction

Control section (0% RAP) was designed for two cases: weak and strong subgrade, following the AASHTO 1993 design method for flexible pavements [50]. The input parameters for the AASHTO 1993 are reliability level (R) = 95%, overall standard deviation (So) = 0.50, change in serviceability (ΔPSI) = 1.5 listed in Table 6. A traffic level of three million ESALs and a design life of 20 years were chosen, and the base material moduli were based on the testing results for 0% RAP [24, 51]. The pavement layers’ thickness is shown in Fig. 2.

Table 6 Input and output parameters for the AASHTO 1993 design method
Full size table
Fig. 2
figure 2

Pavement structures used for analysis

Full size image

Two levels of design speeds were chosen to evaluate the effect of traffic loading frequency on the effective |E*| and hence on the predicted pavement performance. The slow speed of 16 km/h (10 mph) simulates the traffic speeds at intersections and toll stations, while the high speed of 80 km/h (50 mph) simulates the typical speed level for rural roads in Egypt. Finally, three different climatic datasets for three Egyptian cities: Alexandria, Cairo, and Aswan, were used to assess the effect of climate on predicted performance. Tables 7 and 8 summarize the inputs for the QRSS for the investigated cases. The HMA properties were taken from Amin [54], and the mix is typically used for wearing surface layers in pavement projects in Egypt. The performance grade (PG) of the binder is compatible with the climatic location.

Table 7 QRSS input parameters for all investigated cases
Full size table
Table 8 Asphalt layer mix inputs [54]
Full size table

Dynamic modulus |E*| prediction

Six QRSS simulation runs were conducted for the different traffic speeds and climate conditions. Table 9 shows the effective frequency, temperature, and AC dynamic modulus for rutting and fatigue. One may surmise from the results that the effective temperature for rutting is higher and hence the effective |E*| is lower compared to fatigue. Also, as the speed increases, the effective |E*| also increases due to the viscous nature of the binder. Moreover, |E*| is lower at the hot climate condition (Aswan) compared to the moderate climate (Alexandria). Finally, the influence of the climate conditions and traffic speed level is more pronounced on the effective |E*| for rutting as compared to fatigue.

Table 9 QRSS outputs for rutting and fatigue
Full size table

KENLAYER

KENLAYER is a Multi-Layer Elastic Analysis software designed to analyze flexible pavement stresses and strains. The pavement section used in the simulation runs along with the loading characteristics is demonstrated in Fig. 3. KENLAYER was used for the prediction of vertical resilient strain at different locations within the pavement section [mid-depth of each layer, subgrade surface, and 6 in. (15.24 cm) below subgrade surface]. The vertical resilient strain was used for the calculation of pavement rutting by transfer functions and MEPDG performance models [29]. In addition, the horizontal tensile strain (at the bottom of the AC layer) was also calculated for the determination of fatigue cracking.

Fig. 3
figure 3

Cross section used for performance analysis

Full size image

The input parameters are material properties and traffic loading. The engineering properties of the base materials are summarized in Table 10. These properties were based on experimental testing results conducted by Mousa et al. [24] and Mousa [51] on different granular base materials blended with different percentages of RAP. A total of 144 runs were conducted.

Table 10 Material properties for performance analysis [24, 51]
Full size table

Traffic repetitions of 3,000,000 18-kips (80-kN) ESALs per the design life, with tire pressure of 120 psi (0.827 MPa) and spacing between the dual tire of 13 in. (33 cm) were applied to the pavement structural sections as displayed in Fig. 3. The nonlinear analysis module was applied for the base layer which is based on the k − \( \theta \) model (\( M_{\text{r}} = K_{1} \theta^{{k_{2} }} \)). The pavement system was simulated in the KENLAYER program as three-layered system with the previous traffic characteristic. The resulted values of the horizontal tensile strains and the vertical resilient strains were used as input parameters in the MEPDG performance models and transfer functions for performance predictions, which are presented in the following sections.

Rutting prediction

The predicted rutting against the RAP percentage for all investigated climate cases at the design speed of 10 and 50 mph is presented in Figs. 4 and 5, respectively. It is obvious from the figures that the RAP percentage has a significant effect on the total rutting. The predicted base rut depth was found to increase with the decrease in the RAP amount in the blend for all investigated cases. It should be noted that the rutting of foundation layers (subgrade and granular base layers) does not affect the rutting of the AC layer.

Fig. 4
figure 4

Predicted rut depth for the investigated blends (weak subgrade case)

Full size image
Fig. 5
figure 5

Predicted rut depth for investigated blends (strong subgrade case)

Full size image

The effect of the climatic condition on rutting prediction is also clear in the figures. As the temperature gets higher (moving from Alexandria to Aswan), the predicted rut depths become higher for all sections at design speed levels of 10 and 50 mph. This observation confirms that the change in asphalt concrete modulus due to the change in temperature has a significant effect on the rutting prediction. Moreover, at high traffic speed level, the values of predicted total rut depths were lower compared to slow traffic speed, which is due to the viscous nature of the asphalt. Thus, the slower the traffic speed (low rate of loading), the larger the rutting in the HMA layer.

Tables 11 and 12 show the reduction in rut depth of each layer due to the use of different RAP amounts in road base for weak- and strong-subgrade soil, respectively. Data show a significant decrease of up to 61% in the base layer rutting and 38% in the subgrade layer rutting when the 100% RAP is used as a base layer compared to the virgin aggregate for Aswan at the design speed of 10 mph. The data in Table 11 clearly show that for all practical purposes, the rutting of foundation layers (granular base and subgrade layers) does not affect the rutting of the AC layer. The increase in rut depth reduction is due to the increase in stiffness (resilient modulus) of base layer with the increase in RAP amount in the blend as indicated previously. A maximum decrease of up to 25% in rutting was achieved with the use of 100% RAP relative to 0% RAP (virgin aggregate only) for Alex at a speed of 50 mph. More reduction in total rutting was evident in the case of weak subgrade as compared to the strong subgrade.

Table 11 Reduction in rut depth due to use RAP in road base: weak-subgrade case
Full size table
Table 12 Reduction in rut depth due to the use of RAP in road base: strong-subgrade case
Full size table

Fatigue cracking prediction

Excel sheets were designed for the prediction of fatigue cracking distress using the KENLAYER outputs along with the MEPDG performance models and transfer functions based on the effective temperature calculation for fatigue. Figure 6 presents the predicted fatigue cracking against the RAP percentage for all investigated climate cases at a design speed of 10 and 50 mph. It can be observed from the figures that the fatigue cracking is strongly dependent on the base layer stiffness. As the stiffness of the base layer underneath the AC layer(s) increases (by increasing the RAP amount in the blend), the tensile strain at the bottom of the HMA layer decreases, and consequently, fatigue cracking decreases. Alam et al. [5] reported the same observation for the decrease in fatigue cracking with the increase in RAP amount. Also as seen from the figure, the fatigue cracking for strong-subgrade cases is lower than weak-subgrade soil at the same speed and climate condition. The reason for this is that as the subgrade stiffness increases, the tensile strain (at the bottom of the HMA layer) decreases, and hence fatigue cracking decreases.

Fig. 6
figure 6

Predicted fatigue cracking for all investigated cases

Full size image

The effect of the climatic condition on fatigue cracking is also obvious in figures. In general, as the temperature gets higher (moving from Alexandria to Aswan), the fatigue cracking gets higher. The significance of climatic conditions on fatigue cracking agrees with the findings of Ezzat et al. [55], Tarbay et al. [56] and Azam et al. [57] studies.

The effect of traffic speed is presented in the same figure; at slow speed, the predicted fatigue cracking values were higher compared to fast speed due to the viscous behavior of the asphalt material. The significance of the selected climates on the values of the predicted fatigue cracking is minor as the Egyptian climate is mostly hot and moderate and.

Table 13 summarizes the reduction in fatigue cracking for the different percentages of RAP in road base for weak- and strong-subgrade soil. A maximum decrease of up to 92% and 88% was observed in the fatigue cracking when the 100% RAP was used as a base layer compared to the virgin aggregate for Aswan at the design speed of 10 mph for weak and strong subgrade, respectively. The increase in fatigue cracking reduction is due to the increase in resilient modulus of the base layer material with the increase in RAP amount as indicated previously.

Table 13 Reduction in fatigue cracking due to the use of RAP in road base for all investigated cases
Full size table

Conclusions

This study combined three powerful tools: MLEA, transfer functions, and MEPDG performance models, and the QRSS software in a simplified methodology to predict the pavements performance constructed using RAP/virgin aggregate blends as base course material for road construction in Egypt. Based on the laboratory and performance prediction results and analyses, the following conclusions can be found:

  1. 1.

    RAP blends showed superior/comparable performance compared to virgin aggregates.

  2. 2.

    The effects of rate of loading (vehicle speed), climatic conditions and subgrade strength were significant on both fatigue cracking of the AC layer and rutting of asphalt, base, and subgrade.

  3. 3.

    The state of stress at the base layer had a significant effect on the modulus of the layer and consequently on the pavement performance. Further laboratory testing is required to determine the dynamic modulus of the HMA, since the prediction of pavement performance is mainly based on the predicted values of |E*|.