An Asphalt cement 40 - 50 grade (obtained from Nasirya refinery) has been used as binder for all mixtures investigated in this research. The rheological properties of binder were determined according to traditional tests shown in Table 1. These tests involved penetration test, softening point test, ductility test and flash point test. The result of these test are shown in Table 1.

Natural crushed aggregate (available locally-Badra site) with maximum size of 19 mm was used for all mixes as suitable for binder layer of flexible pavement (SORB/R9) [17] . The gradations of aggregate were design to compare between Superpave and Marshall procedures for designing aggregate gradation in asphalt concrete mixes. Three aggregate gradations were adopted according to Superpave procedure: above the restricted zone

(S-AR), through the restricted zone (S-TR), and below the restricted zone (S-BR). Similarly, three aggregate gradations were adopted according to Marshall procedure: upper limit (M-UL), mid-specification (M-MS), and lower limit (M-LL). Figure 1 shows the aggregate gradations according to both Superpave and Marshall methods. From Figure 1, it can be noticed that the discontinuity of Marshall aggregate gradation in fine aggregate zone due to missing of three sieves at this region (1.18 mm, 0.6 mm and 0.15 mm).

Optimum Asphalt Contents (OAC) were determined for each aggregate gradation as shown in Table 2. Five percentages of asphalt content were examined to find OAC starting with 4% and ending by 6% at interval of 0.5%, consequently, 90 (3 × 5 × 6) samples were tested for this purpose. The sample of OAC determination is illustrated in Figure 3.

The pavement rutting test has been carried out using wheel track machine manufactured in road laboratory, college of engineering, University of Al-Qadisiyah for the purpose of doing this research and other research in future. The machine consists of metal cabinet with digital control electric heater with heating temperature reached up to 55˚C. Inside the cabinet there is a wheel with frame to carry the desired load and can move up and down only. In the middle of cabinet there is a movable base with five wheels on each side connected to motor with three speeds (34 passes/min., 100 passes/min., and 345 passes/min.). This base moves over two thick plates forward and backward. The mould of specimen is fixed on the movable base after removing the upper part of it. Figure 4 shows the testing machine. A dial gauge is used to measure the depth of pavement rutting after each specific number of wheel passes.

A mould was design to closely represent the real pavement especially the thickness of pavement layer. The mould consists of three parts: base with clear dimensions of 300 × 200 mm, lower sides as one piece with height of 30 mm and upper sides as two pieces with height of 60 mm, consequently the inside dimensions of mould are 300 × 200 × 90 mm. The upper part of mould (with height of 60 mm) is removed after the compaction of the specimens. Figure 5 illustrates the parts of moulds.

The specimens are prepared according to the each specific aggregate gradation by blending the components according to Marshall procedure described in ASTM D 1559-89. After the completion of blending of component, the specimens is put in the mould for compaction process. The compaction was made by pressing the specimens in universal loading machine. The magnitude of load was determined by compacting several specimens with different load values then selected the value which give similar density to that obtained from Marshall specimens (2.33 gm/cm³). The selected load value (240 kN) was adopted for all specimens. Figure 6 shows the compaction process of specimen, as well as Figure 7 shows the selection of compaction load.

After the completion of compaction process, the upper part of mould was removed and the specimens were left

for 24 hours. Next, the specimens were cured in oven for 1 hour to reach the desired testing temperature, then it was placed in testing machine to start testing process. Load was applied on the wheel to produce a tire pressure of 560 kPa. Each specimen was loaded for 10,000 passes and the rutting depth was measured in the middle of specimen after each specific numbers of passes.

In addition to pavement rutting test, Marshall stability and Marshall retained stability were carried out on Marshall specimens for each type of mix according to selected aggregate gradation.

The Marshall stability test was conducted by testing three specimens for each aggregate gradation. The specimens were prepared and tested according to ASTM D 1559-89. The specimens were cured in water bath for 30 minute at 60˚C then they were tested to measure the ultimate load. The test results illustrated in Figure 8 show that the specimens prepared with aggregate gradation according to Superpave method give more Marshall stability than those produced with aggregate gradation according to Marshall method. Also, the coarse aggregates yield more Marshall stability compared with finer particles. This can be attributed to stiffness of coarse aggregate and more interlock between the coarse aggregate particles. The mix that contained aggregate has a gradation pass through the restricted zone (S-TR) gives lower Marshall stability than the other types of Superpave base aggregate gradation mixes. This region is usually recommended to be removed from Superpave aggregate

gradation [18] .

Six specimens were prepared for each aggregate gradation according to [19] to have same specific gravity. Three of each six specimens were cured by immersion them in water bath for 24 hrs at temperature of 60˚C (wet). The other three specimens were left for normal curing in an air bath for 4 hrs at temperature of 25˚C (dry) [20] . Each group (three samples) was tested to measure the ultimate load. The index of retained strength was calculated according to Equation (1).

S1: Compressive strength of wet specimens.

S2: Compressive strength of dry specimens.

Figure 9 presents the results of index of retained strength for all mixes. From that figure, it can be observed that the average percentage of index of retained strength for all specimens prepared with aggregate gradations based on Superpave method is more than these specimens prepared with aggregate gradations based on Marshall method. The other observation is that, the coarser aggregate gradation has less index of retained strength. This may be due to more air voids initiated with coarse aggregate. However, it still gives adequate level of index of retained strength.

As mentioned earlier, six aggregate gradations were adopted for the current research. Three of these gradations were according to Marshall method as lower limit of specification (M-LL), mid-specification (M-MS), and upper limit of specification (M-UL). The other three represented aggregate gradations according to Superpave method as blow restricted zone (S-BR), through restricted zone (S-TR) and above restricted zone (S-AR).

The rutting depths were measured after 250, 500, 1000, 2000, 3000, 4000, 5000, and 10,000 passes. The testing machine was stopped after each specific number of passes at the middle of specimens to measure the rutting depth. Six specimens were tested at the temperature of 55˚C, this temperature represents the maximum prevailing temperature at Summer in most of middle east countries. Also; at such hot weather, the pavement rutting probability increases.

Figure 10 shows the results obtained of the six specimens prepared with different aggregate gradations according to Marshall and Superpave methods. The general trend of results for all specimens shows that for coarser aggregate gradation the rutting depth decreases. This behaviour attributes to better interlock of coarse aggregates that have rough surface texture (crushed) which gives more stability and stiffness for HMA. This finding agrees with other research in literature [21] - [23] .

The most important point can be observed from Figure 10 is that the average rutting depth for specimens prepared according to Superpave aggregate gradations is about 50% of those prepared according to Marshall aggregate gradations. This may be due to the restriction procedure follows by Superpave in selecting of aggregate gradation, which involve three additional sieves (1.18 mm, 0.6 mm and 0.15 mm) control the amount of rounded and finer aggregate which can cause the rutting. These sieves are missing in aggregate gradation according to Marshall procedure. The fine aggregate is graded using five sieves in Superpave procedure which can provide best and well gradation for aggregate at this important zone for rutting occurrence, while only two sieves were included for same zone in Marshall procedure. Unlike Marshall method, the continuous aggregate gradation in Superpave method provides the appropriate sizes of aggregate to fill unnecessary gaps in HMA and creates more contact points between the aggregate then consequently, increase the cohesion between the particles which gives more rutting resistance. The fine aggregate consists about 40% - 50% of total aggregate in mixture, therefore the less control on this zone may allow for more inappropriate sizes to come in.

Figure 11 presents the photos of rutting depth of the six specimens at 55˚C, these photos clearly show the difference in rutting depth for the considered specimens.

Two specimens were tested at 25˚C as shown in Figure 12, No clear differences or trend can be observed between these specimens. This can be attributed to moderate temperature and normal load, which reasonable does not cause rutting.

Figure 13 illustrates the photos of rutting depth of the two specimens at 25˚C. There is no clear difference in rutting depth between the two specimens.