INFLUENCE OF LABORATORY COMPACTION METHOD ON UNBOUND GRANULAR MATERIALS

Inge Hoff 1, Leif J. Bakløkk 2 and Joralf Aurstad 3 INFLUENCE OF LABORATORY COMPACTION METHOD ON UNBOUND GRANULAR MATERIALS ABSTRACT All laboratories that perform cyclic triaxial tests on unbound granular materials for use in pavements have developed their own methods for sample compaction and preparation. It was not possible for the committee that was preparing the European standard to agree on one recommended method for sample compaction. In the GARAP-project the effect of compaction was investigated by testing samples compacted with different methods in the cyclic triaxial test. The Multistage procedure described in the new EN standard [CEN 2003] was followed. The compaction methods used where: Gyratory compaction, Impact hammer (Modified Proctor), Vibratory hammer (Kango) and Vibratory table. Three to five different levels of target density were tested for each method using a well graded material from Askøy, Norway. For the resilient modulus no systematic difference was found between the different methods of compaction. However, for the resistance against permanent deformations significant differences were found between samples compacted to the same density with different methods. Samples compacted using a vibratory compaction method showed about 20-25 % higher resistance to incremental failure and 40 50 % higher limit values for purely elastic behaviour compared to samples compacted using the Modified Proctore impact hammer. 1 Inge Hoff Senior Research Scientist SINTEF, Roads and Transport NO-7465 Trondheim, Norway Inge.Hoff@sintef.no 2 Leif Jørgen Bakløkk Head of Pavement design group Norwegian Public roads Administration, Centre for Road and Traffic Technology NO-7030 Trondheim Leif.Baklokk@vegvesen.no 3 Joralf Aurstad Research Scientist SINTEF, Roads and Transport NO-7465 Trondheim, Norway Joralf.Aurstad@sintef.no

1 INTRODUCTION 1.1 GARAP research project The research project Granular Aggregates for Road and Airport Pavements - GARAP is supported by: - Norwegian Public Road Administration - Avinor - Norwegian Rail Administration - Norwegian Aggregate Producers Association - Nynas - Norwegian Research Council. The aims of the project are to improve laboratory testing, material modelling and design of structures with unbound granular materials. 1.2 Laboratory compaction Laboratory testing of materials can only be an imitation of the real field conditions. For coarse graded material it is not possible to take undisturbed samples from the road into the laboratory for triaxial testing. Hence, the material must be compacted in the laboratory to the same density as in the field or to a range of densities if the field density varies. Several different methods for laboratory compaction have been developed over the years. The most popular method for general sample preparation is maybe the impact hammer (Proctor). This is a very simple method and no expensive equipment is necessary. On the other hand the impact hammer is not a good simulation of the compaction process in the field with a heavy roller often combined with some sort of vibration or oscillation. Other methods have been developed that are supposed to better simulate the field compaction. It is obvious that the compaction level, i.e. dry density, strongly influences the material behaviour. Some research also indicates that samples compacted to the same density with different methods will behave differently. As an example research performed in the KPG project [Hoff 1998] showed dramatically higher CBR-values 4 for samples compacted with gyratory compactor compared to samples compacted with Modified Proctor hammer. 2 TESTING EQUIPMENT In the GARAP-project the samples were tested in a triaxial testing apparatus for 150 mm diameter samples, applying cyclic deviatoric loads and constant confining pressure. The development of the equipment at NTNU/SINTEF was started in 1975 and has been going one since. New methods for load control and strain measurements have repeatedly been taken into use to secure an increasing accuracy. The equipment now has the possibility to also cycle the confining pressure in phase with the deviatoric load, but this ability has not been used in the tests described here. Figure 1 shows a photo of the equipment with a sample ready for testing. 4 CBR-loading procedure was used on samples with non-standard compaction and without soaking in water.

Figure 1 Cyclic triaxial test apparatus at NTNU/SINTEF 3 SAMPLE PREPARATION Four different compaction methods were used for sample preparation. Some information about the different compaction equipments is given in Table 1. Table 1 Characteristics of the different compaction methods Type Principle Manufacturer Characteristics Gyratory compactor Kneading Invelop oy ICT 150 RB 0 2 o gyratory angle 0 700 kpa vertical pressure 0 500 cycles Modified Proctor Impact Own Weight 4.8 kg Vibratory table Vibration AEG VT 360/630 cy Vibratory hammer Vibration/impact Kango 950 X Free fall 0.450 m Frequency: 50 Hz Amplitude: 1.0 mm Dead weight 4.95 kg Total weight 35 kg Frequency: 25 60 Hz Amplitude: 5 mm The gyratory compactor can produce samples as high as 220 mm. This height was used as target height also for the other methods. The EN standard requires a 2:1 ratio between sample height and diameter. For a 150 mm diameter sample this implies a height of 300 mm. This height was not possible to archive in the gyratory compactor. To compensate for the lower samples, a special

procedure was used to secure low friction against the end-platens. Teflon foils and silicon oil were used for lubrication. The samples were extruded from the mould using a special technique keeping the sample confined by internal vacuum until the confining pressure in the triaxial chamber could be applied. 3.1 Material In this investigation only one type of material has been used. The gneiss from Askøy, outside of Bergen, is of good quality and typical for materials used in road construction in Norway. Some of the properties are listed in Table 2. Table 2 Characteristics of the Askøy material Specific density 2690 kg/m 3 Los Angeles value 14.5 Modified Proctor density 2186 kg/m 3 Modified Proctor optimal moisture content 5 % The material is a metamorphic granitic gneiss that consists of quarts and feldspar with smaller amounts of amphibole, titanite and mica. The material was sieved and combined to a well graded Fuller curve (Figure 2): p = d D n Where: p = percentage passing sieve d = sieve size D = maximal grain size, 22 mm n = 0.5 SILT Medium Coarse Fine SAND Medium Coarse Fine GRAVEL Medium Coarse 100 90 80 70 60 50 40 30 20 10 0 0.01 0.02 0.06 0.075 0.125 0.200 0.250 0.500 0.600 1.00 2.00 4.0 6.0 8.0 16 22 11.2 19 32 60 Figure 2 Grain size distribution of tested material

4 PROCEDURES FOR TRIAXIAL TESTING It is believed that the compaction method particularly influence on the development of permanent deformations. Hence, it was decided to follow a triaxial procedure that has been developed to determine the resistance against permanent deformation. The multi-stage procedure described in the new EN-standard [CEN 2003] is well suited for the purpose of relative comparison between different materials or, as in this case, studying different sample preparation methods. The procedure is similar to procedures used earlier at NTNU/SINTEF [Hoff 1999] The load is applied stepwise in five sequences for each level of confining pressure. Table 3 shows the load levels used for strong samples. A similar table exists for use on weaker samples with lower stress levels. Table 3 Stress levels for multistage loading procedure (high stress) Confining stress, σ 3 (kpa) Sequence 1 Sequence 2 Sequence 3 Sequence 4 Sequence 5 Deviator Confining Deviator Confining Deviator Confining Deviator Confining stress, σ d stress, σ 3 stress, σ d stress, σ 3 stress, σ d stress, σ 3 stress, σ d stress, σ 3 (kpa) (kpa) (kpa) (kpa) (kpa) (kpa) (kpa) (kpa) Deviator stress, σ d (kpa) Constant min max Constant min max Constant min max Constant min max Constant min max 20 0 50 45 0 100 70 0 120 100 0 200 150 0 200 20 0 80 45 0 180 70 0 240 100 0 300 150 0 300 20 0 110 45 0 240 70 0 320 100 0 400 150 0 400 20 0 140 45 0 300 70 0 400 100 0 500 150 0 500 20 0 170 45 0 360 70 0 480 100 0 600 150 0 600 20 0 200 45 0 420 70 0 560 For each sequence the test is interrupted when all steps are completed or an axial strain of 0.5 % is reached. The test is then continued by applying the next sequence. For each step 10 000 continuous sine shaped load pulses are applied at a 10 Hz frequency. This procedure reveals the resistance against permanent deformations and the resilient modulus for a range of stress levels. 5 RESULTS FROM TESTING 5.1 Resilient behaviour The resilient modulus can be calculated using the following formula for cyclic triaxial tests with constant confining pressure: E r σ d = ε a Where: Er = Resilient modulus σ d = Applied deviatoric stress ε a = Measured axial resilient strain

The resilient modulus is highly stress dependent as illustrated in Figure 3. Each point represents the stabile average value from each load step (10 000 pulses). 700 600 Resilient modulus (MPa) 500 400 300 200 100 0 0 50 100 150 200 250 300 350 400 Max mean stress (kpa) Figure 3 Resilient modulus as a function of mean stress for one of the samples Several different models for stress/strain dependent resilient models have been proposed. The simplest models use two material parameters for resilient modulus and a constant value for Poisson s ratio. More complex models using four or more parameters have also been proposed. The results from these tests could be interpreted using one of these models. However, for simple comparison between the samples in this investigation the resilient modulus for maximum mean stress of 200 kpa has been plotted versus density as shown in Figure 4.

Resilient modulus for mean stress = 200 kpa (MPa) 700 600 500 400 300 200 100 0 1.9 1.95 2 2.05 2.1 2.15 2.2 Dry density (kg/dm3) Gyrator Vibratory table Vibratory table Impact hammer Figure 4 Resilient modulus as a function of dry density It seems clear from Figure 3 that there is a correlation between the dry density of the samples and the resilient modulus. However, the different compaction methods seem to give similar values for resilient modulus. The differences observed are smaller than scatter for cyclic triaxial testing on this type of material. 5.2 Permanent deformation behaviour Characterisation of a materials resistance against permanent deformations is not straightforward. Several procedures for laboratory testing and interpretation have been proposed. One obvious problem is that the development of permanent deformation is highly dependent on the load history. This means that a new sample should be used for every topical single stress condition. However, this would lead to an unpractical and expensive testing program; the material is exposed to a wide range of stress conditions at different locations in a pavement structure. To reduce the number of tests, a multistage procedure has been developed. This procedure is now adopted in the EN-standard for cyclic triaxial testing. Figure 5 shows the development of permanent axial strain for one load sequence for one of the samples compacted with vibratory table. A similar response is recorded for the other four load sequences applied to the same sample.

Permanent axial strain (o/oo) 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0-0.5 0 1000 2000 3000 4000 5000 6000 Time (sec) Figure 5 Development of permanent axial strain for one sequence Each load step are then analysed to determine if the material is in one of the following stages: A Almost purely elastic response B Some permanent deformation, but stabilizes towards the end of the step C Incremental failure Obviously the development of permanent deformations is faster in the beginning of a load step than at the end. The average strain rate (strain per pulse) for the last 5 000 cycles in each step has been used to limit the different stages. The limits used correspond to what Werkmeister [Werkmeister 2003] proposed using an average strain rate for cycle 3 000 to 5 000. Table 4 Limits between permanent deformation regions p ε& 1 2.5 10 2.5 10 8 8 Category A p 7 ε& 1 1.0 10 Category B p 7 ε& > 1.0 Category C 1 10 All steps were characterized in this way and plotted in a deviatoric stress vs. confining stress diagram (Figure 6). Then best fit lines for the boundary between the different stages could be found using the Mohr Coloumb parameters (apparent attraction a and friction angle φ failure and ρ elastic limit.) Because the difference in a was relatively small, the tests were reinterpreted using a fixed value for a to make it easier to visualize the differences between the different compaction methods. The results are illustrated in Figures 7 and 8.

700 600 Deviatoric stress (kpa) 500 400 300 200 100 0-50 0 50 100 150 200 Confining pressure (kpa) Figure 6 Determination of failure and elastic limit lines 0.9 0.85 0.8 Failure angle (sin φ ) 0.75 0.7 0.65 0.6 0.55 0.5 0.45 0.4 1.9 1.95 2 2.05 2.1 2.15 2.2 Density (kg/dm 3 ) Gyrator Vibratory hammer Vibratory table Impact hammer Figure 7 Failure limit angel interpreted using a=15 kpa for all samples

0.9 Elastic limit angle (sin ρ ) 0.8 0.7 0.6 0.5 0.4 Gyrator Vibratory hammer Vibratory table Impact hammer 0.3 1.9 1.95 2 2.05 2.1 2.15 2.2 Density (kg/m 3 ) Figure 8 Elastic limit angle interpreted using a=15 kpa for all samples 6 CONCLUSIONS The tests show that different compaction methods can give different resistance against permanent deformation for samples compacted to the same dry density. Samples compacted using the Modified Proctor hammer show less resistance to permanent deformation compared to the samples compacted with one of the two methods based on vibration. It should be noted that this observation is only valid for this specific material and it is possible that ranking between the different method could be different for other materials or other conditions. However, it is likely that different compaction methods will produce significantly different results also for other materials. A more trivial conclusion is that both the resilient modulus and the resistance against permanent deformation increase with increasing compaction effort. 7 REFERENCES CEN EN 13286-7 Unbound and hydraulically bound mixtures Part 7: Cyclic triaxial test for unbound granular materials Werkmeister, S. Permanent Deformation Behaviour of Unbound Granular Materials in Pavement Constructions Dr. thesis. Der Fakultät Bauingenieurwesen der Technischen Universität Dresden 10. Februar 2003 Hoff, I. Material Properties of Unbound Aggregates for Pavement Structures

Dr. thesis Norwegian University of Science and Technology. Trondheim, 1999 Hoff, I. GARAP - Improvement of equipment for cyclic triaxial testing SINTEF report STF22 O04321 Trondheim 2004 Hoff, I and Bakløkk, L.J. Materialegenskaper for grus- og pukkmaterialer SINTEF report STF22 A98459 (Delprosjektrapport KPG 18) October 1998