Category WATER IN ROAD STRUCTURES

Relationships between bearing capacity and moisture content were studied in the Danish Road Testing Machine (ALT-facility) (Krarup, 1995). A typical Danish pavement was built in the facility, and for almost one year the only condition that was changed was the level of the water table. No load was applied except for FWD – measurements. The layer stiffness moduli (E-values) of the pavement layers were calculated from FWD-measurements with the program ELMOD (Dynatest, 1989). Based on the test results, a relationship between E-value and suction, pore pressure and degree of saturation was established. Simple linear regression describes the

Suction head, s, (cm)

Fig. 10.18 Correlations between the resilient modulus and the suction for the two soils ana­lyzed for a confining pressure of 20 kPa and a deviatoric stress of 20 kPa (BrUll, 1983). Figure courtesy of the Belgian Road Research Centre

Note: Mr = Resilient Modulus (MPa), G = Shear Modulus (MPa), R = coefficient of correlation, N = number of data points, s = suction (cm), light dotted lines show 95% confidence limits for Mr.

relation between the value of the E-value and suction measured by the tensiometer in the form: E = B1 + B2 x s, where

• E is the layer E-value in MPa from the ELMOD calculation;

• B1 and B2 are constants; and

• s is the measured suction in kPa.

The regressions reveal high levels of R-squared within the range of the measured suction values for all unbound layers. The measured values and the linear regression lines from three of the tensiometer depths are plotted in Fig. 10.20.

When calculating the stress in a pavement (a continuous body), the stress accord­ing to Boussinesq is independent of the E-modulus of the material. Combining this statement with Terzaghi’s principle of effective stress, it becomes unpredictable to what extent E-values from FWD-tests relate to positive pore pressure measured in standpipes. Positive pore water pressure only appears in what is considered to be a saturated condition. In the Krarup study, the upper pavement layers only became saturated for very short periods and water/suction versus time series exist for this granular base course layer data.

The available measured data, linear and linear-exponential regressions were tested on the data to investigate any dependency. The linear regression came up with a R-squared value ranging from 0.56 to 0.88, whereas the linear-exponential regression had the best fit at 0.72. The data and the linear regressions were plot­ted in Fig. 10.21. Note the much lower values of stiffness were determined in-situ than those typically measured in laboratory triaxial assessments (cf. Fig. 10.12 for example).

The reason for the decrease of the lower unbound layer material stiffness over the monitoring period is expected to be the water that adheres to the surface of the mineral’s granular materials, more than any pore water pressure phenomenon.

Saturation: Saturation and E-values are not expected to reveal a linear relation­ship, as the change in E-value dependent on degree of saturation occurs at a certain level of saturation determined by the void size distribution. The results plotted in Fig. 10.22 are derived from time series and are not scattered data points. During the monitoring period the saturation increased and decreased, so the curves should be read from left to right and then back again so as to follow time.

The two time series of saturation data from 20 to 40 cm below surface are plotted with the E-values assigned to the sub-base layer, and therefore plotted with the same

set of E-values. Figure 10.23 shows that granular base course and sub-base layers tended to have threshold E-values in the test pavement. Keeping the low E-values of the subgrade at the construction time in mind, the E-value of the natural till subgrade might continue to decrease as long as the water table is above the material.

As a rather strong relationship between E-value and suction was confirmed from the measurements, the relation between suction and saturation becomes interesting. In soil water research the relationship is the so-called soil-water characteristic curve or the retention curve. From experiments in the laboratory the relationship can be found for small soil samples. Data measured in the test pavement were plotted as Fig. 10.23.

The deterioration of strength, stiffness and resistance to the development of per­manent deformation, or the reduction in pavement life, with increasing moisture levels is a common observation. Trial pavement studies in which the water content of the construction has been changed and reduced performance observed are quite numerous. In recent years work in Finland has been reported by Korkiala-Tanttu and Dawson (2007) showing the much faster rutting of a test pavement with a high water table than one in which it was lower. In an earlier study by accelerated trafficking, Vuong et al. (1994) found that the life of a crushed-rock base was very dependent on the degree of saturation in the aggregate base course. Assuming a water content for optimal behaviour, then a 5% change increase in relative water content could lead to a 400% reduction in pavement life. Sharp et al. (1999) reported significant deterioration of in-situ moduli values at an accelerated pavement testing site in lat – eritic gravel bases and sub-bases upon wetting (or improvement on drying) by more than a factor of 2. The silty sand subgrade at the test site also changed stiffness to a similar degree.

Thus site studies broadly support the laboratory and theoretical work reported elsewhere in this book. As an example of a particular study of the effects of moisture change on bearing capacity, the following case record is instructive.

Examples of variation of resilient modulus and permanent deformations of subgrade soils with moisture content are presented in Figs. 10.15 and 10.16. Figure 10.15 presents results obtained on a clayey sand (14% fines, optimum moisture content

wOPM = 8%).

Figure 10.16 presents results obtained for a silt (85% fines, optimum moisture content wOPM = 14%). For the 2 soils, the resilient modulus (determined for two different levels of stress) decreases by a factor of 3-4 when the water content in­creases from wOPM — 2% to wOPM + 2% (typical in-situ moisture contents). For the same change, the permanent axial strains (determined after 200 000 load cycles with cyclic stresses p = 26 kPa and q = 80 kPa) increase considerably.

Brull (1983) performed triaxial tests on 2 different soils, a loam (from Sterrebeek in Belgium) and a slightly clayey sand (from Noucelles in Belgium). The tests were carried out for different dry volumetric masses, for suctions between 0 and 600 kPa, for different confining pressures (between 10 and 50 kPa) and for different deviatoric stress (between 5 and 25 kPa). The soil water characteristic curves (also known as retention curves) as a function of the volumetric mass of the samples are presented in Fig. 10.17.

The Resilient Modulus, Mr, of the different samples were determined from the triaxial tests. The influence of the volumetric mass and stresses have been analysed. For the tested materials, Brull attempted to establish a linear correlation between the Resilient Modulus, Mr, and the suction, s (Fig. 10.18).

a.

Cui (1993) tested a remoulded Jossigny silt in an unsaturated state and gave an evaluation of Resilient Modulus, Mr, and shear modulus, Gr, as a function of confining pressure and of suction level (Fig. 10.19).

Unbound granular materials, which are continuously graded materials containing fines, are also sensitive to moisture. Examples of influence of moisture on the

resilient modulus and on the permanent deformation of 3 French unbound granu­lar materials, of different mineralogy (hard and soft limestone, micro-granite) are shown in Fig. 10.10. All 3 materials present a decrease of their resilient modulus as the water content approaches the modified Proctor Optimum (wOPM), but the sen­sitivity to moisture is much more important for the limestone than for the igneous rock material (micro-granite). The permanent axial strains become very large for all 3 materials when the water content approaches wOPM.

Other examples, showing the influence of water content on the permanent de­formation and modulus of elasticity of several unbound granular materials from Slovenia, are presented in Fig. 10.11. Again, the permanent strains appear to be more sensitive to moisture than the resilient modulus.

Several studies carried out in France on a large number of different unbound granular materials have shown that their sensitivity to moisture is strongly related to their mineralogical nature, and is particularly important for soft limestone materials. This is illustrated in Fig. 10.12 which presents values of resilient modulus obtained for different natures of granular materials and different water contents.

The igneous materials present relatively low resilient moduli (generally between 300 and 500 MPa), but are not very sensitive to moisture. The soft limestone materi­als present significantly higher moduli at low water contents (up to 1000 MPa), but these moduli drop when the water content approaches the optimum (wOPM).

Ekblad (2004) investigated the influence of water on the resilient properties of coarse unbound granular materials in the saturated as well as the unsaturated state. This study was limited to one type of aggregate of different gradings (with

I In situ water content

Fig. 10.10 Influence of water content, w, on the resilient modulus, Mr, and permanent axial strains, A1c, of 3 French unbound granular materials: hard limestone, soft limestone and microgranite (Hornych et al., 1998)

Note: A1c is a level of strain anticipated once plastic strain has stabilised.

maximum particle size 90 mm). The aggregate comes from Skarlunda in Ostergotland in Sweden. Ekblad’s tests on unbound granular materials of differ­ent granulometric curves showed that the influence of water content on resilient properties depends on the material grading.

First, the dependency of resilient modulus, Mr, on confining stress was estab­lished (Fig. 10.13). Increased confining pressure leads to a substantial increase in resilient modulus. Confining pressures of 100 kPa were reached by Ekblad. Triax­ial tests at different water contents were also performed. The water content was successively increased from an initially low water content to a soaked condition (representing full saturation) and then the sample was allowed to drain freely. All these triaxial tests were performed at a confining pressure of 40 kPa (Fig. 10.13).

Finally, to achieve a summary comparison, the resilient response for a mean nor­mal stress of 100 kPa at a confining pressure of 40 kPa was calculated as a function of the degree of saturation. From Fig. 10.14 it can be observed that the relative reduction in modulus seems to depend on the grading coefficient, with a lower

Fig. 10.12 Sensitivity to moisture of unbound granular materials of different origin (Hornych et al., 1998)

grading parameter (i. e. a higher proportion of fine particles) yielding a larger mod­ulus reduction upon saturation.

10.4.1 Laboratory Results

Wetting of unsaturated soil reduces the suction in the soil, the pore pressure ap­proaches the pore air pressure and the effective stress is reduced. Because of this, increasing moisture is associated with decreases in shear strength, stiffness and re­sistance to plastic deformation in all soils and aggregates and we can observe a decrease in bearing capacity and lower moduli of elasticity and increases in de – formability under the same applied loading.

10.4.1.1 CBR Tests

This influence of moisture on the bearing capacity of soils can be easily observed in the simple CBR test. This test measures the resistance of the compacted soil to the penetration of a piston, to evaluate its bearing capacity. CBR values for three typical soils related to water content are presented in Fig. 10.8. The sensitivity to moisture variations is particularly important for the silty sand (a).

Decrease of bearing capacity and of unconfined compression strength of clay with increasing water content can also be observed in Fig. 10.9.

The CBR test gives a good indication of the moisture sensitivity of subgrade soils, but its results are only qualitative. The influence of moisture on the resilient modulus and resistance to permanent deformations of soils and unbound granular materials can be studied using repeated load triaxial tests.

There are many ways of evaluating pavement structural capacity or adequacy and it is very common to perform deflection measurements with non-destructive testing equipment (COST Action 325, 1997). Once again, there is usually no knowledge of the pore pressure or pore suctions in the soil or pavement layer being assessed, so a total stress interpretative framework is necessary even though an effective stress framework would be more desirable.

There are several reasons for deflection measurements to be carried out: for quality assurance, to evaluate the bearing capacity of the unbound material of the granular base, sub-base and subgrade layers, to identify weak parts of the road, to investigate reinforcement requirements, to establish priorities for road strengthening and for research purposes.

Deflection measurements can be performed in various ways, using:

i) static deflection measurement equipment;

ii) automated beam deflection measurement equipment;

iii) dynamic deflection measurement equipment;

iv) deflection instruments with a harmonic load; and

v) deflection measurement equipment with an impulse load.

i) Static devices include: static and dynamic plate loading (bearing) tests and Benkelman beam.

Plate bearing test. This involves measurement of the deflection caused by a known static load applied through a hydraulic jack on the pavement layer surface by a circular plate. Circular plates are of specified diameters -300,450 and 600 mm with loads up to 60 kN. It is a slow and laborious test and needs a heavy vehicle acting as stationary reaction frame against which the hydraulic jack reacts. The plate test device may be mounted inside a van or at the rear of a lorry.

Dynamic plate bearing test. This is performed by lifting and dropping a known load from a known height onto the circular plate and measuring deflections. In this way, the dynamic effect of vehicles to the pavement is simulated.

The two preceding plate bearing tests are generally only used for mea­surement of bearing capacity of unbound layers.

Benkelman beam. This device consists of two main parts: a stand and a beam (Fig. 10.4). One end of the beam rests on the road surface but the other one is connected to a dial test indicator. The beam is suspended on

Fig. 10.4 Benkelman beam: (1) dual tyres of a loaded vehicle, (2) tip of the beam, (3) ball bearing, (4) adjustable support legs, (5) dial test indicator, (6) stand (Tehnicne specifikacije za ceste, TSC 06.630, 2002). Reproduced by permission of Direkcija Republike Slovenije za Ceste

the stand at two thirds of its length from the end being in contact with the road surface. In this way, when this end moves in one direction, the other end moves in the opposite direction half of that movement and it will be measured by dial gauge. The tip of the beam is placed between the dual tyres of a loaded vehicle. The vertical movement of the surface is then recorded as the truck moves slowly away from the loading area as this results in the rebound of the deflection that was caused by the application of the loaded vehicle. The rebound deflection/deflection ratio depends on the condition of the underlying road layers. For example, high water content caused by thawing can result in a relatively low ratio. If the deflection bowl is large, as in the case of weak subgrades, and the supporting legs are placed in the deflection area, this may produce inaccurate measurements.

ii) Automated beam deflection measurement equipment is represented by the Lacroix Deflectograph. It was developed in order to carry out measurements more quickly than by the time-consuming Benkelman beam. The measurement beam is automatically displaced along the measurement direction, while a mea­surement vehicle proceeds at slow speed. Equipment includes a T-shaped frame towed between the axles of lorry, enabling deflection to be measured in both wheel paths simultaneously (Fig. 10.5).

The measurement starts when the rear wheels of lorry are at a certain distance behind the tip of measuring beam and ends when the wheels reach a certain distance in front of it. Then the entire T-frame is moved forward a specified distance in the direction of movement. The axial load can vary from 80 kN to 130 kN.

iii) Curviameter records the surface deflection under a dynamic load. The vertical displacement is determined in the right-hand wheel path by means of velocity sensors when the loaded twin-wheel rear axle of the measuring lorry passes over. Like the Lacroix deflectograph, loads can vary from 80 kN to 130 kN. The High Speed Deflectograph also belongs to this category. At a driving speed of 80-90 km/h, laser Doppler sensors measure the movement of the pavement surface under a 10-tonne axle.

iv) Deflection measurement by harmonic load can be performed by equipment that can exert a sinusoidal vibration in a road pavement. A well-known example of this type of equipment is the Dynaflect, in which the power is transferred to the road by means of crank at a specified frequency via two steel wheels. The deflections are measured by five geophones spaced at various offsets from the centre of load.

v) Bearing capacity assessment. The most common device to measure bearing capacity of roads is the Falling Weight Deflectometer (FWD). In a FWD test an impulse loading is applied to the pavement which is similar in magnitude and duration to that of a single heavy moving wheel load. The vertical response of the pavement system is measured on the surface with deflection sensors at different distances from the loading. Usually only the peak values are registered, see Fig. 10.6.

Frequently six or seven sensors are used but the spacing between the sensors can vary. Typical values are given in Table 10.1.

The deflections can be used to back-calculate the layer moduli of the structure. This requires information on the number of layers in the pavement structure and

their thicknesses. This is an inverse process where a single, correct solution does not exist. A number of software solutions exist to perform such back-analysis.

The deflections can also be used to estimate some simple parameters which are related to the condition of the pavement, some of the more common ones are:

SCI = D0 – D2 (10.1)

BDI = D2 – D4 (10.2)

BCI = D4 – D5 (10.3)

1 N-1

AREA = (Di – + Di )(n – n_0] (10.4)

D0

where SCI = the Surface Curvature Index, BDI = the Base Damage Index, BCI = the Base Curvature Index and AREA = the area of the deflection basin (all assuming radii as per the light pavement in Table 10.1).

A falling weight deflectometer is mounted on a trailer or in a van, so it needs a stable surface. On soft ground or in trenches the smaller portable light weight FWDs (Fig. 10.7) can be used. These equipments measure the central deflection, resulting in a surface modulus.

A general overview of conventional and some advanced numerical models used in practice has been given in Chapter 9, Section 9.4. Therefore, tests needed for the parameters of these models will now be presented.

• Resilient behaviour models

о Routine pavement design model: in practice much routine pavement design is carried out as catalogue – based design. The pavement is considered as a multi-layered elastic system with constant stiffness parameters in each layer.

о Advanced pavement models: RLT tests are required with variable confining pressure (CEN standard EN 13286-7 (2004)), they correspond to strain sta­bilization. Parameters are determined with curve fitting by the least squares method applied to the equations of the model and to RLT tests results in the stress – elastic strain planes (p, ev) and (q, eq). Both phenomenological models which describe behaviour from an observational standpoint (such as the k—6 and “Universal” Models) as well as theoretically-derived models (such as the Boyce and modified Boyce model) require this kind of testing. The more complex models require VCP test procedures (see Section 10.2.1, above).

• Permanent deformations models

о Analytical models: they require 3 monotonic triaxial tests for the rupture parameters and 3 VCP tests (q /p = 1, 2 and 3 for example) for the plas­ticity parameters. Curve fitting can be used with the least squares method applied to VCP tests results and to the analytical equation of the model

(evertical f(N)) in the plane (N, e^cal).

о Plasticity theory based models

• Bonaquist and Desai models

– Elastic behaviour: the parameters require RLT tests (q /p = 3) at various confining pressures (AASHTO T307-99 (2000)).

– Monotonic plasticity parameters: 3 monotonic triaxial tests till rup­ture for the rupture characteristics and hardening parameters.

– Approximate accelerated analysis: 1 RLT test with one stage at (q /P = 3) is required.

• Chazallon and Hornych model

– Elastic behaviour: RLT tests are required with variable confining pressure (CEN standard EN 13286-7 (2004)), they correspond to strain stabilization.

– Monotonic plasticity parameters: 1 oedometric test, and 3 triaxial tests till rupture are required.

– Cyclic plasticity parameters: 1 RLT test with one stage is required

(q /p = 2).

о Elasto-visco-plastic equivalent models

• Suiker elasto-visco-plastic model

– Elastic behaviour: these parameters are required for the initial state of the material. They require at least 2 monotonic triaxial tests at two different confining pressures when 100 cycles have been per­formed (q / p = 3).

– Monotonic plasticity parameters: they are required for the initial state of the material: 3 monotonic triaxial tests are required.

– Cyclic plasticity parameters: RLT tests (q/p = 3) are required.

о Mayoraz visco-plastic model

– Plasticity parameters require 3 triaxial tests till rupture and a RLT test

(q/p = 3).

о Shakedown models

• Perfectly plastic models

– Elasticity parameters: RLT tests (CEN standard EN 13286-7 (2004)) are required with variable confining pressure till stabilization.

– Plasticity parameters: 3 triaxial monotonic tests till rupture are re­quired.

• Kinematic hardening models

– Elasticity parameters: RLT tests (CEN standard EN 13286-7 (2004)) are required with variable confining pressure till stabilization.

– Plasticity parameters: 3 triaxial monotonic tests till rupture are re­quired for rupture parameters and 3 VCP tests (q /p = 1,2 and 3 for example) with 3 stages for each stress path.

As shown in Chapter 9, Section 9.5, a complete description of a material’s be­haviour necessitates an effective stress approach with the pore pressures (or pore

Number of load pulses Number of load pulses

Fig. 10.3 Typical results from permanent deformation testing where the accumulated axial strain is shown as a function of the number of load pulses in the permanent deformation tests (Erlingsson, 2000) suctions) being separately controlled, or monitored, from the applied pressures. Be­cause most road materials are coarse grained and partially saturated and/or above the ground water table, it usually proves impossible – and certainly it is impractical in most situations – to monitor the pore suctions during each transient pulse. For this reason, almost all testing programs determine parameter values for resilient and incrementally-developed plastic strain models in terms of total, not effective, stresses. Instead, test procedures typically seek to control the moisture or suction conditions so that data is collected at representative conditions.

As described above, unsaturated soils exert an attraction on water, either by cap­illarity in the pores, between grains of the soil, or by the physico-chemical effects. The energy necessary to extract water from the unsaturated soil, for example from a dense compacted swelling clay used for engineering barriers in nuclear waste, can be higher than 10 MPa. However, a granular soil will have a suction that is
several tens of kPa (Delage 2001). The following techniques are commonly used

for suction/moisture control:

i) Axis translation system

This system is based on ceramic low porosity stone, called High Air Entry Value (HAEV) porous stone. The principle of the system is that the pores of the ce­ramic are too small to de-saturate, even when a high air pressure is applied. In other words, the capillary air-water menisci located at the surface of the ceramic can resist an air pressure applied to it, thereby maintaining a continuous water column from stone to specimen. The system has been employed for determi­nation of the water retention curve (i. e. the soil water characteristic curve) (as in the Richards cell) as well as for suction control in oedometer and triaxial testing (Peron et al., 2007; Cuisinier & Laloui, 2004). The retention curve can be determined as follows. A saturated sample should be placed in the cell and the air pressure increased in a step-by-step progression. For a given air pressure, suction equilibrium should be achieved in a time period of a few days (typically 3-10). At the end of equilibrium phase, the sample should be quickly withdrawn and weighed to determine the water content. The sample is then placed into the cell and the air pressure increased to bring about a new equilibrium with a higher suction and lower water content and lower degree of saturation. The process is then repeated as desired.

ii) Osmotic technique

This was initially used by Kassif and Ben Shalom (1971) in an oedometer test to study expansive soils. Several authors extended this technique to a standard triaxial apparatus (Laloui et al., 2006; Komorik et al., 1980; Delage et al., 1987). The technique is based on drainage of the specimen caused by the process of osmosis. The soil sample is placed in a tube-shape cellular semi-permeable mem­brane, which is immersed in an aqueous solution of PolyEthyleneGlycol (PEG). Since the PEG molecules are too big to cross the semi-permeable membrane, the difference between concentration of the PEG solution and that of pore water results in an osmotic pressure. Thus, the suction value depends on the concen­tration of the solution: the higher the concentration of the polyethylene glycol (PEG) solution, the higherthe suctionof soil specimeninthe semi-membrane after the equilibrium state is reached (Dineen & Burland, 1995; Delage et al., 1998).

iii) Vapour phase method

Usually, this method is used to determine the water retention curve of soils with high suctions, though it can be used to bring a triaxial or other test specimen to a particular moisture-suction condition prior to mechanical testing. The method is based on the theoretical standpoint that water potential (i. e. suction) is related to a particular relative vapour pressure of the water in the soil-water system (Qian et al., 2006). The relative vapour pressure of water in equilibrium with the system is characterised by its relative humidity. Therefore, suction can be established by creating the relative humidity that is related to the concentra­tion of a solution identical with the composition of the soil water. The soil specimen should be placed in a desiccator containing an aqueous solution of a given chemical compound. Depending on the physico-chemical properties of the compound, a relative humidity is imposed within the desiccator. Water exchange occurs by vapour transfer between the solution and the sample via the vapour and a given suction is imposed when vapour equilibrium is reached. It should be noted that the technique generally requires a very long equilibrium time that can be a few months (1-2 and up to 6). Marcial et al. (2002) introduced an air circulation technique to reduce the equilibrium time from 6 months down to 2-4 weeks. Also, it should be noted that either the same products at various concentrations or various saturated saline solutions may be used.

In conventional practice, osmotic and vapour phase methods have quite often been combined to bring soil specimens to any desired point on the soil-water charac­teristic curve. The osmotic technique should be used to control low matrix suction (less than 2 MPa) and the vapour phase method for measurements of higher suctions (> 2MPa).

The repeated load triaxial testing (RLT) (also known as the cyclic triaxial test) method is commonly used to establish the mechanical characteristics of granular materials. During the testing, a cylindrical specimen is compacted to a desired level and then tested by applying confining and vertical stresses. Two variants exist:

• a constant confining pressure (CCP) method; and

• a variable confining pressure (VCP) method.

In the CCP-method the sample is initially subjected to a hydrostatic confining pres­sure ac, which induces an initial strain ec (unmeasured in the test, but it is the same in

all directions for isotropic material behaviour and, thus, can be estimated). The axial stress is then cycled at a constant magnitude q, which induces the cyclic resilient axial strain As. In the VCP-method both the axial and the radial stresses are cycled. The RLT can be used to obtain both the stiffness characteristics as well as the ability of the material to withstand accumulation of permanent deformation during pulsat­ing loading (Gomes-Correia et al., 1999; Erlingsson, 2000). Figure 10.1 illustrates the general stress regime experienced in an unbound granular layer in a pavement structure as a result of a moving wheel load within the plane of the wheel track. Due to the wheel load, pulses of vertical and horizontal stress, accompanied by a double pulse of shear stress with a sign reversal, affect the element (Brown, 1996).

This stress regime associated with the vertical as well as the horizontal stress pulses can be established using the VCP-method in the RLT. Using the CCP-method the variation of the horizontal stresses is neglected, as the confining pressure is kept constant. For further details please refer to CEN standard EN 13286-7 (2004).

Resilient testing of granular material is usually divided into two phases: (i) a conditioning phase and (ii) a testing phase. During the conditioning phase 20,000 symmetric haversine load pulses are applied with the frequency of 5 Hz to stabilize the response from the specimen. Thereafter, different stress paths are applied to estimate the specimen’s response. Since the unbound granular materials show stress dependency behaviour, it is very important to apply a number of stress paths in order to observe such behaviour. During each stress path 100 symmetric haversine load cycles are applied with a rise time of 50 ms (total length of pulse 0.1 s) followed by a 0.9 s rest time. During the last ten load cycles data from the transducers as well as the axial load are collected to evaluate the specimen response, see Fig. 10.2.

For the permanent deformation estimation, constant amplitude symmetric haver – sine load pulses are applied in the axial direction usually with a frequency of 5 Hz, without any rest time between the pulses (Fig. 10.3). The stresses, cyclic numbers as well as the axial and sometimes the radial deformations are recorded during the test at appropriate intervals.

Although there are a wide variety of tests for assessing soils and road materials, any serious investigation will need to know the mechanical behaviour of these materials when subjected to repeated loading that simulates the effects of trafficking and under moisture conditions (water content and suction) that simulate that found in the layers of the road construction and embankment. Various devices have been developed including the k-mould (Semmelink et al., 1997) and the Springbox (Edwards et al., 2004), but the cyclic triaxial test has secured the greatest following for material assessment over many years and is now the subject of European, US and Australian standards (CEN 2004, AASHTO 2000, Standards Australia 1995). It is the use of this test that is described in this section.

In laboratory testing procedures it is well known that the size of the sample may have a very important influence on the results. If the size of the sample is not appro­priate for a test procedure, the results obtained may be corrupted and not valuable. In order to have a continuum condition in the sample, it is necessary to satisfy some conditions with regard to its microstructure and sample size. For fine soils, it can be assumed that the test sample should be some centimetres in diameter, roughly from 2 to 6 cm. However, for granular materials, the sample diameter should be much higher, from about 5 to 10-15 cm, depending on maximum grain size.