Evaluation of Interlayer Shear Bond Devices for Asphalt Pavements

For testing the bond between asphalt pavement layers quite a variety of methods were proposed and used during the last 30 years. In many publications some of these test methods and devices were described by presenting photographs and sketches. Often photographs show the functioning of the devices only insufficiently and detailed information regarding the test devices (e.g. gap width) and test conditions (e.g. loading function, normal force) are difficult to retrieve. The following paper summarises the test methods and devices for the determination of the bond between asphalt pavement layers regarding shear testing. Direct and simple shear tests from all over the world are presented and their mode of operation shown. Furthermore, the range of applications is described and information regarding test evaluation and results are given.


In troduction
Due to the increasing demands of the road users as well as increasing rehabilitation costs and decreasing budgets the design and construction of long lasting asphalt pavements is becom ing more and more important. Extensive research efforts are still under way world -wide, focusing on the optimisation of the mechanical proper ties of mixes in the individual layers. However, it was often neglected that not o nly the material properties of the individual laye rs but also the interlayer bond play an important role in achieving optimal long-term structural performance of a pavement (Raab, Part! 2004).
As shown in Fig. 1 the bo nd between asphalt layers is extremely important for the bearing capaci ty and the long term performan ce of pavements, a fact that has beco me more widely accepted during recent years and led to adhesion testing as a subject of study and a development of many different test methods and procedures to evaluate the bond between pavement layers over the last decades.
111e reason why it has taken long to formulate qualitative requ irements fo r the bond between the layers of an asphalt pavem ent may certai nly have to do with the great number of parameters influencing this bond as well as thei r interactions. 111e complexity of these interactions is also the reason for the difficulties to quant ify the single parameters. Fig. 2 names some of the most impo rtant para meters fo r a durable bond between the layers. By listi ng the ISSN I R22-427X print I ISSN 1822-4288 online /1tt1>:!/www./)jr/Je. vgtu.11 1111111111111 d ifferent parameters separately it becomes clear that thcrl' are many interactions between them. For example mineral aggregate size, binder properties and m ix tu re compositio11 are influenced by the chosen pavem ent type, while they an· responsible of the friction and the inte rlock properties. Consequently it is not surprising that a lot of d iffer ent methods have been proposed to determ ine the bond between pavement systems. 111e following figure (Fig. J) gives a schematic overview of possible test methods and their application ranges.
111e choice of a certai n test m ethods depends on th1 assumed loading mode and the type of applicatio n (e.g. ill situ, laboratory), the problem area (e.g. bond failure due ht tensile stresses, such bl isters o r fai lure due to shear stres~ es) as well as the accu racy and repeatabi lity of a cer1ai11 test method. DOI (Raab, Partl 1999) During recent years many European countries as well as the United States and Canada have established methods and equipment for testi ng the interlayer bond. On lhc one hand, there are methods commonly used in different cou ntries, such as the Leutner shear tesl ( Lcutncr 1979) which was taken into the national test speci fical ions in Germany shortly after its standardisat ion in Switzerland and Austria. On the other hand specifi c solut io ns such as the wedge splitting test (Tschegg et al. 2007) or I he torsion test British Board of Agreement (B B/\). Cuiddine document for the assessment and cerl ifi calion ol lhin suriacing systems for highways, 2004, Choi et 11/. 2005) were proposed. In Italy the ASTRA shea r apparatus (Canestrari, Santagata 2005) was developed and will shortly become a national specification. In the US /\ int erlayer bond testing has become a serious isSlll'. As a potential option fo r testing the bond bclwccn asphalt laye rs, (Mohammad et al. 2002) designed a <. :uslom made shearing apparatus fo r use in th e Su pcrpavc Shear Tesler. In Canada, Carleton Uni-Ill versity, has also been working for many years on the development of an in-situ shear tester (Abd El Halim 2004).
TI1e different test methods, including the various equipment, have been presented in numerous publications (Canestrari et al. 2005;Kruntcheva et nl. 2006;Raab, Partl 1999;Stockert 2001;West et al. 2005) in such a way that photographs of the various devices are depicted. But often from these photographs the functioning of the devices is not too clear and detailed information regarding test devices (e.g. gap width) and test cond it ions (e.g. loading function, normal force) are difficult to obtain.
The following paper tries to give a more complete overall overview of the most important test method fo r interlayer bond testing, the shear testing, highlighting the differences in terms of test d evices, testing specifications and test results for the different devices and cou ntries. TI1e concentration on shear testing was chosen since that test method has been by far the most common method to determine the bond between asphalt pavements. Although, there are many different devices, some of them have already been standardised and common test specifications (deformation rate, test temperature) have already existed in a few countries for some time.

Sh ear testi ng
The construction of shear testing devices for asphalt pavements originally was derived fr om shear testi ng in soil mechanics and already in the late I 970ies different equipments such as the Leutner test (Leutner 1979) in Germany or similar tests in th e US were developed (Uzan et nl. 1978). There are two fundamentally d iffe rent systems: the direct and the simple shear test.
The direct shea r test, in general, is a guillotine type test where the shear force is induced directly at one side part and not at the front su rface of the specimen (Fig. 4).
a The direct shear testing d evices, as depicted in Tables I and  2, can be divided in devices which use a clamping or fitting system to hold the test specimen {Partl, Raab 1999;Romanoschi, Metcalf 2002;Sholar et nl. 2004;West et al. 2005;Zeng et al. 2008) and devices which utiliz.e a bending mechanism (3 or 4 point shear tests) to apply the shearing (De Bondt 1999;Miro Recasens et al. 2003).
In the simple shear tests {Table 3) the upper part of the test specimen is sheared against the bottom part of the test specimen and the shea r force is induced at the specimen front surface of the specimen. In the case of a th ree layered specimens (De La Roche 1996;Milien et al. L 996) the middle part is sheared against both outer parts. For the simple shear test, as depicted in Table 3 the mechan ism of the different devices is similar, di ffering mainly in the way the shear forces are applied and how both parts of the test specimen are moved against each other (Canestrari et al. 2005;Sanders et al. I 999).
As opposed to the direct shear tests, where the test specimens can either be clamped or fitted into steel moulds. the test specimens in the simple shear test are always fitted into the shear mould by glue or tight fixtures. Therefore, here the application of a norm al force vertical to the shear plane is always an optio n. Whereas in shear test devices using clamping mechanisms, normal forces are often not taken into consideration. Another possibility to include a normal force was developed by Romanosch i whose testing device allowed for the long itudinal axis of the test s pecimen being at a 25.5° angle with the vertical (Romanoschi,Metcalf 200 I).

111111
Since in Europe most direct shear devi ces where designed to be mounted in a servo-hydraulic Marshall testing machine, the tests were normally co nducted defo rmation controlled at a rate of 50 mm/min.
Mostly, cylindrica l test specimens of I 00 mm (Austria) or 150 mm (Germany, Switzerland ) taken either directly from th e road or laboratory specimens were tested (Stockert 2001 ). Some devices, such as the modified Em pa direct shear device LPDS, could also be used to measure the bonding of rectangular test specimens (Raab, Partl 2007). Normally the specimens were conditioned at lest temperatures betwee n 20 °C and 25 °C. Only in the case of research projects have other temperatures of between 10 °C and 40 °C been looked at (Part!, Raab 1999). However, they were fo und inappropriate for quality assessment, since the higher the temperature, the more diffi cult to fi nd distinct differe nces between the different asphalt pavements. Furthermore, specimens may already be damaged dming conditioni ng or during testing (clampi ng of the specimen).
Most European countries (besides Germany, Austria, Switzerland and th e UK) adopted the Leutner equipment, modifying il slightly, for qualit y assurance of constructi on sites (Aus trian Standard RVS BS. 04. 11: 2004) Bending type test set ups where developed for research purposes in Spain and the Netherl ands (De Bondt 1999). The Spanish device known as the LCB shea r test was developed at the Technical University of Catalonia, Spain (Miro Recasens et al. 2003 ). Here, cyli ndrical test specimens were tested at a deformation rate of only 1.27 mm/min. At Delft University in the Netherlands de Bondt (De Bond t 1999) developed a four point shear test where bending effects were minimized through special arrangement of loading and supporting points.
At the Techn ical University of Dresden th e development of a dynamic version of the Leutner shear test, is under way. 111is dyn amic device was constructed by Ascher and also allows for a normal force (Ascher, Wellner 2007). In the dyn amic testi ng of the bond different parameters such as temperature (-10°C,+ 10 °C and +30 °C), norm al load (O to 1.11 N/mm 2 ) and the loading function (sinusoidal fun ction with amplitudes from 0.005 to 0. 1 mm and a frequency from J to 15 Hz) were included. 111e purpose of the project was to find a "bonding factor" which can be used for pavement design in BISAR or in finite element programs.
For the simple shear test deformation rates between 1.5 mm/min in the UK (Sanders et al. 1999), and 2.5 mm/min in Italy (Canestrari, Santagata 2005) were used. In the UK the direct shear test was normally conducted in the dynamic mode, where the speci mens were tested under a sinusoidal shear stress with a frequency of 2 Hz. Wh ile the vertical stress was kept constant at 200 kN/m 2 , the shear stress was increased in 5 levels (SO, 100, 200 and 250 kN/m 2 ) until the specimen fail ed. If the specimen did not fail during dynamic testing, a static test was performed using the above mentioned deformation rate of 1.5 mm/mi n. In Italy shear tests were conducted in a static mode using different normal loads (O, 0.2 and 0.4 MPa).
The specimens in the simple shear test, were found to be either prismatic (320x200 mm) (Sanders et al. 1999) or rectangular (max cross section of 1OOx 100 mm) an d cylindrical with a diameter between 95 mm and 99 mm (Canestrari et al. 2005).
A simple dynamic shear test for glued three layered specimens, known the Modi fied Compact Shearing (M CS) test (Millien et al. 1996;Diakhate el al. 2006) was developed at the Laboratory "Mechanic and Modelling of Materiab and Structures in Civil Engineering (3MsCE) of the Uni versity of Limoges in Fran ce. 111e device allowed conduct ing static or dynamic tests on glued three laye red specimem test specimens with the dimension of 70x I OOx30 mm . 111e specimen was placed in a metal frame where the sick parts of the sample are fi xed while its central part was sub jected to a sinusoidal displacement, causing a shear force at both interfaces. The aim of the test program was th~· investigation of shear fatigue tests of asphalt concrete lay er interfaces with emulsions at a constant temperature ol 5 °C and a frequency of l Hz.
In the US direct shea r testi ng was generally used in quality assess ments and research projects, where the ma in focus was on the evaluation of bonding properties of di ! ferent tack coat types and tack coat application rates. Di f ferent DOTs, asphalt pavement institutes or universitie\ evaluated or modified various guillotine type shear test devices using different clamping and fixi ng mechanism (Leng et al. 2008;Sholar et al. 2004;West et al. 2005). A~ depicted in Table I the device differed in the fixing mech an ism of the specimen as well as in the specimen di ameti:r and the deformation rate of the testing machine. 111e Iowa Department of Transportation shearing device, a modi Ii cation of the shearing device for Portland Cement Con crete (PCC)  for Determining the Shea ring Strength of Bonded Concn:ll' by lowa Department of Transportation Highway Divi sion), was built fo r 100 mm diameter cylind rical speci mens (either roadway cores or laboratory specimens) and with a gap width of 3.1 75 111111 between its steel sheari 1111 platens. Further modifications used aluminium rings of 150 mm and a width of 4.8 mm between them to hold tll!' specimen (Sholar et al. 2004).
Some devices, such as the so called NCAT bo11d strength device (West et al. 2005), where the specimrn was held in a metal half cups, also allowed the applica tion of normal forces, which were chosen between 0 and 550 kPa (80 psi). For direct shear testing, specimen dialll eter generally varied between 95 mm and 150 mm and tlw deformation rate between 2.5 mm/ min, 12 mm/min and 50 mm/min, often depending on the available testing ma chine.
In the course of another research project Roman11 schi (Rornanoschi, Metcalf 2002) used a direct shear tt·~t device with no rmal load. 1he cores (0 95 m m) were fi1 ~I fixed in a steel split ring, with the interface positioned at the end of the ring. The half outside the steel ring"'"" then placed and fixed in a steel cup positioned vertically and welded to a ver ti cal supporting plate. The positinn of the in terface was adjusted at the rim of the cup u" ing a screwing piston placed inside the cu p. To generate the shear at th e interface, the vertical actuator pushed on top of the steel split ring with the constant d isplacem ent ( 12 mm/ min) until a shear displacement of 12 mm was reached. To this day in the United States different modified Leutner type shear test devices such as (Leng et al. 2008) have been developed and va rious research projects are still underway. For his research Al-Qadi (Leng el al. 2008) developed a fixture where the test specimens were housed in a special steel camber with a diameter of about 100 mm. TI1e device was designed to apply shear force in the vertical direction and normal fo rce in the horizo ntal.
To simulate the repetitive load of moving vehicles, in another study Romanoschi an d Metcalf (2002) proposed a test configuration to conduct shear fatigue tests on asphalt concrete laye r interfaces. The longitudinal axis of the specimen was tilted 25.5° to the vertical. A vertical load was appli ed with 10% of the max load and with a frequency of 5 Hz. So, the total period was 0.2 s an d the length of the pulse was 0.05 s, simulating the pass of a vehicle at 50 km/ h. The corresponding normal stresses at the inte rface, 0.5, 0.75, 1.0 and 1.25 MPa were within the range of no rmal stress values for interfaces of road and airfield pavements.
The elastic and permanent displacements at the interface in normal and tangential directions were recorded for each cycle and the dynamic tests were stopped when the permanent shear displacement (PSD) at the interface reached 6 mm o r when it was co nsidered th at the number of cycles corresponding to a PSD of 6 mm could be extrapolated.
In the course of the American research program SHRP (Sousa et al. 1994) a relatively complicated test device for performing simple shear tests, the so call ed Superpave shear tester (SST) was developed. Originall y the device was not used to evaluate the interlayer shear properties between pavement layers, but to determine permanent deformation and the modulus of asphalt layers.
The SST consisted of shear and axial actuators, load cells and deformati on measurement systems, computer control and data acquisition systems, a temperature co ntrol and a hyd raulic pump. TI1is machine uses closedloop computer drive n control hyd raulic pistons connected to vertically and h orizontally operating platens. The specime n was normall y glued onto aluminu m "caps" which were hydraulically clamped to platens inside the temperature control chamber. Mohammad et al. (2002) p er fo rmed simple direct shear tests on various types of tack coat materials at several spread rates using laboratory fab ricated asphalt specimens. A custom made shearing appara tu s was designed and fabricated for use in the SST. Specimens were fabricated in the gyrator y compactor in two lifts wi th a tack coat applied prior to compaction of the second lift. The apparatus was mounted inside the SST and the tests we re conducted in constant load mode (222.4 N/min).
No normal load was applied to the specimens. ·n1e tests were co nducted at 25 °C and 55 °C.
As opposed to bo nd testi ng using pull -off or to rque devices, shear testing is generall y performed in the laboratory. In the early 1980s Empa developed a method for shear testing in situ. TI1e shea r test with a truck tire was used to test the adhesion between bituminous surface courses and cement concrete laye rs. Additional to the horizont al shear fo rce a vertical force induced by a single truck tire was app lied duri ng the test and the caused d eformations were measured (Empa rep ort 1985, not for publi c use).
In some European countries bond testing was stan dard ised during the 1990s. Although the requ irements often stayed below the limits sh own in diffe rent research projects, standard isation was a fi rst step usi ng shear bond testing on a regular bases in quali ty cont rol. Research by Raab and Part I ( 1999;2008) fo r example showed that for pavements with stone mastic asphalt (SMA) and asphalt concre te surface courses, a max shear fo rce of 21 kN or 18 kN fo r the adhesion between the base courses could easily be obtai ned fo r 150 mm cores. Nevertheless, Swiss specification only required a max shear force of 15 kN be twee n surface and binder course a nd 12 kN between a binder and a base course or between two base courses. TI1eses values correspond to 1.3 N/mm 2 for the adhesion between surface and base course and l. l N/mm 2 be tween two base courses when using the shear strength the values. In Germany a research project lau nched by the German Road Authorities in 200 l (Stockert 200 I) and based on approx 500 cores with SMA o r AC surface course d elivered similar results and proposed the following require ments for the adhesion between the laye rs: 25 kN for the adhesion surface cou rse/binder course; 20 kN for the adhesion binder course/base course; 16 kN for the adhesion surface course/base course. In Austria the adhesion testing according to Leutner was conducted on I 00 mm specimens and a test temperature o f 20 °C ± 1°C. According to t he RVS BS.04. I I: 2004 fo r SMA a nd AC surface and the binder course a m in shear strength o f 0.8 N/ mm 2 was required when using a non m odified b inder and 1.2 N/mm 2 when using a modified binder tack coat. For binder and base courses or two base course layers the requirements were 0.5 N/mm 2 fo r non modified and 1.0 N/mm 2 for polymer modified tack coats. The shear strength in Aust ria had to be measured p ara llel to the direction of the traffic.
ln Tables I to 3 schematic drawings of the different direct and simple shear devices are presented. Si nce the shear equipment was ofte n not included in standa rds a nd testing specifications, th e main test parameters such as specimen dimension (core diameter), deformation rate (test speed), test mode (static, dynamic), normal fo rce, temperature, and o th ers parameters such as the gap width between the shear plates accordi ng to special are also given.

Discussion
As Tables I to 3 depict there is a great variety of test devices to test the shear bond of asphalt pavements. TI1e shear tests are inspired by shear testing in soil mechanics and application with or without normal force are used. The application and influence of normal force is one of the issues which have been under debate for quite some time. Many researchers argue that the normal force, representing the wheel load on the road, has to be included in interlayer bond testing. Regarding its influence (e.g. the magn itude of normal force) different opinions and find ings are being d iscussed (Romanoschi, Metcalf2002;Uzan et al. 1978).
Furthermore, when looking at the presentation and interpretation, as well as the comparison of the test results from different shear devices, no uniform opinion is available. Although some common statements such as the dependency of adhes ion tests on temperature or deformation rate are not debated among researchers, there are many divergent results regarding the influence of normal st ress, tack coat and surface roughness on the adhesion properties (Raab, Partl 2004;Romanoschi, Metcalf2002;Uzan et al. 1978;Ziari , Khabiri 2007).
Especially for quality assurance, standards and testing specification only require the interlayer bond values in form of forces since lest specifications prescribe specific test specimen diameters. This method is easy for comparison of speci mens of equal size, but has a disadvantage for the comparison with other results.
Another distinction between different test devices is their workability and the simplicity of performing a test. Here, devices using clamping mechanisms are preferable over devices where the test specimens have to be glued into moulds. TI1e more time is needed fo r specimen preparation, and the more cumbersome a test set up becomes (e.g. the MCS device), the greater the influence of unknown variables on the test results and test devices are not likely to be used for daily quality assurance. Looking at the guillotine devices, the different clamping mechanisms play an important role for the workability but they are also important for a defined pressure with which specimens are held during the test (e.g. as in the Empa test device). Furthermore, devices usi ng prismatic specimens are more practical especially fo r quality assurance since field specimens are mostly dri lled cores and even a lot of laboratory specimens such as Marshall and gyratory specimens are prismatic. Some devices are flexible in a way that they allow for the testing of either prismatic or rectangular specimens (e.g. Empa test device, ASTRA test device). Anot her advantage of the guillotine (Leutner) type devices is that they are very flex ible since they can be installed in a common universa l testing machine requiring no special test set ups and constructions.
111at the comparison of di fferent test devices as well as their results and outcome becomes a more and more important issue shows the inter-laboratory test program initiated by RJLEM. Here, research and materials testing institutions from Europe and North America were asked to perform shear tests on pre-selected and defined mate-1111111 rial under certain test conditions using their specific shear test equipment (Piber et al. 2009). First investigations show that a comparison of results in case of the Leutner device (or some of its modified versions) leads to similar findings and tests using 100 mm or 150 mm specimens provide similar results.

. Conclusions
TI1e paper presents an extended overview on the existing test shear test devices and gives detailed information on the fun cti oning mechanisms (device figures) and test specifications.
Looking at the different publications and devices the followi ng statements and conclusions can be drawn.
Shear testing seems to be a good and effective method fo r testing the interlayer bond of asphalt pavements.
In many publications some of these test methods and devices are described by presenting photographs and sketches. Often photographs show the fun ction ing of the dev ices onl y insufficiently and detailed information regarding the test devices (e.g. gap width) and test conditions (e.g. loading function, normal force) are diflicull to retrieve. TI1erefore, detailed drawings showing the mechanism of a device as depicted in this paper are preferable.
For the construction of test devices it is important that the test set up is not complicated and the installation of test specimens is simple. Clamping mechanisms are often preferable over set ups where specimens have to be glued or fixed into special moulds. Wh en clamping the specimen, care has to be taken that this procedure docs not damage the specimen and does not influence the test results. TI1erefore, it is important that a defined pressure is used and that the specimen is not tilted during the test.
The gap between upper and lower part of the shear moulds has to be small enough not to induce a bendi ng moment. The device itself has to be sufficiently stiff to enable the occurring forces to be accommodated.
Although shear failure normally occurs in warm cl imate, moderate test temperatures (around 20 °C) seem to be preferable, as compared to testing at hot temperature the danger of damaging the specimen during testing is smaller.
For the comparison of different test devices, it is important that test parameters such as normal force and deformation rate are comparable. TI1e application of normal forces has an influence on test results and more research is necessary to clearly work out in which way.
Regard ing the results from interlayer shear bond testing it is important to compare the outcome of different devices and methods in a detailed way. TI1e above mentioned Rilem interlaboratory test provides a first step in this d irection, but here definitely more research is needed.