Bridge decks expand and contract during a year due to temperature variations, as any other non-restrained structure. The amplitude of this movement depends on the type of bridge, its length and the climatic circumstances. The standard method of connecting the pavement of a bridge deck to the pavement on an embankment, is by making a joint, often consisting of steel grips or a soft asphalt mixture. This has always been problematic with respect to comfort (for drivers as well as people living in the neighbourhood) and performance. Reasons for this are, that joints restrict runoff of rain and ice, create noise, cause bumps, show ruts and allow water penetration (and subsequently erosion of subbase/soil layers). The paper describes a specific project where extensive three- dimensional finite element analyses, in combination with sufficient material testing, have resulted in a cost-effective proposal for a jointless pavement structure; this including tendering specifications.
Since the introduction of continuous pavement structures such as the ones built by means of the elasto-viscoplastic material bitumen, the transition between the pavement on the bridge deck and the pavement laying on the natural soil has been a problem. If the asphalt pavement is simply paved without measures onto the bridge deck, one can expect after a few or even during one severe winter, that wide cracks become visible at the bridge end. This is due to the large strains which are generated in the asphalt concrete layers, especially at its bottom fibre. These wide cracks will not only allow water penetration into the foundation layers, but will also cause ravelling to occur at the crack edges at the pavement surface. Also the faces of the crack itself will deteriorate by the repetitive shearing actions of trucks, implying progressive crack widening in time, de Bondt (1999).
Given the problem described above, several types of joints have been developed and applied over the past years. However, these have in common that their lifetime is short and difficult to asses. This means that quite rapidly and often unexpectedly (costly) maintenance, in the form of replacement is needed, Maijenburg (2000). In the situation such as in the Netherlands, where the current motorway system is already loaded beyond its capacity, closing lanes for joint maintenance causes a lot of disturbance and is unacceptable from the user point of view.
It is clear that long lasting jointless asphalt pavements at bridge ends should be developed. More specific, the wearing course layer should be a continuous layer with a maintenance interval for the aspect cracking, which is at least similar to the maintenance interval of the bridge itself (often planned every 50 years). This is a much more stringent
1 Ooms Avenhorn Holding bv PO Box 1, 1633 ZG Avenhorn, the Netherlands, adebondt@ooms.nl 2 Unihorn bv PO Box 58, 1633 ZH Avenhorn, the Netherlands, jschrader@unihorn.nl 2 requirement than simply demanding that the wearing course layer should be free of cracks as long as the wearing course layer itself lasts; the latter period is in most cases limited by ravelling, rutting or lack of skid resistance and normally varies between about 8 and 15 years. It is important to realize this aspect, because placing a new wearing course on an underlying pavement, which shows (active!), wide cracks implies certainly within a very short period of time (a few or even one single summer/winter cycle), that a reflection of the crack pattern becomes visible.
In the paper, a specific project will be described where based on three-dimensional finite element analyses, supported with asphalt mixture tensile testing, layer interface (bond) testing and reinforcement single-end tensile testing, tender specifications for a durable jointless pavement have been drawn-up. These specifications are of a so-called functional nature (the asphalt mixture is defined in terms of mechanical properties instead of in terms of composition).
PROBLEM DESCRIPTION
The motorway A50 in the Netherlands, which runs from Eindhoven to Emmeloord, has a missing section between Eindhoven and Oss. At the moment construction has started to built this part (opening is planned at the end of 2003). A number of special type of bridges have been designed for this stretch by the Engineering Office on Bridges and Tunnels of the Dutch Road Administration (Bouwdienst Rijkwaterstaat). Some of these bridges have already been built, in order to ease the road construction process (transport of subbase material). The so-called integral bridges, varying in length between about 20 and 70 m, have in common, that they have a special type of support at the ends. More specifically, the (continuous) concrete bridge deck only rests on steel bearing piles. This method is attractive from several points of view (construction costs, esthetics, etc.), as summarized by Maijenburg, (2000). It is clear that given the relatively low rotational stiffness of the supports, as compared to the bridge power, a considerable thermal movement at the bridge ends needs to be taken into account, when designing the transition to the road pavement. Figure 1 shows a photo (dated end of 1999) of the 70 m long bridge at Son (crossing the Wilhelmina-canal), in the phase that earth construction work of the road has still not been started. This bridge was finished in 1997.
Figure 1 Overview of Situation of the Bridge at Son at the End of 1999 3 The area in which the road is being made, has good supporting conditions; at least from the Dutch point of view, since the soil consists of sand. However, given the fact that an embankment needs to be made, still so-called approach slabs will be utilized. The function of an approach slab is to create a smooth vertical pavement surface profile in case of settlements; in other words to avoid sudden bumps when hitting the bridge deck pavement. For this reason, these slabs are placed on an angle (see figure 2). At the bridge Son this angle is equal to 2.6 .
Figure 2 Explanation of the Function of an Approach slab
The connection between the bridge deck itself and the approach slab is via steel bars, which are embedded in such a way that only rotations are possible (in case of settlements). This configuration implies that the approach slab will be subjected to the thermal expansion and contraction process of the bridge. Given the length of the bridge (70 m) and the length of the approach slabs on both sides (each side 5 m), it is clear that the amplitude of the thermal movements (summer/winter cycle) is extremely large.
From the foregoing it can be concluded that if the wearing course layer should be a continuous layer with a maintenance interval for the criterion cracking (caused by the thermally induced bridge movement), which is at least similar to the maintenance interval of the bridge itself (a period of 50 years), a complicated design problem would arise. The Research & Development department of Ooms Avenhorn Holding took this challenge at the end of 1999. All in all, the challenge was defined as follows:
Design a cost-effective jointless pavement near a bridge end (including the preparation of tendering specifications), which can sustain 2 mm daily movement (day/night) and 20 mm seasonal movement (summer/winter); this for a period of 50 years (under Dutch climatic conditions)
It is obvious that this goal could not be achieved without finite element modelling, given the complex geometry. The development work started by preparing a three-dimensional finite element mesh of a composition of the pavement structure which was at that time thought to be adequate. In the end, it became apparent that two major (and time consuming!) changes in the mesh configuration had been necessary. 4 FINITE ELEMENT MESH
A sketch of the superelement configuration of the final three-dimensional mesh is presented in figure 3 (note that the horizontal and vertical scale are different). The mesh has a width of 3500 mm and consists of 60 superelements, representing 40 different materials or interfaces (encircled in figure 3); the latter simulating friction, adhesion or bond. The superelements were eventually subdivided into 6515 elements: 5265 cubic elements and 1250 interface elements. The program CAPA-3D, Scarpas and Karsbergen (1999) was used for the analyses. 36 35 3 27 5 29 7 31 9 33 13 22 34 32 30 28 33 31 29 27 10 8 6 4 9 7 5 3 14 24 1 1 1 1 18 11 15 17 26 2 1 2 1 2 1 2 1 19 18 12 11 16 20 21 11 25 25 25 26 23 20017 mm 3 6 5 0
m m Road Pavement Bridge Pavement
Figure 3 Sketch of the Superelement Configuration (not on scale!)
The geometry is quite complicated. Table 1 gives an explanation of the material / interface numbers given in figure 3 (note that the reinforcement elements were not shown).
Material / Interface Number Description 1/2/11/12/15/16/18/19 Asphalt Concrete Layers 3/4/5/6/7/8/9/10/27/28/29/30/31/32/33/34/35/36 Pavement Layer Interfaces (Bond) 13/14/20 Stress-Relieving System 17 Unbound Granular Base Course 21 Approach Slab (PCC) 22 Dry Friction Simulation 23 Cement Stabilized Sand 24/25 Air (Simulation of No Contact) 26 Sand Subbase Course 37/38/39/40 Asphalt Reinforcement
Table 1 Explanation of Superelements 5 The solution which was finally adopted, can be described as follows: on a gravel asphalt ne of the advantages of this solution is, that it fits well with the road pavement (left side in Figure 4 concrete layer (which acts as working platform) and the PCC-approach slab, a stress- relieving system, consisting of 3 mm polymer modified bitumen (PMB) is sprayed. After this, an extremely ductile (but still stable enough) 30 mm thick special type of asphalt concrete (called Thermifalt) is layed down. On top of this, a stiff and strong asphalt reinforcement is attached. Then two more of these special asphalt layers are placed, each with reinforcement put on their surface. On top of this, three standard asphalt mixtures of each 60 mm thickness are layed (including the use of a high quality PMB). Again, reinforcement is placed on their surface, except for the top layer of these three. This means that all in all, five layers of asphalt reinforcement are planned (note that the model has four reinforcement layers). Finally, a 70 mm thick porous asphalt wearing course will be paved (also on the road pavement).
O figure 3) and the bridge deck pavement (right side in figure 3). Figure 4 shows the entire subdivided mesh utilized for the final analyses (note that the angle of the approach slab is clearly visible).
Mesh used for the Final Analyses he main purpose of using the finite element method was the investigation if cracking would occur (and to what extent) in the cross-section at the end of the PCC-approach slab (see figure 5 for a detail of the mesh at the centre location).
T 6
Figure 5 Detail Mesh Focussed on the Important Location OADING wo major loading cases have been discerned: an extreme daily temperature drop and an asonal temperature drop. The extreme daily one has been represented by a hange in temperature from 0 to 15 C; for the seasonal drop these values were +35 and h slab (at the location of the connection with the ridge itself, where the steel bars are). This approach saves a lot of elements (reduces the
L
T extreme se c 15 C. Based on analyses performed by Maijenburg (2000), it can be deduced that the bridge movements (at the end of the approach slab!) are: 2 mm for the extreme daily case and 20 mm for the extreme seasonal case. Note that because of the time needed to warm- up the bridge, these displacements are not related to the temperature interval (15, respectively 50 degrees Celsius) in the same way. Only in case of the seasonal case (the long period), the bridge movement can be directly related to the length (70 m), by simply using the well-known equation T L.
The simulation of the bridge movement has been carried out via setting prescribed displacements at the end of the approac b mesh size) and is allowed from the reliability point of view, if the displacements along the approach slab itself do not vary. Given the difference in length between the bridge itself and the approach slab the latter is no problem. 7 Note that this approach is not allowed for the analysis of reflective cracking in bituminuous surfacings which are laying on slabs in pavements. Inputting the actual temperature drops, the individual coefficients of thermal contraction of each layer and modelling half the slab quality subgrade material and) and the fact that only standard truck traffic passes. linear elastic approach has been followed. Since some of the material and interface ture and displacement (strain) rate dependent, it meant that anual interations had to be carried out, in order to have representative input parameters d roach Slab and Cement Stabilized Sand ven by CUR (1998). These rules ere transformed in such a way that they were applicable for a linear elastic approach. (N/mm)/mm 2 for the ch was carried out at e Research & Development Laboratory of Ooms Avenhorn Holding. These cores were were made in a routine way by means of standard spraying seasonal mperature drop. These stiffness values are based on a temperature of 15 C and a rate (using symmetry) is then necessary, see de Bondt (1999, 2000).
The effect of passing traffic has not been taken into account, given the considerable thickness of the entire pavement structure (340 mm), the good (s
MATERIAL PROPERTIES
A characteristics are tempera m for the analyses. For the first run, this meant that interface displacements (slip values) had to be assessed from the given bridge movements. From table 1 it is clear that an extensive list of material data was inputted. The most important data is discussed in detail below.
Approach Slab (PCC) The Youngs modulus of this material was set to 35000 MPa.
Cement Stabilized San A value of 4500 MPa was used for the Youngs modulus.
Dry Friction between App The input data has been based on dry friction rules, as gi w Shear stiffnesses used for the interface elements were equal to 0.025 daily drop and 0.0025 (N/mm)/mm 2 for the seasonal drop. The so-called normal stiffnesses of the interface (perpendicular to its plane) were 4500 (N/mm)/mm 2 .
Stress-Relieving System The input data originated from interface shear testing on cores whi th drilled from sections which trucks. Figure 6 shows typical data from this testing (4 specimens). It can be seen that the interface shows a ductile behaviour, even at this low temperature (-10 C). It is mentioned that in this situation the end of the test is caused by limitations of the current (standard available) test equipment; when the test ends the two opposite faces of the stress-relieving system are still sticking together; this even at a slip of nearly 8 mm. Different series of tests were performed. Using the whole data set, a mastercurve was derived which enables to deduce (extrapolate) input data for the extremely slow movements which occur during the summer/winter cycle. Details can be found in de Bondt and van Rooijen (2002).
The inputted shear stiffnesses for the interface elements were 0.3 (N/mm)/mm 2 for the extreme daily temperature drop and 0.1 (N/mm)/mm 2 for the extreme te of displacement (slip) of 1 mm in 6 hours (daily) or 10 mm in 2000 hours (seasonal). The normal stiffnesses of the interface were equal to 1350/300 (N/mm)/mm 2 for the daily drop, respectively 80/15 (N/mm)/mm 2 for the seasonal drop.
8 Interface Shear Testing - Overview Series i01 Shear Rate 0.0001 mm/s, Temperature -10 C 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 Slip (mm) S h e a r
S t r e s s
( M P a )
Figure 6 Example of Shear Test Results on Stress-Relieving System iven the typical nature of bitumen, mechanical properties of asphaltic mixtures highly epend on temperature and strain rate. Within this project, stiffness values for several binations were obtained by means of a so-called frequency ase of the daily temperature rop. These values were 70, respectively 15 MPa for the seasonal temperature drop. the inforcement was placed (embedded): 0.6 (N/mm)/mm 2 for the extreme daily temperature eme seasonal temperature drop. Note that this value with a specific xial stiffness, which is located (embedded) between two interface elements. The axial al terminology product stiffness, depends on the stiffness of the
Asphalt Concrete Layers G d temperature/frequency com sweep procedure; this for three different asphaltic mixtures. In addition, stiffness data was obtained from slow monotonic uniaxial tensile tests, which were carried out at several temperatures and strain rates. The latter experiments also provided information on critical strain value (strain at maximum load) and fracture energy. By making use of the whole data set, mastercurves could be derived, which enable extrapolation to the slow strain rates which are typical for seasonal thermal cycles. It is noted that this process is explained in more detail in van Rooijen and de Bondt (2003).
The stiffness values used for the design process were equal to 900 MPa (for the 60 mm PMA-layer) and 300 MPa (for the 30 mm Thermifalt layer) in c d
Pavement Layer Interfaces (Bond) The following values were used to decribe the pavement layer interfaces where re drop and 0.2 (N/mm)/mm 2 for the extr was used for each interface element (above and below the reinforcement).
Asphalt Reinforcement In CAPA-3D, reinforcement is modelled as a continuous, equivalent sheet a stiffness, or in more practic reinforcing material and the cross-sectional area of the strands per unit width. This parameter can easily be obtained from single-end tensile testing, de Bondt (2000). The required product stiffness value (EA) eq,rf was in this project found to be 4000 N/mm. 9 SHORT OVERVIEW OF COMPUTATIONAL RESULTS
An extensive series of computations on all kind of different alternatives has been carried ut, de Bondt and Schrader (2001); this during a period of almost two years. In this Figure 7 o section, a short overview of this work will be given; mainly focussed on the finally adopted solution at the bridge Son. First of all, a better understanding of the mechanisms behind the analyzed phenomenon has been obtained by plotting the deformed mesh from different perspectives. Figures 7 and 8 show in an exaggerated way (10 times enlarged), the deformations due to the seasonal temperature drop of the entire mesh, respectively of a detail at the critical location (at the left end of the approach slab!)
Overview of Deformations of Entire Mesh (Exaggerated)
10
Figure 8 Deformations around Critical Location (Exaggerated) is clearly visible how the approach slab slides over the cement stabilized sand and that e asphalt is brought under tension. rameters such as layer interface slip, asphalt tensile train and reinforcement force will be given; this for two computational cases where the
It th
In the figures below, distributions of pa s seasonal temperature drop has been applied (characterized by 20 and 26). It is mentioned that in the cases shown, the composition of the asphalt cross-section, etc. was similar. The only difference is that computation 20 simulates the behaviour of the final solution in its virgin phase, whereas in case of computation 26, the situation is represented that the bottom asphalt layer (one of the three layers with the special type of asphalt) has cracked.
11 First of all, figures 9 and 10 present the computed slip values at the location (depth) of the Figure 9 stress-relieving system. This at the road side of the approach slab (left of the critical location), respectively on the approach slab (right of the critical location); this for the mentioned cases.
Layer Interface Slip along Stress-Relieving System Left of Critical Location 0 1 2 3 4 5 6 7 -15000 -13500 -12000 -10500 -9000 -7500 -6000 -4500 -3000 -1500 0 Distance to Critical Location (mm) S l i p
( m m ) Computation 20 Computation 26
Slip along Stress-Relieving System (Left of Critical Location) Figure 10
Layer Interface Slip along Stress-Relieving System Right of Critical Location -7 -6 -5 -4 -3 -2 -1 0 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 Distance to Critical Location (mm) S l i p
( m m ) Computation 20 Computation 26
Slip along Stress-Relieving System (Right of Critical Location) can be seen that the plots of the slip distributions are quite similar left and right of the critical location. Furthermore, it can be observed that, as expected, cracking of the bottom layer implies that a lower degree of slip along the interface occurs.
It 12 Figure 11 gives the strains in the asphalt layers along the pavement cross-section.
Tensile Strain Distribution in Asphalt at Critical Cross-Section 0 50 100 150 200 250 300 350 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Tensile Strain (%) D e p t h
( m m ) Computation 20 Computation 26
Figure 11 Computed Asphalt Strains for Several Cases also be concluded that e cracking of the bottom layer hardly influences the strain distribution. The computed rce in the deepest layer of reinforcement (on top of the first layer of 30 mm special
Figure 12
The transition between each asphalt layer is clearly visible. It can th fo asphalt, which means 310 mm under the pavement surface) is given in figure 12. It can be concluded that the required anchorage length is several meters. This is due to the large amplitude of the (slow) movements. It can also be seen that cracking of the bottom layer causes that this reinforcement is more activated. Distribution of Reinforcement Force Reinforcement at Depth 310 mm 0 10 20 30 40 50 60 70 80 90 -15000 -12500 -10000 -7500 -5000 -2500 0 2500 5000 Distance to Critical Location (mm) F o r c e
( N / m m ) Computation 20 Computation 26 Computed Force in Reinforcement for Several Cases 13 EVALU C is obvious that each alternative solution had to be evaluated whether it would meet the ith respect to the stress-relieving system, the ultimate slip was used as a criterion, de ll in all, it became clear that the summer/winter case was more damaging than the
Figure 13 ATION OF OMPUTATIONAL OUTPUT
It 50 year lifetime requrement. This meant for the asphalt concrete layers a check on the critical strain; given the severe demands for this bridge, this parameter (which is strain rate dependent) was chosen as criterion and not the (less severe) fracture energy. It was found that after an extremely strong winter, the bottom asphalt layer would crack. However, the strain in the asphalt layers directly on top of this layer, is low enough to expect no further progress of cracking (see also figure 11).
W Bondt and van Rooijen (2002). The reinforcement was checked via its tensile strength value (100 kN/m per layer).
A day/night case. In order to illustrate the mechanisms which occur, an evaluation of the forces which are acting on the approach slab and the asphalt is given in figure 13. Values are given per meter width of the bridge (note that these are rounded off).
4 6 0 kN / m A p p ro a c h S la b Asphalt 5 5 kN / m 2 5 0 kN / m 2 5 0 kN / m 5 5 kN / m 5 5 kN / m 2 1 0 kN / m 250 kN / m 45 kN / m 1 1 1 1 1 1 1 1 1
Sketch of Equilibrium of Forces (Free Body Diagram)
It can be seen that with the current configuration (and input data) roughly half the restraint force is developed by the jointless asphalt pavement and roughly half the restraint force by friction between the approach slab and the cement stabilized sand underneath. An interesting aspect is that the generated force in the steel bars which connect the approach slab and the bridge was higher than expected by the bridge engineers. This had led to some design changes.
14 Figure 14 presents a sketch of the forces along the critical cross-section in the asphalt.
Figure 14
Detailed Sketch of Forces along Critical Cross-Section
It can be deduced that the asphalt takes 50 % of the generated force in the cross-section and the reinforcement 50 %, in the phase represented by computation 20; after the bottom asphalt layer has cracked (computation 26) these values are 40, respectively 60 %. Computation 20 35 kN / m 1 10 kN / m 1 60 kN / m 1 35 kN / m 1 N
/
m 1 15 kN / m 1 20 kN / m 1 20 kN / m 1 25 kN / m 1 30 kN / m 1 4 6 0
k Computation 26 40 kN / m 1 20 kN / m 1 20 kN / m 1 20 kN / m 1 25 kN / m 1 30 kN / m 1 10 kN / m 1 75 kN / m 1 0 kN / m 1 4 5 0
k N
/
m 1 15 CONCLUSIONS AND RECOMMENDATIONS Based on the work described above, it can be concluded that via adequately detailed three-dimensional finite elements modelling, in combination with sufficient material testing and the use of high quality materials, it is possible to develop durable (long-lasting) jointless asphalt pavement structures even for bridge ends which move 20 mm during a summer/winter cycle. With respect to the design process itself, it has to be mentioned that with the current finite element software, it is time consuming to change the model (e.g. adding or removing asphalt concrete layers), because of new insights (e.g. ease of construction, costs, etc.).
It is strongly recommended to monitor the behaviour of the jointless pavement which has been developed; this during several (different) summer/winter cycles. The monitoring programme should include measurements of movements (strains) at a wide range of locations as well as of the force which is transferred at the connection between the bridge and the approach slab (via the steel bars).
ACKNOWLEDGEMENT
For their stimulating discussions during the whole project and their willingness to innovate, ir. W.A. de Bruijn and ing. F.A.M. van Gestel of the Engineering Office on Bridges and Tunnels of the Dutch Road Administration (Bouwdienst Rijkwaterstaat) are highly acknowledged.
REFERENCES
de Bondt, A.H. (1999). Anti-Reflective Cracking Design of (Reinforced) Asphaltic Overlays. Ph.D.-Thesis, Delft University of Technology. Maijenburg, A.T.G. (2000). Integral Bridges (in Dutch). Dutch Road Administration / Delft University of Technology. Scarpas, A. and Kasbergen, C. (1999). CAPA-3D User's Manual. de Bondt, A.H. (2000). Effect of Reinforcement Properties. 4 th RILEM Conference on Reflective Cracking, Ottawa. CUR (1998). FLOOR 1.0 - Concrete Floor Design Background Report (in Dutch), Gouda. de Bondt, A.H. and van Rooijen, R. (2002). Layer Interface Shear Testing, Internal Report. van Rooijen, R. and de Bondt, A.H. (2003). Development and Use of Functional Asphalt Tender Specifications, Paper prepared for RILEM - PTEBM, Zurich. de Bondt, A.H. and Schrader, J.G.F. (2001). Overview of Finite Element Analyses on Jointless Asphalt Alternatives for Bridge Son (in Dutch), Period July 1999 April 2001.
A Short Guide to the Types and Details of Constructing a Suspension Bridge - Including Various Arrangements of Suspension Spans, Methods of Vertical Stiffening and Wire Cables Versus Eyebar Chains