Mechanical analysis of a mixed mode debonding test for “composite” pavements

TL;DR: In this article, a four-point bending test was performed on bi-layer structures to investigate interlayer debonding near skrinkage cracks or joints of composite pavements, and the strain energy release rate was calculated.

About: This article is published in Construction and Building Materials.The article was published on 2013-03-01 and is currently open access. It has received 36 citations till now. The article focuses on the topics: Strain energy release rate & Shear stress.

Summary

1. Introduction

• To ensure a long lasting pavement, high bond strengths between pavement layers are obviously required.
• In order to investigate the properties of pavement interfaces, a number of experimental works carried out in the past mainly for reflective crack studies [8-13] may contribute to the knowledge.
• As regards the mechanical characteriza ion of interfaces between the cement concrete overlay and the bituminous material layer, the data given in [1] illustrate (see Table 1 below) the results derived from the literature on both field and laboratory measurements.
• For this last type of interface in particular however, the problem has not yet been fully investigated.
• The advantage of this test is to be able to investigate the mechanical properties of interfaces under mixed mode conditions without using a y supports or applied any loads directly on the bituminous material.

2. Simulations

• This section will discuss the elastic simulations conducted for multilayer structures made from homogeneous and isotopic materials.
• The first subsection will present the specific model used to derive the mechani al interface field at the edge location before its application to the four-point bending test.
• Next, for the materials tested herein, simulations will be run for the purpose of designing bi-material specimens.
• These simulations will also be useful for the developments in Section 3, as well as for the discussion provided on experimental results under static conditions for two different temperature values.

2.1 The multi-particle model of multilayer materials with 5 equilibrium equations per layer (M4-5n)

• An analysis of the delamination phenomenon on a multilayered system leads to studying a singular stres field at both the interface and edge location betwen two materials.
• This non-convergence is due to oscillatory singularities around the crack tip [36].
• Some solutions may be found in the literature [38-41]; however, the use of conventional crack criteria remains quite complex [42-43] and finite element simulations are usually "cumbersome" to implement and expensiv for common application by engineers.
• This modeling approach avoids singularities and reduces the real one-dimensional problem that accelerates the equation resolution process compared to other modeling approaches.
• These polynomial approximations offer the advantage of defining the out-of-interface plane normal ( )yxii ,1, +ν and shear stresses ( )yxii ,1, +ατ at interface i,i+1 (and similarly i1,i), between layers i and i+1 (and similarly between layers i-1 and i).

2.2 M4-5n equations applied to the 4PB test

• In order to simplify the analysis, the 4PB test on a bi-layer specimen, as illustrated in Figure 1, has been simulated under the assumption of plane strain conditi s. Subsequently, the mechanical fields of the M4-5n depend solely on variable x.
• The materials are considered to be elastic, homogen us and isotropic.
• The problem can be divided into three zones (Fig. 1a).
• For specific tension problems occurring with the composite materials, each M4 strain energy release rate relative to each fracture mode (GI , GII , GIII ) is expressed as a quadratic function of interfacial stresses at the crack tip [46, 49-50].

2.3 Validation of the numerical M4-5n results

• Boundary conditions of the central bi-layer zone where system (1) is solved are listed in Table 2.
• The equations are numerically solved by applying a non-dimensional method along with the finite difference method, according to the Newmark scheme used in [32, 54] and implemented in the French open source software for numerical computations known as Scilab.
• Along the x-axis of the bi-layer zone II, for two different temperature conditions applied in the following tests, their intensities are well within the range of values given previously in the literature (see Table 1).
• Between the two external force points (i.e. B and C in Fig. 1), M4-5n interface stress intensities (Fig. 2), especially for normal stresses (Fig. 2b), are quite small and remain constant in comparison with the values from zones I and II near the edges.
• Each simulation consumes no more than 2 seconds (in CPU time).

2.4 Influence of specimen geometry and material char cteristics on the stress field

• M4-5n simulations will be run on the global specimen geometry, which takes into account both the space constraints of the test and the heterogeneities of the material.
• For the displacement-controlled condition at a 0.7-mm/min loading rate, under the assumption described in Section 3.2 below, the equivalent modulus value of the bituminous material equals approximately 2,000 MPa at 20°C and 11,000 MPa at 5°C.
• This M4-5n parametric analysis reveals that the tensil stress at the bottom of concrete layer 2 is in competition with interface stresses; this stress depends on the equivalent elastic value of the bituminous material modulus and hence on the test temperature.
• The parametric analysis confirms that t e intensity of interface stresses at the edge (x = L-a2) increases from 20% to 60%, in comparison with stresses on the other side when only the length a2 is increasing.

3. Experimental results on bi-material beam for composite pavement

• In [35,51], the modeling was first validated for tests on bi-layer specimen composed of aluminum as the first layer and PVC as the second.
• Debonding between these wo materials was observed according to expectations.
• The elastic M4-5n simulation fits the elasticpart of experimental load-displacement curves.
• This section will present the initial experimental results obtained on a bi-material specimen for the composite pavement made with cement concrete and bituminous materials, as described above.
• The second subsection w ll examine the load displacement curves, as well as the load vs. strain measured at the bottom of the bituminous layer, focusing solely on delaminated specimens.

3.1 Description of test specimens and initial observations

• For the bituminous material and cement concrete matrials described above in Tables 2 and 3, two types of interfaces have been tested.
• For type I specimen, the cement concrete layer was cast directly onto the prefabricated bituminous slab.
• Both types I and II specimens were delaminated according to this proposed test protocol (see Table 5, Figs.
• The crack started in the middle of the specimen and at the bottom of the cement concrete layer.
• Furthermore, some observations have displayed the presence of a non-negligible effect from the bituminous layer self-weight on type II interfaces (i.e. with use of a tack coat).

3.2 Experimental curve data

• Only the curves associated with tests that ultimately produced delamination failure under controlled displacement conditions are reported for the two width values of bi-layers (i.e. b = 100 and 120 mm).
• To properly test this type of interface, which is not exactly the ultimate objective of this work, the present test would need to be improved.
• Compared to results obtained at ambient temperature (see Fig. 10), the authors obs rve on Figure 12, that small visco-elactic effects xists in Type I specimen in regards to Type II ones.
• As a general indication for the type I specimen, the maxi um force value of this test carried out at 5°C was of the same intensity as the value obtained for tests at 20°C.
• The M4-5n curve perfectly fits the experimental displacement and strain result before failure occurs in the cement concrete layer.

3.3 Discussion

• M4-5n simulations (Fig. 4), in correlation with cemnt concrete material characteristics (Table 3) and initial experimental observations (Table 5, Fig. 8a), indicate that a competition exists between tensile stress at the bottom of the cement concrete layer and interface normal and shear stresses.
• The curves at ambient temperature match the maximum load value (Fig. 13).
• To complete this first effort and provide some indicators, if no cracks occur at the bottom of the cement concrete lay r before delamination (i.e. case 2), then additional values of the M4-5n interface stresses at point D (or A) in Figure 1 and the strain energy release rat h ve been listed in Table 6.
• First, one shows that the elastic model used is really effici nt to design the geometry of the specimen.
• These preliminary experiments have been done to check the capability of the four-point bending test to distinguish between the two types of interface.

Figures

Figure 13. Different failure scenarios of the bi-layered 4 pt bending test for pavement material

Table 5: Dimension of bi-layer specimens, test conditions and visual observations

Table 4: Cement material characteristics (IFSTTAR -476)

Table 2: Numerical boundary conditions of the bilayer modelled (zone II of Figure 1)

Table 3: Bituminous material characteristics with high level of void content (IFSTTAR A476)

Figure 4. Effect of Young modulus ratio between layers (F= 5kN, a1=a2=70mm, b=120mm): (a) on the tensile stress at the bottom of layer 2, (b) on the interface normal stress ( )22,1 aL −ν and the shear stress ( )22,1 aL −τ

Figure 5. Variation effect of length a2 (E 2/E1 = 17.4, F= 5kN, a1=40mm, b=120mm): (a) on the tensile stress at the bottom of the cement concrete layer, (b) on the normal stress ( )22,1 aL −ν and the shear stress ( )22,1 aL −τ

Figure 1. (a) Four-point bending test on bilayer materi ls (b) anti-symmetrical debonding case

Figure 3. Evolution of strain energy release rate in function of the normalized crack length af/LF (for FType I = 12kN, for FType II = 5kN, E 1/E2= 17.4, b= 120mm, a1=a2=70mm): (a) Comparison of the two methods: energy and VVCT; (b) GI and GII strain energy release rate compared to the total G.

Figure 2. Interface M4-5n stresses and numerical convergence of calculations at point 2aLx −= (F= 5kN, b= 120mm, a1=a2=70mm): (a) Interface shear stress ( )x2,1τ , (b) Interface normal stress ( )x2,1ν

Figure 8. Bending parasite failures at the: a) Middle of the beam (I-PT-3-2) b) Edge of layer 2 due to non regular thickness of layers (Type I-PT-2.1 example)

Figure 6. (a) Four-point bending test arrangement (b) First debonding observations of Type I specimen (IPT-2-3 under load controlled condition 100N/s)

Figure 7. Visual debonding fracture aspect of Type I specimen: a) I-PT-2-3 b) I-PT-1-3

Table 2: Numerical boundary conditions of the bilayer modelled (zone II of Figure 1)

Figure 10. Load versus midspan deflection curves of bi-layer specimen tested under controlled displacement rate (0.7mm/min)

Figure 9. Typical debonding fracture aspect of Type II specimen (II-PT-2-2)

Figure 11. Load strain results for bilayer specimens tested under controlled displacement conditions (0.7mm/min)

Figure 12. Load displacement results for bi-layer specimen tested under a lower temperature fixed at 5°C and a controlled displacement rate (0.7mm/min)

Table 6: Interface stress intensity and strain energy release rate results for ambient temperature (around 20°C)

Frequently asked questions

Q1. What are the contributions in "Mechanical analysis of a mixed mode debonding test for composite pavements" ?

Results are discussed relative to both data provided in the literature and testing campaigns. 

Q2. What future works have the authors mentioned in the paper "Mechanical analysis of a mixed mode debonding test for composite pavements" ?

In looking to future studies, additional simulations for the purpose of introducing, under loading points of the specimen, an initial crack at the bottom of the concrete layer into the model should help determine the exact cracking and debonding mechanisms of these specimens. 

Q3. What is the role of modulus between layers in the performance of the bond?

The modulus ratio between layers, due to the variation in temperatureeffect and the loading rate on mechanical properties of bituminous material as well as to the position of the load close to the vertical crack, plays an important role in the long-term performance of the bond [3-4]. 

Q4. What is the main reason for the debonding between pavement layers?

During the pavement service life however, due to the structural heterogeneity of multilayer systems, debonding between pavement layers can occur, especially near the edges or vertical cracks through one layer. 

Q5. What type of test can simulate the interface stress in cracked pavements?

A specific elastic Multiparticular Model which can simulate the 3D interface stress in crackedpavements [33] may be used in this investigation. 

Q6. How many emulsions were used to cover the concrete layer?

The emulsion, i.e. 0.4 kg/m² of residual binder, was applied to the concrete layer and left for 24 hours before being covered by the bituminous material layer. 

Q7. How many tests have been performed on each interface?

Under the same controlled displacement rate (0.7 mm/min) as for previous results, only one test on each interface has been performed at a lower temperature, i.e. set around 5°C (Fig. 12). 

Q8. What type of interface is used for a bituminous material?

The type II interface corresponds to a bituminous material with the cement concrete material being bonded by means of a tack coat layer. 

Q9. How long is the test duration determined by the experimental results?

From these loads, the test duration is determined by reviewing the experimental time-load result; it amounts to roughly 5.3 s for the two specimens tested at high and low temperature. 

Q10. What is the M4 model for the pavement bending problem?

The M4 selected herein for the pavement bending problem contains five kinematic fields per layer i ( { }ni ,...,1∈ , where n denotes the total number of layers): average plane displacements ( )yxU i ,α ; the average out-of-plane displacement ( )yxU i ,3 ; and average rotations ( )yxi ,αΦ , where ( )yx, represent the layer's plane coordinates and the α-plane directions { }( )2,1∈α . 

Q11. What is the tensile stress intensity at the bottom of the concrete layer?

In Figure 5a, as length a2 increases, the tensile stress intensity at the bottom of concrete layer 2 increases under loading point C and the interface normal and shear stresses increase at the edge (x = L-a2) (Fig. 5b). 

Q12. What was the first validation for the modeling?

In [35,51], the modeling was first validated for tests on bi-layer specimen composed of aluminum as the first layer and PVC as the second.