Sunteți pe pagina 1din 9
Journal of Adhesion & Adhesives 59 (2015) 53 – 61 Contents lists available at ScienceDirect International

Contents lists available at ScienceDirect

International Journal of Adhesion & Adhesives

journal homepage: www.elsevier.com/locate/ijadhadh Progressive damage modeling of adhesively bonded lap joints

Progressive damage modeling of adhesively bonded lap joints

Iordanis T. Masmanidis, Theodore P. Philippidis n

Department of Mechanical Engineering & Aeronautics, University of Patras, P.O. Box 1401, GR 26504 Panepistimioupolis Rion, Greece

P.O. Box 1401, GR 26504 Panepistimioupolis Rion, Greece article info Article history: Accepted 1 February 2015

article info

Article history:

Accepted 1 February 2015 Available online 8 February 2015

Keywords:

B.

Composites

D.

Cohesive zone model

E.

Joggle lap joint

Progressive damage

abstract

A continuum damage model for simulating damage propagation of bonded joints is presented, introducing a linear softening damage process for the adhesive agent. Material models simulating anisotropic non-linear elastic behavior and distributed damage accumulation were used for the composite adherends as well. The proposed modeling procedure was applied to a series of lap joints accounting for adhesion either by means of secondary bonding or co-bonding. Stress analysis was performed using plane strain elements of a commercial nite element code allowing implementation of user dened constitutive equations. Numerical results for the different overlap lengths under investigation were in good agreement with experimental data in terms of joint strength and overall structural behavior.

& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

The increasing size of structures built exclusively from light- weight composite materials, such as wind turbine rotor blades, has raised the need for more advanced design tools to optimize the joints between the various components. Moreover, the fact that maintenance and repair of such structures is becoming a major issue, due to the high replacement cost, increases the demand of more efcient joint design techniques. Numerous studies on the analysis of bonded joints with compo- site adherends, using the Finite Element method, have been pub- lished, see the recent reviews [1,2]. These studies can be categorized by their approach for predicting the strength of the adhesive joints. The continuum mechanics approach, that assumes perfect bonding between the adhesive and the adherends, suffers from the bi- material singularities inherent in a bonded joint and as a result maximum stress and strain for such a model will vary greatly with mesh renement. The fracture mechanics approach addresses the singularity issue but still there are limitations such as the difculty of the nite element modeling procedure to calculate the stress state at the crack tip and the need for measuring the fracture properties of the materials. Finally, there is the Cohesive Zone Modeling (CZM) approach which simulates the macroscopic damage along a pre- dened crack path by specication of a traction-separation response between initially coincident nodes on either side of the path. The great advantage of the cohesive models is their ability to simulate onset and non-self-similar growth of adhesive damage. However, except from the downside of predening the damage path in CZMs,

n Corresponding author. Tel.: þ 30 2610 969450, 997235; fax: þ 30 2610 969417. E-mail address: philippidis@mech.upatras.gr (T.P. Philippidis).

0143-7496/ & 2015 Elsevier Ltd. All rights reserved.

fracture characterization experiments must be performed to specify the cohesive law parameters. In the present work a modeling procedure for simulating adhe- sive joint behavior is presented. Degradation models, simulating damage propagation, without predening the failure path, are intro- duced for both composite adherends and polymeric adhesives. Material nonlinearity of the adherends is also implemented, while the presented softening procedure accounts for energy dissipation during debonding. A series of secondary bonded and co-bonded lap joints, of varying overlap length, were analyzed by means of the Finite Elements method to verify the predictive capabilities of the proposed model. The mesh renement issue is addressed by corre- lating element size with the softening law parameters so that the FE results are independent from the mesh density. Validation of the FE modeling procedure was performed by comparing predictions with experimental results and numerical predictions using the CZM approach.

2.

Material nonlinearity and progressive damage model

2.1.

Composite nonlinear behavior

Mechanical properties of both, the glass ber composite adher- ends and adhesive paste, used for coupon manufacturing, were determined experimentally in a comprehensive material character- ization campaign. The epoxy resin system used was HUNTSMAN Araldite s LY 3505 / Hardener Aradur s 3405 and laminate unidirec- tional reinforcement was AHLSTROM E-glass bers of an areal weight equal to 700 g/m 2 . Slight material non-linearity was found parallel to the bers, whereas a more pronounced one was mea- sured transverse to the bers, different in tension and compression.

54

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

As expected, a highly non-linear behavior under in-plane shear was observed as well. To account for material non-linearity, incremental stress strain relations were implemented, retaining the validity of the general- ized Hooke law for each individual interval as described in [3,4] :

E

1 t

ν 12 E 2 t

2 t

12

12

d

d

ε 2

ε 2

d σ 1 ¼

1 ð E 2 t = E 1 t Þν 2

E

d σ 2 ¼

1 ð E 2 t = E 1 t Þν 2

ð 1 Þ

The tangential elastic moduli in the principal coordinate system of the orthotropic material, E 1 t (parallel to the ber), E 2 t (transversely), G 12 t (in-plane shear) were derived as follows by adopting the nonlinear constitutive model introduced by Richard and Blacklock [5]:

d σ 6 ¼ G 12 t d ε 6

1 ð E 2 t = E 1 t Þν 2

12

ν 12 E 2 t

1 ð E 2 t = E 1 t Þν 2

12

d ε 1 þ

d ε 1 þ

σ i ¼

E o i ε i

h

1 þ E o i ε i = σ o i

n i

i ð 1 = n i Þ ;

i ¼ 1 ; 2 ; 6

ð 2 Þ

By differentiating Eq. (2) one has:

E it ¼ d σ i

d

ε i

¼

G 12 t ¼ d σ 6

d

ε 6

E o i 1

¼ G o 12

σ

i

σ

o i

n i

ðð 1 = n i Þþ 1 Þ

i ¼ 1 ; 2 ;

1

σ

6

σ

o 6

n 6

ðð 1 = n 6 Þþ 1 Þ

ð

3 Þ

A summary of the numerical values for all constants in elasticity expressions can be found in Table 1 ; they were derived through non-linear regression on the experimental data. Mean values for tensile and compressive strength properties in the ber direction (X T , X c ), transversely to the bers (Y T , Y c ) and in shear (S) for the composite tested are given in Table 2. The relatively low elastic properties of the adherend are due to the wet hand-layup manu- facturing technique, characteristic of the in-situ patching procedure of the industrial partner; typically this results in a ber weight fraction of ca. 51%. Mean values were deduced from 5 tests for each specimen type while engineering elastic constants were derived as suggested by relevant standards.

2.2. Polymer matrix and adhesive resin properties

The epoxy resin, used as adhesive for the secondary bonded specimens, is HUNTSMAN XD 4734 with XD 4741-S hardener cured at 80 1C for 1 h. The response of both the, previously described, polymer matrix of the adherends and adhesive resin was found to be slightly non-linear especially under shear stressing. In this work the epoxy resins are assumed to have linear behavior until failure since

Table 1 Elasticity constants for the non-linear model, Eq. (3) , of UD Glass/Epoxy composite.

 

ν 12 ¼ 0 : 26

E o i ½ MPa

σ o i ½ MPa

n i

E 1 t

 

26,870.00

9,016.00

1.00

E

ð T Þ

9,478.00

49.00

3.04

2

t

E

ð C Þ

10,473.00

178.00

2.54

2

t

G 12 t

 

2,760.00

44.00

1.87

Table 2 Failure stresses for the Composite material (in MPa).

 

X T

 

X C

Y T

Y C

S

558.60

411.12

40.00

128.14

38.42

the experimental stressstrain curve deviates from linearity close to coupon failure. Properties of the two resins are listed in Table 3.

2.3.

Progressive damage model

2.3.1.

Composite adherends

Besides non-linear mechanical response, progressive damage

mechanics were also implemented in the FE modeling procedure.

To account for the composite adherends progressive failure, the

Puck criterion [6] with the associated property degradation strategy is used. Details of the failure mode dependent stiffness degrada- tion were described in [3] and are summarized for completeness in Table 4 . According to Puck theory, there are 5 ply damage modes, two associated with either tensile or compressive ber failure (FF) and three describing matrix cracking or inter-ber failure (IFF); IFFA, -B, -C resulting mainly from a combination of transverse to the ber normal stress and in-plane shear. Index (k) in the above relations refers to an arbitrary load step after failure has been detected. The degradation factor, η r 1, multi- plying the engineering elastic constants to account for damage growth in the ply is given by [6]:

η ð k 1 Þ ¼

1 η r

1 þ c ð f

ð k 1 Þ

ð IFF Þ

E

1 Þ ξ þη r

ð 4 Þ

where f E(IFF) is the failure effort as calculated by Puck's matrix failure criterion while c ¼ 5, ξ¼ 3 and η r ¼ 1 10 6 are the values of the parameters of Eq. (4).

2.3.2. Polymer resin

To account for the adhesive paste progressive damage (micro- cracking), since a brittle isotropic adhesive material is assumed, the paraboloid failure surface criterion by Stassi D' Alia [7], adapted for generalized plane strain, is implemented:

σ x σ y

2 þ

σ y σ z 2 þ σ z σ x

ð

Þ 2 þ 6 τ xy þ 2 σ u ð R 1 Þ σ x þσ y þσ z 2 R σ u r 0

2

2

ð 5 Þ

where σ u represents the adhesive tensile strength and R , expressing the strength differential effect, is the ratio of compressive to tensile failure stress. Here R is calculated in terms of the measured values

Table 3 Engineering elastic constants and failure stresses for the polymer systems.

 

E [GPa]

G [GPa]

σ u [MPa]

τ u [MPa]

Araldite LY3505/Aradur 3405 XD 4734/XD 4741-S

3.98

1.48

56.94

51.64

4.01

1.39

35.29

39.94

Table 4 Progressive stiffness degradation model for the composite adherends.

Failure mode

FF(T) or FF(C)

E ð k Þ

1

¼ 10 10

E 1

 

E

k Þ

ð

2

¼ 10 10 E 2

 

G

ð k Þ

¼ 10 10 G 12

12

IFF(A)

E

k Þ

ð

2

¼

η ð k 1 Þ E 2

 

G

ð k Þ

¼ η ð k 1 Þ G 12

12

IFF(B)

G ð k Þ

12

¼ η ð k 1 Þ G 12

IFF(C)

E

k Þ

ð

2

¼ 10 10 E 2

G ð k Þ

12

¼ 10 10 G 12

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

55

of shear strength τ u and σ u of the adhesive. As it can be readily derived from Eq. (5):

R ¼ 3

τ

u

σ

u

2

ð

6 Þ

When failure is detected a softening process is imposed to the

adhesive to account for damage evolution. For the k th step, after failure has been detected, the degraded moduli are calculated by:

A continuum damage approach has been also introduced in [8] , where a stress/strain softening relationship is used as well. There the area under the stress/strain curve was related to the critical

energy release rate by introducing a characteristic length in order to transform displacements to strains.

3. Coupon description and FE implementation

E ð k Þ ¼

1 d ð k 1 Þ

E

The lap joint geometry presented here, see Fig. 2, is usually referred to as joggle lap joint (JLJ) and is often encountered in joints where a smooth nal surface is required, such as wind turbine rotor blade restoration procedures. The presence of the joggle increases the complexity of the problem due to the curvature; a study in which the JLJ con guration was also considered can be found in [9] . Total length and width of the coupons are equal to l ¼ 500 mm and

equals to a ¼ 15 mm. In both

ð secondary bonded and co-bonded joint types the overlap length (2c) varied from 50 to 200 mm in steps of 50mm, see Fig. 3. An alternative approach concerning coupon geometry could be

to keep constant the ratio of overlap to total specimen length;
ð

another, to keep constant the distance of the joint edges from the clamp area, both resulting in specimens of varying overall length. Nevertheless, a preliminary numerical investigation has revealed no signi cant differences in the ultimate load capacity and thus results concerning joint behavior from the comparison between the various overlap lengths presented in Section 4 are believed unaffected of said geometry variations. All JLJ coupon adherends consist of 2 layers of the Glass/Epoxy

UD previously mentioned, bers directed along the coupon long axis, with a nominal total thickness equal to t c ¼ 1.744 mm.

ð 7 Þ

G ð k Þ ¼ E ð k Þ = 2 ðÞ1 þν

The damage index, d , is calculated by considering the linear softening process shown in Fig. 1 for the equivalent stress and strain:

σ eq ¼

ε eq ¼

r

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1

2

σ x σ y

2

þ

σ y σ z

2

þ ð

σ z σ x

Þ

2

2

xy

þ 6 τ

1

p

ffiffiffi

2 ð 1 þνÞ

r

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ε x ε y

2 þ

ε y ε z 2 þ ε

ð

z ε

x

Þ 2 þ 3

2

γ

2

xy

8

9

Þ

Þ

As it can be readily proved from the above equations, σ eq ¼ E ε eq and thus the scalar parameter d of Eq. (7) expressing the damage accumulation for the current load step ( k ) is given by:

d ð k Þ ¼

ε

eq u

ε

ð

eq

k Þ

ε

eq

o

ε

k Þ

eq

ð

ε

eq

u

ε

eq

o

ð 10 Þ

is the

where ε

equivalent strain of the material at the ( n th) load step when the failure criterion was satis ed and the softening process began and

nally ε eq u is its maximum value.

Therefore, by updating the stiffness matrix, the stress for this

step is calculated by:

eq

is the equivalent strain at that load step, ε eq o ¼ ε

eq

ð

k Þ

ð

n Þ

ð

σ i

k Þ

¼ C

k Þ

ð

ij

ð

j

ε

k Þ

with

i ; j ¼ 1 ; ; 6

ð 11 Þ

The slope of line AB in Fig. 1 de nes the new stiffness of the degraded material for a load step during the softening process. The value σ eq o is of the equivalent stress when the failure criterion is satis ed ( n th load step) and therefore, it can have different values at the various elements depending on the current stress combina- tion when failure is detected. On the other hand, the value of the maximum equivalent strain,ε eq u , for which complete failure occurs is assumed dependent only on joint geometry and the adhesive material; it is derived by adapting numerical simulations on the observed experimental behavior of an arbitrary overlap length specimen from each joint type. See Section 3.3 for details.

w ¼ 25 mm while the curvature length

3.1. Secondary bonded coupons

In the coupons manufactured with secondary bonding the adhesive thickness was measured with the use of a digital caliper and by analyzing coupon free edge pictures with Mathworks Matlab Image Processing toolbox, resulting in an average thickness of t a ¼ 1.30 mm for all coupons. For each overlap length, the curved adherend plate was manufactured by the industrial partner using a mold and was bonded to the already cured at adherend plate using spacers to ensure even distribution of adhesive thickness.

spacers to ensure even distribution of adhesive thickness. Fig. 1. Softening stress-strain curve after failure onset
spacers to ensure even distribution of adhesive thickness. Fig. 1. Softening stress-strain curve after failure onset

Fig. 1. Softening stress-strain curve after failure onset of the adhesive.

Fig. 2. Geometry of the JLJ repair coupons.

the adhesive. Fig. 2. Geometry of the JLJ repair coupons. Fig. 3. Adhesively bonded JLJ coupons

Fig. 3. Adhesively bonded JLJ coupons of varying overlap length.

56

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

Specimens were then cut using water jet. The adhesive paste spew, present in all coupons, was included in the FE models, see Fig. 4 , since it was found to drastically in uence the predicted joint behavior.

3.2. Co-bonded coupon modeling

While the previously described progressive damage modeling can be implemented straightforwardly for secondary bonded joints, the same does not apply for the co-bonded ones since there is no distinct adhesive layer present. Nevertheless, since adhesion is mainly achieved by the interaction between the matrix resin located at the joined faces of the composite adherends, a model is introduced that replaces the ply of the composite adherend by a two-layer effective material, consisting of a modied composite and a distinct polymer layer, see Fig. 5. Motivation was provided by an early concept by Puppo and Evensen [10]. More speci cally, the total thickness, t c , of the composite adherend is assumed to be equal to the sum of t ec , thickness of an equivalent composite layer and t r / 2 , that of the polymer resin

layer and t r / 2 , that of the polymer resin Fig. 4. Close-up view

Fig. 4. Close-up view of the 50 mm coupon model showing mesh density.

view of the 50 mm coupon model showing mesh density. Fig. 5. Co-bonded JLJ coupon model.

Fig. 5. Co-bonded JLJ coupon model.

determined by:

t r ¼ t c

2

A f

ρ f

and

t ec ¼ t c t r

2

ð 12 Þ

where t r and t f are calculated as the thicknesses of the polymer and ber layer respectively of a single UD ply as shown in Fig. 6 . In the above relations A f is the areal mass of the UD glass fabric, equal to 0.7 kg/m 2 and ρ f the density of the E-glass bers, 2560 kg/m 3 . The ber elastic properties were back calculated using the ply and resin effective properties E o i of Tables 1 and 3 and the micromecha- nics equations (38a-d) of VDI 2014 Part 3 [11]. Then, by means of the same equations the elastic constants for the higher ber volume fraction equivalent composite layer, with t ec ¼ 1.445 mm, were derived and are shown in Table 5. The nonlinear trend, dened by the other parameters of Table 1, and failure stresses of the equivalent composite layer were assumed to be equal to the ones of the original composite. Finally, the resin layer of thickness t r /2 ¼ 0.299 mm has the properties of the polymer matrix in the UD composite ply, Araldite LY 3505, already described in Table 3.

3.3. FE mesh optimization

Non-linear material behavior and progressive damage models were implemented in ANSYS commercial FE code using the user programmable features of the PLANE182 element. In order to reduce computational effort, generalized plane strain analysis was chosen since relatively small strains are expected in the width direction of the model. Geometric nonlinearity was also included in the analysis by taking into account large strain effects. Traditional nite elements have dif culties in resolving the stress state at bi-material wedges due to the existence of singula- rities and their results vary with mesh re nement. The current work addresses this issue by correlating mesh density with the maximum equivalent strain, ε eq u of Eq. (10) . Square elements are used around the overlap region, as it can be seen in Fig. 4 , and by changing the mesh density the value of ε eq u , for which satisfactory agreement for joint strength was reached between numerical predictions and test results, was calculated. The study was performed with element size ranging from ¼ 0 : 1 mm to ¼ 0 : 4 mm in steps of 0.05 mm. The relationship between ε eq u and was derived by non-linear regression; for the epoxy resin of the co-bonded joints it was found that ε eq u ¼ 1 : 28 10 5 1 : 174 while for the adhesive paste of the secondary bonded joints ε eq u ¼ 4 : 65 10 5 1 : 031 , in [m]. Calculations were performed for the coupon geometry with the largest overlap, i.e. 200 mm, for each joint type and have proven to be valid for all overlap lengths resolving the mesh convergence issue. For the results presented later on this work an element size of ¼ 0 : 3 mm was used.

t r /2 t f t ec t c /2 t r
t r /2
t f
t ec
t c
/2
t r

Fig. 6. Composite adherend modeled as (a) two homogeneous UD plies (b) a set of resin and ber layers and (c) as an equivalent composite layer and a resin layer at the interface.

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

57

Table 5 Elasticity constants for the adherend equivalent composite.

E o 1 [MPa]

E

ð T Þ

[MPa]

E

ð C Þ

[MPa]

G o 12 [MPa]

ν 12

 

o

2

o

2

 

31,606.00

11,981.00

13,932.00

3,097.00

0.24

31,606.00 11,981.00 13,932.00 3,097.00 0.24 Fig. 7. Loading and boundary conditions. 3.4. Boundary

Fig. 7. Loading and boundary conditions.

3.4. Boundary conditions

The FE models boundary conditions were selected to properly simulate the testing procedure. For all coupon models stepwise loading is introduced by applying axial displacement increments on the nodes at the right end of the coupons. To avoid the dependence of load step convergence to mesh density, every time an element failure is detected the load is not increased for the following solution steps, until element failure stops propagating to adjacent elements and then the next increment is applied. After a thorough load step convergence study, including a range of displacement increments of 0.06 mm to 0.005 mm, it was found out that for a step equal to 0.01 mm, all models have fully converged. The above loading strategy resulted in the calculation of an average of more than 1000 steps until joint nal failure. All other nodal displacements were con- strained for a distance of d ¼ 50 mm at either end of the coupon model to account for the gripping of the test machine, Fig. 7.

3.5. Cohesive zone models

In order to further validate the modeling procedure introduced in the previous section of this manuscript, the co-bonded coupon behavior has been also simulated by modeling the progressive debonding using ANSYS contact elements and its integrated bilinear Cohesive Zone Material model [12] . Values for the critical strain energy release rate for the mixed mode CZM model are taken equal to G Ι c ¼ 1160 N = m and G ΙΙ c ¼ 2030 N = m [13,14], while the relative displacement for damage initiation is equal to δ ο ¼ 1 10 6 m . All FE runs using the CZM model have fully converged regarding mesh density and number of load increments.

4.

Results and discussion

4.1.

Test procedure

Seven samples from each joint con guration were tested on a 100 kN MAYES DH 100 S test rig equipped with a 407 MTS controller with its crosshead speed set to 5.8 mm/min. All coupons were inserted in the grips for 50 mm at each side leaving a gauge length of 400 mm. Load and displacement were measured by acquiring the test machine load cell and LVDT signals respectively. Strain monitoring on several key positions was also performed in a number of JLJ specimens in order to validate the predictions of the numerical models.

4.2. JLJ coupons joined by secondary bonding

The average failure loads and coefcient of variation (COV) from 7 coupons for each JLJ con guration are summarized in Table 6 along with the ultimate load from 7 continuous 2 layer UD coupons, of same geometry with the JLJ specimens, as a reference.

The progressive damage models described earlier were used to predict ultimate loads at failure for all the series of coupons. The term failure at the numerical procedure denotes the complete

separation of the adherends. Even though localized failure at areas with high stress concentration and along the bondlines was observed for the adherend elements, the stress developed until coupon failure was low for extensive composite damage. This agrees with the experimental results where the extent of damage at the composite increased with overlap length but only reached a failure mode of moderate ber tear out. More specically the failure mode for the 50 mm overlap is pure adhesive failure; for the 100 mm overlap, while mostly adhesive, some spots of cohesive failure are also observed. The dominant failure modes observed by further increasing overlap length to 150 and 200 mm, were light and moderate ber tear-out respectively, as it can be seen in Fig. 8 . Even though average stress at the adherends is much lower than the respective failure stress, localized damage in the composite plies is observed at the overlap regions due to high stress concentrations induced by the joint geometry and dissimilar material interfaces. While increasing overlap length, a plateau in joint strength was reached since the majority of the coupons with 200 mm overlap failed at almost the same loads with the 150 mm ones, see Table 6. Similar observations for single lap joints with a brittle adhesive were reported in [15]; the increase of strength restoration was limited to a certain overlap length, accompanied with more extensive adherend damage as well. A comparison between the numerically predicted load-dis- placement curves and the experimental data is presented in Fig. 9; the FE results compare well with the test data even though the strength of the coupons with smaller overlap is overestimated. The trend for the coupons to reach their maximum strength for an overlap length of 150 mm, observed in the tests, is fairly corrobo- rated by the FE results. This behavior is driven by the geometry and material non-linearity combined with stress concentration at the edges of the overlap during debonding progression, leading to the variation in failure mode as observed in Fig. 8. A systematic and

Table 6 JLJ coupons failure loads.

Overlap length

(mm)

Secondary bonding

Co-bonding

 

Reference

FEA

Test

COV

FEA

Test

COV

Test

COV

 

(kN)

(kN)

(%)

(kN)

(kN)

(%)

(kN)

(%)

24.88

3.10

50

10.27

6.64

22.10 15.60

11.61

9.70

100

13.62 10.51

8.80 16.36

16.31

9.23

150

14.10

14.83

4.67 16.74

16.82

7.63

200

14.11

14.65

4.20 16.60

16.33

3.28

50 mm overlap 100 mm 150 mm
50 mm overlap
100 mm
150 mm

Fig. 8. Typical fracture surfaces of the JLJ secondary bonded coupons.

58

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

Journal of Adhesion & Adhesives 59 (2015) 53 – 61 Fig. 9. Comparison of numerical predictions

Fig. 9. Comparison of numerical predictions and test results for the adhesively bonded JLJs.

predictions and test results for the adhesively bonded JLJs. Fig. 10. Location of strain gauges overlaid

Fig. 10. Location of strain gauges overlaid on the 2c ¼ 150 mm overlap FE model.

gauges overlaid on the 2c ¼ 150 mm overlap FE model. Fig. 11. Measured vs. FEA

Fig. 11. Measured vs. FEA calculated strains at the edge of lower adherend.

thorough study of this mechanism should be undertaken in the future. The sudden load drop observed in the experimental curves at a magnitude around 4 kN marks the onset of debonding. The observed

around 4 kN marks the onset of debonding. The observed F i g . 1 2

Fig. 12. Measured vs. FEM calculated strains at the middle of the overlap.

slope change in the FEA curve indicates that debonding onset is predicted at somewhat lower loads by the numerical model; to further verify and validate its predictions, strain measurements were performed in some specimens of 150 mm overlap.

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

59

Journal of Adhesion & Adhesives 59 (2015) 53 – 61 59 Fig. 13. Comparison of numerically

Fig. 13. Comparison of numerically predicted and experimental damage patterns. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

the reader is referred to the web version of this article.) Fig. 14. Progressive debonding of

Fig. 14. Progressive debonding of the JLJ coupon at 20, 50, 75, 95 and 100% of the ultimate load.

Strain gauges 5 mm long were placed on the upper adherend in the middle of the overlap length, i.e. c ¼ 75 mm, and on the lower adherend at a distance of s ¼ 5 mm from the edge as shown in Fig. 10. Their measurements were compared with the strains calculated from FEA at the corresponding nodes of the JLJ model. The strain gauge placed at the lower adherend edge measures zero strain when total debonding of the rst 10 mm of the joint has occurred. As it is seen in Fig. 11 , complete debonding of the overlap edge area was indeed predicted by the FE model. However the softening procedure, eventually leading to the complete debonding, begins at lower load levels than the experimentally observed abrupt failure of the area. The strain gauge placed on the upper adherend was strained up to 14 kN, see Fig. 12 , indicating that the middle of the overlap is close to the region to be debonded just before complete joint failure. This was veri ed by the FE model strain calculations and also by damage progression patterns, as those shown in Fig. 13 where the debonding pattern is very similar to the experimental observation. The red colored elements in the FE model correspond to failed adhesive material while other colors correspond to the various failure modes of the adherends

according to Puck criteria; FFT and FFC denote Tensile and Compressive Fiber Failure respectively while IFFA, IFFB and IFFC refer to the 3 matrix failure modes as described in Section 2.3.1 . It should be noted that even the trend for the adhesive spew to remain attached to the upper adherend after complete debonding, see Fig. 8 , was reproduced for all joints. A more detailed sequence of damage progression in the JLJ coupon of 150 mm overlap is also displayed in Fig. 14 showing consistency with the observed fracture surfaces of the repair coupons indicating light ber tear out, see Fig. 8 .

4.3. JLJ coupons joined by co-bonding

Comparison of experimental results in the form of load displacement curves from tests of JLJ coupons produced by co- bonding and numerical predictions are presented in Fig. 15 , where FEA curves compare well with experimental ones. The predictions of the Cohesive Zone Model (CZM), described in Section 3.5 , are also plotted along with the ones of the Continuum Damage Model (CDM), introduced in this work.

60

I.T. Masmanidis, T.P. Philippidis / International Journal of Adhesion & Adhesives 59 (2015) 53 61

Journal of Adhesion & Adhesives 59 (2015) 53 – 61 Fig. 15. Numerical against test results

Fig. 15. Numerical against test results for JLJs manufactured with co-bonding.

The trend to reach maximum strength restoration at 100 mm was closely predicted as well by the FE results of both the proposed model and the use of cohesive elements. It is of great signi cance to note that the time needed by the solver for the fully converged CZM analysis approach, in some cases was even 7 times greater than the one required for the proposed procedure.

5. Conclusions

A bilinear softening model combined with a failure criterion

suitable for brittle polymers has been introduced to account for damage progression and accumulation phenomena associated with

lap joints failure. Interaction of all stress components was taken into account using equivalent stressstrain response at the crack tip to drive the softening procedure.

In addition to the proposed model, implementation of a progres-

sive damage strategy for the composite adherends along with material and geometrical nonlinearity resulted in loaddisplacement curves that showed good agreement with the experimental response. Esti- mating for an arbitrary overlap length the only model parameter that is not related to the adhesive effective properties, resulted in satisfac- tory predictions for the other cases as well. Comparison of numerical predictions from cohesive zone models and the current approach was also favorable. Furthermore, the lack of need for a predened crack path and use of generalized plane strain with a relatively coarse mesh make the proposed FEA procedure adaptable for potentially modeling large scale repair patches of complex geometries. Numerical simulation provided satisfactory results for ultimate loads in most cases. Test results demonstrate the existence of a limit

at the strength recovery that can be achieved using JLJ joint cong- uration. In the current study experiments showed no increase in joint strength for overlap length greater than 150 mm or even 100 mm for the co-bonded coupons and this behavior was closely predicted by the FE analysis.

Acknowledgments

Financial support by Compblades Ltd. ( www.compblades.com ) through contract D357-2011 with the Research Committee of the University of Patras is gratefully acknowledged. The research work was also partially funded by EYDE-ETAK of the Greek Ministry of Development, in the frame of SYNERGASIA 2011 under contract Σ YN11_7_1000 (REWIND).

References