Practical Applications of Improvements in FML Crack Bridging Theory

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

25th ICAF Symposium Rotterdam, 2729 May 2009

PRACTICAL APPLICATIONS OF IMPROVEMENTS IN FML CRACK BRIDGING THEORY


Greg Wilson,1,2 Ren Alderliesten,2 Riccardo Rodi,2 and H.J.K. Lemmen2 Alcoa Technical Center, Alcoa Center, PA, USA. Corresponding author: gregory.wilson@alcoa.com 2 Delft University of Technology, Faculty of Aerospace Engineering, Chair of Aerospace Materials, Delft, The Netherlands
1

Abstract: Fiber metal laminates (FMLs) have excellent crack growth performance compared to monolithic metals thanks to crack bridging by intact fibers in the wake of a fatigue crack. Calculating the distribution of bridging loads in the fibers is key to analyzing and predicting the crack growth of FMLs. Most analytical approaches to modelling this phenomenon do so by imposing compatibility between the deformation of cracked metal layers and the elongation of the bridging fibers. In doing so, they assume that the crack opening displacement is equal to the displacement of the metal sheets at the boundary of the delamination between the cracked and the bridging layers. This paper derives a solution to the crack bridging problem that accounts for the deformation of the metal between the crack flanks and the delamination boundary. The results of doing so show that neglecting that deformation is acceptable for FML crack growth prediction, but the solution incorporating the exact displacement of the metal layers enables the application of the crack bridging method to a variety of additional situations as well as the extension of its applicability to more complex FMLs. This paper surveys a number of such applications, including examples of how to apply crack bridging theory to these problems.

M. Bos (ed.), ICAF 2009, Bridging the Gap between Theory and Operational Practice, 539558. Springer Science+Business Media B.V. 2009

540

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

NOMENCLATURE
a as b E K Ktip K Kbr R v vbr f pp ,al max Crack length Saw-cut or notch length Delamination height Youngs modulus Stress intensity factor K of the cracked metal layers in an FML Portion of Ktip due to farfield metal layer stress Contribution of crack bridging loads to the reduction of Ktip Stress ratio of fatigue loading Vertical displacement due to farfield stress Vertical displacement due to crack bridging loads Elongation of reinforcing fibers Shear deformation of adhesive at delamination boundary Poissons ratio Farfield stress in metal layers Maximum applied cyclic stress in fatigue loading

INTRODUCTION
Fiber metal laminates (FMLs) such as Glare [1], ARALL [2], and CentrAl [3], consist of alternating layers of metals and fiber-reinforced polymer composites. FMLs are known for exceptionally slow fatigue crack growth compared to monolithic metallic materials, while still exhibiting many desired properties of metals, such as formability, impact resistance, machinability, and reparability. The slow crack growth of FMLs is a result of crack bridging by the fibers. While the metal layers may develop fatigue cracks over the life of a structure, the fibers in the composite layers remain intact. Load that would otherwise contribute to the stress intensity of the cracks in the metal is borne by the fibers, reducing the stress intensity and therefore the crack growth rate. Another way of thinking about crack bridging is to consider the fibers to be resisting the opening of the cracks. Delamination between the fiber and metal layers develops around cracks due to the high degree of load transfer over this interface. As cracks grow across the metal layers, perpendicular to the main loading direction, the delamination grows in the loading direction, away from the cracks. This delamination behaviour is critical in understanding crack bridging, since large delamination size and shape reduces the effectiveness of fiber bridging. Small delaminations bridge better than large ones. Several models [4, 5, 6] have been developed to characterize the bridging effect of the fibers in FMLs. These models all employ a compatibility constraint in their

Practical applications of improvements in FML crack bridging theory

541

bridging solutions, noting that since the cracked metal layers and the intact fiber layers are attached at the delamination boundary, the deformation of the metal layers at the delamination boundaries must equal the elongation of the fibers in the wake of the crack. It is possible to calculate what the bridging loads must be in order to satisfy this condition. Once the bridging loads are known, the influence of the bridging loads on the crack tip stress intensity factor of the metal layers and on the energy release rate at the delamination boundary can be determined. These values then give a prediction for the crack and delamination growth behavior of the laminates. One feature common to these models is that they avoid an exact calculation of the displacement of the cracked metal layers at the delamination boundary. The models of Alderliesten [4] and Wu and Guo [5] use the crack opening in the compatibility solution, assuming that the additional deformation of the metal between the crack flanks and the delamination boundary is negligible. Marissen [6] solves compatibility with two bounding scenarios, one in which the crack opening is used, with the stiffness of the fibers contributing to this displacement, even in the delaminated region, and one in which the displacement at the edge of a delamination-shaped hole in the laminate is calculated. It is indeed possible to calculate the exact displacement of the metal layers along the delamination boundary. The first section of this paper will briefly explain how, then show that the decision to approximate this displacement by using the crack flank opening displacement had little effect on the outcome of the calculation. While it is therefore not necessary to go to the extra effort to include the exact calculation in these models, the exact displacement calculation adds additional versatility to these tools. The rest of this paper focuses on several examples of new practical uses for a crack bridging model using the exact displacement solution.

EXACT DISPLACEMENT SOLUTION


Brief review of the Alderliesten method In order to appropriately apply fracture mechanics-based crack growth models to FMLs, the stress intensity factor of the crack tip in the metal layers, Ktip, must be determined. By superposition, the stress in the cracked metal layers can be separated into components due to remote loading and internal residual stress in the laminate, and the stress reduction caused by the bridging action of the fibers. The former is treated entirely as a remote load, and the latter is analyzed as a line load acting along the boundary of the delamination between the metal and fiber layers. Similarly, the stress intensity factor is split by superposition, giving

K tip = K K br ,

(1)

542

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

where K represents the stress intensity factor due to the farfield stress applied as a remote load, and Kbr is the stress intensity factor due to the bridging load. In order to determine Kbr, the distribution of bridging load along the delamination boundary must be known. Compatibility is employed as a means of calculating the bridging load. Since the metal and fiber layers are connected at the delamination boundary, the displacement there must be the same in each layer. The left side of Eqn. 2 gives the crack opening of the cracked metal layer, which Alderliesten assumes is nearly equivalent to the displacement at the delamination boundary. Here, superposition is employed to separate that due to the farfield stress, v, and that due to bridging, vbr. The right side of Eqn. 2 gives the displacement of the reinforcing fiber layers at the delamination boundary. The term, f, represents that displacement due to the elongation of the fibers, and pp accounts for shear deformation in the prepreg layer at the delamination boundary. All of the terms in Eqn. 2 are functions of x since the bridging stress, crack opening, and fiber elongation may vary along the boundary, and compatibility must hold at all x.

v ( x) vbr ( x ) = f ( x) + pp ( x )

(2)

The vbr(x) and f(x) terms are both functions of the bridging load, and Equation (2) can therefore be solved for the bridging load distribution. In turn, Ktip can be determined from these results, and the growth of a crack in an FML can be predicted. For the crack opening due to the farfield stress in the metal layers Alderliesten uses the following equation, from [7], in which ,al represents the farfield stress in the metal layers and Eal is the metal stiffness:

v ( x) = 2

,al
Eal

a2 x2

(3)

The crack opening due to the bridging load is expressed as an integral of the crack opening due to point loads along the delamination boundary:

vbr ( x) = v( x, xP ) dxP
s

(4)

In Eqn. 4, s is the width of the saw cut, where no fiber bridging force is present, and v(x,xP) is the crack opening displacement at horizontal location x due to a point load applied at location xp. Alderliesten approximates v(x,xP) as

Practical applications of improvements in FML crack bridging theory

543
2 a 2 xP 2

v( x, xP ) =

4 P ( xP ) tanh 1 E

(1 + ) b a2 x2 + 22 2 2 2 a x +b xP x 2 + b 2
1 2

a x +b
2 2

(5)

for x<xP, and


v( x, xP ) = 4 P ( xP ) tanh 1 E

(1 + ) b a2 x2 + 22 2 2 2 2 a xP + b x xP + b 2
1 2

a2 x2 2 a 2 xP + b 2

(6)

for x>xP. In Eqns. 5 and 6, P represents the point bridging load, normalized by the thickness of the metal sheet. This approximation combines the solutions for crack opening displacement due to point loads above and below the crack plane on the centerline and point loads on the crack flanks to the left and right of the centerline from [7] in such a way that Eqns. 5 and 6 correspond exactly to those handbook solutions in the limits of x0 and b0, respectively. The stress intensity factor due to the bridging loads is similarly approximated by combining the solutions for the stress intensity factor due to point loads above the crack flanks along the centerline and due to point loads along the flanks, symmetric about the centerline.

K I ( xP ) =

2P

1 b2 1 + (1 + ) 2 2 2 2 2 2 a xP + b a xP + b 2 a

(7)

Because the integral in Eqn. 4 includes P, an unknown function of xP, Eqn. 2 cannot be solved explicitly for the bridging stress distribution. Instead, it is solved numerically by treating the reinforcing material as a series of bar elements with width, wi, height, bi, and location at the center, xi, as shown in Figure 1.

544

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

Figure 1 Bar element scheme for solving the compatibility equation for the bridging stress Under this scheme, the crack opening due to bridging at a given bar element, i, is given by:

vbr ( x) = v( x, xi )wi
i

(8)

Eqn. 2 can now be written as a series of linear equations one equation for each bar element, in which its crack opening is a linear function of the bridging load at every bar element, as well as the farfield stresses and solved with linear algebra. Derivation of the exact displacement method It is possible to derive exact solutions for the displacement along the delamination boundary due to point loads along the boundary and for the stress intensity factor due to the same point loads, using the Westergaard stress functions given on page 5.7 of Refs. [7] and [8] 1 . These functions relate to the loading conditions depicted in Figure 2.

It should be noted that Ref. [8] includes extra terms in the Westergaard functions that affect the displacement, but not the stress intensity factor. Referring only to the Westergaard functions given in [7] will give erroneous results.

Practical applications of improvements in FML crack bridging theory

545

Figure 2 Symmetric point load scenario from [8] The Westergaard stress functions describing this scenario are
2 2 a 2 z02 P 1 a z0 + q ( z , z0 , z0 ) Z I = 1 y0 2 y0 ( z 2 z0 ) ( z 2 z02 ) 1 a 2 (z)

(9)

and
ZI = P 1 1 y0 tan y0 z2 a2
2 a 2 z0

+ tan 1

z z tan 1 i tan 1 i z0 z0 a 2 z02 (10) z 2 a2

with z = x + iy , z0 = x0 + iy0 , and z0 = x0 iy0 , where (x,y), and (x0,y0) are the coordinates of the stress function and of the point of load application, respectively; with

1 (1 + ) plane stress ; = 2 1 1 2 ( 1 ) plane strain

(11)

546

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

and with

z 1 z q ( z , z0 , z0 ) = 2 0 2 2 0 2 i z z0 z z 0

y0 y0 + . = 2 2 2 2 ( z x0 ) + y0 ( z + x0 ) + y0

(12)

The Westergaard functions of Eqns. 9 and 10 can be used to solve for the vertical displacement of the metal sheets using Eqn. 13 for plain stress.

vbr =

1 1 y + Im Z I Re Z I 2G G

(13)

Similarly, Westergaard functions can be used to find the exact displacement at the delamination boundary due to the farfield stress in the metal layers, using those for a remote biaxial stress:

ZI =

1 ( a ) z
2

(14)

ZI = z2 a2

(15)

The displacement due to the remote load can be calculated using a formula similar to Eqn. 13, but corrected for the fact that the loading of interest is uniaxial, rather than biaxial [9]:

v =

3 1 +1 Im Z y Re Z y 2G 2 2 2

(16)

where = ( 3 ) / (1 + ) for plane stress. Substituting the exact expressions for displacement at the delamination boundary, from Eqns. 13 and 16, into the compatibility equation, Eqn. 2, in place of Eqns. 8 and 3, respectively, allows the bridging stress to be solved with the exact displacement method. Comparison of results Figures 3, 4 and 5 show the results of an example calculation using this method. The results of a calculation using the original crack flank opening method of Alderliesten are compared to those of a calculation in which the exact solutions for vbr, Kbr, and v at the delamination boundary are used.

Practical applications of improvements in FML crack bridging theory

547

The example calculation is for a laminate of Glare 3-4/3-0.4, meaning the reinforcing fiber layers are composed of one 0 and one 90 S2-glass prepreg ply each, there are four metal layers and three fiber layers, and each metal layer is 0.4 mm thick. The initial notch size, as, is 1.5 mm and the starting crack size for the calculation is 2 mm, with an initially elliptical delamination. The prediction was run with an applied loading of max = 80 MPa and R = 0.05. The bridging stress distributions of the initial scenario at the maximum applied load calculated with each method are shown in Figure 3. Here, the bridging stress is defined as the total bridging load carried in a given bar element divided by the total thickness of fiber layers. The bridging stress calculated with the exact displacement method is significantly higher than that calculated with the original crack flank opening method of Alderliesten along the entire delamination boundary, with the discrepancy highest at the crack tip.

Figure 3 Fiber bridging stress calculated for initial parabolic delamination shape in crack growth prediction of Glare 3-4/3-0.4 at an applied stress of 80 MPa The prediction was carried out, allowing the crack to grow from 2 mm to 14 mm, using both the original crack flank opening method and the new exact displacement

548

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

method proposed here. Figure 4 shows the bridging stresses calculated with each method for the maximum applied load and the final delamination shape from the respective predictions. Again, the bridging stress calculated by the exact displacement method is consistently higher, but the differences between the two solutions are smaller. The final predicted delamination shapes are also compared in Figure 4. Since the exact displacement method calculates a higher bridging stress along the delamination boundary throughout the prediction, the delamination grows to a larger size.

Figure 4 Fiber bridging stress calculated for the respective final predicted delamination shape of each method, after crack growth prediction of Glare 3-4/30.4 at an applied maximum stress of 80 MPa and a stress ratio of R = 0.05

Practical applications of improvements in FML crack bridging theory

549

Figure 5 Predicted crack growth of Glare 3-4/3-0.4, max = 100 MPa, R = 0.05 The crack growth predictions obtained using the two methods are shown in Figure 5, compared with test data of Glare in the same configuration and loading. The exact displacement prediction is slightly more conservative than the prediction using the original method. The latter part of the test data has been transposed to the right, aligning it with the two predicted curves, to illustrate that at both the beginning and second half of the test both methods predict the behaviour of the crack accurately. The only discrepancy comes in the middle of the testing. Figure 6 compares the predicted crack growth rates using both method with test data for the above scenario, Glare 3-4/3-0.4 with a maximum applied stress of 80 MPa and R = 0.05, as well as for Glare 3-6/5-0.4 with a maximum applied stress of 100 MPa and R = 0.05. Both methods result in predictions that match the experimental crack growth rates once the cracks have grown to around 10 mm. Because of the small difference between the predictions in these regions compared to the scatter in the data, it is not possible to conclude that either method is significantly better than the other from these comparisons.

550

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

Figure 6 Comparison of crack growth predictions against experimental results

In the end, the crack growth prediction obtained using the exact displacement formulas is similar to that using the original assumptions of Alderliesten, and slightly closer to the experimental results in the two scenarios examined here. The added complexity of including the exact displacement solution may not be worth enduring for the purposes of FML crack growth prediction alone. As the following sections demonstrate, however, there is a great deal more that can be done by employing the exact displacement solutions in ways beyond the crack growth of Glare.

ARBITRARY LAMINATE
A fiber metal laminate is made up of a number of layers, and describing it mathematically by assuming all the metal layers have the same displacement, and all the fiber layers have the same elongation in the wake of the crack, limits the applicability of the model to situations in where this is the case. Of course, there are many situations in which it may be beneficial to consider each lamina separately. Bending of the laminate, a corner flaw in one sheet at a hole, a laminate with layers of different thickness or composition, and many more scenarios are not adequately described using Eqn 2. A more complex approach is required. Cracked metal layer between two bridging layers Consider the scenario depicted in Figure 7, in which a single cracked metal sheet is reinforced with one fiber layer on each side, but the delaminations at each interface are not identical. In this case, the assumption that the crack opening displacement can be used in place of the actual displacement at the delamination boundaries yields a poor solution.

Practical applications of improvements in FML crack bridging theory

551

Figure 7

In this scenario there are two unknown functions, the distribution of bridging load along delamination 1 that the first reinforcing layer carries and the distribution of bridging load along delamination 2, carried by the other reinforcing fiber layer. Compatibility can be imposed along each delamination boundary, since the metal sheet and the respective fiber layers must be in contact there. Therefore, it is reasonable to write Eqn. 2 twice, once for each delamination boundary, giving two equations which can be solved for the two unknowns. However, applying Eqn. 2 using only the crack opening for v and vbr for compatibility means that the left side of both compatibility equations must be equal. This gives the result that the displacement of the bridging layers at both delamination boundaries, plus their respective adhesive shear deformations, must be equal. Two, more independent, equations can be written and solved if the compatibility equation (Eqn. 2) is written once for the displacement of the metal layer and the first bridging layer along the boundary of the first delamination and once for the displacement of the metal layer and the second bridging layer along its delamination. Eqn. 17 demonstrates the form this solution takes. The top equation enforces compatibility along delamination 1, and the bottom equation does so along delamination 2. Notably, since the bridging stresses along both delamination boundaries affect the displacement of the single metal sheet, the displacement due to the bridging stress transferred at delamination 2 must be included in the calculation of the deformation of the metal sheet along delamination 1, and vice versa.

552

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

v ( x) b ( x ) vbr ( x, br ,1 ( x) )
1

b1 ( x )

vbr ( x, br ,2 ( x) ) vbr ( x, br ,2 ( x) )

b1 ( x )

= f 1 ( x, br ,1 ( x) )

b1 ( x )

+ pp ( x)

b1 ( x )

v ( x) b ( x ) vbr ( x, br ,1 ( x) )
2

b2 ( x )

b2 ( x )

= f 2 ( x, br ,2 ( x) )

b2 ( x )

+ pp ( x)

b2 ( x )

(17) By discretizing the distribution of bridging stress with the same scheme shown in Figure 1, Eqn. 17 can be rearranged to solve for the bridging stress:
v* ( x , x , b ) br 1 1 1,1 b1 ( x ) v*br ( xn , x1 , b1,1 ) b1 ( x ) v* x , x , b br ( 1 1 1,1 ) b2 ( x ) v*br ( xn , x1 , b1,1 ) b ( x ) 2

v*br ( x1 , xn , b1,n ) v*br ( xn , xn , b1,n ) v*br ( x1 , xn , b1,n )


v*br ( xn , xn , b1,n )

b1 ( x )

v*br ( x1 , x1 , b2,1 ) v*br ( xn , x1 , b2,1 )


v*br ( x1 , x1 , b2,1 )

b1 ( x )

b1 ( x )

b1 ( x )

b2 ( x )

b2 ( x )

b2 ( x )

v*br ( xn , x1 , b2,1 )

b2 ( x )

v*br ( xn , xn , b2,n ) b1 ( x ) v*br ( x1 , xn , b2,n ) b2 ( x ) v*br ( xn , xn , b2,n ) b2 ( x )


b1 ( x )

v*br ( x1 , xn , b2,n )

* f 1 ( x1 ) v ( x1 ) f 1, ( x1 ) b1 ( x ) b1 ( x ) b1 ( x ) P ( x1 ) 1 * f 1 ( xn ) b ( x ) P ( x ) v ( xn ) b1 ( x ) f 1, ( xn ) b ( x ) 1 n 1 1 + I = * f 2 ( x1 ) b2 ( x ) P2 ( x1 ) v ( x1 ) b2 ( x ) f 2, ( x1 ) b2 ( x ) P2 ( xn ) * f 2 ( xn ) b ( x ) v ( xn ) b ( x ) f 2, ( xn ) b2 ( x ) 2 2

(18) In Eqn. 18 above, the large matrix can be divided into four quadrants. The upperleft quadrant represents the component of the vertical displacement of the metal sheet along delamination 1 due to the bridging load of delamination 1. The upperright quadrant represents the component of the vertical displacement of the metal sheet, measured along the boundary of delamination 1, due to the bridging stress of delamination 2. Likewise, the lower-left and -right quadrants give the displacement of the sheet along the boundary of delamination 2 due to the bridging loads transferred across delamination 1 and delamination 2, respectively. These components are defined as

v*br ( x n , x m , bk ,m ) b ( x ) =
l

wk v br ( x n , x m , bk ,m ) b ( x )
l

t metal Pk ( x m )

(19)

where vbr is that from Eqn. 13, xn and bl(x) refer to the horizontal and vertical coordinates where the displacement is measured (in bar element n and at the height

Practical applications of improvements in FML crack bridging theory

553

of delamination l), xm and bk,m are the coordinates where load Pk is applied (at bar element m and at the height of delamination k). The wk term is the width of bar element k, and tmetal is the thickness of the central metal sheet. Also in Eqn. 18, the * fk ( xn ) b ( x ) terms added to the diagonal of the matrix (by multiplication with the
k

identity matrix) represent the component of the elongation of fiber layer k at bar element n due to the bridging force, normalized by the bridging force. On the right of Eqn. 18 are the metal layer deformation due to the farfield stress, from Eqn. 16, and the component of fiber elongation due to the farfield stress in the fibers layers. Eqn. 18 can be solved for the unknown Pk(xn)s, which are the bridging forces at delamination k and bar element n. Example crack bridging results for such a scenario are depicted in Figure 8. There is a clear difference between those using the adapted Alderliesten method, with crack flank opening, and using the exact displacement at the boundary layer. The calculated Kbr from each method are 570 Mpa-mm1/2 and 659 Mpa-mm1/2, respectively.

Figure 8 Fiber bridging stress with elliptical delaminations of unequal size

554

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

Extending this solution to an arbitrary laminate and other practical applications Complex FMLs, with metal and fiber layers of differing thickness and composition, have been observed to have cracks of different lengths in every layer and different delamination sizes and shapes throughout [10]. Accurate modelling of such situations depends on obtaining the best estimates of bridging stresses possible. Using the exact displacement method, it is possible to apply compliance between the metal and reinforcing composite layers at every delaminated interface. In conjunction with a means of describing the extension of the bridging material in an arbitrary laminate, which can include the metal as well as fiber layers, the exact displacement solution provides a route to a complete description of the bridging problem in any fiber metal laminate.

SUPPLEMENTING DIGITAL IMAGE CORRELATION


Digital image correlation (DIC) is an experimental technique which measures displacements and strain on an objects surface [10-11]. In DIC, grid points are selected in an initial image of an unloaded test specimen. In subsequent images, when the specimen is subject to load, the positions of those grid points are tracked. By comparing the relative positions of each grid point throughout the test, the displacement at each point can be found, and the strain can be calculated. DIC allows full field surface strain measurements to be made simply. DIC has been used recently to study cracked FMLs [12], with the demonstrated capability to measure the crack tip strain fields, the delamination shape, and even the fiber bridging stress of inverted FMLs with the fibers on the outside. Since DIC provides data on the displacement fields of FMLs, it is natural to compare DIC results to predictions of displacement made with the methods of this paper. Several comparisons were made against DIC data from cracked FMLs. First, the delamination shapes were estimated using the method described by Rodi et al. [12], in which the difference of the strain between the delaminated and undelaminated areas of the outer metal sheets makes the shape of the delamination recognizable. The measured delamination shape was used in a bridging compatibility calculation for each FML, the result of which was the predicted bridging load along the delamination boundary. With Eqns. 13 and 16, the displacements of the metal layers due to bridging and the farfield stresses, calculated using classical laminate theory for the applied loads and thermal residual stresses, were calculated and summed to get the displacement over a grid of points. In order to compare the results to the DIC measurements, in which the photos taken of the specimens under load are compared with those taken with no applied load, a second bridging and displacement calculation was made with no applied load, but with residual stresses,

Practical applications of improvements in FML crack bridging theory

555

and the displacement results were subtracted from those calculated with an applied load. Figures 9, 10, and 11 show the results of three comparisons between the predicted displacement and DIC, made by comparing displacement along horizontal lines at given distances, y, above the crack plane. In Figure 9, a comparison is made with Glare 3-2/1-0.3. In Figure 10, a comparison is made with an FML composed of three 0.4 mm thick aluminum sheets and two layers of carbon fiber composite, each made up of one prepreg with fibers in the loading direction and one with fibers in the transverse direction. Figure 11 compares the predicted and measured displacements in Glare 3-5/4-0.4. In all three cases, the predicted trends along each line agree with the DIC-measured displacement, however the exact values are not quite the same. In general, predictions along lines closer to the crack plane are in better agreement with the experimental results. Notably, the prediction fails to capture the observed behavior that, in the delaminated region, points closer to the crack plane have greater vertical displacements than those above them, meaning there is negative strain in the loading direction in this region. This discrepancy is likely a result of the way residual stress is treated in the model. Since residual stress due to curing is treated as part of the remotely applied stress of the metal layers, the relaxation of the residual stress in the delaminated region is seen in the model as an increase in the bridging stress along the delamination boundary, whereas in reality, the residual stress is introduced to the metal sheets via shear stress from the fiber layers, making relaxation of the residual stress observable as shortening of the metal, negative strain, in the delaminated region. Further work may allow the residual stress to be more realistically incorporated into the model.

556

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

Figure 9 Glare 3-2/1-0.3, a = 11.7 mm, = 100 MPa, comparison of experimental and predicted displacements

Figure 10 Aluminum/Carbon 3-3/2-0.4, a = 10 mm, = 120 MPa, comparison of experimental and predicted displacements

Practical applications of improvements in FML crack bridging theory

557

Figure 11 Glare 3-5/4-0.4, a = 14 mm, = 110 MPa, comparison of experimental and predicted displacements

CONCLUSION
The exact displacement solution presented in this work adds accuracy to calculations of the bridging stress in cracked fiber metal laminates; however, the similarity between the crack growth solutions with and without the exact displacement shows that the assumption of previous models, that the deformation of the metal layers between the crack tip and delamination boundary is negligible, is acceptable. The true value in the exact displacement solution comes not from improving old bridging models, but from adding flexibility to those models, allowing crack bridging solutions to be applied in new ways to a variety of situations. This work provides an overview of several such situations, with examples demonstrating how crack bridging analysis, enhanced with the exact displacement solution, can be applied to advance the study of FMLs. Each case can and should be explored in much greater depth, both to provide the confidence that the approach is appropriate and to gain further insight into ways the exact displacement method can improve the analysis of cracked structures and materials.

558

G. Wilson, R. Alderliesten, R. Rodi, and H. J. K. Lemmen

REFERENCES
[1] Vermeeren, C. (2002), Around Glare: a new aircraft material in context, Kluwer Academic Publishers. [2] Wu, H. F., Bucci, R. J., Wygonik, R. H., and Rice, R. C. (1993). AIAA J. of Aircraft, Vol. 30, no. 2, pp. 275282. [3] Roebroeks, G. H. J. J., Hooijmeijer, P. A., Kroon, E. J., and Heinimann, M. B., (2007). In: First International Conference on Damage Tolerance of Aircraft Structures, Benedictus, R., Schijve, J., Alderliesten, R. C., and Homan, J. J. (Eds.), Delft, The Netherlands. [4] Alderliesten, R. C. (2007), Intl. J. of Fatigue, Vol. 29, pp. 628-646. [5] Wu, X. R. and Guo, Y. J. (2002), Fatigue Fract. Eng. Mat. Struct., Vol. 25, pp. 417-432. [6] Marissen, R., Fatigue crack growth in ARALL. A hybrid aluminium-aramid composite material: Crack growth mechanisms and quantitative predictions of the crack growth rates, Ph.D. thesis, Delft University of Technology, Delft, The Netherlands, 1988. (available at http://repository.tudelft.nl/file/755663/375263) [7] Tada, H., Paris, P.C., Irwin, G.R. (2000), The Stress Analysis of Crack Handbook, 3rd ed., The American Society of Mechanical Engineers, New York. [8] Tada, H., Paris, P.C., Irwin, G.R. (1973), The Stress Analysis of Crack Handbook, 1st ed., Del Research Corporation. [9] Sun, C. T., Farris, T. N. (1989), Intl. J. Fract., vol. 40, pp. 73-77. [10] Lemmen, H.J.K., Alderliesten, R.C., Benedictus, R., Hofstede, J.C.J., and Rodi, R. (2008), The power of Digital Image Correlation for detailed elasticplastic strain measurements. EMESEG 08 conference, Engineering mechanics, structures, engineering geology, Crete. [11] Po-Chin Hung and A.S. Voloshin, In-plane strain measurement by Digital Image Correlation, J. of the Braz. Soc. of Mech. Sci. & Eng., Vol 25, no.3, 2003. [12] Rodi, R., Alderliesten, R. C., and Benedictus, R. (2009), In: Design for durability in the digital age, Proceedings of the 21st ICAF Symposium.

You might also like