You are on page 1of 11

On the Design of the Cracked Sandwich

Beam (CSB) Specimen


L. A. CARLSSON
Department of Mechanical Engineering
Florida Atlantic University
Boca Raton, FL 33431

ABSTRACT: A design analysis of the cracked sandwich beam specimen for bondline
shear fracture characterization is presented. The analysis enables selection of the appro-
priate core thickness, crack length and span length in order to achieve the desired debond-
ing failure mode prior to core crush or core shear failure.
INTRODUCTION
boat hulls
designed utilizing concept the of
Esandwich constructionstrong
XTREMELY STIFF AND are

[1-4]. State-of-the-art yacht hulls employ high stiff-


ness and strength facings of composite materials bonded to a lightweight core
material. The sandwich construction is the plate analogy to an I-beam where ma-
terial is placed in the flanges farthest away from the neutral axis, with enough ma-
terial left in the web to make the flanges interact to resist shear forces.
It is recognized that failure of the bondline between facing and core in a sand-
wich leads to loss of the &dquo;I-beam effect&dquo; because the facings would no longer in-
teract. Such a failure could initiate at a disbond introduced during manufacture
or service. In order to characterize debonding failure of sandwich construction a
test specimen, namely &dquo;the cracked sandwich beam (CSB) specimen&dquo; has been
recently introduced [5-7]. The CSB test geometry, shown in Figure 1, contains
a pre-existing disbond that is propagated in three-point flexure loading. From the
critical load for crack propagation it is possible to establish the shear fracture en-
ergy, GIlc, of the bondline. As discussed in [5-7], however, it is difflcult to propa-
gate the crack in the CSB geometry prior to shear failure of the core or core crush
at the crack tip. Actual tests of CSB specimens show that beams with very short
cracks may fail due to core shear while beams with long cracks may fail in a core
crush mode rather than the desired crack extension. Only by a trial and error ap-
proach is it possible to achieve a beam geometry where the desired crack propa-
gation mode of failure occurs prior to core shear failure or core crush.
It is the objective with this study to present design analysis of the CSB speci-
men so that the appropriate CSB geometry (core thickness, crack length and span

length) can be established to achieve debonding failure prior to shear or com-


pressive failures of the core or excessive beam deflections.
434 Journal of REINFORCED PLASTICS AND COMPOSITES, Vol. 10-,luly 1991

0731-6844/91/04 0434-11 $600/0


@ 1991 Technomic Publishing Co., Inc.

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
435

Figure 1. Cracked sandwich beam (CSB) geometry and loading.

ANALYSIS
Fracture toughness determination with the CSB specimen requires crack prop-
agation to occurprior to shear or compressive failure of the core. Shear failure
would occur if the ultimate shear strength of the core, T~,~i, is exceeded. If it is
assumed that the shear force is entirely carried by the core [1,6] the core shear
stress, 7c is calculated from:

where P is the the core thickness and b is the CSB width.


applied load, t, is
Compressive failure of the core, &dquo;core crush,&dquo; tends to occur at the crack tip of
the CSB geometry. To calculate the compressive strain at the crack tip a laminate
beam analysis is presented. This simplified analysis does not incorporate the in-
fluence of the crack tip singularity, but is considered accurate enough for initial
design of a CSB. As shown in Reference [6] it is possible to partition the load P/2
acting at the left support of the CSB specimen (Figure 1) into two concentrated
forces P, and ~2 that are assumed to act on the top and bottom facing, respec-
tively (see Figure 2). An expression for P2 as a function of P is provided in the
Appendix.
In the freebody diagram in Figure 3 it is easily visualized that the force PZ act-
ing the sub-beam 2 causes compressive stresses in the core at the crack tip.
on
To calculate the crack tip strain, laminate shear deformation beam theory [8] is

Figure 2. Load partitioning in the cracked region of the CSB specimen.

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
436

Figure 3. Free-body diagram of sub-beam 2 of the cracked region of the CSB geometry.

employed. With the coordinate system of Figure 3, the axial displacement u of


any cross-section of the beam is:

where u (x) is the displacement of the midsurface (z 0) and ~(x) is a function =

that describes the rotation of the cross-sections [8].


The bending moment per unit width acting on the beam sub-element is, ac-
cording to Figure 3,

The constitutive relation for the bending moment is [6]:

where Diis an element of [D]:

in which [A], [B] and [D] are the extensional, coupling and bending stiffness
matrices of classical lamination theory [9]. The calculation of the term Diis
quite involved because it involves extensive matrix manipulations. In this study
a computer program was employed to calculate Di for certain combinations of
materials and geometry.
To obtain that axial strain E, in the beam, Equation (2) is differentiated with re-
spect to x:

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
437

where Ex is the midsurface strain (E, =


duldx). Combination of Equations (4)
and (6) yields,

Stress Resultants
The stress and moment resultants Nx and Mx are obtained by integrating the
stress over the beam cross-section.

where the stress Qx is given by

where (Qll)k is a stiffness matrix element of ply k,

in which El is the axial Youngs modulus of a ply and v12 and V21 are the major
and minor Poisson ratios of a ply.
Substitution of Equations (7) and (10) into Equations (8) and (9) yields after in-
tegration

where All, B&dquo; and D1, are elements of the [A], [B] and [D] matrices [9], and xX
is defined as:

For the case considered (Figure 3), the beam is free of axial force N,, and Equa-
tion (12) yields,

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
438

The axial strain can thus be written as:

Substitution of the curvature xx given by Equation (14) into Equation (16) yields,

At the crack tip x = a and

It is thus recognized that magnitude of the compressive strain is increasing with


crack length in a linear manner. The load P is not a strong function of crack
length (see Appendix), and Equation (18) shows that longer crack lengths would
promote larger compressive strains at the crack tip at a given applied load P.
Based on the geometry shown in Figure 3 and the definitions of A,1 and B11 given
by:

it can be shown that the ratio

in which tF is the thickness of the facings. Subscripts F and C denote facing and
core, respectively.
Because(Qll)C < (Qll)F the ratio Bll/ All is always negative. By inspection of
Equation (18), it is thus recognized that the asymmetry of the lower beam,
(Figure 2) enhances the compressive strain at the crack tip. Examination of Equa-
tion (21) reveals that as the core thickness diminishes (h approaches tF), B,1/All i

approaches zero. Equation (18) shows that as a result of the decreased ratio
BI,IAI,, the magnitude of the compressive strain is decreasing, but as will be
shown subsequently the term Diis increasing with decreasing thickness h,
which increases the magnitude of compressive strain.

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
439

Design Considerations
Based on this analysis the magnitude of the compressive strain can be calcu-
lated for any applied load P. It is required that crack propagation occurs prior to
excessive core crush which may be defined by a strain allowable:

where Ea is the core compressive strain allowable and Pc is the critical load for
crack propagation.
To calculate the critical load for debonding, the fracture toughness GIIC is re-
quired. This value must be established by testing. For design it is recommended
that a literature value for a similar combination of core and facing materials is
chosen as an initial guess. Based on GIIC, Pc can be calculated from the following
relation derived in Reference [6]:

where subscripts AB and BC refer to the delaminated and intact regions, respec-
tively, of the CSB (see Figure 1). An expression for (D;l)AB is provided in the Ap-
pendix.
Based on this equation, it is thus recognized that the critical load is dependent
on the square root of the fracture toughness. This means that changes in GIIC do
not lead to substantial changes in the critical load. Furthermore, P, is inversely
proportional to crack length. Increased crack lengths reduce the critical load, but
increase the compressive strain at the crack tip according to Equation (18).
Another design requirement is that crack propagation should occur prior to ex-
cessive beam deflections. Otherwise, the small deflection analysis employed for
data reduction would not be valid. Small deflections are considered when the ver-
tical displacements are less than the beam thickness. It is expected that the largest
deflection of the CSB is approximately equal to the deflection 6 at the center load
nose. The requirement of small deflections prior to crack propagation may thus
be formulated as:

where 6, is the deflection corresponding to the critical load and Aror is the total
thickness of the sandwich. 6, is simply calculated from

where an expression for the compliance, C, is given in the Appendix.

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
440

Table 1. Mechanical properties of


E-glass/polyester facing and balsa wood core [7].

Design Examples
The analysis outlined will be illustrated on a sandwich representative for a
yacht hull. Continuous E-glass/polyester facings with a layup of [0/ d=45/ 45/0]
combined with a balsa wood core of density 151 kg/m3 were experimentally
characterized in [7]. Mechanical properties are provided in Table 1. To enable
calculations, a fracture toughness, G,1~ 400 J/m2, a facing thickness tF
= 3.3 =

mm, a beam width of 25.4 mm and a core compressive strain allowable, Ea .02 =

were assumed. The core shear strength, Tc,ult, is about 3 MPa [10]. Calculations
of allowable and critical loads were performed according to Equations (18) and
(23) for CSB specimens with a total span (2L in Figure 1) of 30.5 cm and core
thicknesses of 12.7, 25.4 and 50.8 mm, and crack lengths a 50.8 and 76.2 mm. =

Critical displacement and core shear stress were calculated from Equations (25)
and (1), respectively. The results of the calculations are provided in Table 2. It is
observed that for the combination of material properties and geometry con-
sidered here only the thickest core constitutes an acceptable design (Pa >_ Pc).
Increasing the crack length at a given core thickness reduces the critical and
allowable loads in exactly the same proportion leaving the ratio Pal Pc unchanged.
This is contrary to experimental experience [7], where it was found that core
crush occurs prior to crack extension at long crack lengths. This discrepancy is
evidently a result of the simplified assumptions in the beam modeling presented
here. A higher order beam analysis or a numerical analysis would be required to
more accurately obtain the distribution of strain close to the crack tip. In agree-
ment with experimental observations, however, is the increasing core shear stress

Table 2. Critical load for crack propagation, Pc, allowable load to avoid
core crush, Pa, critical displacement 6r and core shear stress at Pc
(tF = 3.3 mm).

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
441

T~, with decreasing crack length. At very short crack lengths core shear failures
were observed prior to crack extension [7].
The examples presented concern one span length (2L) only. Inspection of the
governing equations show that span length primarily affects the compliance of the
CSB. An increase in L would thus increase the beam deflection but would not
alter the critical and allowable loads.
To investigate how facing stiffness affects the results all facing moduli in Table
1 were divided by two (Poissons ratio was kept constant) and new calculations
were performed. The results are given in Table 3. Comparison with Table 2
shows that reduction of facing moduli brings down the critical loads for crack
propagation somewhat. The allowable load, however, is reduced to a larger extent
which indicates that a CSB specimen with low modulus facings would be more
prone to core crush. In this case, all facing moduli were reduced. It is easily
verified that the modulus of most importance is the axial modulus E.,. Changes
in the transverse and shear moduli and Poissons ratio have very little influence
on the results.
To investigate the importance of the core stiffness, the core modulus Ex = Key
was increased by a factor of two. Results of the calculations are shown in Table
4. Comparison with Table 2 shows that increased core modulus is favorable for
crack propagation. Almost all geometries in Table 4 are acceptable designs based
on the material properties assumed. It is recognized that the data cannot be ra-
tionalized on a simple facing-to-core modulus concept. Comparison of Tables 2
and 3, shows that reduced asymmetry of the lower beam of the debonded region
caused by decreased facing modulus does not promote crack propagation. Table
4, however, shows that asymmetry reduced by increased core modulus favors
crack propagation.
Another way to reduce asymmetry of the lower beam in the delaminated region
is to decrease the facing thickness, tE. Table 5 shows the results computed based
on the baseline mechanical properties in Table 1. Comparison between Tables 3
and 5 shows that reduction of the facing thickness by a factor of two has a similar
effect as reducing the facing moduli a factor of two. This indicates that, similar
to an uncracked sandwich, the primary property of the facing is the tensile stiff-
ness (E x t).

Table 3. Critical and allowable loads for a CSB


with 2 x reduced facing moduli. Core mechanical
properties are those of Table 1 (tF = 3.3 mm,
a = 50.8 mm).
~ --~- z

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
442

Table 4. Critical and allowable loads for a CSB


with 2 x increased core modulus. Facing
mechanical properties are those of Table 1
(tF = 3.3 mm, a = 50.8 mm).

Prior to manufacture of CSB specimens for characterization of a given set of


core and face sheets, it is recommended that the specimens are sized according
to the methodology outlined herein.

CONCLUDING REMARKS
A design analysis of the cracked sandwich beam (CSB) specimen, based on
shear deformation beam theory, has been presented. The most important require-
ment for successful bondline fracture toughness characterization is that crack
propagation occurs prior to core crush or core shear failure. Geometric
nonlinearities due to large deflections can also be avoided by appropriate design.
Based on the analysis and examples presented here, it was found that large core
thickness and large facing modulus promote the desired crack propagation failure
mode. A large core thickness, however, elevates the critical load for crack propa-
gation and this may lead to core compressive failure at the load introduction. This
failure mode is not considered herein, but it may be avoided by local reinforce-
ment and/or a large diameter load nose. It was found that reduced facing modulus
and facing thickness promote the undesired core crush failure mode. Increased
core modulus was found to promote debonding rather than core crush failure.

Table 5. Critical and allowable loads for a CSB


with 2 x reduced facing thickness. Core and
facing mechanical properties are those of Table 1.
(tF = 1.65 mm, a = 50.8 mm).

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
443

ACKNOWLEDGEMENTS
The author is grateful for programming assistance by Mr. Feridun Ozdil.
Thanks are also due to Mrs. Rosemarie Chiucchi, Mrs. Evelyn Hoffman, and
Mrs. Susan Cury for typing the manuscript.

APPENDIX
Load partitioning in the delaminated region, Figure 2, is according to Refer-
ence [6] given by:

where subscript 1 and 2 refer to the sub-beams shown in Figure 2.


Effective flexural flexibility of the debonded region, AB, of the CSB specimen,
Figure 1, is given by:

where the ratio 2 P2/ P is obtained from Equation (Al).


The compliance, C, of the CSB is [6]:

where subscript BC refers to the intact region of the CSB, and A 5 is a shear flex-
ibility element defined in Reference [6].

REFERENCES
1. Allen, H. G. 1969. Analysis and Design of Structural Sandwich Panels. Pergamon Press
2. Plantema, F. J. 1966. Sandwich Construction. N. J. Hoff, ed. Wiley.
3. Olsson, K. A 1987. "GRP Sandwich Design and Production in Sweden," Report No. 87-3, The
Royal Institute of Technology, Stockholm, Sweden.
4. Reichard, R. P. "The Design of FRP Sandwich Panels for Ship and Boat Hulls," Proc. of the 1st
Intl. Conf. on Sandwich Constructions, Stockholm, Sweden, June 19-21, 1989.
5. Sendlein, L. S. 1989 "Investigation of Debonding Failures in Sandwich Constructions," M. S.
Thesis, Department of Ocean Engineering, Florida Atlantic University.
6. Carlsson, L. A., L. S. Sendlein and S. L. Merry. 1989. "Characterization of Face Sheet/Core
Shear Fracture of Composite Sandwich Beams," J. of Composite Materials, 25(1):101-116.
7. Sendlein, L. S. and L. A. Carlsson "Experimental and Numerical Evaluation of the Cracked
Sandwich Beam (CSB) Specimen," paper presented at 3rd International Conference on Marine

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015
444

Applications of Composite Materials, Florida Institute of Technology, Melbourne, FL, March


19-21, 1990.
8. Whitney, J. M. 1987. Structural Analysis of Laminated Anisotropic Plates. Lancaster, PA. Tech-
nomic Publishing Co., Inc.
9. Jones, R. M. 1975. Mechanics of Composite Materials. McGraw-Hill.
10."Why Baltec Core: The Properties of End-Grain Balsa Core," Product Literature, Baltec Corp.,
Northvale, NJ.

Downloaded from jrp.sagepub.com at NORTH DAKOTA STATE UNIV LIB on May 30, 2015

You might also like