Repair-Based Design of Composite Structures: Scarf Repair

Samaneh Tashi a, Ali Abedian a, *

a
Aerospace Engineering Department, Sharif University of Technology, Azadi Street, Tehran Iran PO Box:

11365-11155

*
Corresponding Author: Ali Abedian, Associate Professor, Email: abedian@sharif.edu, Phone Number: +98- 21-

66168166, Mobile Number:

First Author: Samaneh Tashi, PhD Student, Email: Samaneh_tashi@ae.sharif.edu, Phone Number: +98- 21-

66166322, Mobile Number:

Abstract

Satisfying Design Limit Load for scarfed laminate and Design Ultimate Load for repaired laminate are required for

certifying adhesively bonded repair. These regulations in association with contradictory influence of scarf angle on

DLL and DUL makes the certification of a scarf repair a sophisticated procedure. Current study is dedicated to

obtain ultimate strength of quasi-isotropic pristine laminates and their scarf joints with the aim of investigating the

effect of scarf repair performance of a laminate on its design considerations using FEM. The results showed scarf

joint strength is substantially affected by the way plies shuffle in quasi-isotropic laminates. Following the

conventional design guideline to stack plies of composite laminate cannot favorably affect the strength of scarf joint

of that laminate. Considering the scarf repair efficiency as one of the principles to design a laminate provides the

opportunity for satisfying DUL and DLL, enhance the probability of approval of a scarf repair.

Keywords: Scarf Joint, FEM, Design Philosophy, Airworthiness Regulations, Composite Structure

1
 1. Introduction

The demand for lower fuel consumption has made composite materials popular in the aerospace industry to the

extent that the primary structures of new-generation aircrafts have mostly been composed of composites. Although

primary and secondary composite components are rapidly replacing their metallic counterparts in the aviation

industry, commercial aircraft authorities have identified several safety issues that are mainly categorized as limited

knowledge on composite parts behavior, technical concerns related to the unique properties of composite materials,

limited standardization of composite materials, repair techniques, and lack of training and awareness of composite

materials. As the structural design rules of composite parts including manufacturing processes and joining

technologies are not as mature as metals, satisfying the safety and performance of composite airframes substantially

demands for advanced repair techniques [1]. The scarf repair is a viable method for restoring the strength and

stiffness of thick damaged parts, does not affect the aerodynamic or stealth characteristics of the airplane. However,

a scarf joint, which is a 2-D projection of full 3-D scarf repair, is widely used as a representative for scarf repair in

most of the available experimental and numerical research works.

Jones and Graves [2] conducted a set of experiments on load carrying capacity of scarf joint and scarf repair

subjected to tension and compression loadings, respectively. A comparison of test results showed that the efficiency

of scarf repair under compressive load was 20% higher than efficiency of the equivalent scarf joint under tensile

load. Soutis and Hu [3] obtained the strength of a scarf repair and its equivalent scarf joint using the FEM and

compared their simulation result with experiments of Jones and Graves [2]. Harman and Wang [4] developed an

analytical technique to optimize design of isotropic adherends of scarf joints and validated their method using the

FEM. Gunnion and Herszberg [5] adopted a thin slice model of a scarf joint with solid elements to study influence

of laminate thickness, stacking sequence, mismatched adherends, adhesive thickness, and overply laminates on

adhesive stresses. Wang and Gunnion [6] experimentally obtained the tensile strength of scarf joints of laminates

0 / 45 / 45 / 902s and 90 / 45 / 45 / 02s , followed by a simulation based on the Generalized Plane Strain

assumption and an elastic-perfectly plastic behavior for adhesive. Utilizing Abaqus software, plane strain solid

elements, and a cohesive mixed-mode damage model for adhesive bonds, Campilho et al. [7] simulated two-

dimensional tensile-loaded scarf CFRP joints of unidirectional laminate with various scarf angles. The obtained

shear and peel stress distributions were in excellent agreement with the analytical results presented by Erdogan and

2
Ratwani [8]. Campilho et al. [9] investigated buckling of scarf joint of cross ply laminates experimentally and

numerically using plane stress solid elements of Abaqus software. Trapezoidal and triangular separation-traction

laws were applied to the adhesive/adherends interface and laminate ply interfaces. Good correlations were reported

between the numerical predictions and experimental results for the elastic stiffness, strength, and failure modes of

the joints.

A 3D FEM of scarf repair was performed by Pinto et al. [10] using the Cohesive Zone Model for the adhesive/patch

interface to study the influence of scarf angle and overplies on adhesive bond strength. Goh et al. [11] conducted a

set of experiments to obtain the scarf joint tensile strength with different flaw sizes embedded in the bondline.

Experiments were followed by simulation approaches, including average shear stress, Linear Elastic Fracture

Mechanics, Virtual Crack Closure Technique, and CZM in Abaqus, to estimate the strength of the joints

numerically. Their results showed the CZM can accurately predict the strength of flawed and perfectly bonded scarf

joints. Patel et al. [12] experimentally obtained the residual compressive strength of scarfed laminates with scarf

angles of 6 and 10 for a variety of soft and stiff laminates. The results demonstrated that the reduction in the

residual strength of a scarfed laminate, in comparison to its pristine laminate, depends on the laminate stiffness ratio.

Yoo et al. [13] experimentally studied the static and fatigue strengths of scarf joints of unidirectional laminates. The

influence of scarf angle, doubler overlap length, and patch size were investigated using tensile static tests. One

million cycles fatigue test was also performed for two shallow scarf angles. The test results showed that the fatigue

strength of the scarf joint specimen, in comparison to static strength, was significantly affected by the scarf angle.

Cohesive failure was the dominant failure mode in specimens with steeper scarf angles, whereas mixed-mode failure

was observed in specimens with shallower scarf angles.

Ridha et al. [14] performed a numerical study to analyze the progressive failure of scarf repair of 8 plies quasi-

isotropic laminate. A stepped repair geometry was adopted to take advantage of assigning continuum shell elements

to the model geometry. Continuum Damage Mechanics was applied to predict damage initiation and damage

progression of the fiber and matrix, while one row of cohesive elements was used to model adhesive bond. The

influence of the adhesive traction-separation criteria, adhesive strength, and softening law on the load-carrying

capacity of scarf repair was investigated. Hayes et al. [15] experimentally obtained the tensile strength of scarf joint

specimens with different disbond lengths located at the tip or the center of adhesive bondline under Room

3
Temperature, Hot-Wet, and Cold Dry conditions. The results showed that the strength of specimens under HW was

not affected by the disbond length, whereas the residual strength of the scarf joint under RT and CD decreased at a

faster rate than the bondline area reduction. Verified by the experiments presented by Hayes et al. [15], Hayes et al.

[16] improved scarf joint strength prediction using Abaqus Explicit solver. The contribution of various failure

mechanisms, including adhesive/adherend bond failure, adherend ply failure, plies delamination, and adhesive

plasticity, to damage initiation and damage progression of scarf joints were investigated. A 3D thin slice of scarf

joint with one element-size width under quasi-static tensile load was simulated. Hashin criteria were assigned to the

composite materials to simulate fiber and matrix degradation. An elastic-perfectly plastic behavior was assigned to

the adhesive to model its degradation and failure. For laminate  45 / 0 / 0 / 45 / 903s , the damage sequence analysis

validated by experiments, showed that the damage initiated at the 45 / 90 interface and progressed towards the

adhesive/adherend bondline and eventually caused separation of the bondline.

Pitanga et al. [17] experimentally investigated a ply-wise variable scarf angle to reduce the removal of damaged

laminates. The 0 plies were scarfed with 1:20 scarf ratio and the 45 and 90 plies were scarfed with 1:2 scarf

ratios. The results showed this scarfing method can achieve 64% of tensile strength of the straight scarf with 1:30

scarf ratio while reducing the scarf removal by 60%. Roy et. al. [18] obtained the tensile strength of scarf joints

experimentally and numerically. Scarf joints were tested at 25C , and 75C . The tensile strength of specimen at

75C showed 86% lower strength than that ones at 25C . Finite Element Analysis using cohesive elements were

conducted for 4 different 24 plies quasi- isotropic laminates, showed the joints strength varied up to 15%. Vadean et

al. [19] investigated the optimized shape of scarf repair under uniaxial and biaxial tensile loads at different loading

ratios using FEM. Their results showed the elliptical scarf patch allows removing up to 41% less of the parent

structure in comparison to circular patch. The optimum scarf geometry is dependent on the loading ratio, when

y b y b y
0  0.25 , the  0.25 and when the 0.25   1 , the  . Sonat and Ozerinc [20] investigated the
x a x a x

failure behavior of woven CFRP scarf joints with scarf angles of 1.9 , 2.8 and 5.7 , experimentally and

numerically. The Hashin criteria were used to model the intra-laminar failure of scarf joints. A good agreement

between strength obtained from tests and simulation was reported. Their study showed that the dominant failure

mode for scarf joint with 5.7 and 1.9 were respectively the cohesive failure and intra-laminar failure, while the

4
specimen with 2.8 scarf angle exhibited a mixed mode failure of the composite laminate and adhesive. Moreira et

al. [21] conducted experimental and numerical works to study the static strength and high cycle fatigue life of a scarf

joint of 16 plies UD CFRP, under three-point bending load. The experimental results that were obtained for scarf

angle of 10 used to validate the finite element model which was a 2-D model consisting of CZM elements for

adhesive. The results showed the fatigue life of a scarf repaired laminate was more affected by the scarf angle in

comparison to the static strength. Hoang et al. [22] obtained the strength of scarf joint with the stacking sequence of

[45 / 45 / 45 / 0 / 90 / 0 / 45 / 45 / 90 / 0]S , an over-ply layer, and four different scarf ratios (

1/ 5, 1/10, 1/ 20, 1/ 30 ) using experiments and FEM. T-Sai Wu criteria and CZM were used to capture damage

initiation and progression in composite laminate and adhesive. An exponential function used to fit the experimental

data to correlate the scarf angles and failure loads. The results showed the discrepancies between experimental and

simulation strengths depends on scarf angle. Tashi and Abedian [23] used Abaqus to simulate scarf repair and joint

under uniaxial and equibiaxial loadings for various lay-ups and stacking sequences. The results showed ply shuffling

in a quasi-isotropic laminate significantly affects the Stress Concentration Factor (SCF) of the adhesive bondline.

The results raised a key design question: Should scarf joint/repair performance be a design criterion for composite

laminates in early airframe design?

A general design guideline for most aerospace structures suggests considering the following recommendations

when selecting a laminate stacking sequence [24].

 Stacking sequence have to be symmetric.

 Stacking sequence have to be balanced.

 No more than a limited number, generally four plies of the same orientation should be stacked.

 The difference between fiber orientations of consecutive plies should not exceed 45 .

 No 0 ply on the free surface of the laminate is allowed.

 At least, 10% of plies should be placed in each direction of 0 , 45 and 90 .

The quasi-isotropic laminates with same number of plies are supposed to have similar strength under uniaxial load

based on Classical Laminate Theory. But, as proven by experiments and numerical studies, their tensile strengths are

different from each other due to the out-of-plane normal and shear stresses occurring at free edges of the laminates

5
[25-31]. Stress analysis of the free edge of the laminates showed that the positive out-of-plane normal stress at the

edge causes the delamination and subsequently a lower strength, while negative out-of-plane normal stress at free

edges suppresses the delamination [25, 27]. Ogihara et al. [29] experimentally determined the static and fatigue

strengths of two quasi-isotropic laminates, 0 / 45 / 90 / 45s and  45 / 0 / 45 / 90s , and showed that the latter has a

lower strength. Tessema et al. [28] applied Damage Image Correlation to capture the damage initiation and

propagation in three quasi-isotropic laminates: 0 / 45 / 45 / 90 s , 0 / 45 / 90 / 45s , and 0 / 90 / 45 / 45s . The

laminate 0 / 90 / 45 / 45s exhibited the highest strength. Chen et al. [30] experimentally determined the tensile

strength of six quasi-isotropic laminates presented in Table 1 and discussed the effect of free edge stresses on

delamination using ANSYS. Hesabi et al. [31] obtained the strengths of six stacking sequences under a static tensile

test and ranked the laminates based on their strengths, presented in Table 2. Wisnom et al. [32] experimentally

obtained the strength of quasi-isotropic laminates with  45 / 90 / 45 / 0ns known as dispersed plies laminates and

45n / 90n / 45n / 0n s known as blocked plies laminates with “n” includes 1, 2, 4, 8. As demonstrated by the

experiments, laminates with dispersed plies exhibited much higher strengths than their blocked counterparts.

So far, many innovative scarf repair designs have been reported to increase the scarf repair strength and decrease the

removal of pristine laminates. However, none of the previous research discussed on scarf repair performance as a

parameter to design a composite laminate [33]. This study aims to analyze how the initial design of a pristine quasi-

isotropic laminate is affected by scarf repair efficiency when the laminate is damaged and needs repair. FEM is used

to obtain the ultimate tensile strengths of pristine laminates and their scarf joints. To lower computational costs,

scarf joints are examined instead of repair geometry. Quasi-isotropic laminates and their scarf joints are studied to

only measure the effect of stacking sequence on the strength of pristine laminate and its scarf joint.

2. Finite Element Model, Material Properties, Analysis Method

The 16 plies pristine laminates and their scarf joints with 5 scarf angle were modeled using ABAQUS. The
material and geometric properties of the models are listed in Table 3, Table 4, and

Table 5. Pristine specimens under tensile loads were modeled as shown in Fig. 1. Composite plies were individually
modeled using continuum shell elements and bonded together by cohesive contact interactions. 2D Hashin Failure
criteria were used to model the composite failure modes. Cohesive damage properties based on the quadratic traction
criterion for damage initiation and mixed-mode behavior for damage evolution were used to capture the plies
delamination. Details of intralaminar and interlaminar failure properties are presented in

6
Table 6, Table 7, and Table 8. The ABAQUS Explicit solver was used to obtain strength of pristine laminates with

time period of 0.01 second. The tensile load was applied by smooth step amplitude. A mass scaling factor of 10 was

assigned to the elements to decrease the run time without compromising calculation accuracy. The performed mesh

study which was in good agreement with mesh study results of Divse et al. [34] showed that the mesh size of

0.25 mm 0.25 mm is refined enough to have an accurate results.

A thin slice of the scarf joint specimen, which was extruded to the size of an element was considered as the scarf

joint model. The front and rear surfaces of the specimen were constrained against the displacement normal to these

surfaces to simulate conditions close to the plane strain assumption. The geometry and boundary conditions of the

model are depicted in Fig. 2, and are similar to the simulation of Hayes et al. [16]. The simulation approaches

considered for the scarf joint in the current study are verified by the simulation and experimental results of Hayes et

al. [16]. The strengths of scarf joint specimens were obtained using two different modeling approaches. In modeling

approach “A” only one cohesive interface was defined between one adherend and adhesive and strength of the joint

specimens were obtained using Abaqus Standard. In modeling approach “B,” using Explicit solver of Abaqus,

cohesive contact properties were assigned to all the plies interfaces as well as the interfaces between plies and the

adhesive to understand the influence of interlaminar failure on scarf joint performance. For simulation approach

“B”, the time period, mass scaling factor and amplitude of applied load were considered 0.01 second, 10 and smooth

step as well. Neglecting other damage mechanisms that contribute to scarf joint failure is acceptable as Hayes et al.

[16] reported that joint failure is substantially influenced by delamination propagated towards the

adhesive/adherends interfaces. Based on previous study recommendation [16, 11, 35] the element size of a ply

thickness with one row of elements through the thickness of each ply was considered to mesh composite plies.

C3D8R and C3D6 elements were assigned to the plies. Four rows of solid elements were used to mesh the adhesive

through its thickness. The damage stabilization coefficients required to stabilize the solutions were set to the 104

and 105 for composite material damage and cohesive damage, respectively.

2.1.Simulation Verification
The accuracy of the pristine laminate model was verified using the experimental results of Kechai et al. [36]. The

strength of the pristine laminate with the  04 lay-up was considered for the verification. The cohesive contact and

composite failure properties were borrowed from the work of Divse et al. [34] who verified the experiments of

7
Kechai et al. [36]. The strengths predicted by current study and the experimental results of Kechai et al. [36] shows

good agreement as depicted in Fig. 3. To investigate the scarf joint model accuracy, the experiment and simulation

results of Hayes et al. [16] for a scarf joint with a 3 scarf angle were considered. Comparison of the strength

results obtained by Hayes et al. [16] and the strengths obtained by the simulation approaches “A” and “B” are

presented in Fig. 4. There is a very good agreement between the current simulation results and those of Hayes et al.

[16] when 1 cohesive interface between adhesive/adherends was defined. However, assuming only one cohesive

interface to simulate scarf joint failure leads to strength overestimation. The results obtained from approach “B” is in

a very good agreement with experimental result of Hayes et al. [16]. However, a slight difference between

simulation approach “B” and Hayes et al. [16] is observed which can be caused by the discrepancy between load

amplitude, mass scaling factor, and density assumed in these two models. Moreover, Hayes et al. [16] considered all

damage mechanisms contributing to scarf joint failure, while the current simulation was limited to interlaminar and

adhesive/adherend interface failure modes.

2.2.Strength of Pristine Laminates

The Last Ply Failure criterion was considered to determine the ultimate tensile strength of the pristine laminates.

Despite the fact that LPF is equivalent to the unstable growth of damage and large deformations of the composite

laminate, it was intended to have a common understanding of failure of the pristine laminates.

The considered stacking sequences did not necessarily follow the composite laminate design guidelines. The stress-

strain curve of 16 plies laminates is presented in Fig. 5. The quasi-isotropic laminates, before damage initiation,

behaved similarly in the region of elastic loading, while their differences commenced as the interlaminar damages

were initiated and propagated through the laminates. Reported in previous studies [25-31], the free edge stresses that

vary with ply placement are responsible for the strength discrepancies of quasi-isotropic laminates under tensile

loads. Ranking of quasi-isotropic laminates according to their ultimate tensile strengths are presented in Table 9.

Considering Fig. 5, the difference in strength for laminate 45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s is highly

distinguishable compared to the others. This could be explained by the two neighboring 0 stiff plies close to the

free surface. Interestingly, the results for laminate 0 / 45 / 45 / 902 s shows that the effect of neighboring stiff plies

is much higher than having the 0 plies on the free surface. As Fig. 5 shows, 0 / 45 / 45 / 902 s behaves almost the

8
same as the 16 plies laminate with 0 ply closest to the plane of symmetry (i.e. 90 / 45 / 45 / 02 s ). The reason

could be in load carrying compensation that the 0 ply inside the laminate provides for the failed 0 ply on the free

surface.

The simulation results presented in Table 9 are in good agreement with the trend reported by Wisnom et al. [32],

who demonstrated that a quasi-isotropic laminate with dispersed plies has a higher tensile strength than a quasi-

isotropic laminate with blocked plies.

2.3.Strength of Scarf Joint Model

2.3.1. Strength Prediction with 1 cohesive interaction (Approach “A”)

Following the approach “A,” tensile strength of scarf joints of 16 plies laminates were obtained when only one of

the interfaces between adherends and adhesive considered as cohesive interface. The scarf joint failure was

dependent solely on the failure of adherend/adhesive interface. The tensile strengths and contour plots of cohesive

damage propagation through the interface of the adhesive/adherend are presented in Fig. 6 and Fig. 7. As shown and

Fig. 6, the stacking sequence significantly affects the tensile strength of the scarf joints of quasi-isotropic laminates.

2.3.2. Strength Prediction with all cohesive interaction (Approach “B”)

To gain insight into the interlaminar damage influence on scarf joint performance, tensile strengths of scarf joints

were also obtained using approach “B.” The results are shown in Fig. 8. The presented results in Fig. 8 confirm that

the recommended design guidelines for stacking sequence do not necessarily are in favor of strength recovery of

scarf joints. The scarf joints with 0 ply next to the symmetry plane do not show a good strength. Worth to

mention that the scarf joint  45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s showed the worst strength among all the laminates.

This is explained by neighboring effects of two 0 stiff plies close to the free surface.

The guideline recommendation of keeping the angle difference between neighboring plies to 45 is not effective for

enhancing the strength of scarf joints. As shown in Fig. 8, the scarf joints with stacking sequences of

45 / 0 / 90 / 452s and 45 / 45 / 0 / 902s have higher strengths than stacking sequence of  45 / 0 / 45 / 90s ,

despite not meeting the minimum angle difference rule. Moreover, the scarf joint  45 / 45 / 0 / 90 / 90 / 0 / 45 / 45s ,

despite having a stacking sequence with 0 / 90 neighboring plies, ranked 3rd among the scarf joints.

9
According to Table 10, strengths of 0 / 45 / 45 / 902 s , 45 / 90 / 90 / 45 / 0 / 45 / 0 / 45s and

45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s obtained from approaches “A” and “B” are close, revealing that dominant failure
mechanism of these stacking sequences is adhesive/adherend failure.

Because several quasi-isotropic stacking sequences were studied here and a one-by-one investigation of their

damage initiation and progression is beyond the scope of the current study, the failure mechanism investigation is

limited to stacking sequences with the lowest and highest tensile strengths.

The cohesive damage initiation and propagation for scarf joint of 45 / 45 / 0 / 902s are presented in Fig. 9.

Cohesive damage began at the tip of the 0 ply and was then slowly propagated. Following the increase in the

applied load, disbond occurred at the adherends/adhesive interfaces adjacent to the 0 ply tips. Delamination

observed between the 0 and 45 plies of the left adherend and the 0 and 90 plies of the right adherend.

Increasing the applied load caused adherend/adhesive disbond resulted in final failure. As shown in Fig. 10, the

failure of the scarf joints of the  45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s stacking sequence was expectedly driven by the

high SCF produced at the tip of the 0 plies. The disbond that occurred at the tip of the 0 plies rapidly propagated

along the bondline and caused the final failure.

A comparison of the damage initiation and propagation of the upper and lower ranks of the understudied stacking

sequences in scarf joints confirmed that the plies placement, even in quasi-isotropic laminates, significantly affects

joint performance. Ply placement in a quasi-isotropic laminate can cause premature failure of the

adhesive/adherends bondline, whereas the composite adherends are still capable of bearing extra load. On the other

hand, there are stacking sequences of quasi-isotropic laminates that can provide a more uniform stress field in the

joint which leads to global failure of the joint, where the capacity of composite adherends also contributes to

carrying the applied load. Although the analysis of damage initiation and damage progression in scarf joints is a

broad and interesting topic that requires further study, it is not the primary objective of the current research. Instead,

the focus of this failure analysis is to address the limitations of the current design procedure for composite laminates

and its impact on their repair process, while also providing new insights into the design and optimization of

composite laminates.

10
 3. Discussion

Referring to Table 10, the scarf joint performance of quasi-isotropic laminates is much more sensitive to ply

placement than that of the pristine laminate. For the pristine laminates, the difference in ultimate tensile strength for

the 1st and last ranked understudied stacking sequences is 22.5%, while this difference for the best and worst scarf

joints is 125%.

Based on Table 10, the maximum strength ratio (scarf repair strength-Approach “B”/Pristine Laminate ultimate

strength) is 52%. It is emphasized that the ratios are calculated according to the ultimate strengths obtained from the

failure of the last ply of the pristine laminates, and they are limited to the discussed stacking sequences.

The large disparities in strength between the quasi-isotropic scarf joints and their corresponding pristine laminates

indicate that the presence of an inclined thin layer of adhesive as a bonding medium, combined with plies

placement, can remarkably influence the stress field adjacent to the bondline in a way that the strength recovery of

the scarf repair is reduced beyond the expectations.

One of the most significant conclusions that can be drawn from Table 10 is that the influence of stacking sequence

on the strength of pristine quasi-isotropic laminates is not the same for their corresponding scarf joints. In some

cases, stacking sequence has a reverse effect on the strength of the pristine laminate and its scarf joint.

In a hypothetical scenario in which a quasi-isotropic laminate is required to carry tensile loads, the stacking

sequence of  45 / 45 / 0 / 902 s is excluded from the potential candidate selection pool in the first stage of the

laminate design. This is based on the recommended design guidelines, despite the fact that, the neglected stacking

sequence showed the best tensile performance as a scarf joint rather than the other discussed stacking sequences

(Table 10).

Early exclusion from the laminate selection pool can occur for [45 / 45 / 0 / 90 / 90 / 0 / 45 / 45]s with low tensile

strength. The stacking sequence does not obey the design guidelines, because the placement of plies in a pristine

laminate result in early delamination. As shown in Fig. 11 (a), the delamination of the pristine laminate of the

stacking sequence occurred at 55% of the final failure load, resulting in premature failure of the laminate at a

relatively low strength. Nevertheless, the scarf joint performance of the stacking sequence was comparatively high,

ranked 3rd among the examined scarf joints. As shown in Fig. 11 (b), the disbond of the adhesive/adherends and

11
plies delamination initiated and propagated slowly. No disbonds or delamination was observed at 90% of the final

failure load.

Among the examined stacking sequences, 90 / 45 / 0 / 452 s is the most favorable stacking sequence, as its tensile

strength is ranked 1st among pristine laminates and 2nd in scarf joint specimens, implying that manufacturing a

laminate with acceptable performance in its pristine and scarf joint configurations is not unlikely. To determine the

global optimum for stacking sequences or provide new design guidelines for laminates, the design philosophy

should include the acceptable repair performance of the composite structures, as the use of composite materials is

growing in aerospace and other industries. Additionally, the Philosophy of Design for Repair (PDR) for composite

laminates should also be considered for recycling, to minimize the impact on the environment after the service life

of the parts.

The contradictory effects of the stacking sequence on the pristine laminate and its scarf joint warrant further

investigation as it can affect the Repair, Maintenance and Overhaul program (MRO) of composite parts used in an

airframe structure. Damage to composite structures during their design life is not unexpected as composites have a

brittle nature and are susceptible to impacts. However, as proven by the current study, the performance of scarf

repair can vary significantly with the stacking sequence. Thus, involvement of the repaired laminate performance in

the design procedure of a composite laminate is necessary to avoid penalties when strength recovery by scarf repair

is required for a damaged part.

The exclusion of stacking sequences that do not have an acceptable tensile performance at the early stage of a

structural part design can result in difficulties with repair design and decision making on repairing or replacing the

part. As demonstrated by the results in Table 10, most of the stacking sequences with reasonable tensile performance

in their pristine shape that were probably kept in the selection pool showed a lower strength recovery as scarf joints.

This means that the scarf joints of the selected laminates need shallower scarf angles to meet the DUL requirement.

Traditionally, defined by aviation regulators and airworthiness authorities, the strength of adhesively bonded repair

of a laminate is compared to the DUL, which is equivalent to the tensile strength of Open-Hole Tension specimens

with the hole diameter of 6.35 mm of that particular laminate. The residual strength of the scarfed laminate is

2
compared to Design Limit Load which is equal to of DUL. The shallower scarf angles impose greater removal of
3

12
pristine portions of the laminate, making it difficult to obtain approval for the scarf repair as the DLL requirement

would not be met, which is obliged by airworthiness regulations for adhesively bonded repairs.

The controversial influence of scarf angle on the DUL and DLL is a complicated condition that can lead to the

following decisions: a) approval of steeper scarf angles for scarf repair with probable use of a thick doubler to

restore the laminate strength. However, thick doublers may raise additional concerns including a local increase in

stiffness and secondary bending effects. b) rejection of adhesively bonded scarf joints and probable use of bolted

repairs for strength recovery of the damaged part, which may impose unwanted stress concentrations on brittle

composite laminate. c) rejection of all potential repair candidates and replacement of the part, which may be

environmentally and financially a costly decision.

Therefore, it is necessary to reconsider the design procedure of pristine composite laminates by considering their

repair performance in the early stages of their design. A small defect, even a minor damage, or existence of an

inclined thin adhesive film for joining purposes in composites, can disrupt the symmetry and balance of the

laminate, creating unexpected couplings that may in turn change the laminate functional behavior. Such incidents

can be detrimental in some cases by causing high SCF at ply stations in adhesively bonded repairs. Reported by

Tashi and Abedian [23], the SCF for 3-Dimensional 8 plies scarf repair changes from 1.53 to 2.40 for various quasi-

isotropic stacking sequences, while the SCF of an isotropic repair with E  E11 of the composite lamina is 1.10. Their

results demonstrated that a composite part under complex circumstances, such as adhesively bonded repair, is

considerably different from an isotropic part. Considering the efficiency of repaired laminates as a design principle

for stiffer laminates commonly used in the aerospace industry is much more emphasized by the authors of current

study. This is because, as proven by Wang and Gunnion [6], scarf joints of stiffer laminates with higher bondline

SCF have less potential for strength recovery; therefore, a shallower scarf angle is required (meeting DUL). On the

other hand, as demonstrated by Patel et al. [12], scarfed hole specimens of stiff laminates require steeper scarf angles

to provide sufficient residual strength (meeting DLL). This makes it more conundrum to meet airworthiness

regulations for the approval of scarf repair of stiff laminates.

To examine the concept of PDR, another FEM analysis was conducted. A new geometry as shown in Fig. 12 (a),

was modelled with a width of one element under tension. The material, interface properties, plies, and adhesive

thicknesses were the same as those in the scarf joint model. The failure analysis was limited to the disbonding of

13
adhesive/adherends by assigning cohesive behavior to the interfaces between the adhesive and adherends. The

stacking sequence of  45 / 45 / 0 / 902 s was considered because it had the best scarf joint performance under tensile

load among the discussed stacking sequences. This laminate has a higher chance of recovering the required strength

when steeper scarf angle is required. The tensile strengths of the scarf repair cross-sections for scarf angles of 5 and

8 were obtained using Abaqus/Explicit, presented in Table 11 and Fig. 13.

As expected, the strength of the scarf repair cross-section for an 8 scarf angle is approximately 32% less than the

strength of the 5 scarf repair cross-section. Assuming that the 8 scarf angle was steep enough to meet the DLL,

adding doublers to the repair configuration was required to enhance the tensile strength of the 8 scarf repair cross-

section to meet the DUL. Three different configurations of doublers, as presented in Table 11, were considered to

enhance the load-carrying capacity of the scarf repair cross-section with an 8 scarf angle. The dimensions of the

two-sided unsymmetrical doublers are schematically presented in Fig. 12(b). The ply thickness and material

properties of the doublers were similar to those of the repair and parent structures. Thin layers of adhesive with 0.13

mm thickness were used to bond doublers to the top and bottom surfaces of the scarf repair cross-section. Cohesive

contact was considered for all interfaces between adhesives and adherends. The tensile strengths of the repair-

doubler configurations are shown in Fig. 13 and Table 11. The strength of the scarf repair cross-section with 8

scarf angle is 142 MPa lower than that of the scarf repair cross-section with a 5 scarf angle. However, the scarf

repair cross-section with an 8 scarf angle of  45 / 45 / 0 / 902 s still yields a higher strength than some of the

stacking sequences with a 5 scarf angle ranked in Table 10. Increasing the doubler thickness stiffened the repair.

Moreover, two-sided doubler configurations, both symmetric and unsymmetric are more effective than a one-sided

doubler in restoring the strength of a scarf repair cross-section. Nevertheless, none of the examined repair-doubler

configurations for an 8 scarf angle, recovered the tensile strength up to a strength of 5 scarf angle of

45 / 45 / 0 / 902s stacking sequence. The scarf repair-doubler configuration with an 8 scarf angle and stacking

sequence of [[45 / 45 / 45 / 45]  [45 / 45 / 0 / 90]2s  [45 / 45]] restored the strength to 399.7 MPa, which is

higher than the tensile strength of most of the understudied stacking sequences with a 5 scarf angle presented in

Table 10. This example confirms that designing a pristine laminate considering its future repair performance can

14
enhance its chance of meeting DLL and DUL requirements when adhesively bonded repair of the laminate is

needed.

It should be emphasized that no parametric study has been conducted on the design parameters of a doubler.

Studying the DLL of scarfed composite laminates accompanied by probabilistic analysis is required to gain a better

understanding of how the repairability of a composite laminate can affect its design procedure, as long as the major

concern for adhesively bonded repair is the residual strength of the damaged part. In addition, a more comprehensive

FEM analysis in association with experimental work that includes other loading conditions can lead to improved

conclusions.

Afterwards, the authors did not emphasize any of the stacking sequences as the best possible one for repair

performance, but insisted that the design procedure of composite laminates must include their repair efficiency as

one of the design principles.

4. Conclusion

A slight change in balanced symmetric composite laminates produces a 3D complex stress field with a high SCF

owing to the coupling of longitudinal and flexural stiffness matrices. Various stacking sequences of 16 plies quasi-

isotropic laminates were studied to understand the effect of ply placement on the tensile strength of a pristine

laminate and its scarf joint. Failure analysis of pristine laminates included the failure of composite materials utilized

by Hashin criteria as well as delamination modeled by cohesive interactions between plies. The load carrying

capacity of the equivalent scarf joints was obtained by two simulation approaches: A) cohesive interaction was only

defined for one adhesive/adherend interface and B) cohesive interactions were defined for all the interfaces. The

pristine composite laminates and their scarf joints were ranked based on their tensile strengths. The free edge effect

caused up to 22.5% discrepancy between ultimate tensile strengths of various stacking sequences. The tensile

strength of the scarf joints was highly affected by the ply placement when the discrepancy between the 1st ranked

and the last ranked understudied stacking sequences was 122.5%. A stacking sequence with a higher tensile strength

in the pristine configuration does not necessarily result in a higher performance of its scarf joint. In some cases, the

stacking sequence that can be excluded from the selection pool in the first step of the laminate design, demonstrated

higher tensile strength recovery in its scarf joint configuration. This observation led to the idea of considering the

repair performance of a composite part in its design process (PDR) to decrease the risk of part replacement when

15
damage occurs. The selection of a stacking sequence whose scarf repair provides a higher strength recovery

increases the probability of meeting the DUL and DLL requirements of airworthiness regulation for adhesively

bonded repairs. To evaluate the accuracy of the idea, the tensile strength of a scarf repair cross section with an 8

scarf angle that was elaborated with one-sided and two-sided doublers were obtained. The results showed that the

configuration with a steeper scarf angle carried equal or more tensile load than most of the understudied stacking

sequences with a shallower scarf angle of 5 . This example shows how seeing the scarf repair performance as a

design consideration for a composite laminate can provide an opportunity for the approval of adhesively bonded

repair. Scarfing damaged parts with steeper angles causes lower removal of the pristine portions of the laminate,

thereby increasing the chance of meeting the DLL requirement. Concurrently restoring the strength and stiffness of

the damaged part and satisfying the DUL requirement become feasible by utilizing a doubler-scarf configuration for

the repair.

5. Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or none-profit

sectors.

6. Data Availability

Not Applicable

7. Authors Biography

Dr. Ali Abedian is an associate professor at aerospace engineering department of Sharif university of

technology. His research interests and activities are focused on creep analysis of metal matrix composites, stress

analysis of free surface damage of unidirectional composites, application of genetic algorithms (gas) to the

optimum design of laminated composites, stress analysis of delamination phenomena in laminated composite,

FEM analysis of super plasticity in composites, FEM analysis of contact problems, optimum material selection

and Nano technology. In addition to Academia, he has actively worked in the industry in the field of design,

manufacture and repair of composite structures, structural health monitoring and Life prediction and extension

of metallic structures.

16
 Samaneh Tashi is a PhD graduate of Sharif university of Technology. Her research activities and interests are

focused on design and analysis of composite structures, mainly the adhesively bonded joints. In recent years,

she has been working as an R&D project manager, an author of technical- financial proposals, and a lecturer of

undergraduate courses.

8. References

]1[Katnam K.B., Da Silva L., Young T., "Bonded repair of composite aircraft structures: A review of
scientific challenges and opportunities, Progress in Aerospace Sciences", 61, pp. 26-42, (2013).
]2[Jones J.S., Graves S., "Repair Techniques for Celion/LARC-160 Graphite/Polyimide Composite
Structures", Rockwell international downey ca space transportation systems Div., (1984).
]3[Soutis C., Hu F., "Strength analysis of adhesively bonded repairs", In Recent Advances in Structural
Joints and Repairs for Composite Materials, L. Tong, C. Soutis, pp. 141-72, Springer Nature BV,
Dordrecht, Netherlands, (2003).
]4[Harman A.B., Wang C.H., "Improved design methods for scarf repairs to highly strained composite
aircraft structure", Composite structures, 75, pp. 132-44, (2006).
]5[Gunnion A.J., Herszberg I., "Parametric study of scarf joints in composite structures", Composite
structures, 75, pp. 364-76 (2006).
]6[Wang C.H., Gunnion A.J., "On the design methodology of scarf repairs to composite laminates",
Composites Science and Technology, 68, pp. 35-46, (2008).
]7[Campilho R . D. S. G. de Moura M.F.S.F., Domingues J. J. M. S., "Stress and failure analyses of scarf
repaired CFRP laminates using a cohesive damage model", Journal of Adhesion Science and Technology,
21, pp. 855-70, (2007).
]8[Erdogan F., Ratwani M., "Stress distribution in bonded joints", Journal of Composite Materials, 5, pp.
378-93, (1971).
]9[Campilho R.D., De Moura M., Ramantani D., et al., "Buckling behaviour of carbon–epoxy adhesively-
bonded scarf repairs", Journal of Adhesion Science and Technology, 23, pp. 1493-513, (2009).
]10[Pinto A., Campilho R., De Moura M., et al., "Numerical evaluation of three-dimensional scarf repairs
in carbon-epoxy structures", International Journal of Adhesion and Adhesives, 30, pp. 329-37, (2010).
]11[Goh J., Georgiadis S., Orifici A., et al., "Effects of bondline flaws on the damage tolerance of
composite scarf joints", Composites Part A: Applied Science and Manufacturing, 55, pp. 110-9, (2013).
]12[Patel M., Daynes S., Wang C.H., "The Influence of Orthotropy and Taper Angle on the Compressive
Strength of Composite Laminates with Scarfed Holes", Advanced Materials Research, Trans. Tech. Publ.,
pp. 178-84, (2014).
]13[Yoo J.-S., Truong V.-H., Park M.-Y., et al., "Parametric study on static and fatigue strength recovery
of scarf-patch-repaired composite laminates", Composite structures, 140, pp. 417-32, (2016).
]14[Ridha M., Tan V., Tay T., "Traction–separation laws for progressive failure of bonded scarf repair of
composite panel", Composite structures, 93, pp.1239-45., (2011).
]15[Hayes-Griss J., Gunnion A., Khatibi A.A., "Damage tolerance investigation of high-performance scarf
joints with bondline flaws under various environmental, geometrical and support conditions",
Composites Part A: Applied Science and Manufacturing, 84, pp. 246-55, (2016).

17
 ]16[Hayes-Griss J., Orifici A ,.Khatibi A.A., "An improved progressive failure modelling and damage
tolerant design methodology for composite scarf joints with bondline flaws", Composites Part A: Applied
Science and Manufacturing, 131, 105776, (2020).
]17[Pitanga M.Y., Cioffi M.O.H., Voorwald H.J., et al., "Reducing repair dimension with variable scarf
angles", International Journal of Adhesion and Adhesives, 104, pp. 102752, (2021).
]18[Roy R., Kweon J., Nam Y., "Tensile strength of multi-angle composite laminate scarf joints with
FEM", IOP Conference Series: Materials Science and Engineering, IOP Publishing, 784, 012007, (2020).
]19[Vadean A., Abusrea M., Shazly M., et al., "Improvement of scarf repair patch shape for composite
aircraft structures", The Journal of Adhesion, 99, pp. 1044-70, (2023).
]20[Sonat E., Özerinç S., "Failure behavior of scarf-bonded woven fabric CFRP laminates", Composite
Structures, 258, 113205 (2021).
]21[Moreira R., De Moura M., Silva F., et al., "High-cycle fatigue analysis of adhesively bonded
composite scarf repairs", Composites Part B: Engineering, 190, 107900, (2020).
]22[Hoang V.-T., Lee D.-S., Nam Y.-W., et al., "Numerical Prediction of Failure Load of Scarf-Patch-
Repaired CFRP Composite Using Damage Zone Model and Cohesive Zone Model', International Journal
of Aeronautical and Space Sciences, 24, pp. 419-29, (2023).
]23[Tashi S., Abedian A., "A Comprehensive 2 Dimensional and 3 Dimensional FEM Study of Scarf Repair
for a Variety of Common Composite Laminates Under In-Plane Uniaxial and Equibiaxial Loadings",
International Journal of Adhesion and Adhesives, 103092, (2022).
]24[Irisarri F.-X., Laurin F., Leroy F.-H., et al., Computational strategy for multiobjective optimization of
composite stiffened panels, Composite structures, 93 (3), pp. 1158-1167, (2011).
]25[Pagano N., Pipes R.B., "The influence of stacking sequence on laminate strength, Mechanics of
Composite Materials", Springer, pp. 246-54, (1994).
]26[Smith P., Pascoe K., "The effect of stacking sequence on the bearing strengths of quasi-isotropic
composite laminates", Composite Structures, 6, pp. 1-20, (1986).
]27[Stevanović M., Marković D., Pešikan-Sekulić D., "Effects of stacking sequence on the strength of
cross-ply and quasi-isotropic carbon/epoxy laminates", Materials Science Forum, Trans. Tech Publ ,pp.
465-72, (2004).
]28[Tessema A., Ravindran S., Wohlford A., et al., "In-Situ observation of damage evolution in quasi-
isotropic CFRP laminates", Fracture, Fatigue, Failure and Damage Evolution, Volume 7, Springer, pp. 67-
72, (2018).
]29[Ogihara S., Takeda N., Kobayashi S., Kobayashi A., "Effects of stacking sequence on microscopic
fatigue damage development in quasi-isotropic CFRP laminates with interlaminar-toughened layers",
Composites Science and Technology, 59, pp. 1387-98, (1999).
]30[Chen J., Takezono S., Nagata M., et al., "Influence of stacking sequence on the damage growth in
quasi-isotropic CFRP laminates", Journal of the Society of Materials Science, 50, pp. 178-85, (2001).
]31[Hesabi Z., Majidi B., Aghazadeh J., "Effects of stacking sequence on fracture mechanisms in quasi-
isotropic carbon/epoxy laminates under tensile loading, Iranian Polymer Journal, 14(6), pp. 531-538
(2005).
]32[Wisnom M., Khan B., Hallett S., Size effects in unnotched tensile strength of unidirectional and
quasi-isotropic carbon/epoxy composites, Composite Structures, 84, pp. 21-8, (2008).
]33[Orsatelli J.-B., Paroissien E., Lachaud F., et al., "Bonded flush repairs for aerospace composite
structures: A review on modelling strategies and application to repairs optimization, reliability and
durability", Composite Structures, 304, p. 116338, (2023).
]34[Divse V., Marla D., Joshi S.S., "Finite element analysis of tensile notched strength of composite
laminates", Composite Structures, 255 p.112880, (2021).

18
 ]35[Wang C.H., Gunnion A.J., Orifici A.C., et al., "Residual strength of composite laminates containing
scarfed and straight-sided holes", Composites Part A: Applied science and manufacturing, 42, 1951-61,
(2011).
]36[Khechai A., Tati A., Guerira B., et al., "Strength degradation and stress analysis of composite plates
with circular, square and rectangular notches using digital image correlation", Composite structures,
185, pp. 699-715, (2018).

19
Table 1. Tensile strength of T800H/Epoxy quasi-isotropic laminate by Chen et al. [30]

Table 2. Tensile strength of U.F.C200 carbon/epoxy quasi-isotropic laminates by Hesabi et al. [31]

Table 3. Summary of pristine laminate and scarf joint geometry

Table 4. Material Properties for unidirectional IM7/977-3[16]

Table 5. Room temperature material properties for FM 300-2 K Adhesive [16]

Table 6. In-plane material failure properties for a single ply of IM7/977-3 [16]

Table 7. Damage parameters of IM7/977-3 [16]

Table 8. Cohesive failure properties of IM7/977-3 [16]

Table 9. Tensile Strength of quasi- isotropic laminates

Table 10. Comparison of the simulation results for quasi-isotropic scarf joints and pristine laminates

Table 11. Scarf-Doubler configuration for quasi-isotropic scarf repair with  45 / 45 / 0 / 902 s stacking sequence

20
Fig. 1. Pristine laminates boundary and loading conditions

Fig. 2. Boundary and loading conditions of scarf joint model

Fig. 3. Comparison of current simulation strength perdition with experimental results of Kechai et al. [36] for [0]4

Fig. 4. Comparison of current simulation results with results of Hayes et al. [16]

Fig. 5. Stress- Strain curve for pristine quasi-isotropic laminates

Fig. 6. Comparison of stress-strain curves of scarf joints of quasi-isotropic laminates obtained from Approach “A” model

Fig.7. Contour Plot of Cohesive Damage for adhesive/adherend interface of a quasi-isotropic scarf joint (Approach “A”)

Fig. 8. Stress-Strain curve of scarf joints obtained from Approach “B” model

Fig. 9. Contour plot of cohesive damage parameter for  45 / 45 / 0 / 902 s at a) 80% of failure load, b) 90% of failure load, c)

failure load d) Post failure (Approach “B”)

Fig. 10. Contour plot of cohesive damage parameter for  45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s at a) 80% of failure load, b) 90%

of failure load, c) failure load d)Post failure (Approach “B”)

Fig. 11. a) Delamination initiation at 55% of final failure load for  45 / 45 / 0 / 90 / 90 / 0 / 45 / 45s pristine laminate, b)

Cohesive damage state at 90% of final failure load for  45 / 45 / 0 / 90 / 90 / 0 / 45 / 45s scarf joint

Fig. 12. Schematic for cross section of a scarf repair a) without doubler, b) with two sided unsymmetric doublers

Fig. 13. Tensile strength of scarf repair cross section of  45 / 45 / 0 / 902 s stacking sequence for various Scarf-Doubler
configurations

21
 Table 1.

Stacking Sequence Tensile Strength (MPa)

0 / 45 / 90 / 45s 804

0 / 45 / 45 / 90s 693

45 / 0 / 45 / 90s 675

45 / 0 / 90 / 45s 665

45 / 90 / 45 / 0s 657

45 / 90 / 0 / 45s 615

Table 2.

Stacking Sequence Tensile Strength (MPa)

90 / 45 / 0 / 45s 486

45 / 0 / 45 / 90s 467

45 / 90 / 0 / 45s 476

90 / 0 / 45 / 45s 463

0 / 45 / 45 / 90s 432

45 / 45 / 0 / 90s 424

Table 3.

Pristine Laminate dimension 100  mm  25  mm 

Scarf joint model (length  width) 100  mm  0.13  mm
Number of plies 16
Scarf angles 5
Ply thickness 0.13  mm 
Adhesive thickness 0.2  mm 

22
 Table 4.

Property IM7/977-3
E1 (GPa) 168
E2  E3 (GPa) 9.86
G12  G13 (GPa) 4.95
G23 (GPa) 2.944
12  13 0.33
23 0.34

Density Tonne / mm 3
 1.57e9

Table 5.

Property FM 300 Adhesive

E  GPa  1.577
 0.3

Density Tonne / mm3  8.75e10

Table 6.

Failure Type Symbol (MPa)

Longitudinal Fiber Tensile Strength Xt 2825
Longitudinal Fiber Compressive Strength Xc 2275
Transverse Matrix Tensile Strength Yt 66
Transverse Matrix Compression Strength Yc 275
In-Plane Shear Strength S12 110
Out-of-Plane Shear Strength S23 130

Table 7.

G ft kJ / m 2  G fc kJ / m 2  Gmt kJ / m 2  Gmc kJ / m 2 

       
100 100 0.22 0.22

Table 8.

Interlaminar Fracture Property G  kJ / m 2  K  N / mm3  σ , τ  N / mm 2 

     

23
 Mode I 0.22 105 30
Mode II 0.9 5 60
10
Mode III 0.9 5 60
10
Damage Initiation and Evolution Specifications
Cohesive damage criterion Quadratic traction
Mixed Mode behavior Power law   1.5

Table 9.

Laminate Tensile Strength

Stacking Sequence
Number (MPa)
1 90 / 45 / 0 / 452s 932.8
2 45 / 0 / 90 / 452s 888.6
3 45 / 0 / 45 / 902s 866.1
4 90 / 45 / 45 / 02s 865.7
5 0 / 45 / 45 / 902s 862.2
6 45 / 45 / 0 / 902s 831.0
7 45 / 90 / 90 / 45 / 0 / 45 / 0 / 45s 803.8
8 45 / 45 / 0 / 90 / 90 / 0 / 45 / 45s 793.3
9 45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s 760.0

Table 10.

Nonlinear Abaqus
Repair strength/pristine

Repair strength/pristine
laminate (Explicit) %
Abaqus Explicit
laminate (Static) %

Static Explicit
Solver
Solver Solver
Scarf Joint Scarf Joint
Stacking Sequence Approach Approach Pristine laminate
“A” “B”
Tensile Tensile
Tensile Strength
Strength Strength
(MPa)
(MPa) (MPa)
45 / 45 / 0 / 902s 488.9 437.6 831.0 58.8 52.7
90 / 45 / 0 / 452s 447.9 418.8 932.8 48 44.9
45 / 45 / 0 / 90 / 90 / 0 / 45 / 45s 446.9 401.5 793.3 56.3 50.6
45 / 0 / 90 / 452s 436.7 404.2 888.6 49.1 45.5
45 / 0 / 45 / 902s 433.7 395.5 866.1 50.1 45.7
90 / 45 / 45 / 02s 327.6 300.6 865.7 37.8 34.7
45 / 90 / 90 / 45 / 0 / 45 / 0 / 45s 318.5 315.8 803.8 39.6 39.3

24
0 / 45 / 45 / 902s 283.3 294.8 862.2 32.9 34.2
45 / 0 / 0 / 45 / 90 / 45 / 90 / 45s 210.2 194.1 760.0 27.7 25.5

Table 11.

Scarf Bottom Tensile Strength

Scarf-Doubler Repair Configuration Top Doubler
Angle Doubler (MPa)
45 / 45 / 0 / 902s -5deg 5 N/A N/A 446.4
45 / 45 / 0 / 902s -8deg 8 N/A N/A 304.0
45 / 45 / 0 / 902s -8deg-1SD 8  45 / 45 N/A 339.7
45 / 45 / 0 / 902s -8deg-2SD-Symm 8  45 / 45  45 / 45 381.2
45 / 45 / 0 / 902s -8deg-2SD-Unsymm 8 45 / 45 / 45 / 45  45 / 45 399.7

Fig. 1.

25
 Fig. 2.

Kechachi-Test 1
Kechachi-Test 2
Kechachi-Test 3
Kechachi-Test 4
700 Current Simulation-MeshElmSize=0.25mm

600
Tenile Strength (MPa)

500

400

300

200

100

0
0.000 0.010 0.020 0.030 0.040 0.050
Strain(mm/mm)

Fig. 3.

26
 Hayes Experiment

Hayes-Standard Solver-1 Cohesive Intreface

Hayes-Explicit Solver-All Interfaces are cohesive+Composite

Damage+Adhesive Plasticity
Current Simulation -Approach "A"

Current Simulation-Approach "B"

600 563.3 567.5
Tensile Strength (MPa) 496.4 501.2 487.7
500

400

300

200

100

Fig. 4.

[90/45/0/-45]2s [45/0/90/-45]2s
[45/0/-45/90]2s [90/-45/45/0]2s
[0/45/-45/90]2s [-45/45/0/90]2s
[-45/90/90/45/0/45/0/-45]s [45/-45/0/90/90/0/-45/45]s

1000

900
Tensile Strength (MPa))

800

700

600

500

400

300

200

100

0
0 0.01 0.02 0.03 0.04
Strain (mm/mm)

Fig. 5.

27
 500 [-45/45/0/90]2s
[90/45/0/-45]2s
450 [45/-45/0/90/90/0/-45/45]s
[45/0/90/-45]2s
400 [45/0/-45/90]2s
[90/-45/45/0]2s
350
[-45/90/90/45/0/45/0/-45]s
Tensile Stress (MPa)

300 [0/45/-45/90]2s
[45/0/0/-45/90/-45/90/45]s
250

200

150

100

50

0
0 0.002 0.004 0.006 0.008 0.01
Longitudinal Strain (mm/mm)

Fig. 6.

Fig. 7.

28
 500 [-45/45/0/90]2s
[90/45/0/-45]2s
450 [45/0/90/-45]2s
[45/-45/0/90/90/0/-45/45]s
400 [45/0/-45/90]2s
[-45/90/90/45/0/45/0/-45]s
350 [90/-45/45/0]2s
[0/45/-45/90]2s
Tensile Stress (MPa)
300 [45/0/0/-45/90/-45/90/45]s
250

200

150

100

50

0
0 0.002 0.004 0.006 0.008 0.01
Strain (mm/mm)

Fig. 8.

(a)

(b)

29
 (c)

(d)

Fig. 9.

(a)

30
 (b)

(c)

(d)

Fig. 10.

31
(a)

(b)

Fig. 11.

(a)

(b)

Fig. 12.

32
 500 [-45/45/0/90]2s-5deg
[-45/45/0/90]2s-8deg
450 [-45/45/0/90]2s-8deg-1SD
[-45/45/0/90]2s-8deg-2SD-Symm
400 [-45/45/0/90]2s-8deg-2SD-Unsymm
350
Tensile Stress (MPa)
300

250

200

150

100

50

0
0 0.002 0.004 0.006 0.008 0.01 0.012
Strain (mm/mm)

Fig. 13.

33