Le collage structural se présente comme une alternative intéressante aux méthodes classiques d'assemblages par ajout d’éléments mécaniques pour alléger les structures aéronautiques. Cependant, l'utilisation de cette méthode soulève de nombreuses questions en termes de conception, caractérisation ou modélisation. Ce travail, à fort caractère expérimental, visent deux grands objectifs : (i) sélectionner les moyens d'instrumentation les plus adaptés pour la caractérisation du comportement jusqu'à rupture des assemblages collés et (ii) prédire le comportement des assemblages collés à partir d'une caractérisation et d'une modélisation complète sous chargements quasi-statiques bi-axiaux, en utilisant un type d’essai simple et industrialisable. Dans un premier temps, pour atteindre ces objectifs, un nouvel essai appelé Scarf modifié a été proposé, afin de caractériser le comportement mécanique du joint de colle, en minimisant les effets de bord par la présence de becs. La deuxième partie de la thèse porte sur le développement d'une stratégie de caractérisation simplifiée du comportement de joint de colle, sous certaines hypothèses (états de contraintes/déformations, continuité du champ de contraintes). Cette stratégie a permis d'obtenir le comportement intrinsèque de l’adhésif sous la forme de courbes contraintes/déformations. La dernière partie abordée dans ces travaux, s'intéresse quant à elle à l'identification et la sélection des moyens de métrologie les plus appropriés, pour accompagner la caractérisation expérimentale. / Today, structural bonding presents an interesting alternative to conventional methods of assembly, in order to reduce the weight within aeronautical structures. However, the use of this method raises many questions in terms of design, characterization or modeling. This work presents a robust experimental work, aiming two main objectives: (i) select the most suitable instrumentation for the characterization of the behavior of bonded joints up to failure and (ii) predict the behavior of bonded joints from complete characterization and modeling under quasi-static bi-axial loading, using a simple test, directly integrated to industry. In a first stage, to achieve this goal, modified Scarf test has been proposed to characterize the mechanical behavior of adhesive joints. This type of specimen allows applying multiaxial loadings without having high-stress concentrations near the edges. The second part of the thesis deals with the development of a strategy for simplifying the characterization of a bonded joint using some hypotheses (stress/strain states). This strategy allows obtaining the intrinsic behavior of the adhesive in the form of the stress/strain curves. The last part of this work deals with the identification and selection of the most appropriate instrumentation systems in the experimental characterization.

Adhesively bonded joints have been investigated for several decades. In most analytical studies, the Bernoulli-Euler beam theory is employed to describe the behaviour of adherends. In the current work, three analytical models are developed for adhesively bonded joints using the Timoshenko beam theory for elastic material and a Bernoulli-Euler beam model for viscoelastic materials. One model is for the peeling test of an adhesively bonded joint, which is described using a Timoshenko beam on an elastic foundation. The adherend is considered as a Timoshenko beam, while the adhesive is taken to be a linearly elastic foundation. Three cases are considered: (1) only the normal stress is acting (mode I); (2) only the transverse shear stress is present (mode II); and (3) the normal and shear stresses co-exist (mode III) in the adhesive. The governing equations are derived in terms of the displacement and rotational angle of the adherend in each case. Analytical solutions are obtained for the displacements, rotational angle, and stresses. Numerical results are presented to show the trends of the displacements and rotational angle changing with geometrical and loading conditions. In the second model, the peeling test of an adhesively bonded joint is represented using a viscoelastic Bernoulli-Euler beam on an elastic foundation. The adherend is considered as a viscoelastic Bernoulli-Euler beam, while the adhesive is taken to be a linearly elastic foundation. Two cases under different stress history are considered: (1) only the normal stress is acting (mode I); and (2) only the transverse shear stress is present (mode II). The governing equations are derived in terms of the displacements. Analytical solutions are obtained for the displacements. The numerical results show that the deflection increases as time and temperature increase. The third model is developed using a symmetric composite adhesively bonded joint. The constitutive and kinematic relations of the adherends are derived based on the Timoshenko beam theory, and the governing equations are obtained for the normal and shear stresses in the adhesive layer. The numerical results are presented to reveal the normal and shear stresses in the adhesive.

The effects of high loading rates (HLR), and nano reinforcement on the mechanical response of adhesively-bonded SLJs with composite adherends, subjected to different loading (strain) rates are systematically investigated. The results are then compared to those of neat thermoset resin and thermo-plastic adhesive. More specifically, nano-reinforced and neat resin bonded joints mating carbon/epoxy and glass/epoxy adherends were subjected to tensile loadings under 1.5 and 3 mm/min and tensile impacts at a loading rate of 2.04E+5 mm/min. In some cases, additional tests were conducted under 15, 150, and 1500 mm/min to obtain additional properties gained using the nano-reinforcements for use in the further numerical investigations. The HLR tests were conducted, using a modified instrumented pendulum equipped with a specially designed impact load transfer apparatus. The dispersion of nanoparticles was facilitated using a mechanical stirrer and a three-roll mill machine. The failure mechanisms were studied with a scanning electron microscope.

Using experimental data obtained from standard fracture test configurations, theoretical and numerical tools are developed to mathematically describe non-self-similar progression of cracks without specifying an initial crack. A cohesive-decohesive zone model, similar to the cohesive zone model known in the fracture mechanics literature as the Dugdale-Barenblatt model, is adopted to represent the degradation of the material ahead of the crack tip. This model unifies strength-based crack initiation and fracture-mechanics-based crack progression. The cohesive-decohesive zone model is implemented with an interfacial surface material that consists of an upper and a lower surface that are connected by a continuous distribution of normal and tangential nonlinear elastic springs that act to resist either Mode I opening, Mode II sliding, Mode III sliding, or a mixed mode. The initiation of fracture is determined by the interfacial strength and the progression of the crack is determined by the critical energy release rate. The adhesive is idealized with an interfacial surface material to predict interfacial fracture. The interfacial surface material is positioned within the bulk material to predict discrete cohesive cracks. The interfacial surface material is implemented through an interface element, which is incorporated in ABAQUS using the user defined element (UEL) option. A procedure is established to formulate a rate dependent model based on experiments carried out on compact tension test specimens. The rate dependent model is incorporated into the interface element approach to capture the unstable crack growth observed in experiments under quasi-static loading conditions. The compact tension test gives the variation of the fracture toughness with the rate of loading, this information is processed and a relationship between the fracture toughness and the rate of the opening displacement is established. The cohesive-decohesive zone model is implemented through a material model to be used in an explicit code (LS-DYNA). Dynamic simulations of the standard test configurations for Mode I (Double Cantilever Beam) and Mode II (End Load Split) are carried out using the explicit code. Verification of these coupon tests leads to the crash analysis of realistic structures like the square composite tube. Analyses of bonded and unbonded square tubes are presented. These tubes shows a very uncharacteristic failure mode: the composite material disintegrates on impact, and this has been captured in the analysis. Disadvantages of the interface element approach are well documented in the literature. An alternative method, known as the Extended Finite Element Method (XFEM), is implemented here through an eight-noded quadrilateral plane strain element. The method, based on the partition-of-unity, is used to study simple test configuration like the three-point bend problem and a double cantilever beam. Functionally graded materials are also simulated and the results are compared to the experimental results available in the literature. / Ph. D.

PR-2930 was developed by PPG Industries, Inc. to meet the challenging performance requirements of MIL-PRF-32662 Group-I-classified adhesives. PR-2930 is a high-strength, high-toughness, epoxy-based adhesive intended for automotive and aerospace applications. As PR-2930 functions as a structural adhesive, quantification of its mechanical properties and limit-states is a necessary task for designing joints bonded with the adhesive. The combination of both strength and ductility results in material non-linearities, making experimental characterization and numerical analyses more challenging. This work explores the quantification of fracture energy for PR-2930 bonded joints. Fracture can occur in one of three different modes, or in some combination. Many practical adhesive joints fail in the mixed-mode region involving both opening (mode I) and shearing (mode II) displacements. Mode I fracture was evaluated with double cantilever beam (DCB) tests, mode II fracture was characterized by end-notched flexure (ENF) tests, and varying degrees of mixed mode I/II fracture were assessed through single leg bend (SLB), single-lap joint (SLJ), and asymmetric DCB and SLB tests. Test specimens were fabricated by bonding Al 2024-T3 adherends, ranging from 1.6 mm to 25.4 mm thick, with a 0.25 mm thick PR-2930 adhesive layer. Digital image correlation (DIC) was used to experimentally measure local displacements and surface strains on the adherends. Standard data-reduction methods often used to determine fracture energies of bonded joint specimens were used to numerically analyze test results. These methods included the Corrected Beam Theory (CBT), the Compliance-Based Beam Method (CBBM), and the Paris and Paris J-Integral approach. Linear elastic fracture mechanics (LEFM) conditions must be valid to correctly apply these methods, however plastic deformations were observed in some adherends. Drawbacks of these approaches and their validity for analyzing PR-2930 joints were discussed. To account for non-linearities, more advanced numerical analysis was performed using finite element analysis (FEA) with cohesive zone models (CZMs) to model the adhesive layer. CZM parameters such as fracture energies and traction separation law (TSL) shapes were determined from experimental data and published literature. Results from CZMs were compared to experimental load, displacement, and strain data. Recommended TSLs for mode I and mode II fracture were formed in this work as well as a mixed-mode relationship using a Benzeggagh-Kenane damage evolution law. More ideal analytical methods were suggested to simplify analysis of joints using the same or similar material compositions. / M.S. / Structural adhesives are used to safely transmit loads in our furniture, automobiles, aircraft, and buildings. PR-2930 is a newly developed epoxy that exhibits top-of-the-line strength and ductility. To safely design joints utilizing PR-2930, the bonding material and its limit states must be defined. The most pertinent mechanical limit state for adhesively bonded joints is its resistance to fracture, also known as fracture toughness. Fracture often occurs due to a combination of opening (mode I) or shearing (mode II) displacements. In this work, standard and novel advanced fracture characterization techniques are employed and subsequently compared. Adhesive joints using a 0.25 mm layer thickness are bonded to Al 2024-T3 adherends varying from 1.6 mm to 25.4 mm of thickness and tested in quasistatic conditions. Mathematical models of mode I, mode II, and combined mode I/II stress displacement responses (AKA a traction-separation laws) of PR-2930 are developed and compared with experimental data. Future experimental and numerical methods for fracture analysis of structural adhesives are discussed.

No description available.

This dissertation is to obtain an overall understanding of the crack path selection in adhesively bonded joints. Using Dow Chemical epoxy resin DER 331® with various levels of rubber concentration as an adhesive, and aluminum 6061-T6 alloy with different surface pretreatments as the adherends, both symmetric and asymmetric double cantilever beam (DCB) specimens are prepared and tested under mixed mode fracture conditions in this study. Post-failure analyses conducted on the failure surfaces indicate that the failure tends to be more interfacial as the mode II component in the fracture increases whereas more advanced surface preparation techniques can prevent failure at the interface. Through mechanically stretching the DCB specimens uniaxially until the adherends are plastically deformed, various levels of T-stress are achieved in the specimens. Test results of the specimens with various T-stresses demonstrate that the directional stability of cracks in adhesive bonds depends on the T-stress level. Cracks tend to be directionally stable when the T-stress is compressive whereas directionally unstable when the T-stress is tensile. However, the direction of crack propagation is mostly stabilized when more than 3% mode II fracture component is present in the loading regardless of the T-stress levels in the specimens. Since the fracture sequences in adhesive bonds are closely related to the energy balance in the system, an energy balance model is developed to predict the directional stability of cracks and the results are consistent with the experimental observations. Using the finite element method, the T-stress is shown to be closely related to the specimen geometry, indicating a specimen geometry dependence of the directional stability of cracks. This prediction is verified through testing DCB specimens with various adherend and adhesives thicknesses. By testing the specimens under both quasi-static and low-speed impact conditions, and using a high-speed camera to monitor the fracture sequence, the influences of the debond rate on the locus of failure and the directional stability of cracks are investigated. Post-failure analyses suggest that the failure tends to be more interfacial when the debond rate is low and tends to be more cohesive when the debond rate is high. However, this rate dependence of the locus of failure is greatly reduced when more advanced surface preparation techniques are used in preparing the specimens. The post-failure analyses also reveal that cracks tend to be more directionally unstable as the debond rate increases. Finally, employing interface mechanics and extending the criteria for the direction of crack propagation to adhesively bonded joints, the crack trajectories for directionally unstable cracks are predicted and the results are consistent with the overall features of the crack paths observed experimentally. / Ph. D.

Large scale structures such as those in aerospace flight vehicles are made in parts and assembled. Joints are inevitable in these systems and they are potential threats to the structural integrity of the flight vehicles. Fastener and bonded joints are the most commonly used methods of joining in these structures. Among these, adhesive bonding has become more popular with the advent of composite structures, due to the presence of less number of points of stress concentration and the resulting benefit for static strength and fatigue life. In modern aircraft in which maximum percentage of composite materials are being employed due to several benefits, designers are contemplating to replace discrete joints with adhesively bonded joints wherever possible. A detailed literature survey shows that the field of adhesively bonded joints has been extensively studied in the past. Initial publications appeared in late 1950’s and early 1960's, but many of the initial attempts were based on one dimensional (1-D) approximation of the adherents due to lack of computing power. With the current day emphasis on safety and damage tolerance, there is a definite need to study these joints with 2-D and 3-D idealization. In spite of valuable contributions in the literature from several researchers in past 4-5 decades, one finds that there are gaps to be filled, in particular, with reference to static strength prediction and de-bond growth to failure under fatigue loading. This thesis is intended as a modest contribution in this direction covering the methods of strength prediction and also correlations between de-bond growth and fracture parameters. Most commonly used bonded joints are single lap joints. The primary issue in their analysis is the geometric nonlinearity resulting in large deformations due to eccentricity of load path between the adherents. Further, adhesives have very low yield strength and plastic deformation in thin adhesives could affect the mechanics of load transfer. The current work is initiated by carrying out geometric and material nonlinear analysis of adhesively bonded single lap joints between metal-metal (aluminum-aluminum) adherents using standard NASTRAN finite element software. Modified Newton-Raphson iterative technique has been used to economize the computer time and also achieve fast convergence. A convergence study has been conducted to determine the order of mesh size required. Preliminary results are obtained on configurations analysed by earlier workers and the current results are compared with their results. Later, extensive experimental and numerical studies have been taken up on the numerical strength prediction of these joints correlating them to the experimental values. Cohesive failure along the centre line of the adhesive is assumed under both static and fatigue loading. The bonded joints are studied with both 2-D plane stress and plane strain nonlinear FE analysis. The issue in this type of analysis is the presence of theoretical elastic singularity at the ends of the lap length. The normally used maximum stress criterion can not be used in such circumstances. There were attempts in the past to use point stress or average stress criteria for this purpose. In point stress criterion the shear stress (or von-Mises stress) is picked at a characteristic distance away from the ends of the lap length and compared with the corresponding strength value to predict failure. In the average stress criterion the stresses are averaged over a characteristic distance from the ends of lap length and this is compared with the corresponding strength to predict failure. Determination of the characteristic distance in both the cases needs extensive experimental results on static strength of joints. The static strength data is to be correlated with numerical results to determine the characteristic distance in various specimens. In the current thesis a series of specimens with aluminum-aluminum, aluminum-CFRP composite and CFRP-CFRP composite adherents were tested to determine the static strength. In all the specimens the adhesive used was Redux 319 A. These experimental strength data was used to determine characteristic distance using point stress criterion. The consistency of estimates of the characteristic distances in all the specimens shows that the approach is capable of predicting the static strength. The above approaches are capable of predicting the strength of joints with linear material and nonlinear geometric analysis. But when the adhesive yield strength is low, a novel approach is required to predict the static strength. Numerical analysis is conducted using a combined material and geometric nonlinear analysis in NASTRAN software. The plastic zone size from the ends of the lap length is determined at different load levels. Combining the numerical results with experimental failure load data, a failure criterion based on plastic zone size (PZS) is proposed in this thesis and validated. It has been observed that the validation is with limited testing carried out and further experimental programs are required to complete the validation. To the best of the knowledge of the author PZS criterion is used for the first time for failure prediction of bonded joints. The structural integrity of the joints also requires a study of de-bond growth and damage tolerance assurances in the presence of de-bond type of defects. The first step in this direction is to estimate the fracture parameters at the tips of de-bond in the adhesive of lap joints between various adherents. Modified virtual crack closure integral (MVCCI) technique has been developed in the past for estimating Strain Energy Release Rates (SERR) in several crack problems. Large contributions for developing this technique have come from the group where the author has worked. This technique is simple and has the ability to estimate individual SERR components GI and GII in cases of mixed mode fracture. It is seen clearly that the de-bond growth in bonded joint is one of mixed mode. The mode-II component is because of shear stresses transferring the load across the adherents and mode-I component is due to peel stresses developed during the deformation. The mode I SERR component is primarily responsible for de-bond growth and the effect of mode II component on de-bond growth is insignificant. The mesh details for accurately estimating the SERR components are evaluated and those meshes are used to estimate these values for the cases of aluminum-aluminum, aluminum-CFRP composite and composite-composite joints. Obviously, when the adherents are dissimilar, mode I SERR components are the highest and assist faster de-bond growth. Painstaking fatigue de-bond growth experiments were conducted and de-bond growth rate with number of cycles of fatigue loading was determined. MVCCI method is used to estimate SERR components at maximum load and zero load in the fatigue cycle, to determine the SERR range in the fatigue cycle. Since the stress ratio, R of the loading cycle is -1, the minimum load for estimating SERR components is taken as zero. From the experimental data and numerical estimates, a Paris type of equation was developed for the de-bond growth. The thesis concludes with a summary of the achievements in the current work with respect to the structural integrity of adhesively bonded joints and also with suggestions for future work.

Στην παρούσα διπλωματική εργασία αναπτύχθηκε τρισδιάστατο αριθμητικό μοντέλο με βάση την μέθοδο των πεπερασμένων στοιχείων για την προσομοίωση της μηχανικής συμπεριφοράς συνδέσεων με κόλλα πολύστρωτων πλακών. Το μοντέλο αναπτύχθηκε χρησιμοποιώντας το εμπορικό πακέτο πεπερασμένων στοιχείων ANSYS. Για την προσομοίωση της συμπεριφοράς της κόλλας χρησιμοποιήθηκαν δύο προσεγγίσεις: η μοντελοποίηση της ζώνης συνοχής και η μοντελοποίηση της βλάβης του συνεχούς μέσου. Οι δύο αυτές προσεγγίσεις συγκρίθηκαν τόσο ως προς την αξιοπιστία τους, η οποία καθορίζεται από την σύγκριση με πειραματικά αποτελέσματα, όσο και ως προς την ευκολία εφαρμογής τους, η οποία καθορίζεται από τα δεδομένα που απαιτούνται και τον υπολογιστικό χρόνο. Η σύγκριση των δύο μεθοδολογιών έγινε στην βάση της εφαρμογής τους για την προσομοίωση της μηχανικής συμπεριφοράς σε φόρτιση τύπου Ι σύνδεσης με κόλλα μεταξύ δύο ψευδοισότροπων CFRP πολύστρωτων πλακών. Το συγκεκριμένο πρόβλημα επελέγη διότι υπήρχαν διαθέσιμα πειραματικά αποτελέσματα προς σύγκριση στο Εργαστήριο. Οι πολύστρωτες πλάκες μοντελοποιήθηκαν χρησιμοποιώντας το στρωματικό στοιχείο του ANSYS SOLID185. Στο στοιχείο αυτό κάθε στρώση μοντελοποιείται ξεχωριστά ως ορθότροπο υλικό. Η εφαρμογή της μοντελοποίησης της ζώνης συνοχής έγινε μέσω της χρήσης του στοιχείου του ANSYS INTER205. Για την εφαρμογή της μοντελοποίησης της βλάβης του συνεχούς μέσου αναπτύχθηκε μακρο-ρουτίνα χρησιμοποιώντας την γλώσσα προγραμματισμού του κώδικα ANSYS. Τα αριθμητικά αποτελέσματα έδειξαν ότι και οι δύο μεθοδολογίες προσομοιώνουν με ικανοποιητική ακρίβεια την καμπύλη δύναμης-μετατόπισης της σύνδεσης. Σχετικά με την ευκολία εφαρμογής των δύο μεθόδων, η σύγκριση έδειξε ότι η μέθοδος της μοντελοποίησης της ζώνης συνοχής υπερτερεί έναντι της μεθόδου μοντελοποίησης της βλάβης του συνεχούς μέσου διότι απαιτεί μικρότερο αριθμό δεδομένων, μειονεκτεί όμως ως προς τον απαιτούμενο υπολογιστικό χρόνο. Και οι δύο μέθοδοι κρίνονται κατάλληλες για χρήση στην αριθμητική σχεδίαση συνδέσεων με κόλλα. / This work is based on the development of three-dimensional numerical model based on the finite element method to simulate the mechanical behavior of adhesive bonded joints in composite materials. The model was developed in finite element procedures under the framework of the commercial software ANSYS. To simulate the behavior of the adhesive used two approaches: the cohesive zone modeling (CZM) and the continuum damage modeling (CDM). These two approaches are compared both in terms of reliability, which is determined by comparison with experimental results, and applicability, which is determined by the parameters required and the computational time. The comparison between the two methodologies was the basis of their application to simulate the mechanical behavior under mode-I fracture behavior of adhesively bonded joints between two CFRP plates. This problem was chosen because there were experimental results to compare in the laboratory. The sandwich plates are modeled using the stromal element of ANSYS SOLID185. This item each layer separately modeled as orthotropic material. The adhesive is modeled using the interface element of ANSYS INTER205. For the purpose of modeling the failure of continuous medium developed macro routine using the programming language code ANSYS. The numerical results showed that both methodologies simulate sufficient precision the curve force-displacement of the connection. About applicability of the two methods, the comparison showed that the process of cohesive zone modeling outweighs the process of continuum damage modeling because it requires less number of parameters, but falls to the computational time. Both methods are suitable for use in numerical design of adhesively bonded joints.

Structural adhesives are widely used with great success, and yet occasional failures can occur, often resulting from improper bonding procedures or joint design, overload or other detrimental service situations, or in response to a variety of environmental challenges. In these situations, cracks can start within the adhesive layer or debonds can initiate near an interface. The paths taken by propagating cracks can affect the resistance to failure and the subsequent service lives of the bonded structures. The behavior of propagating cracks in adhesive joints remains of interest, including when some critical environments, complicated loading modes, or uncertainties in material/interfacial properties are involved. From a mechanics perspective, areas of current interest include understanding the growth of damage and cracks, loading rate dependency of crack propagation, and the effect of mixed mode fracture loading scenarios on crack path selection. This dissertation involves analytical, numerical, and experimental evaluations of crack propagation in several adhesive joint configurations. The main objective is an investigation of crack path selection in adhesively bonded joints, focusing on in-plane fracture behavior (mode I, mode II, and their combination) of bonded joints with uniform bonding, and those with locally weakened interfaces. When removing cured components from molds, interfacial debonds can sometimes initiate and propagate along both mold surfaces, resulting in the molded product partially bridging between the two molds and potentially being damaged or torn. Debonds from both adherends can sometimes occur in weak adhesive bonds as well, potentially altering the apparent fracture behavior. To avoid or control these multiple interfacial debonding, more understanding of these processes is required. An analytical model of 2D parallel bridging was developed and the interactions of interfacial debonds were investigated using Euler-Bernoulli beam theory. The numerical solutions to the analytical results described the propagation processes with multiple debonds, and demonstrated some common phenomena in several different joints corresponding to double cantilever beam configurations. The analytical approach and results obtained could prove useful in extensions to understanding and controlling debonding in such situations and optimization of loading scenarios. Numerical capabilities for predicting crack propagation, confirmed by experimental results, were initially evaluated for crack behavior in monolithic materials, which is also of interest in engineering design. Several test cases were devised for modified forms of monolithic compact tension specimens (CT) were developed. An asymmetric variant of the CT configuration, in which the initial crack was shifted to two thirds of the total height, was tested experimentally and numerically simulated in ABAQUS®, with good agreement. Similar studies of elongated CT specimens with different specimen lengths also revealed good agreement, using the same material properties and cohesive zone model (CZM) parameters. The critical specimen length when the crack propagation pattern abruptly switches was experimentally measured and accurately predicted, building confidence in the subsequent studies where the numerical method was applied to bonded joints. In adhesively bonded joints, crack propagation and joint failure can potentially result from or involve interactions of a growing crack with a partially weakened interface, so numerical simulations were initiated to investigate such scenarios using ABAQUS®. Two different cohesive zone models (CZMs) are applied in these simulations: cohesive elements for strong and weak interfaces, and the extended finite element method (XFEM) for cracks propagating within the adhesive layer. When the main crack approaches a locally weakened interface, interfacial damage can occur, allowing for additional interfacial compliance and inducing shear stresses within the adhesive layer that direct the growing crack toward the weak interface. The maximum traction of the interfacial CZM appears to be the controlling parameter. Fracture energy of the weakened interface is shown to be of secondary importance, though can affect the results when particularly small (e.g. 1% that of the bulk adhesive). The length of the weakened interface also has some influence on the crack path. Under globally mixed mode loadings, the competition between the loading and the weakened interface affects the shear stress distribution and thus changes the crack path. Mixed mode loading in the opposite direction of the weakened interface is able to drive the crack away from the weakened interface, suggesting potential means to avoid failure within these regions or to design joints that fail in a particular manner. In addition to the analytical and numerical studies of crack path selection in adhesively bonded joints, experimental investigations are also performed. A dual actuator load frame (DALF) is used to test beam-like bonded joints in various mode mixity angles. Constant mode mixity angle tracking, as well as other versatile loading functions, are developed in LabVIEW® for use with a new controller system. The DALF is calibrated to minimize errors when calculating the compliance of beam-like bonded joints. After the corrections, the resulting fracture energies ( ) values are considered to be more accurate in representing the energy released in the crack propagation processes. Double cantilever beam (DCB) bonded joints consisting of 6061-T6 aluminum adherends bonded with commercial epoxy adhesives (J-B Weld, or LORD 320/322) are tested on the DALF. Profiles of the values for different constant mode mixity angles, as well as for continuously increasing mode mixity angle, are plotted to illustrate the behavior of the crack in these bonded joints. Finally, crack path selection in DCB specimens with one of the bonding surfaces weakened was studied experimentally, and rate-dependency of the crack path selection was found. Several contamination schemes are attempted, involving of graphite flakes, silicone tapes, or silane treatments on the aluminum oxide interfaces. In all these cases, tests involving more rapid crack propagation resulted in interfacial failures at the weakened areas, while slower tests showed cohesive failure throughout. One possible explanation of this phenomenon is presented using the rate-dependency of the yield stress (commonly considered to be corresponding to the maximum traction) of the epoxy adhesives. These experimental observations may have some potential applications tailoring adhesive joint configurations and interface variability to achieve or avoid particular failure modes. / Ph. D.