This dissertation has been submitted by a student. This is not an example of the work written by our professional dissertation writers.
CHAPTER 1
INTRODUCTION
1.1 Introduction
The use of sandwich structures has been increasing in recent years because of their lightweight and high stiffness. Commonly, the naval industry and transportation uses the Eglass fibers while the aerospace industry uses composite structures such as carbon fiber. The use of sandwich panels with composite facesheet in the naval industry is particularly appealing because they have better corrosion and environmental resistance and reduced magnetic signatures when compared to doublehull construction steel ships. On the other hand, composite sandwich panels are easily susceptible to damage by a strange object impact. The damage may be visible, penetration or perforation, or invisible, internal delamination and debonding. Both types of damages will result in stiffness and strength reduction. It is then important to study the impact behavior of composite sandwich panels.
Failure in composite structures can be caused by low, high and extremely high or localized impact. An impact caused by a foreign body initiates two waves from impact point in a panel: a throughthickness wave and a transverse shear wave. Whether or not these waves play an important role in the impact response of the panel depends on the actual contact duration between the projectile and panel and the time it takes the transverse shear wave to reach the panel boundary. Figures 1.1 (a)(c) show threeimpact scenarios: lowvelocity, highvelocity and ballistic impact. In lowvelocity impact, the contact force duration is long compared to the time it takes the transverse shear wave travel to reach the plate boundary. Many waves reflect back and forth across the side dimension of the panel. In highvelocity impact, the contact force duration is much shorter than the transverse shear wave travel time through the panel. Usually highvelocity impact is the same with perforation and localized damage of the panel. Ballistic impact deals only with throughthickness wave propagation. During ballistic impact, there is complete perforation of the panel with little or no panel deformation. The contact force duration is approximately the wave travel time through the panel thickness. Ballistic impact usually involves the study of penetration mechanics.
 Lowvelocity
 Highvelocity
 Ballistic Impact
The projectile to panel mass ratio will control whether wave propagation effect dominates the panel impact response and then suggested that a mass ratio be use as a parameter to determine impact response. It was shown that small mass impacts produce more damage than highmass impacts having same kinetic energy. While smallmass impacts were defined by wavecontrolled response, large mass impacts were defined by boundarycontrolled response.
Common examples of lowvelocity impact are of bird strikes, collision with floating object, and dropped tools, may cause damage. Underwater blast or debris from a faraway explosion and air was considered as a highvelocity impact situation. Examples of ballistic impact would be a bullet or fragments from a nearby explosion hitting the panel.
Another important factor governing the impact on composite structures is the ballistic limit. The ballistic limit is defined as the highest velocity of the projectile to cause perforation. When the residual velocity (exit) of the projectile is zero, then the initial velocity of the projectile that causes perforation is the ballistic limit of the sandwich panel. The ballistic limit may be calculate analytically or determined experimentally. In the experimental method, sandwich panels are shoot with projectiles over narrow range of velocities to either just cause penetration or to just perforate the panel. There exists a striking velocity at which 50% of the panels are completely perforate above this value and remaining 50% are partly penetrate below this value. This striking velocity is expresse as V50, which is the ballistic limit of the panel. In the analytical approach, the ballistic limit is determined by the conservation of energy principle. The approach is complex because it includes a variety of factors like core thickness, facesheet thickness, shape of the projectile, core crushing stress, and so on.
1.2 Problem Statement
This topic was an expansion of the Wan Awis research. He has done only an experimental work. For impact application, we need to predict skin and core material thickness. Since impact phenomena depend on numerous parameters such as material properties or projectile geometry, a numerical model, validated experimentally, is necessary to allow the study of the influence of several parameters without making costly experimental tests. This will definitely enhance the development of our military technology and achievements in the future because of the ability of this software to cut production cost and time consuming of the experimental work. The numerical figures have been compared to modal test results aiming mainly to validate the studies. Simulation based on finite element analysis (FEA) must not exceed ± 15% error or this simulation could be claimed not acceptable.
1.3 Objective
 To simulate the damage of composite sandwich structures subjected to highvelocity impact using finite element analysis.
 To determine the energy absorption capability of the components on the behavior of the sandwich panel under impact load using ANSYS AUTODYN 13.0
 To validate a numerical model with actual experiment.
1.4 Scope of Works
 To characterize a mechanical behavior of carbon fiber panel by using tensile and determine the fiber volume force and density.
 Design and validate the numerical model.
 Conduct a ballistic impact test simulation.
 Using the experiments data to calculate the energy absorption on the impact.
CHAPTER 2
LITERATURE REVIEW
2.1Introduction
A great deal of research has been conducted in the area of impact of composite structures. In this chapter, previous work done on the impact response of laminated composite plates and composite sandwich panels will be reviewed.
2.2Impact of Composite Laminates
A detail study of impact of composite laminates in the three impact regimes ballistic impact, lowvelocity and highvelocity is presented in this section.
2.2.1Lowvelocity Impact
Abrate, 1998 give a specific review on different analytical models of impact on composite laminates. He classified impact models into four groups: impact on infinite plate model, energy balance models, springmass models, and complete models. In the energy balance model, the initial kinetic energy of the projectile is used to calculate the deformation of the composite laminate. The velocity of the projectile reaches zero at the maximum deflection of the composite laminate. At this point, all of the kinetic energy of the projectile is converted to strain energy needed to deform the composite laminate. Energy balance model assumes that the structure behaves in quasistatic manner. The time history of force and deflection are obtained using the springmass model representing the composite laminate. The model shown in Figure 2.1 consists of nonlinear contact stiffness (K), one spring representing the linear stiffness of the structure (Kbs), another spring for the nonlinear membrane stiffness (Km), effective mass of the structure (M2) as well as the mass of the projectile (M1). Equations of motion are written from a free body diagram. The infinite plate model is used when the deformation wavefront has not reached the boundary but if the wave reaches the plate boundary then this model is not an appropriate one to use. In the complete model, the dynamics of the structure and projectile are taken into explanation. Appropriate plate theory has to be selected and used. In many cases the classical plate theory can be used but when transverse shear deformations become significant, higherorder theories must be used.
One of the earliest studies on the impact of composite laminates was by Goldsmith et al, 1995, who conducted highvelocity and quasistatic impact tests on carbonfiber laminates by using a cylindroconical projectile. Three different specimen of varying thickness were considered. Energy balance principle was used to predict the dynamic penetration energy, static penetration energy, and also the ballistic limit of the composite laminate. The fiber failure accounted for most of the energy absorbed. The predicted theoretical energy was in good agreement with measured energy for thin laminates but not for the thick laminates. This was approved to the fact that transverse shear deformation played an important responsibility in thick laminates subjected to lowvelocity impact. The effect of transverse shear deformation was not dominant due to its quick occurrence in the highvelocity impact of laminates. Therefore, the predicted energy in the dynamic case was always close to but less than the measured energy for the thin and thick laminates. The predicted ballistic limit was less than measured values due to the nonlinear factors.
Cantwell, 2007 studied the influence of target geometry in the lowvelocity impact of composite laminate. The tests were performed on GFRP plates with hemispherical indenter on either circular or square supports. He used energybalance model to predict the plate deflection and the delamination area of the laminated structure. His study stated that there is little or no influence of target geometry on the failure modes. It also suggested that delamination was dependent on interlaminar shear stress and increasing the plate diameter required more energy for damage initiation.
Hou et al., 2000 predicted impact damage in composite laminates using LSDYNA 3D. The numerical results were compared to experimental results on lowvelocity impact on composite laminate with an initial velocity of 7.08 m/s The ChangChang failure criteria was modified taking the shear stress into consideration and the model was implemented in DYNA 3D.
2.2.2 Highvelocity Impact
In 1988, Cantwell performed highvelocity impact tests of CFRP laminates with 6 mm diameter, 1g steel ball as the projectile. The influence of fiber stacking sequence and target geometry was study. The experiments reveal that varying the target geometry had no significance on initial damage caused. While the damage initiated in the distal facesheet in thin laminates, however, in thick laminates it initiated from incident facesheet.
Zhao et al., 2007 investigated the failure modes in composite laminates subjected to highvelocity impact. Three different laminates were subject to impact by hemispherical projectile in the range of 10300 m/s. An energy balance was considered and equations for residual velocity for the laminates were given in terms of the mass of the projectile and striking velocity. The thickness and stacking sequence were finding to play an important role in the energy absorption.
Cheng et al., 2007 developed an analytical model based on the springmass model for highvelocity impact of a blunt ended and a sharpended projectile on thick composite laminates. They considered the effect of moving boundary due to the propagation of shear wave. The analysis was modeled using series of quasistatic events. At the end of each quasistatic step, the failed layers were remove based on punch shear damage and fiber damage criteria, and the wave front was moved outwards. While the first spring stiffness constant was measure based on the penetration depth of the projectile, the second spring stiffness constant was measured based on the bottom node of the plate.
2.2.3 Ballistic Impact
Silva et al., 2005 performed numerical simulations of ballistic impact on thin Kevlar 29 composite laminates using a fragmentsimulating projectile. The laminate material model was simulating using AUTODYN and the projectile was modeled using JohnsonCook strength model. Finite element mesh for both laminate and projectile was generating using True Grid. Accurate predictions of ballistic limit (V50) and the failure modes were made. Ballistic limit is the minimum velocity of impact at which a given projectile just perforates a given target. On occasion, the term is also used to identify the maximum impact velocity at which the projectile can penetrate into the target with perforation. It is often defined statistically as the impact velocity for which the projectile has a 50% probability of perforating the target; it is then denoted by V50.
Guild et al., 2007 conducted numerical simulations of ballistic impact on composite laminates and compared them with experimental results. The laminates were made of Eglass/vinyl ester resin with varying thickness and ball bearings of varying mass were use as projectiles. The damage modes included fiber failure, matrix failure, penetration, and delamination. Hashin failure criteria was use to determine the damage mode. Delamination was modeled using an interface between the two plies. As the force increased between two nodes above the specified value, the nodes were untied and the delamination increased. The ballistic limit from experiments was in good agreement with numerical results
Naik et al., 2008 studied the ballistic impact behavior of thick composites. Eglass/epoxy laminates of varying thickness were subject to highvelocity impact. The effects of projectile diameter, projectile mass and laminate thickness on the ballistic limit were studied. Wave theory and an energy balance were use to predict the ballistic limit of the laminate. The contact duration of the projectile with the laminate was maximum when the initial velocity was equal to ballistic limit and decreased when the initial velocity increased beyond the ballistic limit.
Deka et al., 2008 conducted ballisitic impact on Eglass/polypropylene composite laminates with cylindershaped projectiles. The experimental results were validating with numerical analysis using LSDYNA. Although the laminate was modeling in Hypermesh, LSDYNA was used to analyze failure mechanisms. The analytical model was base on energy conservation and failure in the numerical analysis was predicted based on Hashin's failure criteria.
2.3 Impact of Composite Sandwich Panels
In this section, a detail study of impact of composite sandwich panels in the three impact regimes lowvelocity, highvelocity and ballistic impact is presented.
2.3.1 Lowvelocity Impact
Mines et al., 1998 investigated quasistatic loading and lowvelocity impact behavior on two different composite sandwich panels. While the first panel was made up of Eglass/vinyl ester skin and Coremat® core, the second panel was made of Eglass/epoxy skin and aluminium honeycomb core. The first panel with Coremat® core had failed in the sequence of core shear, debonding, and distal facesheet damage and incident facesheet failure. The second panel failed by core shear, debonding, incident facesheet failure and then distal facesheet failure later. In the lowvelocity impact tests, the failure pattern remained the same in both the panels as of the quasistatic tests. The core properties and impact velocity govern the energy absorption capability of the sandwich panel.
Wen et al., 1998 investigated the penetration and perforation of composite laminates and sandwich panels under quasistatic, dropweight and ballistic impact tests by flatfaced, hemisphericalended and conicalnosed indenters/projectiles. They categorized the impact on laminates and sandwich panels into lowvelocity impact and wavedominated (highvelocity/ballistic impact) response. It was also stated in the research that sandwich panels subjected to lowvelocity impact have similar loaddisplacement characteristics as of quasistatic loading case. The perforation energy required by flat faced projectile was more than hemisphericalended and conical shaped projectiles in highvelocity impact.
Schubel et al., 2005 investigated quasistatic and lowvelocity impact behavior of sandwich panels with woven carbon/epoxy facesheets and PVC foam. The lowvelocity impact model behaved similar to quasistatic loading case when loads and strain levels were same. The static indentation response was compared to the numerical results obtained using ABAQUS and were in good agreement. A membrane solution, assuming membrane in the core affected region and plate on elastic foundation in the rest of sandwich panel was in poor agreement with the numerical results.
Hoo Fatt et al., 2001, developed static and dynamic models of sandwich panels subjected to lowvelocity impact. They investigated the behavior of sandwich panels having carbon/epoxy skins and a Nomex honeycomb core with a hemispherical indenter under various support conditions such as simply supported, fully clamped, and rigidly supported. Springmass models were considered to determine the loaddisplacement curve. They also investigated the damage initiation of sandwich panels under lowvelocity impact loading. The initial mode of damage depended upon the panel support conditions, projectile nose shape, geometry of the specimens, and material properties of the facesheet and core. Various failure patterns were studied and solutions based on them were derived separately. The analytical solution for the ballistic limit was also found and results for thick laminates were in better agreement than thin laminates.
Suvorov et al., 2005 performed numerical analysis on sandwich panels with foam core and studied the effect of interlayer in between the top facesheet and foam core. The foam core was modeled as crushable foam in ABAQUS. While the polyurethane (PUR) interlayer reduced the deformations in both the core and the composite facesheets, the elastomeric foam (EF) interlayer offered a better protection for the foam core alone.
Besant et al., 2001 performed numerical analysis on sandwich panels with aluminium honeycomb core. The metal honeycomb core was modeled as elastic perfectly plastic material. A quadratic yield criterion was proposed for the core material, which included both normal and transverse shear stresses. The importance of core plasticity in finite element analysis was explained.
2.3.2 Highvelocity Impact
A great deal of work has been done in the area of lowvelocity impact of laminates and sandwich panels and highvelocity impact of laminates but limited work has been presented in the domain of highvelocity and ballistic impact of sandwich panels. The following describes some recent studies on the highvelocity impact of composite sandwich panels
Velmurugan et al., 2006 studied the projectile impact on composite sandwich panels in the range of 30100 m/s. The sandwich models in this study were not the typical sandwich panels in the conventional sense. They had a core height comparable to the facesheet thickness and acted as a bonding agent between the facesheets. Energybalance model was used to determine the ballistic limit of three different sandwich panels. They assumed the sandwich panel as a single plate since the foam layer was thin and comparable to facesheet thickness. Also uniform failure mechanism along the through thickness direction was assumed in their model.
Skvortsov et al., 2003 developed an analytical model using energybalance principle to determine the ballistic limit of composite sandwich panels subjected to high velocity impact. Two different sandwich panels were subjected to high velocity impact using three different projectiles. These tests were conducted on simply supported and rigidly supported boundary conditions, and the initial velocity was varied in the range of 7095 m/s. The predicted panel energy was close to the experimental values and the error was due to the strainrate effects, plastic behavior, and hardening phenomena, which are not consider in the analysis.
2.3.3 Ballistic Impact
Kepler et al., 2007 conducted ballistic impact on sandwich panels consisting of GFRP plates and Divinycell H80 core, with three different projectiles. Lumped spring mass model was use to calculate force histories and panel response. Concentric rings connected by shear springs represented the sandwich panel. In this model, core shear deformation was assumed as the single significant contributor to the sandwich panel stiffness. The facesheet orthotropic was neglected in the panel response. Four different force histories: constant force, triangular force, sine series, and combination of sine and triangular force were used to calculate the energy loss in the panel. Of these, triangular and combined force gave results in better agreement with experimental results.
2.4Aluminium Honeycomb
For design and construction of lightweight transportation systems such as satellites, aircraft, highspeed trains and fast ferries, structural weight saving is one of the major considerations. To meet this requirement, sandwich construction is frequently use instead of increasing material thickness. This type of construction consists of thin two facing layers separated by a core material. Potential materials for sandwich facings are aluminium alloys, high tensile steels, titanium, and composites depending on the specific mission requirement. Several types of core shapes and core material are been applied to the construction of sandwich structures. Among them, the honeycomb core that consists of very thin foils in the form of hexagonal cells perpendicular to the facings is the most popular.
A sandwich construction provides excellent structural efficiency, i.e., with high ratio of strength to weight. Other advantages offered by sandwich construction are elimination of welding, superior insulating qualities and design versatility. Even if the concept of sandwich construction is not very new, it has primarily been adopt for nonstrength part of structures in the last decade. This is because there are a variety of problem areas to be overcome when the sandwich construction is applied to design of dynamically loaded structures. Other investigators have previously carried out noteworthy theoretical and experimental studies on linear elastic and nonlinear behavior of aluminium sandwich panels.
Kelsey et al., 1985 derived simple theoretical expressions of the shear modulus of honeycomb sandwich cores. Witherell, 1977 performed an extensive theoretical study for structural design of an air cushion vehicle hull structure using aluminium honeycomb sandwich panels. Okuto et al., 1991 showed the validity of the socalled equivalent plate thickness method in which a honeycomb sandwich panel subjected to inplane loads is approximately replaced by a single skin panel with equivalent plate thickness. Kobayashi et al., 1994, studied Elasto plastic bending behavior of sandwich panels. An experimental study was undertaken by Yeh et al., 1991 to investigate the buckling strength characteristics of aluminium honeycomb sandwich panels in axial compression. Kunimo et al., 1989 both, have studied the characteristics of the energy absorption capacity of bare honeycomb cores under lateral crushing loads theoretically and experimentally.
2.5 Ballistic Limit
The ballistic limit may also be defined as the maximum velocity at which a particular projectile is expected to consistently fail to penetrate armor of given thickness and physical properties at a specified angle of obliquity. Because of the expense of firing tests and the impossibility of controlling striking velocity precisely, plus the existence of a zone of mixed results in which a projectile may completely penetrate or only partially penetrate under apparently identical conditions, statistical approaches are necessary, based upon limited firings. Certain approaches lead to approximation of the V50 Point, that is, the velocity at which complete penetration and incomplete penetration are equally likely to occur. Other methods attempt to approximate the V0 Point, that is, the maximum velocity at which no complete penetration will occur
2.6 Energy Absorption Mechanism of Composite Materials
The research was done by Naik and Shrirao at 2004. Impact loads can be categorized into three categories which is lowvelocity impact, highvelocity impact and hypervelocity impact. This classification is made because of change in projectile's velocity will result in different mechanisms in terms of energy transfer between projectile and target, energy dissipation and damage propagation mechanism.
Basically, ballistic impact is considered as lowmass high velocity impact. In this impact event, a lowmass projectile is launched by source into target at high velocity. It is unlike lowvelocity impact that involved highmass impactor impacting a target at low velocity. In view of the fact that ballistic impact is high velocity event, the effect is localized and near to impact location.
According to Naik et al. (2006), seven possible energy absorbing mechanisms occur at the target during ballistic impact. Those mechanisms are cone formation at the back face of the target, deformation of secondary yarns, tension in primary yarns/fibres, delamination, matrix cracking, shear plugging and friction between the projectile and the target. Then, the researchers formulated all these energies into equation whereby the total energy absorbed by the target is summation of kinetic energy of moving cone EKE, shear plugging ESP, deformation of secondary yarns ED, tensile failure of primary yarns ETF, delamination EDL, matrix cracking EMC and friction energy EF.
ETOTALi = EKEi + ESPi + EDi + ETFi + EDLi + EMCi + EFi
Mines et al. (1999) identified three modes of energy absorption when analysed the ballistic perforation of composites with different shape of projectile. These energy absorptions are local perforation, delamination and friction between the missile and the target. However, the contribution of friction between the missile and the target in energy absorption is low compared to the other two. In terms of local perforation, three throughthickness regimes can be identified, namely: I  shear failure, II  tensile failure and III  tensile failure and delamination. Out of these three regimes, the throughthickness perforation failure is dominated by shear failure. Similar observation has been made by other researcher for thick graphite epoxy laminates whereby the perforation failure is dominated by shear failure. The third main energy absorption mechanism is delamination. Delamination can propagate under Mode I (tensile) and Mode II (shear) loading and each mode can dominate each other depending on structural configuration of the composite as well as material properties. Therefore, it can be predicted that the total perforation energy is a summation of energy absorption due to local perforation, delamination and friction between the missile and the target.
Epred = Ef + Esh + Edl
where Ef = friction between the missile and the target; Esh = local perforation; Edl = delamination
Apart from that, Morye et al. (2000) has studied energy absorption mechanism in thermoplastic fibre reinforced composites through experimental and analytical prediction. They considered three mechanisms that involved in absorbing energy by composite materials upon ballistic impact. The three energy absorption mechanisms are tensile failure of primary yarns, elastic deformation of secondary yarns and the third mechanism is kinetic energy of cone formed at back face of composite materials. They concluded that kinetic energy of the moving cone had a dominant effect as energy absorption mechanism for composite materials. However, they neglected a delamination as one of the factor contributed to the failure of composite materials during ballistic impact.
2.7 Kinetic Energy Equation
Kinetic energy (KE) attack is a penetration of the residual energy of a projectile. A projectile can give a certain amount of energy to attack and damage a vehicle if the projectile sufficient residual energy when it arrive at the target. This residual is very important to overmatch the capability and strength of the target material to resist penetration, and then it will penetrate. Kinetic energy shot can be presented with the simple law of physic.
K.E = ½ Mprojectile Vprojectile2
Increasing the mass (Mprojectile) of the shot increases its energy, but the real payoff comes from increasing its velocity (Vprojectile). If the diameter of the shot fills the whole gun barrel, the projectile becomes heavier and difficult to accelerate to required velocity with the length of the barrel. Additionally, a large diameter solid shot will provide more energy to penetrate the armour plate compared to a projectile which has the same mass but a smaller diameter. Consequently, the larger shot is not only less effective at the target but it is difficult to give it the necessary velocity.
According to Chang et al., 1990, depth of penetration at the target will depend not only on residual energy, but also on shape and size of the projectile. The curve shape at the projectile head is more important, as it must not only able to pierce the armour but the shoulders of the shot must also support the remainder so that it does not break up on its way through the armour. If for given mass the diameter of the shot is reduced and is length increased, then for the same residual energy the shot will penetrate further, as it is working on a smaller cross section area of armour. The ratio of lengthtodiameter is called slenderness ratio. Any projectile with ratio in excess of 7:1 cannot be spin stabilized it is not until they reach a ratio approximately 20:1 that they can call long rod.
So, based on those discussions above, we can conclude that energy absorption can be performed by this relation
Eabsor = Ein  Eout
= [½ Mprojectile Vin2]  [½ Mprojectile Vout2]
= ½ Mprojectile (Vin2  Vout2)
So, Eabsorbed = ½ Mprojectile (Vin2  Vout2)
2.8 TsaiHill Failure Criterion
Hill, 1950 proposed a yield criterion for orthotropic materials:
G+Hσ12+F+Hσ22+F+Gσ322Hσ1σ22Gσ1σ32Fσ2σ3+2Lτ232+2Mτ132+2Nτ122=1
This orthotropic yield criterion will be used as an orthotropic strength or failure criterion in the spirit of both criteria being limits of linear elastic behavior. Thus, Hill's yield stresses F, G, H, L, M and N will be regarded as failure strengths. Hill's criterion is an extension of von Mises' yield criterion. The von Mises criterion, in turn, can be related to the amount of energy that is used to distort the isotropic body rather than to change its volume. However, distortion cannot be separated from dilatation in orthotropic materials, so Equation 2.8 is not related to distortional energy. Unfortunately, some authors still mistakenly call the criterion of TsaiHill a distortional energy failure criterion.
The failure strength parameters F, G, H, L, M and N were related to the usual failure strength X, Y, and S for a lamina by Tsai. If only τ12 acts on the body, then, because its maximum value is S,
2N=1S2
Similarly, if only σ1 acts on the body, then
G+H=1X2
And if only σ2 acts, then
F+H=1Y2
If the strength in the 3direction is denoted by Z and only σ3 acts, then
F+G=1Z2
Then, upon combination of Equations (2.10), (2.11) and (2.12), the following relations between F, G, H and X, Y, Z result:
2F=1Y2+1Z21X2
2G=1X2+1Z21Y2
2H=1X2+1Y21Z2
For plane stress in the 12 plane of a unidirectional lamina with fibers in the 1direction, σ3 = τ13 = τ23 = 0. However, from the cross sectional of such a lamina in Figure 2.3, Y = Z from the obvious geometrical symmetry of the material construction. Thus, Equation (2.8) leads to
σ12X2σ1σ2X2+σ22Y2+τ122S2=1
as the governing failure criterion in terms of the familiar lamina principal strengths X, Y, and S. And, the appropriate values of Xt or Xc and Yt or Yc must be used depending on the signs of σ1, σ2, τ12 (except that the surface is symmetrical about the plane τ12 = 0 because S has only one value).
Finally, for the offaxis composite material example of TsaiHill failure criterion, substitution of the stresstransformation equations,
σ1=σxcosθ2
σ2=σxsinθ2
τ12=σxsinθcosθ
In Equation (2.14) yields the TsaiHill failure criterion for uniaxial offaxis strength,
cosθ4X2+1S21X2cosθ2sinθ2+sinθ4Y2=1σX2
which is one criterion. Because a composite lamina usually has different strengths in tension and compression, the values of X and Y must take on the appropriate values depending on the quadrant of stress space consists of four different segments that are continuous in value but not in slope at the uniaxial strengths.
The TsaiHill failure criterion appears to be much more applicable to failure prediction for this composite material than either the maximum stress criterion or the maximum strain failure criterion. Other less obvious advantages of the TsaiHill failure criterion are:
 The variation of strength with angle of lamina orientation is smooth rather than having cusps that are not seen in experimental results.
 The strength continuously decreases as θ grows from 0° rather than the rise in uniaxial strength that is characteristic of both the maximum stress and the maximum strain criteria.
 The maximum stress and strain criteria are incorrect by 100% at 30°.
 Considerable interaction exists between the failure strengths X, Y, S in the TsaiHill failure criterion depends on whether the material being studied is ductile or brittle. Other composite materials might be better treated with the maximum stress or the maximum strain criteria or even some other criterion.
2.9 Finite Element Analysis
This subheading starts with a brief introduction to AUTODYN and then follows this with an overview of the SPH implementation carried out.
Aizawa et al., 1980 has conducted the research about the AUTODYN software. The AUTODYN software is widely used to simulate nonlinear impact phenomena involving large strains and deformations, plasticity, fracture, and flow. The software, available on PC's to supercomputers, is packaged in an interactive, integrated environment wherein the pre and postprocessing and the analysis are contained in a single menudriven architecture. The software encompasses a number of different numerical approaches for the analysis of impact problems. Within the software, Lagrange, Euler, ALE (Arbitrary Lagrange Euler), Shells, and SPH (Smooth Particle Hydrodynamics) numerical processors (solvers) are available. Impact processes ranging from equipment drop tests to the hypervelocity impact of space debris on a spacecraft can be modeled.
Related study also been done by N.K.et al., 1987. The results of a number of analyses are present to highlight the advantages and disadvantages of each numerical technique for different classes of impact applications. It is shown that the selection of the appropriate numerical technique or combination of technique is critical to achieving both an accurate and computationally efficient solution. Impact case studies presented include:
 Hypervelocity impact of space debris on a shielded spacecraft
 Impact and penetration of ceramic armor by a steel projectile
 Oblique impact and ricochet of a steel sphere on RHA armor
 Impact and crush of a steel girder
 Explosive formation of an oil well perforator with subsequent impact and penetration on a layered steel/concrete structure
The solutions illustrate the use of different numerical techniques with emphasis on efficiency and accuracy. Validation of results with available experimental data is shown. Animations of the numerically simulated impact phenomena can be shown directly from the AUTODYN software.
CHAPTER 3
SIMULATION METHODOLOGY
This chapter provides the detailed description of method of approach used in carrying out the present study. This includes the strategies that have been design to serve guidelines throughout the process of the study and to assist in achieving the desired objectives. An outline of the methodology is first present in a flow chart to provide an overview of the whole process, followed by detailed discussion of the outline and elaboration of key procedures and techniques employed in the simulations.
Flow of my methodology consists the methods that been took in order to accomplish the expected results. Figure 3.1 showed the complete flow of methodology for this research. First thing to do is to make some researches about the topic chosen. All information was gathered through readings from related journals and thesis did some diggings through numerous interviews to get a clear figure about this topic. Next step, study and learn about the simulation program that is going to be used in this research, ANSYS AUTODYN and based on previous researches from literature reviews paper and results, we try to validate whether ANSYS AUTODYN really reliable to be used to continue this research. Then, we simulate in ANSYS AUTODYN. The comparison and analysis were done to validate simulated results with response to actual damage specimen.
3.1Simulation Tools Used
Finite element analysis was conduct on the sandwich structure using ANSYS 2D/3D finite element model was developed and appropriate material properties were given to each component.
For both the cases, ANSYS AUTODYN 2D/3D NONLINEAR hydrocode was used. It is an explicit numerical analysis code, where the equations of mass, momentum and energy conservation coupled with materials descriptions are solved. Alternative numerical processors are available and can be selectively used to model different regions of a problem. The currently available processors include Lagrange, Euler, Euler FCT, ALE and SPH.
3.2 Simulation Justification
Based on much discussion about which software that are available that could conduct the simulation on sandwich structure subjected to high velocity impact, finally the findings showed that ANSYS AUTODYN is the one that is going to be used to simulate the test.
So, to prove that this software is capable to conduct the simulation, some comparison on result between Lagrange solver with SPH solver of ANSYS AUTODYN on one of the literature review topic, Cylinder impacting a rigid wall (Taylor Test) was done.
The comparison was on the results for cylindrical shape impact behavior to see whether results simulated with Lagrange solver showed the more or less the same graph as simulate in SPH solver.
Figure 3.2 (a) showed the experimental deformation of rigid wall. Whereas Figure 3.2 (b) and (c) represent the resulting final deformations and plastic strain contours for the two analyses.
The results which are summarized in Table 3.1 show that both simulations show satisfactory comparison with experiment. The SPH solution compares almost exactly with the Lagrange result.
 Experimental Deformation
 Lagrange Simulation Result
 SPH Simulation Result
Table 3.1 Comparison of experimental and numerical Taylor Test results
Experiment 
Lagrange 
SPH 

Cylinder Length (mm) 
23.13 to 23.59 
23.30 
23.35 
Impact Diameter (mm) 
16.70 to 17.04 
16.78 
16.80 
This validation illustrates that the SPH implementation also works well for cylindrical symmetry. It proved that no special unphysical techniques are used to treat particles close to the axis, or indeed anywhere else in the problem.
3.3 Simulation Setup
In the numerical simulations presented, the projectile is made of steel. The steel material properties are from the AUTODYN material library and are shown in Table 3.2. The size and geometry of the projectile vary with different problems. The projectile was considered rigid. Lagrange solver was applied to the projectile. The laminated composite material was model by Lagrange method.
In order to account for contact/penetration behavior between the Lagrange projectile and the Lagrange laminate, the gap interaction logic of AUTODYN has been activated between the Lagrange cells. AUTODYN has a state of art contact logic wherein objects use a small ‘gap' to determine if interaction exists. This gap defines a detection zone that exists around each interacting cell face or node. If a node enters the detection zone it is repelled by a force that is a function of the intrusion depth. AUTODYN also features another special function “impact/penetration interface” to avoid excessive noise on the impact surfaces. This feature is activated in all the simulations presented.
The Lagrange solver has been implemented in both 2D and 3D. However, AUTODYN2D is used here for simplicity and ease of demonstration. The laminated composite specimens for the penetration experiment were rectangular. In the AUTODYN simulations, an equivalent rectangular laminate plate was modeled in 3D analysis. The size of the rectangular laminate is equivalent to the experiment laminate plate
The detail flow of simulation is shown in Figure 3.3.
3.4 Description of Model
Consider the composite sandwich panel and rigid projectile as shown in Figure 3.4. The dimension of projectile is 23.10 mm in length and had a hemispherical head with 5.56 mm diameter. The effective material properties used in AUTODYN simulations for this material are shown in Table 3.3. The equation of state of the laminate is linear with a bulk modulus of 15.0 GPa. The laminate is treated as a linear elastic material. The sandwich panel consists of thin orthotropic facesheets of thickness h = 3 mm and isotropic crushable polymeric aluminium core of thickness H = 27 mm.
3.5 Failure Theory
For a composite laminate, the TsaiHill Failure criteria are applied. Then, for the projectile, it is consider as rigid body.
3.6 Geometry
Consider the composite sandwich panel shown in Figure 3.4. The core of the sandwich panel and the rigid projectile were created using *Part. The reference node for the rigid projectile was defined at the bottom of the projectile. The mass of the projectile was assigned automatically by the software
3.7 Material Properties
Table 3.2 Material properties of steel projectile
Properties 
Value 
Reference Density (g/cm3) 
7.9 
Bulk Modulus (GPa) 
200 
Shear Modulus (GPa) 
90 
Yield Stress (MPa) 
200 
Ultimate Strain 
0.4 
Erosion Strain 
3.0 
Table 3.3 Material properties of carbon fiber
Properties 
Value 
E1 (GPa) 
68.5 
E2 (GPa) 
68.5 
E3 (GPa) 
9 
G12 (GPa) 
3.7 
v12 
0.11 
Xt (MPa) 
860 
Xc (MPa) 
795 
Yt (MPa) 
860 
Yc (MPa) 
795 
St (MPa) 
98 
ρ1 (kg/m3) 
1430 
εf 
0.02 
σc (MPa) 
60 
Table 3.4 Mechanical properties of aluminium honeycomb core (Boyer et al., 1991)
Properties 
Value 
Density (kg/m3) 
77 
Young Modulus (MPa) 
69000 
Poisson's Ratio 
0.33 
Shear Modulus (MPa) 
25000 
Shear Strength (MPa) 
120 
3.8 Analysis Type
An ANSYS AUTODYN Dynamic Explicit type analysis was performed for a time period of 0.15 ms. Nonlinear geometry was switched on. All the required outputs such as displacements, velocities and stresses were defined in this module.
3.9 Mesh
In finite element analysis, it is preferred to create a mesh with the least number of elements to keep analysis time reasonable while still getting accurate results. The facesheets and core were separate to get a very fine mesh towards the center of the sandwich structure. A biased ratio of 8.5 with 75 elements was defined in the first region of the panel and uniform mesh of 30 elements was defined in the second region. The facesheets had 2 elements through the thickness and the core had 25 elements. Both facesheets and core had default hourglass control and default distortion control.
3.10 Contact, Boundary Conditions
The fixed boundary condition (transverse and radial velocity equal to zero) is applied on the outer boundary of the laminate. The velocity of the projectile is V0 = 287 m/s.
3.11 Simulation Analysis
This simulation was performed using the Lagrange processor with erosion. Another approach available within AUTODYN is the SPH (Smooth Particle Hydrodynamics) solver wherein a gridless Lagrangian technique is used. Figure 3.9 are shown about the runninginprogress in simulation.
Another figure of the simulation progress are shown at Appendix B
CHAPTER 4
RESULT AND DISCUSSION
4.1Introduction
This chapter is the one that showed all the actual experiment results that previous done and simulated results using ANSYS AUTODYN. The simulations that have been do by using Lagrange method. Comparison was done between the results and actual experiment result to see whether simulation results agreed well with the actual penetration done by Wan Awis's experiment. Data for penetration test were presented afterwards.
4.2Actual Penetration/Firing Test Results
These are the data sheets of actual test handle by Wan Awis at 600m closed shooting range at STRIDE. Table 4.1 presents a Summary Penetration Test for Round 5.56mm Steel Core, the profile of specimen used and basic criteria for 5.56mm bullet for this penetration test. Results from this lot size of 5.56 mm bullets are shown in Table 4.2. This data will be compared with the simulated results.
Table 4.1 Actual firing condition
BALLISTIC IMPACT TEST 

Sample 
Hard Panel 
Test weapon 
5.56 mm, 9 mm Test Gun 
Sample Type 
Sandwich Panel 
Shooting Distance 
10m (honeycomb) and 5m (carbon fiber) 
Sample Size (mm) 
100 x 100 x 3.3 
Temperature 
29.3°C  31°C 
Ammunition Calibre 
5.56 × 45 mm 
Rel. Humidity 
88% 93% 
Type of Projectile: 
FMJ (M855) Steel Core 
Shooting Angle: 
0° obliquity 
Table 4.2 Test result on composite sandwich structure with 5.56 mm
Caliber 
Range 
Type of specimen 
Depth of Penetration 
Entry diameter 
Exit diameter 
Entry Velocity 
Exit Velocity 
m 
mm 
mm 
mm 
m/s 
m/s 

5.56mm 
20 
Composite 
33.1 
5.5 
6.8 
287 
220.67 
4.3 Simulation Results by ANSYS AUTODYN
The simulation results were compared with the experimental ones to validate the finite element model. The variables selected to validate the numerical model were the residual velocity, the ballistic limit, and the contact time.
The disadvantage of the experimental impact tests is the limited information concerning the development of the projectile during the impact. The experimental tests provided information only about the velocity of the projectile before the impact over the front skin and after the perforation of the back skin. However, the finite element model showed the progression of the projectile while it was crossing through the sandwich plate. Fig. 4.6 shows the progression of the projectile velocity during the impact (Vimp = 287 m/s).
There are some pictures to show the results for this observation on penetration effect on specimen using ANSYS AUTODYN. The pictures was attached at the Appendix B. Simulation impact observations were done on projectile's back view, projectile's front view, and projectile's side view.
The simulated results produced by ANSYS AUTODYN were done regarding difference parameters and values against time. The simulated results were presented in form of graph related to parameters specified.
4.3.1Residual Velocity
The progress of the velocity shown in Figure 4.6 is representative of each impact. Plotted graphs were extracted from the same node in this simulated analysis. There are three different trends corresponding to the three components of the sandwich (front skin, core, and back skin). In the first region, the composite front skin caused a sudden drop in velocity at the beginning of the impact event, so that the projectile reached the honeycomb core at a velocity of nearly 250 m/s. Secondly, the velocity remained almost constant as the projectile went through the honeycomb core, when the projectile reached the back skin, its velocity was nearly 240 m/s. In the back skin, a new drop in velocity was observed for a residual velocity of over 210 m/s.
4.3.2 Energy Absorbed
The projectile lost 46% of its impact kinetic energy, front and back skins absorbed 46% and 41% of the absorbed energy, respectively, and the honeycomb core absorbed 13%. This analysis was made on each numerical test, calculating the energy absorbed by the three components of the sandwich plate. Figure 4.8 specified the relation of absorbed energy against impact velocity.
The skins were the main factor responsible for the energy absorption, while the energy absorbed by the honeycomb core was lower. The percentage of the energy absorbed by each component was almost constant for impact velocities higher than 250 m/s: the front skin absorbed 45%, back skin 40%, and core 15% of the absorbed energy by the composite panel (refer Figure 4.7). However, when the impact velocity was near the ballistic limit, the front skin absorbed most of the impact energy so that the projectile reached the back skin at a lowvelocity. Thus, the energy absorbed by the back skin was reduced. The energy needed to break high strength carbonfibres is very high, so the projectile underwent a sudden lost of kinetic energy when it penetrated a composite skin.
The main energyabsorption mechanism of the honeycomb core was the plastic strain of the aluminium walls. The experimental tests indicated that the region of the honeycomb over which the projectile impacted had no influence on the results. The energy needed to deform a thinwalled cell of aluminium is very low, so the projectile crossed the honeycomb core with no major loss of kinetic energy.
4.3.3 Depth of Penetration
Figure 4.9 shown the graph of displacement against time. The displacement (depth of penetration) is on specimen's element that experienced the contact with 5.56 bullet projectile. Relation between those parameters was clearly explained by the above graph (refer Figure 4.9). The depth of penetration is increased by the time of penetration. The maximum displacement is about ±3.3 cm which is equivalent to specimen's thickness.
4.3.4 Contact Time
Another analysis was done to see relation about contact time against the impact velocity (refer Figure 4.10). The contact time was determined as the time between the contact of the projectile with the front skin and the immediate at which the projectile fully penetrated the sandwich plate.
4.4 Velocity (mm/s) Against Time (s) Analysis
The value of energy absorb during penetration can be calculated by using this formula
Eabsorbed=12MprojectileVin2Vout2
So, the value for impact velocity for simulated penetration test is 287m/s and value for after impact velocity is around ±210 m/s (Figure 4.6)
Eabsorbed=120.00178×28722102=34.06J
The calculated value of energy absorbed during penetration for the front skin is equivalent with the value in the graph shown. (Figure 4.8)
4.5 Ballistic Limit
The ballistic limit was defined as the minimum impact velocity required for the projectile to completely penetrate the sandwich plate. From the model, the ballistic limit calculated was 147 m/s. The experimental ballistic limit estimated was 139 ± 4.2 m/s, by fitting the equation of Lambert et al., 1976 to the residual velocity versus impact velocity curve. A comparison of the results from the numerical model and the experimental test gave a difference of 6% in the ballistic limit.
4.6 Discussions
The analysis done were clearly told us some methodology used on findings all required results about penetration test for 5.56 mm bullet projectile to the composite sandwich structure with honeycomb core. The value for each method that was performed based on equation involving simulated result.
Table 4.3 contains the comparison between actual and simulation analysis (FEA) regarding the value of residual velocity, and amount of energy absorbed. This table could be highlighted as the final results for this final year degree project on topic Modeling of Sandwich Structure with Honeycomb Core Subjected to High Velocity Impact.
Table 4.3 Table of results
Data Source 
Impact velocity (m/s) 
Residual velocity (m/s) 
Energy Absorb (J) 
Actual test 
287 
220.67 
34.06 
Simulation Result 
287 
210 
34.58 
Percentage of error between results can be calculated using normal method used in many analyses. This percentage error value is very important in order to prove that the simulation software suggested for FEA in this research is reliable to conduct further study on this research in the future. A large difference of error indicates that the simulation doesn't meet it purpose and objective for this research.
Error = [(findings  actual) / actual] × 100%
So, based on this formula we can calculate percentage of errors (Eq 4.3)
Simulation Results (Residual Velocity)
Error = [(210  220.67)] / 220.67 × 100%
= 4.84%
Simulation Results (Energy Absorbed)
Error = [(34.58 34.06)] / 34.06 × 100%
= 1.53%
The percentage errors calculated were presented in Table 4.4. As we can clearly see, the percentage of error is in an acceptable range. The objectives for this research are relevant and proven scientifically using finite element analysis and engineering methods.
Table 4.4 Percentage errors compare to actual penetration test
Task 
Simulation Result (Residual Velocity) 
Simulation Result (Energy Absorbed) 
Percentage Error Between Actual Test and Simulation Analysis (%) 
4.84 
1.53 
CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS
5.1 Conclusions
In this study the perforation of composite sandwich panels subjected to highvelocity impact was analyzed using a three dimensional finite element model implemented in ANSYS AUTODYN. Experimental impact tests were carried out to validate the numerical model. Good agreement was found between numerical and experimental results; in particular, the numerical simulation was able to predict the energy absorption of the sandwich panel with a difference of 1.53% and the residual velocity of the sandwich panel with a different of 4.84%.
The influence of both skins and the core in the energyabsorption capabilities of the sandwich panel was studied in a wide range of impact velocities. Most of the impact energy was absorbed by the skins. For impact velocities above 600 m/s, approximately 45% of the impact energy was absorbed by the front skin and 40% by the back skin. For impact velocities close to ballistic limit, the front skin absorbed almost the 60% of the energy. On the opposite, the honeycomb core absorbed between 10 and 20% of the impact energy by plastic strain, at all the impact velocities analyzed.
Also, the energyabsorption mechanisms in both skins and the core were studied. The main mechanism in the skins was fibre breakage whereas in the core the mechanism was the plastic deformation of the aluminium wall. Both in the skins and the core, the damage was concentrated in a small area around the impact point.