UFDC Home  myUFDC Home  Help 



Full Text  
A MICROMECHANICS METHOD TO PREDICT THE FRACTURE TOUGHNESS OF CELLULAR MATERIALS By SUKJOO CHOI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2002 Copyright 2002 by Sukjoo Choi This dissertation is dedicated to my parents, Sunggu Choi and Jinsil Yang ACKNOWLEDGEMENTS I am very grateful to Dr. Bhavani V. Sankar for providing me the opportunity to complete my M.S. studies under his exceptional guidance and financial support. He is not only my academic advisor but also a great influence in my life. Throughout this research, I have greatly appreciated his consistent encouragement, patience and positive attitude. Also, I would like to thank to Dr. ChenChi Hsu, graduate coordinator, for his advice. Because of great help and advice, studying here was a delightful experience. Many thanks should go to my colleagues, Donald Myers, Nicoleta Apetre and Huadong Zhu. Moreover, I would like to thank my school seniors, Jongyoon Ok, Chungsoo Ha and Kilsoo Mok, who provided me invaluable academic feedback, encouragement and companionship. I would like to thank my girlfriend, Sungshin Kang, who allowed me to devote myself to studying. I would like to express my deepest appreciation to my parents for continuous support and love. I am also thankful to God for giving me the opportunity to extend my education at the University of Florida. TABLE OF CONTENTS page A C K N O W L E D G E M E N T S ...................................................... ................................... iv LIST OF TABLES ............. .................................... ........ ........... vii LIST OF FIGU RES ...................................... .................. ............ .. viii ABSTRACT ........ .............. ............. ...... ...................... xi CHAPTER 1 IN TR O D U C TIO N ..................................................................... .......... .... .... 1 2 ELASTIC CONSTANTS OF THE FOAM...................................................................4 Formulation of MicroMechanics Properties.............. .............................................. 4 Finite Element Verification of Analytical Models ............. ...................................... 7 3 FINITE ELEMENT METHODS OF FRACTURE TOUGHNESS ...............................11 B boundary D isplacem ent at a C rack Tip ....................................................................... 11 M o d el S etu p ................................................................. ......... .............. 12 C onvergence A analysis for M ode I ....................................................................... ... 13 Mode I Fracture Toughness of Open Cell Foam ................................. .............. 17 Mode II Fracture Toughness of Open Cell Foam ................................................... 19 Fracture Toughness of Mixed Mode on Open Cell Foam .......................................... 20 Formulation of Fracture Toughness for Mode I and Mode II.................................. 22 Analytical Model for Mode I Fracture Toughness ............................................ 22 Analytical Model for Mode II Fracture Toughness........................................... 25 Mode I and Mode II Fracture Toughness with Angled Crack.................................. 26 Prediction of the Maximum Strength under Tensile Loading ............................ 27 Prediction of the Maximum Strength under Shear Loading .................................... 29 M ode I Fracture Toughness of Inclined Cracks............... ............... .................. 31 4 FRACTURE TOUGHNESS OF CARBON FOAM........................ ...............34 M material Properties of Carbon Foam ................................... .................... 34 Mode I Fracture Toughness Test (4Pt. Bending Tests)............................................. 34 Finite Element Analysis of Fracture Toughness .............................. ................. 38 Unit Cell of Carbon Foam Solid M odel .......................................................... 38 Micromechanics Analysis for Young's Modulus and Shear Modulus................... 39 Unit Cell of Carbon Foam Beam M odel............................................................ 43 Fracture Toughness Estimation of the Solid Model .............................................. 44 Fracture Toughness of the Beam Model ..................................... .............. 45 5 RE SU L TS A N D D ISCU SSION ......................................................................... ...... 47 Elastic C onstants of O pencell Foam .................................................. ... ................. 47 Numerical Analysis of Fracture Toughness ............... ..................................... 47 Fracture Toughness w ith an A ngle Crack............................................... .... .. .............. 48 Fracture Toughness of Carbon Foam .................................... .......................... ........ 48 APPENDIX A ANALYTICAL METHOD TO ESTIMATE THE SOLIDITY OF SOLID MODEL ..50 B CRACK TIP DISPLACEMENT FIELDS FOR ORTHOTROPIC MATERIAL.......... 52 R E F E R E N C E S ........................................ ........................................................... .. 5 5 B IO G R A PH IC A L SK E TCH ..................................................................... ..................56 LIST OF TABLES Table page 1.1 Material properties of the Zoltex carbon fiber...................... ............................... 3 3.1 Foam m odel properties for convergence analysis.................................................... 13 4.1 Mechanical properties of carbon foam ............................................. ........... 34 4.2 D ensities of various form s of carbon................................................. .. ... ......... 34 4.3 Fracture toughness with specimen properties of carbon foam.................................. 36 5.1 Constants of fracture toughness curves............... ......................................... 48 5.2 R results of fracture toughness .......................................................................... .... 49 LIST OF FIGURES Figure page 1.1 SEM images of low (left) and high (right) density carbon foam............................. 2 1.2 The packing of polyhedra to fill space: tetrakaidecahedra ................ ......... ...... 2 1.3 O pen rectangular cell m odel ............................................... ............................. 3 2.1 Applied stress for a opencell foam .......................... ....... ......................... 4 2.2 Deformed shape of struts on opencell foam ......................................................... 5 2.3 Stress distribution of tensile loading on opencell foam .......................... ................ 8 2.4 Shear applied on opencell foam ........................................................ .............. 9 2.5 Stress distribution of shear loading on opencell foam ............................................... 9 3.1 Opencell foam model with a crack .............................. .............. 12 3.2 Variation of fracture toughness with various model sizes................................... 14 3.3 Axial stress variation from a crack tip for various model sizes ........................... 15 3.4 Loglog plot of axial stress along a crack tip for various model sizes................... 16 3.5 Fracture toughness in various crack lengths........................................................ 17 3.6 Stress distribution of Mode I fracture (c=200 tm, h=20 m) .............. ................. 18 3.8 Stress distribution of Mode II fracture (c=200tm,h=20[tm)............................ 19 3.9 Numerical result of fracture toughness for M ode II ................................ .............. 20 3.10 Stress distribution on the mixed mode fracture (c=200[tm, h=20m)..................... 21 3.11 Mixed mode fracture toughness of constant length of cell edges............................ 21 3.12 Mixed mode fracture toughness of constant cell wall thickness............................ 22 3.13 Crack tip forces and moments in the actual foam and corresponding crack tip stresses in the idealized homogeneous continuum ............................................... 22 3.14 Variation of a (/1c) with relative density for Mode I........................................ 25 3.15 Variation of a ( /c) with relative density for Mode II..................... ............ 26 3.16 Forces in a ligament under combined loading ................................................ 27 3.17 Variation of maximum tensile stress ,y as a function of angle between the xy coordinate system and the principal material directions.................. ........... 29 3.18 Variation of maximum shear stress (shear strength) as a function of angle between the xy coordinate system and the principal material directions ............................ 31 3.19 Stress distribution with various crack angles for Mode I fracture........................... 32 3.20 Mode I fracture toughness for various crack angles with respect to the principal m material direction .................................... ........................... ... ........ 32 3.21 Stress distribution for various angled cracks under Mode II fracture.................. 33 3.22 Mode II fracture toughness for various crack angles with respect to the principal m material direction .................................. ........................... ... ........ 33 4.1 SEN B specim en geom etry ................................................ ............................. 36 4.2 Fourpoint bending test setup on a material testing machine .................................. 37 4.3 Loaddisplacement curves of fourpoint bending tests on carbon foam.................. 37 4.4 U nit cell of solid m odel........ .................................... ..... ...... .. 38 4.5 Boundary conditions on the unitcell surfaces and maximum principal stress distribution when the unitcell is stretched in the ydirection ............................... 41 4.6 Boundary conditions on the unitcell surfaces and maximum principal stress distribution when the unitcell is stretched in the ydirection ............................... 41 4.7 Elastic modulus E* as a function of relative density (solidity) for E, = 2.6 GPa....... 42 4.8 Shear modulus G as a function of relative density (solidity) for E, = 2.6 GPa. ....... 42 4.9 U nit cell of beam m odel ......................................... ..................... .............. 43 4.10 Maximum principal stress distribution of solid and beam models for a unit KI at the cra ck tip .................. ................................... ....... ......... ...... 4 5 4.11 Variation of fracture toughness of carbon foam with relative density................... 46 A 1 U nit cell of solid m odel......................... ................. ....................... .............. 50 B.1 Polar coordinate oriented from crack tip.............. ................... ............. 54 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science A MICROMECHANICS METHOD TO PREDICT THE FRACTURE TOUGHNESS OF CELLULAR MATERIALS By Sukjoo Choi December 2002 Chairman: Dr. Bhavani V. Sankar Major Department: Mechanical and Aerospace Engineering Cellular materials, such as carbon foam, are ideal core materials for sandwich composites in many applications because of their thermal resistance, low density, impact damage tolerance and cost effectiveness. Moreover, carbon foam is a good material for heat exchangers and thermal protection systems. The Mode I, Mode II and mixed mode fracture toughness of a cellular medium is predicted by simulating the crack propagation using a finite element model. For the sake of simplicity, the cellular medium is considered as a rectangular open cell structure, and the struts are assumed to be beams of square cross section. On the macroscale, the cellular solid is considered as an orthotropic material. A crack parallel to one of the principal material directions is assumed to exist in the solid and a small region surrounding the crack tip is modeled using beam finite elements. Displacements are imposed on the boundary of the region such that the crack tip is subjected to a given stress intensity factor for orthotropic materials. The struts are assumed to fail in a brittle manner when the maximum tensile stress becomes equal to the strength of the strut material, and the corresponding stress intensity factor is taken as the fracture toughness of the cellular medium. Based on the finite element results a semiempirical formula is also derived to predict the Mode I and Mode II fracture toughness of cellular solids as a function of relative density. Results are also presented for mixed mode fracture, and a simple mixed mode fracture criterion is derived for cellular solids. It is found that the fracture toughness is a strong function of the relative density, but the cell size also has a significant effect. Cracks inclined to the principal material directions were also considered, and the Mode I and Mode II fracture toughness was calculated as a function of the crack inclination angle. Fracture toughness of carbon foam was estimated by experimental analysis and also finite element method. Mode I fracture toughness of open cell carbon foam was measured using single edge notched fourpoint bend specimens. A micromechanical model was developed assuming a rectangular prism as the unitcell. The cell walls were modeled using threedimensional solid element. Aforementioned finite element micromechanical analysis was performed to predict the fracture. The micromechanical simulations were used to study the variation of fracture toughness as a function of relative density of the foam. The good agreement between the finite element and experimental results for fracture toughness indicates that micromechanics can be an effective tool to study crack propagation in cellular solids. CHAPTER 1 INTRODUCTION Cellular materials are made up of a net work of beam or plate structures leaving an open space or cell in between. Cellular materials, e.g., carbon and polymeric foams, offer several advantages of thermal resistance, durability, low density, impact damage tolerance and cost effectiveness. They have great potential as core materials for sandwich construction, heat exchangers and thermal protection systems in the military and commercial aerospace structures. An excellent treatise on the structure and properties of cellular solids has been written by Gibson and Ashby [1]. While analytical methods of thermal and mechanical properties of carbon foam are well documented, research on fracture behavior of various foams is still in its infancy. Gibson and Ashby [1] have presented approximate formulas for Mode I fracture toughness of cellular solids in terms of their relative density and tensile strength. These are limited to cracks parallel to the principal material direction. Also, fracture behavior under mixed mode was not studied. A SEM was used to capture the images of carbon foam as shown in Figure 1.1. In microscopic observation, the opencell foams are irregularly sized and spaced. For high density carbon foam (300 to 800 kg/m3), the length of cell edges is in the range of 1,000 tm to 2,000 tm. For lowdensity carbon foam (160 to 300 kg/m3), the cell length is in the range of 200 jm to 600 tm.  q Figure 1.1 SEM images of low (left) and high (right) density carbon foam The geometry of the foam is categorized by tetrakaidecahedra cells containing 14 faces, 36 edges and 24 vertices as depicted in Figure 1.2. Figure 1.2 The packing of polyhedra to fill space: tetrakaidecahedra In this study, finite element models are used to predict the fracture toughness of some cellular media under mixed mode conditions. Based on the microscopic observation of carbon foam in Figure 1.1, both cracks parallel to principal material direction and angled crack are considered. For the sake of simplicity, the cellular medium is considered as a rectangular open cell structure, and the struts are assumed to be beams of square cross section of a side h as shown in Figure 1.3. &A L Figure 1.3 Open rectangular cell model On the macroscale level, the cellular solid is considered as an orthotropic material. A crack parallel to one of the principal material directions is assumed to exist in the solid and a small region surrounding the crack tip is modeled using beam finite elements. For carbon foam, material properties of cellular structure are referred from pure carbon. Therefore, the Zoltex Panes 30MF High Purity Hilled carbon fiber (Table 1.1) is chosen because of the high percentage of carbon component weight (99.5%). Table 1.1 Material properties of the Zoltex carbon fiber Density 1750 kg/m3 Modulus of Elasticity 207 Gpa Poisson Ratio 0.17 Ultimate Tensile Strength 3600 Mpa CHAPTER 2 ELASTIC CONSTANTS OF THE FOAM In order to estimate the Young's modulus E1 in the principal material direction we subject the foam to a uniform macrostress din the 1 direction (see Figure 21) where superscript of describes the macromechanics properties and the subscript ofs is micromechanics properties. Formulation of MicroMechanics Properties The total tensile force acting on a strut is given by F = c2 a*, where c is the length of the unit cell. The microstress (actual stress) in the strut can be derived as F C2* a, h2 (2.1) h2 h2 Ik. h T] T Figure 2.1 Applied stress for a opencell foam n lr 1~ where h is the cross sectional dimension of the square strut. It should be noted that both the macrostrain E* and microstrain E are assumed to be equal. The microstrain is given by E= (2.2) E Substituting for os from Eq. (2.1) into Eq. (2.2) we obtain C2 co O h2 C2" ES, .. 2ES (2.3) E E hES where Es is elastic modulus of the strut material. The Young's modulus of the foam can be obtained from Eq. (2.3) as E = 7 .. E s (2.4) E* c2a h 2E B B' o ' o/2 ' A . 7c/2 Figure 2.2 Deformed shape of struts on opencell foam To calculate the shear modulus G12, the unitcell is subjected to a state of uniform shear as shown in Figure 2.2. The deformed shape of the struts is shown in thin lines. Due to antisymmetry the curvature of the deformed beam at the center of the strut must be equal to zero and hence the bending moment at the center of the strut must also be equal to zero [2]. This fact can be used to find a relation between F and M as M=Fc/2. Now halfof the strut, either OA or OB, can be considered as a cantilever beam subjected to a tip force F. The transverse deflection is given by 2 c S= 2 (2.5) 3EI h4 where I, the moment of inertia of the strut cross section is given by I = 12 The macroscopic shear stress c* is related to the shear force F as = (2.6) C2 The macroscopic shear strain y* can be calculated as 28 48 2 (2.7) C c 2 The shear modulus of the foam G22 is defined as the ratio of the macroscopic shear stress c* and macroshear strain y*. Then from Eq. (2.6) and Eq. (2.7) we obtain z* 1 h4 G12 Es (2.8) S2 From the expressions for E* and G*2 a relationship between the shear modulus and the Young's modulus can be derived as h2 E* sc2 C 2 (2.9) G* 1 h4 h2 E, 2 c4 Due to the symmetry of the structure it is obvious that E2=E3=E1 and G23=G31=G12. Further, since we are using beam theory to model the struts, all the Poisson's ratios referred to the principal material directions, v12, V23, and v31 are equal to zero. The relative density p p,, where ps is the solid density or the density of the strut material, is a measure of solidity of the cellular material. The density of the foam can be obtained form the mass m and volume V of the unitcell as shown below: m p(3h2c 2h3 S*3 3h2c 2h3) h 2 h3 p p3 3 c 3 (2.10) Ps Ps Ps c C C If the aspect ratio of the strut h/c <<1, the relative density can be approximated as p* h2 h3 Ih2 == 2 h3 (2.11) P, c c C3 Finite Element Verification of Analytical Models A portion near the crack tip of the foam was modeled using finite elements. Each strut was modeled as an EulerBernoulli beam. Each beam element has 3 integration points. Although we could have one unitcell with periodic boundary conditions by Marrey and Sankar [2], a larger portion of the cellular medium was modeled as the computational cost is not very high for this case. The model shown in Figure 2.3 consists of 100x100 cells. The cell length c of the unitcell is 200 num and the strut is assumed to have a square cross section with a side equal to 20 num. A uniform displacement (2= 0.01 m) is applied at nodes on the top edge of the model. The total force in the 2 direction is computed from the FE results. The average tensile stress (macrostress 02) can be obtained by > F  'T (2.12) Lc where the E sign denotes summation of all nodal forces and L is the width of the foam considered in the FE model (see Figure 2.3). S, Max. Princial Bottom Left Corner (Ave. Crit.: 75%) +1.035e+11 +9.487e+10 +8.625e+10 +7.762e+10 +6.900e+10 . . . +6.037e+10 +5.175e+10 +4. 312e+10 +3.450e+10 +2.587e+10 +1 .725e+10 +8.625e+09 5.356e1 3 Figure 2.3 Stress distribution of tensile loading on opencell foam The Young's modulus E2 of a foam can be determined by the stressstrain definition as g FS E o* ' (2.13) E ]  L For the case considered the FE model gave a value Ea = 2.09 GPa and the analytical model Eq. (2.4) yielded a value of 2.07 GPa. The difference of results is 1%. UB L Figure 2.4 Shear applied on opencell foam S, Max. Principal Bottom Left Corner (Discontinuities) 42 .922e410 +2. 679e+10 2 .435e10 +2 192e+10 +1.948e+10 +1.705e+10 +1.461e10 +1.218e+10 4+9. 742e409 +7. 306e+09 4 871eS409 +2 .435e+09 +0 00Oe+00 2 CRACK PROPAGATION ANALYSTS FOR CELLULAR MODEL ODB: fraba.odb ABAQUS/Standard 6.21 Fri Oct 19 03:14:09 Step: Step Increment 1: Step Time 2.220DE16 Primary Var: S, Max. Principal n) r.r v. n n frmt  , 1 F7A tnr. 1 nnnni Figure 2.5 Stress distribution of shear loading on opencell foam To estimate the shear modulus by numerical analysis, a constant horizontal displacement Us is applied to all the nodes on the topside of the foam as shown in Figure 2.4. The shear stress c* can be obtained by the sum of reaction forces F divided by the area of the top surface: r* (2.14) Lc where L is the length of the foam considered in the FE model and c is the unitcell dimension. The shear strain y* can be calculated as u Y = (2.15) L Thus, shear modulus G2 of the foam can be estimated as IF G L (2.16) Y U us C L The deformed shape of the cellular solid under shear is shown in Figure 2.5. From the FE model the shear modulus of the foam was estimated as 10.25 MPa, whereas the aforementioned analytical model Eq. (2.8) yields 10.35 MPa. The difference between the two results is about 1%. The FE model is slightly compliant because of lack of constraints on the side and applying periodic boundary conditions would have yielded values closer to the analytical solution. CHAPTER 3 FINITE ELEMENT METHODS OF FRACTURE TOUGHNESS In this section, we describe a finite element based micromechanics model for estimating the fracture toughness of the cellular solid. The crack is assumed to be parallel to one of the principal material axes, and Mode I, Mode II and mixed mode fracture are considered. To determine the fracture toughness a small region around the crack tip is modeled using beam elements. A constant mode mixity KI/KII is assumed to be applied. The boundary of the cellular solid is subjected to displacement (ui and u2) boundary conditions corresponding to an arbitrary value of Ki (or KII). The rotational degree of freedom at each node of the beam element on the boundary of the solid is left as unknown and no couples are applied at these nodes. The calculation of boundary displacements for a given stress intensity factor is described in the next section. Boundary Displacement at a Crack Tip The displacement components in the vicinity of a crack tip in a homogenous orthotropic material are derived in Appendix B and they are as follows: Displacement field for Mode I: i = KI 2r Re 1 [sp2 (cos0 +Ssin 0)) s2p1(cosO+s1 sin ) 2] (3.1) Y = K 2r Re 1 [sq2(cos+s2 sin)12 s2q,(cos0+s sin)12 V 7 S1 S2 J Displacement field for Mode II: u = K 2r Re{ [2 (csO1 +cos +ssO p sin O)2 ]}2 (32 Z S1 S2 (3.2) u, K, 2r Re [q2 (cos +s sin ) qi(cos s + sinO') } In deriving the above expressions, the crack is assumed to be parallel to the x axis, and rO is the polar coordinate system oriented at the crack tip. The complex parameters p, q, and s depend on the elastic constants of the homogeneous orthotropic material as described in the Appendix B. Model Setup The commercial FE program ABAQUS is used to conduct numerical analysis. A FE model of open cell foam is taken around the small portion of the crack tip in Figure 3.1. The boundary condition of small portion is determined by using the solution for boundary displacements in the previous section. A crack in the FE model is created by removing beam elements. kd Cell Length c Crack CrackT Length a Figure 3.1 Opencell foam model with a crack To reduce the modeling cost, a FORTRAN code was written to generate nodal and element properties with user specified unitcell configuration. The code calculates boundary displacements at corresponding boundary nodal coordinates. After execution, it exports an ABAQUS input file so that the ABAQUS can read the input directly. The maximum tensile stress in the struts is calculated from the finite element method. The FE analysis outputs axial force, bending moment and shear force at each node of the beam elements, and the maximum tensile stress was calculated using a separate program. Usually the maximum stress occurs at the crack tip strut. From the result the value of Ki (or KII) that will cause rupture of the strut is estimated, which then is taken as the fracture toughness of the cellular solid. Before we describe an analytical model for estimating the fracture toughness, we will discuss the results from the FE simulation. At first, we will check the validity of applying continuum fracture mechanics. It was found that a model consisting of 100x100 cells (total 10,000 cells) gave a converged result for fracture toughness and the same model is used throughout the study. The run time for this model is about 1312 minutes in a 1.7 GHz Intel Pentium 4 computer. Convergence Analysis for Mode I The convergence analysis is performed to evaluate variation of fracture toughness with various sizes of foam models, 1.2cm x 1.2cm (3,600 cells), 2cm x 2cm (10,000 cells), 4cm x 4cm (40,000 cells) and 8cm x 8cm (160,000 cells). The foam model properties are shown in Table 3.1. Table. 3.1 Foam model properties for convergence analysis Length of cell edge 200 tm Cell wall thickness 20 tm Relative density 0.0280 Tensile strength 3600Mpa 14 As the model size increases, fracture toughness becomes stable at approximately 4.67x105 Pa m1/2 in Figure 3.2. For a foam model containing from 10,000 to 160,000 cells, variation in fracture toughness is 2.7%. Therefore, the 10,000 cell model is chosen for further numerical analysis to maintain output accuracy with less CPU time. For the 10,000 cell model, a computer with 1.7GHz Intel Pentium 4 takes 131/2 minutes to complete the job, but the 160,000 cells takes 2 hours more. Fracture toughness variation with number of cells used 4.70E+05 F4.66E+06 I  a 4.65E+05 E .m 4.60E+05  m 4.55E+05  S4.50E+05     S4.45E+05  LL 4.40E+05 I 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000 Total number of cells used for simulation Figure 3.2 Variation of fracture toughness with various model sizes The axial stress along the crack path from a crack tip is shown in Figure 3.3. In the result, smaller models produce higher axial stress at the crack tip. Further from the crack tip, variation in axial stresses becomes insignificant for all model sizes. Maximum principal stress variation alone crack path under Mode 1 fracture 3600 cells (e0603)   10000 cells (e1005) 10000 cells (e1005) 40000 cells (e2010) 160000 cells (e4020)       A Ar lI.f II.U U 1 "I. I U t+UJ I 0.00E+00 4.00E04 8.00E04 1.20E03 Distance from crack tip (m) Figure 3.3 Axial stress variation from a crack tip for various model sizes The loglog plot of maximum strut stress variation from the crack tip is shown in Figure 3.3, the curve is fitted to the equation 41.69 oaz Vr (3.3) Using the relation between micro and macro stresses in Eq. (2.1), Eq. 3.4 can be written as 1.05 o2 r (3.4) indicating that the macro stress in tensile factor is equal to 1.05 which is approximately equal to the applied the stress intensity factor Ki of unity. 2.35E+03 (A . 2.10E+03  4o) S1.85E+03 Q. S1.60E+03 E 2 1.35E+03 1.60E03 2.00E03 16 Loglog plot of maximum principal stress variation alone crack path under Mode 1 fracture 3.5E+00 (C 3.3E+00  IQ S 3.1E+00  5 y= 0.51x +1.62 S* 3,600 cells C* 10,000 cells E 35E+00 2E+00  3E+00  40,000 cells E x 160,000 cells S2.7E+00  02.7E+00 i 0 2.5E+00 3.5E+00 2.9E+00 2.3E+00 Log distance from crack tip (m) Figure 3.4 Loglog plot of axial stress along a crack tip for various model sizes If the fracture toughness estimated by using the present method is truly a material(macroscopic) property, then it should be independent of the crack length. Hence the crack length was varied in the micromechanical method. The results are shown in Figure 3.5. The predicted fracture toughness was plotted as a function of percentage of crack length in the model. The results clearly show that idea of modeling the foam as a homogenous material is quite acceptable. Variation of Crack Length on Mode I fracture 5E+05 $ _ __  _ _ _ __ _ _ .. _..  4E+05 a L. Y" 3E+05 CU 2E+05  CU 1E+05 I OE+OO . 0 20 40 60 80 100 Percentage of Crack Length (%) Figure 3.5 Fracture toughness in various crack lengths Mode I Fracture Toughness of Open Cell Foam Since the relative density depends on the length of the strut c and a cell wall thickness h, the fracture simulation is conducted in two cases. In the first case, c is varied and h is kept constant. The second case, h varied while c is constant. The results of the FE simulation are axial force, bending moment and shear force in each element, which are used to calculate the maximum principal stress at the crack tip, and then the fracture toughness. A sample stress distribution due to Mode I fracture is shown in Figure 3.6. Figure 3.6 Stress distribution of Mode I fracture (c=200 pm, h=20 pm) The variation of fracture toughness with relative density is shown in Figure 3.7. The results of fracture toughness in Figure 3.7 are fitted to a curve with a power of 1.045 for constant wall thickness and 0.7879 for constant length of strut. To obtain a higher fracture toughness, increasing cell wall thickness h is more effective while cell length c is constant. Re 2.0E+06 1.8E+06 1.6E+06 1.4E+06 1.2E+06 1.0E+06 8.0E+05 6.0E+05 4.0E+05 2.0E+05 O.OE+00 lative Densityvs Fracture Toughness for Mode 1 000 0.02 0.04 0.06 008 010 012 014 0.16 018 Relative Density Figure 3.7 Mode I fracture toughness for two cases  y= 1 91E+07X''   y = 7.82E+06x788EO' .    Constant Thickness Constant Length  .  . Power [Constant Thickness S Power (Constant Length]        I Mode II Fracture Toughness of Open Cell Foam The analysis of Mode II fracture toughness is performed in same manner as the Mode I fracture analysis in previous section. Unlikely to Mode I fracture, the boundary displacements are not symmetric about the crack plane as shown in Figure 3.8. Figure 3.8 Stress distribution of Mode II fracture (c=200[tm,h=20[tm) The results of Mode II fracture toughness are fitted in a curve with power of 1.0654 for constant wall thickness h and 1.3231 for the constant cell length c in Figure 3.9. Therefore, to obtain the higher fracture toughness, increasing cell wall thickness h is more effective than increasing cell length c for Mode II. Relative Densityvs. Fracture Toughness for Mode 2 4.0E+05 Constant Thickness 3.5E+05 Constant Length .  _  Power (Constant Thickness) S3.0E+05 Power (Constant Length] r, ' 1.5E+05        I   E ," y1 32. E+00 , 2.5E+05  y =6.95E+06x .......... ........ S2.E+05  __ __      1.5E+05   S 1.E+05     5.0E+04 O.OE+00 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Relative Density Figure 3.9 Numerical result of fracture toughness for Mode II Fracture Toughness of Mixed Mode on Open Cell Foam For the mixed mode fracture analysis, the boundary displacements for Mode I and Mode II are superposed for various ratios of Mode I and Mode II in Figure 3.10. A sample deformation under missed mode is shown in Figure 3.10. Fracture toughness for mixed mode is shown for constant length of cell edge in Figure 3.11 and constant wall thickness in Figure 3.12. The fracture toughness for mixed mode is inversely linearly proportional. K, + = 1 (3.5) where the constants a and b are 1. The combinations of Ki and KII for fracture are shown in Figure 3.11 for constant cell length and in Figure 3.12 for constant cell wall thickness. Figure 3.10 Stress distribution on the mixed mode fracture (c=200rtm, h=20[tm) Fracture Toughness of Constant Cell Length for Mixed Mode 3.OE+05 rel density.104 (1100t20) 2.5E+05  rel density0.028 (1200t20) E rel density0.0127 (1300t20) e rel density0.0073 (1400r20) = 2.0E+05 .) E+04 o 1.5E+05  .  0O.E+0 S1.0E+05 2.E+05 4.OE+05 6..E+05 B..E+05 1.OE+06 1.2E+06 1.4E+06 0c '^ 5 .0 E + 0 4 . . . . .. .....     .. . 0.0E+00 4,, 0.OE+00 2.0E+05 4.0E+05 6.0E+05 8.0E+05 1 .OE+06 1.2E+06 1.4E+06 Fracture Toughness of Mode 1 KI (Pa m1) Figure 3.11 Mixed mode fracture toughness of constant length of cell edges Fracture Toughness of Constant Thickness for Mixed Mode 4 OE+05 rel density0.0073 (1200tl 0) Q 3.5E+05  rel density0.0280 (1200t20) E rel density0.0608 (1200t30) 3.0E+05   rel densit0 1040 (1200t40) S 2.E+05  0 m 2 0E+05 . 0) CD 1 5E+05  0 1.0E+05 I ^~ ~~~ 15 +5 ^ .   j 5.0E+04     O.OE+00 O.OE+00 5.0E+05 1.0E+06 1.5E+06 2.0E+06 Fracture Toughness of Mode 1 KI (Pa m1r) Figure 3.12 Mixed mode fracture toughness of constant cell wall thickness Formulation of Fracture Toughness for Mode I and Mode II Analytical Model for Mode I Fracture Toughness I II Ie II_ F Actual Foam Ideal Homogeneous Material Figure 3.13 Crack tip forces and moments in the actual foam and corresponding crack tip stresses in the idealized homogeneous continuum In order to derive an analytical model for fracture toughness, the stress intensity factor of the homogeneous model should be related to the actual stresses in the crack tip ligament of the foam. This can be obtained by assuming that the internal forces and bending moment in the crack tip strut are caused by a portion of the crack tip stress field ahead of the crack tip in the homogeneous model. Let us define a nondimensional factor a that describes the effective length I as follows: = ac (3.9) The ideal stress distribution cy ahead of crack tip is described as follows: K1 a (3.10) where r is the distance from the crack tip. The axial force in the crack tip ligament is obtained by integrating the tensile stress cy over the effective length 1 K F21 F = cl dr = K c (3.11) The bending moment M is given by KM=cc d Kic 13/2 2 13/2 M =cra dr =r dr 1 c (3.12) f y 2)r r f 2 3 Assuming fracture occurs when the maximum bending stress equals the tensile strength of the ligament material, a relationship between the tensile strength and fracture toughness can be derived as Mh Sy 2 6M 6 2 13/2 22 13/2 a= K I c= K c (3.13) I h2 h3 h3 3 t I h3 12 By substituting for I in terms of a from Eq. (3.9), the fracture toughness can be related to the tensile strength as 2,132 c 3/2 S= Ka3/2 c (3.14) Then the fracture toughness of the cellular material can be derived as Then the fracture toughness of the cellular material can be derived as K, h h (3.15) KI = 2./2a3/2 C5/2 = u 2_/2,3/2 C ) 22a c c2"2 2 a32 c c) The nondimensional distance a can be found from the FE results for fracture toughness KI, and the tensile strength of the ligament material as follows: a = 2 52 2/3 (3.16) K 2 fc The results for a as a function relative density are plotted in Fig. 3.14. It may be noted that a increases with relative density and the variation can be accurately represented by a power law. Interestingly both constant wall thickness case and the constant spacing case we have studied so far fit accurately into a single power law. The effective crack tip distance is about 10% of the cell spacing for lowdensity foam and about 305 for high density foam. Alpha versus Relative densityfor mode 1 0.350 0.250    0.150  m constantwall thickness h 0.050     O constant ellen th  0.200 .  ..    . .   0.100     )K constant wall thickness h 0.050   constant cell length c   Power (constant wall thickness h) 0.000 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 Relative Density Figure 3.14 Variation of a (/1c) with relative density for Mode I Analytical Model for Mode II Fracture Toughness The effective crack tip distance for Mode II can be derived following steps similar to that in the preceding section for Mode I. The shear stress (Txy) distribution ahead of the crack tip is given by: K1 Z (3.17) The total shear force F over the effective distance / can be derived as F=cJr dr=K K c (3.18) 0 ' The maximum bending moment in the crack tip element is the product of force F and the length c/2: 2/ c M = K c (3.19) The maximum bending stress is derived as h My 2 6M 21 c a= 2 3KI, (3.20) I h h3 ,h 12 An expression for fracture toughness KII in terms of strut tensile strength, strut dimensions and the effective distance can be derived as CU.h3 7 n n K11 32 (3.21) Since, fracture toughness can be expressed in terms of the nondimensional distance a: K, h3c2 r (3.22) 3C2 2ac 3 2c From (3.33) an expression for a can be derived as ch,3 h Z T h 6 a= 1 2  (3.23) 3K c2 I2 18K c The constant a can be evaluated using the FE results of fracture toughness. The value of a as a function of relative density is plotted in Fig. 3.15. Again, it can be noted that a power law description is adequate for Mode II also. Further the value of a depends only on the relative density and not on individual cell or strut dimensions. The effective distance seems to be slightly less for Mode II. It is about 5% of the strut spacing for low density and about 22% for high density foams. Alpha versus Relative density for mode 2 y = 0 6765X.472 0.150     0.100   S constant cell length c 0 .0 5 0 ,        Power (constant cell length c) 0.000 0 0.02 0.04 0.06 0.08 0.1 0.12 Relative Density Figure 3.15 Variation of a (1/c) with relative density for Mode II Mode I and Mode II Fracture Toughness with Angled Crack So far our attention has been focused on cracks parallel to the principal material direction. The next step will be to study crack inclined at an angle to the principal direction. Before that, an understanding of how the material will fail under combined loading will be beneficial. In the following we describe methodologies to predict the failure of the cellular medium under combined loading. The results will then be used to predict the tensile and shear strength in an arbitrary coordinate system. Prediction of the Maximum Strength under Tensile Loading When the cellular medium is subjected to combined loading under plane stress conditions, the ligaments in a typical representative volume element will be subjected to axial force, shear force and bending moment as shown in Fig. 3.16. /2 X F Figure 3.16 Forces in a ligament under combined loading From the equation (2.1), axial stress can be obtained as ,h2 a = (3.25) The relation between maximum bending moment and the shear force is given by 3* 3 M = F r=2 C 12 (3.26) 2 2 2 The maximum stress due to the bending moment can be derived as c2 c3 h beg My 2 3 *,c (3.27) Ibending J 4 h 12 It should be noted that we have assumed that only o component of the macrostress exists. The ligament stresses in Eq. 3.25 and Eq. 3.27 can be expressed in terms of macrostresses referred to the xy coordinate system as: a1 = sin2o (3.28a) 72 = "y cos2 U (3.28b) T12 = cay sin 0 cos O (3.28c) Then the microstress ou under combined loading can be derived as oaal = h) =c sin2 (3.29) Bending = 3rz2 =3 sinOcosO (3.30) The microstress ou can be derived as a7 = crsin2 3r sinOcosO =r sin2 03sinOcosO( (3.31) Sh) h h h The maximum stress that can be applied to the cellular medium is obtained by equating the maximum microstress to the strength of the ligament material. The ratio of the macrostress and the microstress for ligament L1 (refer to Fig. 3.16) can be derived as  1 (3.32) Ssin03 sin0 cos0 ( h h) In the same manner, the ratio of the macrostress and the microstress for ligamnet L2 can be derived as cr 1 =2 1(3.33) cos2 0 3 sin cos 0 h h Eq. (3.32) and Eq. (3.33) are used to calculate the maximum stress ,y that ca be applied to the cellular medium before any ligament failure. Then maximum stress under tensile loading is a function of the orientation 0 as shown in Figure 3.17. The strength variation is symmetric about 0=45. Maximum strength with various crack angles when tensile stress is applied 0050 rel density=0.0073(1200t10) 0 040 rel density=O 0280 (1200120) ... 0 040  rel density=002 (120020) S rel density=0 0608 (1200t30) S_ _ rel density=0 1040 (1200t40) S000 0 00 0    1 V  ^I      Z  0 15 30 45 60 75 90 Angle (deg) Figure 3.17 Variation of maximum tensile stress oy as a function of angle between the xy coordinate system and the principal material directions Prediction of the Maximum Strength under Shear Loading The calculation of shear strength at an arbitrary orientation is very similar to that described for tensile strength in the preceding section. The transformation of macro stresses between the global xy coordinates and the principal material coordinates are given by: = r sin 20 (3.34a) 2 = r sin 20 (3.34b) r1 = T cos20 (3.35c) It may be noted that only shear component of the macrostress (Txy) is present. Substituting Eq. (3.34) into the Eq. (3.25) and (3.27) we obtain aial oi = sin 20 (3.36) bendg = 3rz = 3r2 cos20 h (3.37) The microstresses can be derived as c c ~ c 2 c (338 a, = rsin2 3r cos20 = T sin c20 3 cos20 (3.38) h h h h The ratio of the macroshear stress and the microstress in ligament L1 can be derived as rZ 1  =(3.39) sin 20 + 3 cos 20 h h In the same manner, the ratio of the applied shear stress and the maximum stress in ligament L2 can be derived as (3.40) Z c IcI sin 20 3 C cos 20 ( h h) The macroscopic shear strength (maximum shear stress) is plotted as a function of orientation 0 in Figure 5.3. The maximum stress is symmetric about 0=45. Maximum strength with various crack angles when shear stress is applied 0.05  rel density=0.0073 (1200t10)  rel denslty=0.0280 (1200t20) 0.04   .  rel density=0.0608 (1200t30)  rel density=0.1040 (1200t40) 15) 0 03    c to E = 0.02  0.01 .  ^ . V .     0.01 0 15 30 45 60 75 90 Angle (Deg) Figure 3.18 Variation of maximum shear stress (shear strength) as a function of angle between the xy coordinate system and the principal material directions Mode I Fracture Toughness of Inclined Cracks The procedures for predicting the fracture toughness of angled cracks (cracks inclined at an angle to the material principal directions) are very similar to that described in the preceding sections. The only change is in the material elastic constants which have to be transformed from the material principal directions to the global xy coordinate system. The stress fields under Mode I fracture for various crack angles are shown in Figure 3.19. Crack Angle Crack Angle Crack Angle Crack Angle 10 Degree 20 Degree 30 Degree 45 Degree Figure 3.19 Stress distribution with various crack angles for Mode I fracture The variation of Mode I fracture toughness with crack orientation is shown in Fig 3.20 for various relative densities. It may be noted that the variation of Mode I fracture toughness with 0 is very similar to that of tensile strength shown in Fig. 3.17 with the results being symmetric about 0=45. Fracture Toughness in Various Angles for Mode I 2.0E+ 6  rel density = 0.0280 (1200t20)  rel density= 0.1 040 (1200t40) rel density= 0.0608 (1200t30) 1.E+ 6       rel density= 0.0073 (1200tl 0) a 2 1 .2E+G6 .,    , I,  B .2E+C5   y E+ 5  "*      . 0:3  4nFE+05  O.OECO I I I I I I I I 0 5 10 15 20 25 30 35 40 45 Angle (Deg) Figure 3.20 Mode I fracture toughness for various crack angles with respect to the principal material direction Crack Angle Crack Angle Crack Angle Crack Angle 10 Degree 20 Degree 30 Degree 45 Degree Figure 3.21 Stress distribution for various angled cracks under Mode II fracture The stress filed for various angled cracks under Mode II fracture is shown in Figure 3.21. The variation of Mode II fracture toughness is shown in Figure 3.22. Again one can note the similarities in the variation of Kic and shear strength presented in Fig. 3.18 Fracture Toughness in Various Angles for Mode II 1 2E+06 * rel density= 0.0280 (1200120) reldensity= 0.1040 ;i2n1tOi 1.0E+06 rel density= 0.0073 ;200 0 .. ... E rel density= 0.0508 :200I3131i 0E05  OE05    5,BOE+05   ^. ,  I  I j       S2. OE+*05  . .. ^  ^ . ^ 0 OE*00 ~ 0 5 10 15 20 25 30 35 40 45 /Trie '.Le'. Figure 3.22 Mode II fracture toughness for various crack angles with respect to the principal material direction CHAPTER 4 FRACTURE TOUGHNESS OF CARBON FOAM Material Properties of Carbon Foam From the SEM image of lowdensity carbon foam shown in Figure 1.1, one can note that the cells are irregularly sized and spaced. The cell size is measured to be in the range of 1 to 2 mm. Mechanical properties of the carbon foam as reported by Touchstone Research Laboratory, Inc., manufacturers of the foam, are shown in Table 4.1. The solidity (relative density) is determined by dividing, p', density of the foam, by ps, the density of solid carbon that make the struts or the cell walls. The densities of various forms of carbon are given in Table 4.2. The solidity of the carbon foam used in the present study was based on the value of solid density ps = 2,250 kg/m3, and the solidity is determined as 0.1312. Table 4.1 Mechanical properties of carbon foam Elastic Modulus (E*) 123.79 MPa Tensile strength (or) 3.5805 MPa Density (p*) 295.3 kg/m3 Table 4.2 Densities of various forms of carbon Diamond (C Wt. %100) 3,510 kg/m3 Graphite carbon fiber (C Wt. %100) 2,250 kg/m3 Zoltec Pane 30MF carbon fiber (C Wt. % 99.5) 1,750 kg/m3 Mode I Fracture Toughness Test (4Pt. Bending Tests) There are several methods available for measuring the fracture toughness of cellular materials. Compact tension test (CT), Single edge notched bend test (SENB) and Double edge notched tension test (DENT) performed by Fowlkes [6] are some of the tests that are suitable for foam materials. In the present study, we chose the single edge notched fourpoint bend specimens for measuring the fracture toughness of carbon foams. It was thought that the fourpoint bending test would yield more accurate and repeatable results as the crack is in a region under constant bending moment and no transverse shear force. Hence small offset of the loading point with respect to the crack location will not significantly affect the results. The specimen dimensions are depicted in Figure 4.1. The height of the specimen was about 50 mm and the crack length was about 25 mm. Individual specimen dimensions are given in Table 4.3. A notch was cut using a diamond saw, and then a razor blade was used to sharpen the crack tip. The crack length was the distance of the crack tip from the bottom surface edge of the beam. The tests were conducted under displacement control in a material testing machine at the rate of 0.5 mm/min (Figure 4.2). Loaddeflection diagrams are given in Figure 4.3. It may be noted from the curves that the crack propagated instantaneously and the specimens failed in a brittle manner. The fracture loads for various specimens are listed in Table 4.3. The Mode I fracture toughness was calculated from the load at failure using the following formula [3]: a a3 a41 KI = o fa 1.121.39+7.3 213 3 +14 (4.1) where the maximum bending stress o7 in the uncracked beam is determined by MW My 2 6M (4.2) o0 (4.2) I w3 Bw2 B 12 In Eq. 4.2, M is the constant bending moment in the central region, h is the height of the specimen and B is the width. The bending moment Mis given by M=Pd/2, where dis distance between one of the top loading points and the corresponding bottom support as shown in Figure 4.2. The results for fracture toughness are listed in Table 4.3. For the carbon foam samples tested the average Mode I fracture toughness is found to be 0.1337 MPa m1/2 with a standard deviation of 0.011 MPa m1/2 (about 8%). P P  d  0.0733m h=0.0506m 0.2032m a B=0.0254m SSimple supports Figure 4.1 SENB specimen geometry Table 4.3 Fracture toughness with specimen properties of carbon foam Speci Span L Height h Width B Crack Density Fracture KI, men (m) (m) (m) length a (Kg/m3) Load (MPa (m) (N) m12) IF06 0.2284 0.0512 0.0255 0.0264 284 100.9 0.1315 IF07 0.2291 0.0500 0.0255 0.0252 301 112.0 0.1458 IF09 0.2290 0.0507 0.0255 0.0259 292 92.54 0.1201 IF10 0.2290 0.0506 0.0256 0.0261 297 105.8 0.1372 37 Figure 4.2 Fourpoint bending test setup on a material testing machine Loaddisplacement for 4point bending tests 120   IF06 (4pt. Bending) IF07 (4pt. Bending) 100   IF09 (4pt. Bending)  IF10 (4pt. Bending) 80  60  20 O.OE+00 1.OE04 2.OE04 3.0E04 4.0E04 5.0E04 6.0E04 Displacement (m) Figure 4.3 Loaddisplacement curves of fourpoint bending tests on carbon foam Finite Element Analysis of Fracture Toughness Unit Cell of Carbon Foam Solid Model The first step in simulating the crack propagation in carbon foam is to idealize the microstructure of the foam. The unitcell is assumed as a perfect cube of side c in Figure 4.4. The foam model is created by placing a spherical void (bubble) at the center of the cube. By varying the radius of the bubble R, foams in various solidity can be modeled. A relation between the solidity and the R/c ratio can be derived in Appendix A: p 4+7 8 fR3 R)2 = +z 37 (4.3) p, 4 3 a) a Figure 4.4 Unit cell of solid model The average dimension of the unitcell was obtained from the SEM images of the cross section of carbon foam. Then, the radius of the spherical void can be determined from the solidity of the foam. The Pro/Engineering, modeling application, was used to model the unitcell and calculate the solid volume. In the present study, the unitcell dimension c is taken as 1.8 mm and the solidity as 0.1312. The strength of the solid carbon in the foam can be easily estimated from the tensile strength of the carbon foam, which is measured experimentally. The relation between the foam tensile strength and the solid carbon strength is given by 2 = (4.4) Amn where Amin is the minimum cross sectional area of the struts in the carbon foam normal to the principal material axis. It should be noted that the tensile strength of the foam is for a direction parallel to one of the principal material axes. The area Amin was obtained from the modeling software and it was equal to = 7.146 x 108 mm2. Substituting the dimensions of the unit.cell and the measured carbon foam tensile strength, the strength of solid carbon was estimated as 162 Mpa. The procedure for determining the Young's modulus and tensile strength of the solid carbon is as follows. Micromechanics Analysis for Young's Modulus and Shear Modulus The Young's modulus Es of solid carbon was estimated by a trial and error method. An initial value for Es is assumed and the elastic constants of the carbon foam are determined by using the micromechanical methods developed by Sankar et al. [4]. Then the value of Es can be scaled to match the micromechanical results for E* to the experimentally measured E*. The unitcell was modeled by using 4noded tetrahedral solid elements. Due to symmetry only a portion of the unitcell was modeled and periodic boundary conditions were imposed such that only one of the macrostrains is non zero [4]. We assume a value for Es and vs, and the forces required to deform the unit cell are calculated from the nodal reactions. From the forces the macrostresses can be computed. For example (4.5) cx 2 2 C C where EFx represents the sum of all nodal forces on a face normal to the x axis. Similar EFx is the sum of all the nodal forces Fy on the face normal to the y axis. The carbon foam is assumed as an orthotropic material and the macrostresses and strains are substituted in the constitutive relation to obtain the compliance coefficients S,. From the S matrix the elastic constants can be estimated using the following relations: S,, S12 S12 x = S2 S,, S12 y S12 S12 S,, fo 1z 12 12 11 7 Ex E E6 "11 "12 12 xy F = S12 11 S121 E E E. L s s V V 12 12 l11 Ex Ey E The FE model contained approximately 100,000 solid tetrahedral elements. A displacement uy=l was applied to the top surface of a unit cell (Figure 4.5). The contour plot of maximum principal stresses is depicted in Figure 4.5. 41 t Max. Principal y = 0.01 (Ave. crit.: 75%) 43.600e+10 43.300e+10 43.0OOe+10 42.700e+10 42.399e410 42.D99e+10 41.799e+10 +1.498e+10 41.198e+10 + 8.978e+09 Ux=O Ux=O 45.975e+9 X '42.973e+09 =.. 3.941e+07 zJ Figure 4.5 Boundary conditions on the unitcell surfaces and maximum principal stress distribution when the unitcell is stretched in the ydirection For the sake of simplicity no attempt was made to estimate the Poisson's ratio Vs using the micromechanical methods and it was assumed to be equal to 0.17 based on a previous analysis. Based on the foam properties given in Table 4.1 and the unitcell size c of 1.8mm, the Young's modulus of solid carbon E, was estimated to be 2.6 GPa. Using this value for Es the Young's modulus of carbon foams of various solidities was calculated using the FE model and they are shown in Figure 4.7. C Ux c/4 n pal yUy = J4" z Ux/4 =c Figure 4.6 Boundary conditions on the unitcell surfaces and maximum principal stress distribution when the unitcell is stretched in the ydirection a~n~ 42 The shear modulus was also calculated using the micromechanics analysis developed by Marrey and Sankar [4]. The variation of shear modulus as function of solidity is shown in Figure 4.8. Elasiic Modulus of carbon foam for a beam model and a solid model 250 (Es = 2.6 GPa) Ca a O 1u 201 E o 0L 10' o 5 . 5c 0 * 0 .) ....................... .......... .......... .... ....... ....... .. .......... ,'/ Experimental Pesult 0 .................. ......... E* = 124 MPa Beam model , 0  Solid Model i  ........ 0 u U.U3 U. IU U. .1 U.2zU U.5 Relative Density Figure 4.7 Elastic modulus E* as a function of relative density (solidity) for E, Shear modulus G* of foam models using beam elements and solid elements (Es = 2.6GPa) 3.0E+07 2 5E+07 2 0E+07 1 5E+07 1 OE+07 5 E+06 U.Ut+UU  0.00 0.10 0.15 Relative Density = 2.6 GPa. Figure 4.8 Shear modulus G* as a function of relative density (solidity) for E, = 2.6 GPa. / Solid Element y= 1 82E+09661           , . <   ,., Beam Element .' y = 2.78E+08 20 c Unit Cell of Carbon Foam Beam Model In addition to the solid model described in the preceding section a simple lattice model was also attempted. In this model the foam is assumed to be made up of struts arranged in a cubic lattice pattern. The length of the strut was equal to the unitcell dimension c in Figure 4.9. The struts were assumed to be uniform square cross section beam and their dimensions were determined from the relative density of the foam. Figure 4.9 Unit cell of beam model From the Eq. (2.11), the wall thickness h is determined to be 0.4086 mm for a solidity of 0.1312 and c = 1.8 mm. Since this beam model is composed as a simplest shape, FE methods are not necessary to determine the relation between the solid properties and foam properties. Analytical expressions for various elastic properties have been derived earlier section. Based on the Eq. (2.1) and Eq. (2.4), the Young's modulus Es and the strength ous are found to be 2.4 MPa and 69.5 MPa, respectively. The variation of E* and G* with the solidity for the beam model are shown, respectively, in Figures 4.7 and 4.8. Fracture Toughness Estimation of the Solid Model In this section, we describe a finite element based micromechanics model to estimate the fracture toughness of the cellular solid. The crack is assumed to be parallel to one of the principal material axes. The crack is created by breaking the ligaments of the unit cell (Figure 4.10). To determine the fracture toughness, a small region of the foam around the crack tip is modeled using finite elements. Only Mode I fracture is considered in the present study. The boundary of the cellular solid is subjected to displacement boundary conditions (ux and uy) corresponding to a unit KI., i.e., KI =1. The displacement components in the vicinity of a crack tip in an orthotropic solid can be found in Appendix B. The maximum tensile stress in the unitcells corresponding to unit stress intensity factor is calculated from the FE model. In the case of threedimensional solid model the maximum stress is obtained as an output of the FE program. From the result the value of KI that will cause rupture of the strut is estimated, which then is taken as the fracture toughness of the cellular solid. The solid model used 42 cells with 135,000 solid tetrahedral elements as shown in Figure 410. The maximum principal stress distribution is shown in Figure 4.10. When unit KI was applied to the crack tip, maximum principal stress was found equal to 1,463 Pa. Then the fracture toughness is obtained from the strength of the solid carbon (o, = 0.162 MPa) as KIc = 162x106 = .11MPa[m (4.12) max 1463 Comparing Eq. (412) with the results of fracture toughness experiments (Table 4.3), the difference is about 16%. One reason for the difference could be the small number of cells used in the FE simulation. The FE model was used to study the variation of fracture toughness with relative density and the resulting relationship is shown in Figure 4.11. Gibson and Ashby [2] provide analytical results for fracture toughness for opencell foam as given below: KC =0.65 (4.13) where ,us is the tensile strength of the solid. Using the above formula the fracture toughness for the carbon foam considered in this study can be obtained as 0.162 MPa?/m. The variation of KI, according to Eq. (413) is presented in Figure 4.11. At. K  Solid Model Beam Model Figure 4.10 Maximum principal stress distribution of solid and beam models for a unit KI at the crack tip Fracture Toughness of the Beam Model The procedure for simulating fracture using the beam model is the same as that for the FE solid model. The beam model of carbon foam (Figure 4.10) consisted of 46 10,000 cells using approximately 20,000 beam elements. When using the beam model, the rotational degree of freedom at each node of the beam element on the boundary of the solid is left as unknown and no couples are applied at these nodes. The maximum principal stress distribution in the beam model for KI=1 is shown in Figure 4.10. The maximum principal stress for a unit KI was found to be equal to 506 Pa. Therefore, the fracture toughness can be estimated as KI 69106 0.137MPa m (4.14) omax 506 The difference between the experimental result and that from the beam model is only 3%. The beam model was also used to study the variation of fracture toughness with the solidity and it is presented in Figure 4.11. Fracture toughness with various relative density 3.0E+05  2.5E+05   Beam model E pibson and Ashby (1988) /Kc = 1.28x10 (p*0p)1. 0 Kic = 3.39X106 (p, p )15o 2 .0+   ,  , y , t *,' ,*  Experimental Result ,. SStd deviation = 11,194 .: r' 1.5E+05 m e a R s.  .   . S1.0E+405  Solid Model S Kic =1.43x10(p*/p) 7.86x104 .2 5.0E4104 O.OE+00   0.07 0.09 0.11 0.13 0.15 0.17 0.19 0.21 Relative density (p*/p) Figure 4.11 Variation of fracture toughness of carbon foam with relative density CHAPTER 5 RESULTS AND DISCUSSION Elastic Constants of Opencell Foam Young's modulus for carbon foam in the principal material direction is estimated to be 2.09 GPa from the FE micromechanics and 2.07 GPa based on analytical solutions. Shear modulus is predicted as 10.245 MPa from FE analysis and 10.350 MPa from analytical solution. Difference between analytical and numerical results is 0.96% for elastic modulus and 1.01% for shear modulus. For opencell foam model, analytical solutions and FE solutions for elastic constants show good agreement. Numerical Analysis of Fracture Toughness A Finite Element method based micromechanics method was developed to predict the fracture toughness of cellular materials. A portion of material around the crack tip was modeled using finite elements. Boundary displacements calculated using orthotropic fracture mechanics were applied to the FE model. The stress intensity factor corresponding to the failure of the crack tip strut was taken as the fracture toughness of the cellular medium. It has been found that the fracture toughness is a strong function of relative density of the foam, however it also depends on the strut dimensions and spacing. Based on the FE results an analytical model was developed to predict the fracture toughness. A simple empirical formula has been derived for the effective length ahead of the crack tip that contributes to the crack tip ligament forces and bending moment. For the type of cellular medium considered in this study the fracture toughness can be expressed in the form of power law of the type where K is fracture toughness and R is relative density. The constants a and b for Mode I and Mode II are presented in Table 5.1 Table 5.1 Constants of fracture toughness curves Constant length of cell edge with Constant wall thickness with various wall thickness various length of cell edge A b a b Mode I 1.961x107 1.045 7.82x106 0.788 Mode II 6.95x106 1.32 2.76x106 1.070 The effective length I ahead of the cracktip is expresses as l=ac. The constant a was evaluated analytically using the results for fracture toughness obtained using the FE micromechanics models. It has been found that a depends only on relative density and is independent of cell length or strut cross sectional dimensions. Fracture Toughness with an Angle Crack Variation of strength under combined loading for opencell foams was evaluated analytically. The results were used to determine the tensile and shear strength when the loads are applied at an angle to the principal material directions. FE micromechanics simulations were performed for inclined cracks. Mode I fracture toughness is maximum at 0 and Mode II fracture toughness is maximum at 450 to the principal material direction. Fracture Toughness of Carbon Foam Four point bend tests were performed on SENB specimens made of carbon foam, and their Mode I fracture toughness was measured. In addition to the experimental approach, a finite element based micromechanics has been developed to predict the fracture toughness. Two micromechanical models were developed to simulate Mode I fracture. Both models assumed a cube as the unitcell of the foam. In the first model solid finite elements were used to model the foam. The measured density of the carbon foam was used in determining the void size in the micromechanical model. Young's modulus and tensile strength of the solid carbon were also determined from the corresponding values of the carbon foam measured experimentally. A small region surrounding the crack tip was modeled using finite elements. The crack was assumed parallel to one of the principal material directions. Boundary displacements were calculated using linear elastic fracture mechanics for orthotropic materials. From the FE simulation the stress intensity factor KI that will cause the failure of the cracktip elements was determined, and this was taken as the fracture toughness of the cellular material shown in Table 5.2. The agreement between the test results and numerical results are good indicating micro mechanics can be a powerful tool in predicting the fracture behavior of foams and other cellular solids. Table 5.2 Results of fracture toughness Kic ( MPa m1/2) % Difference from Experiment Beam Model 0.137 3 Solid Model 0.110 17 Gibson & Ashby 0.162 21 Experimental 0.132 APPENDIX A ANALYTICAL METHOD TO ESTIMATE THE SOLIDITY OF SOLID MODEL Volume of Top porlion.VB Sphere Volume inside of a cube Volume, VA Figure A. 1 Unit cell of solid model To estimate the solidity of porous medium in the solid model, the volume of the pore (void) left inside the unitcell needs to determined. In order to do that, the volume of top potion VB in Fig. Al is subtracted from entire sphere volume of a void. The volume VA is obtained as (A.1) V = R3 3 where Y is solid angle. The solid angle can be obtained as follows: dA VR=2 dA = 2ir drcosO 2irrdrcosO 2irrhdr h +r2 (h2 +r2) dA h +r2 (A.2) (A.3) (A.4) Integrating both sides, b b = 2 ;irrhdr o o (h +r2)( V=: 2zh I (A.6) 1h R By substitute Eq. (A6) into Eq. (Al), VA is obtained as follows; V = R3 = 2hR3 I 2rR [Rh] (A.7) 3 3 h R_ 3 The volume of the top portion VB can be obtained by subtracting the volume of the cone from VA 2iR 1 2ziR 3 1 3 VB, = [Rh]Ib 2zR2h+lh3 (A.8) [3 3 3 3 The volume of the pore left inside the unitcell is derived as Vc, 47rR3 47rR3 3 2 a3 V, = 6VB = 4rR3 +3rR2Ra (A.9) 3 3 4 Therefore, the solidity can be obtained from its definition as: p* a3 V V 1_ 4zR3 1 1= =1 3 4IR 3+3R2aa3] (A.10) p a a3 a 3 4 For open cell model, the equation is only valid in the range where R is the radius of the sphere in Fig. Al. After simplifying the equation, the solidity can be expressed in terms of the ratio R/a in the form of a polynomial: p 4+7T 8 CR}3 R}2 p 4+ +(R 3R (A.11) p, 4 3 \a) a APPENDIX B CRACK TIP DISPLACEMENT FIELDS FOR ORTHOTROPIC MATERIAL The opencell foam is considered as a homogeneous orthotropic material. The principal material directions are parallel to the 1 and 2 axes. Assuming the Poisson's ratio is negligible, the stressstain relations in the 12 plane are given in the matrix form below. 1 0 0 Er s \=[S[= a (B. 1) 2'12 1 Z'12 *G * When the cellular medium is oriented at an angle to the principal material axes, the stressstain relation can be transformed from the 12 plane to the xy plane by using transformation matrix [7]. The angle 0 is the angle the 1 direction makes with the x axis. cos20 sin20 2sin0cos0 [T] sin2 0 cos2 2sin0cos0 (B.2) sin0cosO sin0cosO cos2 sin2 0 7y = [T R 2 (B.3) By applying the transformation, the compliance matrix in the xy coordinates is written as Ex 7x S11 S12 S16 ox EY = [T [S] [T] a, = S12 S22 S26 a, (B.4) sy xy S16 S26 S66 xy The components of the matrix can be obtained in terms of compliance terms referred to the 12 axes as follows: S11 = Si, cos4 0 + (2S1, + a ) sin2 cos2 + S22 sin4 0 (B.5a) S12 = S12 (sin4 0 +cos4 0) + (S,, + S22 S )sin2 Ocos2 0 (B.5b) S22 = Si, sin4 + (2S2 + S66) sin2 Ocos2 + S2 cos4 0 (B.5c) S16 = (2S, 2S12 S66)sincos3 (2S22 2S12 S66)sin3 cos9 (B.5d) S16 = (2S, 2S2 S66) Sin3 cos (2S22 2S12 S66)sin cos3 0 (B.5e) S11 = 2(2S,, + 2S22 4S2 S66) sin2 cos2 + S~ (sin4 0+ cos4 ) (B.5f) The characteristic equation of the orthotropic material is given by Sllu4 22S +(2S12 + S66)2 2S2601 + S22 = 0 (B.6) where imaginary roots of the characteristic equation are described as (j = 1, 2, 3, 4). The constant (j = 1, 2) can be obtained by using the equation below. The roots and are unequal roots with a positive conjugate value. S, = //, = a + ip, s, = P/2 = a, + i2 The constants, (j = 1, 2) are relates to constants s as shown below. 2 2 P1 = a11s1 +a12 a16s 2 a11s2 + a12 a16s2 2 2 al2s, +a22 a26s a12s2 +a22 a26s2 ql = 2 = S1 S2 Stress components of polar and rectangular coordinates at a crack are shown in Figure B.1. The angle 0 is taken positive in the counterclockwise direction. The distance r is taken from the crack tip at the point O. 54 I I NE RCK I I Figure B. 1 Polar coordinate oriented from crack tip The displacement components for Mode I and Mode II can be defined as The displacement filed near the crack tip for Mode I: u. = K 2r Re (o sp(Cos2+ssOf s 2iI (COS+s sin )1/ (B.7) UY = KI 2r Re I [sq2(cos +s2 sinO) /2s2q,(cos0+s, sin )y/2 Z sI s 2 The displacement filed near the crack tip for Mode II: K 2r p o sin y/2 p, (cos 0 + s, sin Of2 S=Re [2(o0+ S(B.8) I i Z sI 2 L_J 2Z Sr \ 2 l ii REFERENCES 1. Fowlkes, Charles W., "Fracture Toughness Tests of a Rigid Polyurethane Foam," International Journal ofFracture, Volume 10, No. 1, March 1974, pp. 99108. 2. Gibson, L.J. and Ashby, M.F., Cellular Solids: Structure and Properties, Second Edition, Cambridge University Press, Cambridge, United Kingdom, 1988. 3. Hellan, K., Introduction to Fracture Mechanics, McGrawHill, NY, 1984, p. 244. 4. Marrey, Ramesh V. and Sankar, Bhavani V., "Micromechanical Models for Textile Structural Composites," NASA contractor report, 198229, October 1995. 5. Sih, G.C. and Liebowitz, H., "Mathematical Theories of Brittle Fracture," FractureAn Advanced Treatise, Volume 2, Liebowitz, H., Ed., Academic Press, New York, 1968, pp. 67190. BIOGRAPHICAL SKETCH Sukjoo Choi was born in Seoul, Korea, on 1972. He received his Bachelor of Science degree in aerospace engineering in 1997 from the University of Minnesota at Minneapolis. He worked as a mechanical engineer in control surfaces and mechanics at the Cirrus Design, Corp., Duluth, Minnesota from 1997 to 1998. In 1998, he joined the military in Korea and served until 1998. In 2000, he was admitted to the graduate program in the Department of Aerospace Engineering, Mechanics & Engineering Science at the University of Florida, Gainesville, Florida. Sukjoo Choi will be pursuing his doctoral studies in the filed of composite materials and structures in the Department of Mechanical & Aerospace Engineering at the University of Florida. 