Вы находитесь на странице: 1из 9

Th.

Zisis1
e-mail: zisis@metal.ntua.gr

A. Kordolemis A. E. Giannakopoulos
Department of Civil Engineering, Laboratory for Strength of Materials and Micromechanics, University of Thessaly, Volos 38336, Greece

Development of Strong Surfaces Using Functionally Graded Composites Inspired by Natural TeethFinite Element and Experimental Verication
Functionally graded materials (FGMs) are composite materials that exhibit a microstructure that varies locally in order to achieve a specic type of local material properties distribution. In recent years, FGMs appear to be more interesting in engineering application since they present an enhanced performance against deformation, fracture, and fatigue. The purpose of the present work is to present evidence of the excellent strength properties of a new graded composite that is inspired by the human teeth. The outer surface of the teeth exhibits high surface strength while it is brittle and wear resistant, whereas the inner part is softer and exible. The specic variation in Youngs modulus along the thickness of the presented composite is of particular interest in our case. The present work presents a nite element analysis and an experimental verication of an actual composite with elastic modulus that follows approximately the theoretical distribution observed in the teeth. DOI: 10.1115/1.3184038 Keywords: functionally graded composites, surface strength, nite element analysis, FEA, experiments

Introduction

Functionally graded materials FGMs are characterized by a specic distribution of certain properties. The property distribution is the result of a compositional and structural variation over volume. Various approaches based on particulate processing, preform processing, layer processing, and melt processing are used to fabricate the FGMs, resulting in layered materials designed for specic applications. In recent years, FGMs have become the point of interest in engineering due to their performance against deformation, fracture, and fatigue 1 , while several FGMs are manufactured by two phases of materials with different properties. Accordingly, since the volume fraction of each phase gradually varies in the gradation direction, the effective properties of FGMs change along this direction. An important benet of graded materials is the suppression of crack propagation through redistribution of stresses at surfaces or interfaces subjected to mechanical and tribological loading 2 . The present paper builds upon a previous work by Giannakopoulos et al. 3 , where the problem of the application of a normal or tangential line load upon a semi-innite layered elastic substrate was presented. It was assumed that the layered substrate presented a variation in Youngs modulus that follows a distribution of the type E = E0y 0 / y. This type of Youngs modulus distribution emanates from the human teeth, which is characterized as a hard and tough functionally graded composite 3 . The outer surface of the teeth exhibits high surface strength while it is brittle and wear resistant, whereas the inner part is softer and exible. In their work, the analytical stress and distribution elds were presented and compared with the analytical expression given from
1 Corresponding author. Contributed by the Materials Division of ASME for publication in the JOURNAL OF ENGINEERING MATERIALS AND TECHNOLOGY. Manuscript received January 14, 2009; nal manuscript received May 13, 2009; published online November 5, 2009. Review conducted by Hussein Zbib.

Flamants solution. It was shown that the normal loading leads to compressive layer at the surface that can effectively shield the propagation of possible pre-existing cracks. Here, the same elastic distribution is assumed, and the analytical results are compared with numerical solutions. Furthermore, experimental evidences that support the theoretical conclusions are presented.

2 Analytical Formulas for the Loading of a SemiInnite Substrate


The problem of a line load acting on the surface of an elastic homogeneous semi-innite substrate Fig. 1 has been examined previously, and is referred to as the Flamant problem 3 . The stresses in the Cartesian coordinates are
xx

rr

2P sin2 = 2P cos2 =

x2y x2 + y 2 y3 x2 + y 2

2Q

x3 x2 + y 2 xy 2 x2 + y 2

yy

rr

2Q

2.1

xy

rr

2P sin cos =

xy 2 x + y2
2

2Q

x2y x + y2
2

Accordingly, the Cartesian displacements ux , uy are given by ux = P 1+ E P 1 xy 2 x +y E


2

arctan

x y 2.2

x2 + y 2 2P v + 1 x2 uy = 2 2 + ln 2 x +y E y where y is the thickness of the substrate. Giannakopoulos et al. 3 showed that the stress eld in a functionally graded composite for the normal load in the Cartesian coordinates Fig. 1 takes the following forms: JANUARY 2010, Vol. 132 / 011010-1

Journal of Engineering Materials and Technology Copyright 2010 by ASME

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 1 The applied vertical P and horizontal Q line loads acting at the surface of a semi-innite substrate

xx

rr

sin2 =

P P x2 sin2 = r sin y x2 + y 2 P y2 cos2 = y x2 + y 2 2.3


Fig. 2 a Region of the mesh used for the nite element simulations with the appropriate boundary conditions. b Blown up view of the mesh around the center line. The mesh comprises 95517 nodes and 31600 8-node biquadratic plane strain quadrilateral elements CPE8 in ABAQUS notation. Minimum element length: he = 0.12 m.

yy

rr

P P xy sin cos = xy = y y x2 + y 2 Furthermore, the Cartesian displacements ux , uy are ux = P x+ 1+ E0y 0 y arctan x y 2.4 P y 1+ uy = E0y 0 x Py x arctan + y E0y 0

where y 0 is a reference depth at which E = E0 and y is the thickness of the substrate. Note that for plane strain, the following replacements must be taken into consideration: E E 1
2

2.5 1

tances, which are more than three elements around the point of application of the force. A mesh sensitivity study revealed that adequate accuracy is achieved by a nite element FE mesh that comprises about 169,000 nodes and 56,000 8-noded quadrilateral plain strain elements CPE8 in ABAQUS notation . The displacement boundary conditions were: uy = 0 along the bottom of the mesh, ux = 0 for the middle node of the bottom surface of the mesh, and a traction free side boundary. The substrate was loaded at the middle of the top surface by a line force of 1 N/m, acting either normally or tangentially. The minimum element size around the region of the line force was he = 0.12 m. Results are presented for mesh dimensions equal to 6500he 243he. Furthermore, all lengths were in meters m and stresses in Pascal Pa . 3.2 Material Parameters. We performed two sets of calculations. The rst set comprised of three calculations for normal, and three for tangential line load, according to which we considered the substrate to be homogeneous isotropic with Youngs modulus E = 0.666 Pa, 1 Pa, and 5 Pa, and Poisson ratio = 0.3. Then, in a second set of calculations, the substrate was assumed to comprise a stack of layers of equal thickness l. Each layer was treated as homogeneous isotropic with a linear elastic response with Youngs modulus E. In the case of the layered substrate, the Poisson ratio insignicantly affects the attained stress eld 5 , so we assumed it constant and equal to 0.3 for all the layers. The Youngs modulus varied with respect to depth y according to E = A / y, where A = E0y 0. Accordingly, the introduced model described a substrate that comprised of a top surface, which was very stiff, while the bottom of the substrate was very compliant. Simulations were carried out for E0, which is equal to 0.666 Pa, 1 Pa, and 5 Pa, and for layer thickness l = 1 m, 0.5 m, and 0.25 m, while y 0 = 1 m for all the cases considered here, unless otherwise stated. Then, y varied with the thickness as described earlier Fig. 3 . The zero value of y corresponds to the upper surface of the substrate. Preliminary FE simulations, veried that six layers are adequate in Transactions of the ASME

Finite Element Simulations

In the current section, we perform a set of nite element simulations. The extracted stress and displacement elds are presented and compared with the previously presented analytical formulas. The possibility of the propagation of preexisting cracks in the semi-innite composite substrate is explored. 3.1 Finite Element Formulation. In order to approach the problem of a substrate loaded numerically by a line force, we construct a two-dimensional substrate under plane strain conditions. The commercial nite element program, ABAQUS Standard 4 , was employed, and a typical mesh used for the simulations is shown in Fig. 2. Full integration scheme was used. No special types of elements were used, either close to the line force socalled singular elements or at innity so-called innite elements . It is recognized that at distances very close to the force, the solution is not accurate, and only in between the point of application of the force and the outer boundary is sufcient accuracy expected. The mesh was gradually rened to yield results in stresses that converged with available analytical results at dis011010-2 / Vol. 132, JANUARY 2010

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 3 Variation in the continuous Youngs modulus E, used in the FE model, as a function of the depth y for layer thickness a l = 1 m, b l = 0.5 m, and c l = 0.25 m E0 = 1 Pa, y0 = 1 m

Fig. 4 Stress distribution along a path on the x-axis at depth equal to y / y0 = 1.95 from the surface of the substrate. Analytical and FE results for a normal line load and b tangential line load for homogeneous substrate l = 0.25 m, E0 = 1 Pa, y0 = 1 m.

order for the solution with discrete layers to approach the continuous variation in the elastic modulus solution. Furthermore, according to equation E = E0y 0 / y, an innite Youngs modulus corresponds to the top layer, but for numerical purposes, a value of Youngs modulus equal to E0y 0 / l at y = 0 was assigned for all the cases considered here. Thus, as shown in Fig. 3, for l = 0.25 m and E0 = 1 Pa, for example, Youngs modulus at the rst layer will be 4 Pa. 3.3 Finite Element Results. First, we present the results for the homogeneous substrate that is loaded normally and tangentially. From the analysis, it is clear that Youngs modulus E serves as a simple amplication factor for the strains and displacements, but does not affect either the distribution or the magnitude of the stress elds, so results for the case of E = 1 Pa are shown, unless otherwise stated. Figure 4 presents the stress distribution along a path on the x-axis at depth equal to y = 1.95 from the surface of the substrate. Analytical results according to Eq. 2.1 are shown, combined with the FE solution for normal and tangential line load. An excellent agreement is observed for all the presented stress components for both types of loading verifying the validity Journal of Engineering Materials and Technology

of the FE model. Accordingly, results for the case of a graded substrate with l = 0.25 m and E0 = 1 Pa, are shown in Fig. 5. Again, the stress distribution along a path on the x-axis at depth equal to y = 1.95 from the surface of the substrate is presented, and analytical results from Eq. 2.3 are compared with the FE solution for the case of normal line load. A very good agreement is obtained and is concluded that Eq. 2.3 can accurately capture the response of the layered substrate upon normal line loading. Figure 6 presents the normalized vertical and normal displacements ux and uy at y / y 0 = 0.65 y = 1.95 m and y 0 = 3 m for the case of a homogeneous and a FGM substrate. Analytically, Eqs. 2.2 and 2.4 , and numerical results for both cases, are presented. For the case of the FGM substrate, we considered a layer thickness l = 3 m, a Youngs modulus E0 at y 0 = 3 m equal to 1000 Pa, and a substrate thickness y = 243 m. We remind the reader that the increase in Youngs modulus by a factor of 1000 does not affect the attained stress eld but only the magnitude of the attained displacement eld. It is observed that for the homogeneous case, a good agreement is attained between analytical and FE JANUARY 2010, Vol. 132 / 011010-3

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 5 Stress distribution along a path on the x-axis at depth equal to y / y0 = 1.95 from the surface of the substrate. Analytical and FE results for a normal line load and b tangential line load for FGM substrate l = 0.25 m, E0 = 1 Pa, y0 = 1 m.

Fig. 6 Displacement distribution along a path on the x-axis at depth equal to y / y0 = 0.65 from the surface of the substrate. Analytical and FE results are shown for a vertical displacements ux and b normal displacements uy for homogeneous and FGM substrate y = 1.95 m, l = 3 m, E0 = 1000 Pa, y0 = 3 m, y = 243 m.

results. For the case of the FGM substrate, the analytical and FE results present a good qualitative agreement. Quantitatively, they diverge by a factor of about 2. This is due to the compliance of the graded substrate when compared with the homogeneous one, due to second order factors that have not been taken into account into the analytical solution, and nally due to the fact that the displacement eld in a two-dimensional problem has multiple solutions depending on the way that the far eld displacements are constrained. Figure 7 presents contours of equivalent, in-plane normal and shear stresses, at a domain close to the region of the applied normal load. Contours correspond to the mesh at the undeformed conguration, and results are shown for homogeneous and layered substrate with l = 1, 0.5, and 0.25 m. It is concluded that the stress elds shrink normally to the region of the applied load while they expand laterally for reduced layer thickness Fig. 7 a and 7 b . Contours of in-plane yy and xy stresses show that the stress eld is shrinking close to the surface without laterally expanding as the 011010-4 / Vol. 132, JANUARY 2010

layer thickness reduces, but the effect is minor. Furthermore, we note that the xx, and yy stress elds are compressive so a microcrack advance along the x- or y-axis is judged unlikely due to shielding. Figure 8 presents contours of undeformed shape of equivalent, in-plane normal and shear stresses for a layered substrate with l = 1, 0.5, and 0.25 m. Here, we examine the case of tangentially applied load. The homogeneous case is added for reference. As expected, contours of in-plane xx and yy show a compressive stress eld advancing in front of the point of application of the tangential force, while a tensile stress eld forms behind. Again, the attained stress elds shrink normally to the applied load while they expand laterally. Only the xy stress eld seems to shrink in both directions with reduced layer thickness. The tensile stress eld behind the load of application supports the possibility of the advance of a preexisting crack for the case of a homogeneous substrate. Nevertheless, due to the concentration of the attained stress eld that is very close to the surface, with reducing layer Transactions of the ASME

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 7 a Equivalent stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, normally loaded with line force of 1 N/m; b in-plane xx stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, normally loaded with line force of 1 N/m; c in-plane xy stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 P and y0 = 1 m, normally loaded with line force of 1 N/m; d in-plane yy stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, normally loaded with line force of 1 N/m

Journal of Engineering Materials and Technology

JANUARY 2010, Vol. 132 / 011010-5

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 8 a Equivalent stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, tangentially loaded with line force of 1 N/m; b in-plane xx stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, tangentially loaded with line force of 1 N/m; c in-plane xy stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, tangentially loaded with line force of 1 N/m; and d in-plane yy stress eld for i homogeneous ii graded with layer thickness l = 1 m, iii graded with l = 0.5 m, and iv graded with l = 0.25 m substrate, with E0 = 1 Pa and y0 = 1 m, tangentially loaded with line force of 1 N/m

011010-6 / Vol. 132, JANUARY 2010

Transactions of the ASME

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

thickness, it is concluded that the tangential loading does not develop detrimental tensile stresses, and the possibility of the advance of a pre-existing crack minimizes, as expected from the theoretical limit.

Table 1 The mechanical properties of the materials used for the composite specimens 6,7 Properties Elastic modulus GPa Compressive strength MPa Tensile strength MPa Density kg/ m3 Bitumen 40/50 1 6 1.3 1041 Aggregates calsius rock 4075 80120 2700

4 Materializing the Stiffness Distribution as Volume Fraction of Particles


Hashin and Shtrikman 6 and later, Wallpole 7,8 , provided very accurate upper and lower bounds of the elastic properties of composites, based on the second-phase volume fractions and the elastic properties of the constituents 9 . Paul 10 derived some simple and direct approximate formulae for the elastic composite modulus, and Ravichandram 11 gave more rened approximations. More recent approximate average properties for the composites are based on the Mori and Tanaka approach 12 . All the above formulae pertain to composites that are homogeneous in an average sense. Long range inhomogeneous composites can be modeled by assuming spatial variability of the composition 2 . Next, we will present some examples of such approach. Budianski 13 developed an elegant self-consistent composite theory. In the limit case of a composite made of rigid spherical inclusions and an incompressible matrix, the effective elastic properties give an incompressible material with effective shear modulus =
m

5 1 c 2

ci

2/5

4.1

where m is the shear modulus of the matrix, and ci is the volume concentration of the rigid inclusions assumed perfectly bonded with the matrix . Interestingly, for ci = 0.4, the shear modulus is innite. Using Eq. 4.1 , the elastic modulus distribution of the type E = E0y 0 / y can be reduced to a concentration composition ci y = 2 Emy 1 5 E0y 0 4.2

where Em = 3 m is the elastic modulus of the incompressible may . Note that the composition decreases lintrix, and E y = 3 early from ci = 0.4 at y = 0, down to ci = 0 at y = E0y 0 / Em. If H is the total layer thickness, we must have HEm E0y 0 1 4.3

The stiffening of an incompressible matrix by rigid particles has been addressed by Suquet 14 . S = 1 + cp 21c p 1 c2/3 2 1 cp p 5 10 1 c7/3 p
2 1

4.4

phalt mixes, with volume fractions of aggregate ranging from 40% to 85%. The aggregate stiffens the bitumen and it was found that the experimental results compare with a modied version of Hashins composite sphere model. For high values of the bitumen stiffness 1 GPa , the bitumen behaves as an elastic solid. At large strains, the response of the bitumen is nonlinear. For the 50 pen bitumen at 20 C and at stresses above 0.1 MPa, the bitumen can be approximated as a power law material 17 . A mold was manufactured in order to cast and compact the specimens. An oil lubricant mixture of glycerine and valvoline was applied to the inner surface of the mold made from tin foils to prevent sticking of the bituminous mixture. The mixture consisting of the correct amounts of bitumen and aggregate clean, round stones of 10 mm diameter was transferred to the mold in layers, as follows: First, the bitumen was heated to 190 C in an oven for 90 min, and then the aggregates were poured in with the bitumen melt following right after. The ambient temperature was 9 13 C. The mechanical properties of the matrix bitumen and the particles aggregates are shown in Table 1. A wooden hummer was used as the dropping weight of 50 N, falling from a height of 1m and using 25 drops per layer. Air voids were of the order of 4%. Each specimen had dimensions of 1500 480 300 mm3. The specimens were constructed, each with 6 layers of constant thickness of 80 mm. Each layer was made by local mixing of the aggregates and the bitumen in two sublayers. The total time for the casting was around 8 h for the homogeneous specimen, and about 12 h for the graded specimen. In order to keep the surface strength characteristic similar to the two specimens, the surface layer was made the same. The rst specimen was constructed to have constant elastic modulus E E0 by keeping a constant percentage of the aggregate within each layer. The second specimen was constructed to have a variable elastic modulus according to E = E0y 0 / y. This was achieved by the gradual change in the percentage of aggregates within each layer. In order to achieve the variation in the modulus, we used a rule of mixture that is pertinent to the materials we used 16 . 2.5 ca E = 1+ Ebit p 1 ca
p

where c p is the volume fraction of the particles. Enamel is made of needlelike mineral crystals, 1520 nm thick, and 1000 nm long with a ratio of mineral to collagen rich protein matrix on the order of 1:2. Gao et al. 15 gave an estimate of the composite modulus E of such microstructures. 1 12 1 cm 1 + = 2 E c mE c E pc mk 2 4.5

5.1

where E is the composite modulus, Ebit is the elastic modulus of the bitumen, ca is the volume percentage of the aggregates, and p is a parameter given by p = 0.83 log 4 1010 bit Pa 5.2

In Eq. 4.5 , cm is the volume concentration of the mineral, k is the aspect ratio of the mineral plates 40 in the case of enamel , E p is the elastic modulus of the mineral 114 GPa in the case of hydroxyapatite , and Ec is the elastic modulus of the protein 4.3 GPa for keratine . It is clear that if any of the above parameters becomes a function of position, so does the composite modulus.

Experiments

The amount of material used for each specimen is shown in Tables 2 and 3, respectively. The thermal expansion of the bitumen is 2 104 K1. As the bitumen is cooled down from 190 C to 20 C, a 12% volume shrinkage could lead to tensile stresses at the bitumen-aggregate interface. However, creep strains develop, and given sufcient time, they will negate the thermal strains. The bitumen follows Dorns law of creep 18 , and the creep strains are given by
c

5.1 Material Preparation. Deshpande and Cebon 16 performed uniaxial compression experiments on various idealized asJournal of Engineering Materials and Technology

= sst + 0.072 1 exp 26 sst

5.3

JANUARY 2010, Vol. 132 / 011010-7

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 2 Quantities of material used for the homogeneous specimen E = E0 Layer from surface 1 2 3 4 5 6 Total Aggregates kg 162 162 162 162 162 162 972 Bitumen kg 832 832 832 832 832 832 4992 Fig. 10 Overall view of the specimens, loading and layers

2.4

ss = 3.73 104 s1

10 Pa

exp

228 103 J mol1 8.314 J mol1 K1 T 5.4

where c is the creep strain, ss is the rate of thermal strain, is the tensile stress that develops at the aggregate-bitumen interface about 2 kPa , t is the time in seconds, and T is the average temperature 378 K . The minimum time needed for the creep strain to negate the thermal strain can be estimated from
c

12 1 = 4 102 100 3

This would require t = 28 h. Given that the specimens were tested thirty days after casting, we can safely assume that the thermal strains were completely relaxed, and no residual stresses remained into the specimens. 5.2 Specimen Loading and Observations. We designed a self-stressing truss that could apply small normal and high tangential load to the specimen. The truss is shown in Fig. 9. The general two-dimensional theory of Filon 19 assures the global equilibrium and stability of the specimen. The nodal application of the loads were done by using steel T-beams that were inserted to a depth of 4 cm, as shown in Figs. 9 and 10. The surface load was applied at three different points at each specimen at distances of 50 cm from each other, as shown in Fig. 9. The truss geometry was modied accordingly, to keep the loading the same.

For the homogeneous specimen, due to the dead weight of the bars of the truss, a concentrated load of 10 kp at the end of the truss was transferred to a distributed load of P = 333 N / m and Q = 2333 N / m. Upon application of an additional concentrated load of 50 kp, the specimen failed by developing a deep vertical crack, normal to the tangential load direction Fig. 11 . Obviously, the large tensile stresses of the tangential load are responsible for the expansion of the 4 cm initial slit, due to the insertion of the T-beam. We performed the same loading sequence to the graded specimen. The only damage that we observed was a local plastic twist around the surface point of the load application Fig. 12 . Note that the surface layer of the grade specimen was prepared in the same way as the homogeneous specimen. Furthermore, we assured that the loading was fast enough to avoid the so-called thermal softening effect, i.e., development of heat due to fast local plastic straining. In the bitumen used, thermal softening can be avoided provided the strain rate is smaller than 0.03 s1 17 . The largest linear stress that can develop in the bitumen is 2 kPa. This means that the elastic stresses must be smaller than 60 Pa/s. If the change in the tangential load is Q, then within the rst layer, the typical change in the stress is

Table 3 Quantities of material used for the graded specimen E = E0y0 / y Layer from surface 1 2 3 4 5 6 Total Aggregates kg 26 2423 2327 2247 218 21126 1389 Bitumen kg 1018 1137 1222 128 1319 1367 7343

Fig. 9 Side view of the specimens during loading. The selftension mechanism with P = F / b, Q = 7F / b, and b = 30 cm.

Fig. 11 Failure of the homogeneous specimen

011010-8 / Vol. 132, JANUARY 2010

Transactions of the ASME

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

normal loading creates a highly compressive surface layer that can shield the substrate from the opening of existing microcracks. It is concluded that the tangential loading does not develop detrimental tensile stresses, making the substrate particularly strong against frictional forces. The above ndings are supported by the damage mode observed from a series of experiments performed using steel T-shaped beams inserted into the surface of bitumen specimens under different types of loading. The present results can be very useful in the development of armor plates, special bearings, and transportation pavements.

References
1 Suresh, S., 2001, Graded Materials for Resistance to Contact Deformation and Damage, Science, 292, pp. 24472451. 2 Suresh, S., and Mortensen, A., 1998, Fundamentals of Functionally Graded Materials, Cambridge University, Cambridge, England. 3 Giannakopoulos, A. E., Kordolemis, A., and Zisis, Th., 2010, Development of Strong Surfaces Using Functionally Graded Composites Inspired by Natural TeethA Theoretical Approach, ASME J. Eng. Mater. Technol., 132 1 , p. 011009. 4 2001, ABAQUS version 6.2, Users Manual, Hibbitt Karlsson and Sorensen Inc., Pawtucket, RI. 5 Ke, L. L., and Wang, Y. S., 2007, Two Dimensional Sliding Frictional Contact of Functionally Graded Materials, Eur. J. Mech. A/Solids, 26, pp. 171 188. 6 Hashin, Z., and Shtrikman, S., 1963, A Variational Approach to the Theory of the Elastic Behavior of Multiphase Materials, J. Mech. Phys. Solids, 11, pp. 127140. 7 Walpole, L. J., 1966, On Bounds for the Overall Elastic Moduli of Inhomogeneous SystemsI, J. Mech. Phys. Solids, 14, pp. 151162. 8 Walpole, L. J., 1966, On Bounds for the Overall Elastic Moduli of Inhomogeneous SystemsII, J. Mech. Phys. Solids, 14 5 , pp. 289301. 9 Christensen, R., 1990, A Critical Evaluation for a Class of Micromechanics Models, J. Mech. Phys. Solids, 38, pp. 379404. 10 Paul, B., 1960, Prediction of Elastic Constants of Multiphase Materials, Trans. Metall. Soc. AIME, 218, pp. 3641. 11 Ravichandran, K. S., 1994, Elastic Properties of Two-Phase Composites, J. Am. Ceram. Soc., 77, pp. 11781184. 12 Mori, T., and Tanaka, K., 1973, Average Stress in Matrix and Average Elastic Energy of Materials With Mistting Inclusions, Acta Metall., 21, pp. 571 574. 13 Budiansky, B., 1965, On the Elastic Moduli of Some Heterogeneous Materials, J. Mech. Phys. Solids, 13, pp. 223227. 14 Suquet, P. M., 1993, Overall Potentials and Extremal Surfaces of Power Law of Ideally Plastic Composites, J. Mech. Phys. Solids, 41, pp. 9811002. 15 Gao, H. J. B., Jaege, I. L., Arzt, E., and Fratzl, P., 2003, Materials Become Insensitive to Flaws at Nanoscale: Lessons From Nature, Proc. Natl. Acad. Sci. U.S.A., 100, pp. 55975600. 16 Deshpande, V. S., and Cebon, D., 2000, Uniaxial Experiment on Idealized Asphalt Mixes, J. Mater. Civ. Eng., 12, pp. 262271. 17 Cheung, C. Y., and Cebon, D., 1997, Experimental Study of Pure Bitumen in Tension Compression and Shear, J. Rheol., 41, pp. 4573. 18 Amin, K. E., Mukherjee, A. K., and Dorn, J. E., 1970, A Universal Law of High-Temperature Diffusion Controlled Transient Creep, J. Mech. Phys. Solids, 18, pp. 413426. 19 Filon, L. N. G., 1903, On an Approximate Solution for the Bending of a Beam of Rectangular Cross Section Under Any System of Load With Special Reference to Points of Concentrated or Discontinuous Loading, Philos. Trans. R. Soc. London, 201, pp. 63155.

Fig. 12 Failure of the graded specimen

2Q y0

60 Pa/s

2 Q y0 t

60 Pa/s

5.5

In our case, the change in tangential load is Q = 4200 N and y 0 = 0.08 m. This would result in a time change of t 1.02 s. This time scale was less than the time scale of the load application, and so no thermal softening was experienced by the bitumen in the present tests.

Conclusions

In the present work, the analytical approach of the line loading of a semi-innite layered substrate was presented. The substrate was modeled with a varying Youngs modulus with respect to its thickness of the type E = A / y, where y is the depth from the surface. Accordingly, a nite element model was constructed and simulations for the line loading of a graded substrate were carried out. We found that FE and analytical results of the stress eld for a functionally graded material are in good agreement. It was shown that a composite substrate that approximates an elastic modulus distribution of the type E = A / y leads to almost zero tensile stresses due to the application of tangential surface loads. The

Journal of Engineering Materials and Technology

JANUARY 2010, Vol. 132 / 011010-9

Downloaded 30 Nov 2011 to 203.153.32.220. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Вам также может понравиться