Professional Documents
Culture Documents
Finite Element Modeling As A Tool For Predicting The Fracture Behavior of Robocast
Finite Element Modeling As A Tool For Predicting The Fracture Behavior of Robocast
Finite Element Modeling As A Tool For Predicting The Fracture Behavior of Robocast
com
Received 10 January 2008; received in revised form 13 May 2008; accepted 22 May 2008
Available online 5 June 2008
Abstract
The use of finite element modeling to calculate the stress fields in complex scaffold structures and thus predict their mechanical behav-
ior during service (e.g., as load-bearing bone implants) is evaluated. The method is applied to identifying the fracture modes and esti-
mating the strength of robocast hydroxyapatite and b-tricalcium phosphate scaffolds, consisting of a three-dimensional lattice of
interpenetrating rods. The calculations are performed for three testing configurations: compression, tension and shear. Different testing
orientations relative to the calcium phosphate rods are considered for each configuration. The predictions for the compressive configu-
rations are compared to experimental data from uniaxial compression tests.
Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
1742-7061/$ - see front matter Ó 2008 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actbio.2008.05.020
1716 P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724
Fig. 2. SEM micrographs showing the morphology of HA (a, c) and b-TCP (b, d) scaffolds after sintering at 1300 °C for 2 h: (a, b) printing plane view and
(c, d) cross-section view.
Fig. 4. Scheme of the different testing configurations simulated in this study: compression along (a) the printing direction or (b) a rod axis; tension along
(c) the printing direction or (d) a rod axis; shear (e) on the printing plane along a rod axis, or on a plane orthogonal to a rod axis along (f) the printing
direction or (g) a rod axis. The rigid plates used in the FEM simulations to transfer the load to the structure are indicated. The bottom plate is held fixed
while a load is applied to the top plate in the direction indicated by the arrows.
tests) – this implicitly assumes an opening fracture mode the structure was estimated as the maximum applied load
(mode I), which is that most commonly activated in brittle divided by the square external cross-section of the sample,
materials [17]. For the calculation of the maximum value of and the results were compared with the FEM predictions.
rt, the near-contact regions were ignored as they are not More than 10 samples were tested in each case in order
representative of the macroscopic behavior of the structure to get statistically reliable values.
under each load configuration (i.e., uniaxial tension, com-
pression or shear), but rather of the boundary conditions 3. Results and discussion
selected for the simulations. Furthermore, contact stresses
will only produce localized damage [14] and would not be Table 1 summarizes the results of the characterization of
responsible for the ultimate failure of the structure except the individual rods comprising the HA and b-TCP scaf-
under high-loading-rate, e.g., impact, conditions. folds. The elastic modulus, E, of each material was calcu-
Experimental uniaxial compression tests were per- lated from the effective modulus, E* = E/(1 m2),
formed along the aforementioned two directions (Fig. 4a obtained from the indentation tests, assuming a value of
and b) on cubic blocks of about 2 mm side cut from the sin-
tered HA and b-TCP scaffolds. The tests were carried out
on a universal testing machine (AG-IS10kN, Shimadzu Table 1
Intrinsic mechanical properties of HA and b-TCP rods
Corp., Kyoto, Japan) in air at a constant crosshead speed
of 0.6 mm min 1 (the rate of 30 mm min 1 used in bending E* (GPa) ma E (GPa) rF (MPa)
tests was impractical for the compression tests because it HA 83 ± 4 0.28 82 ± 4 68 ± 12
caused premature failure from contact damage and thus TCP 38 ± 8 0.28 36 ± 7 27 ± 9
a
a slower rate was selected). The compressive strength of From Ref. [16].
P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724 1719
0.28 [16] for the Poisson’s ratio. As can be clearly appreci- induce cracking perpendicular to the rod axes over the
ated, the HA rods exhibited more than twice the stiffness entire structure at those points. This type of cracking has
and strength of the b-TCP rods. indeed been observed in experimental compressive tests
These elastic properties were used as input for the FEM with this same orientation [14]. Of course, crack initiation
simulations of the different testing configurations depicted does not necessarily occur at the exact location of the max-
in Fig. 4. The FEM results confirmed that the stresses imum tensile stresses because it also depends on the loca-
developed in the scaffolds for each configuration are obvi- tion of the largest flaws [18].
ously independent of the elastic modulus of the material When the compressive load is applied along one of the
considered, as shown for the case of uniaxial compression rod axes (directions 1 or 2), the maximum tensile stresses
in Fig. 5, where the maximum of rt is represented vs. the are again located on the top and bottom surfaces of the
applied stress for both HA and b-TCP. Consequently, the horizontal rods (Fig. 6b), but now close to the joints with
analysis that follows of the stress fields generated and the the vertical rods (see the inset). The orientation of these
associated damage modes that are to be expected in each maximum tensile stresses is such that it will induce cracking
case is applicable to any scaffold with a similar geometry. (again, assuming mode I fracture) of the horizontal rods
Evidently, the strains do vary from one material to along those joints, effectively detaching them from the ver-
another, and the elastic modulus of the material compris- tical rods that remain largely in compression. This predic-
ing the individual rods will determine the effective modulus tion was also consistent with the curvature of the crack
of the scaffold. surfaces observed in the experimental tests [14]. The rela-
Fig. 6 shows the contours of the greatest tensile stress, tive magnitude of the maximum stresses in the two orienta-
rt, developed in the scaffold during uniaxial compression tions will be discussed in detail below.
at 250 N load (which is around the critical load for fracture For the case of uniaxial tension tests, Fig. 7 shows the rt
in the experimental tests) for the two orientations consid- contours developed in the scaffold at 250 N load (for com-
ered (Fig. 4a and b). When compression is applied perpen- parison with Fig. 6) for the two orientations considered
dicularly to the printing plane (Fig. 6a), i.e., along direction (Fig. 4c and d). Note that, since stresses increase linearly
3, rt presents a maximum at the centers of the unsupported with load (Fig. 5), contours at any other load can be calcu-
segments – specifically at the top and bottom surfaces of lated just by multiplying the stresses in the legend by the
those segments – of all the rods (although not so clearly vis- appropriate factor. When the loading is along direction 3
ible in Fig. 6a for those aligned along direction 2). At those (Fig. 7a), and ignoring the values near the contacts, rt pre-
points, rt is directed along the corresponding rod axis (i.e., sents its maximum values at the joints between adjacent
rt = r11 or rt = r22) and therefore, assuming an opening rod layers, as clearly shown in Fig. 7a for the rods aligned
fracture mode (mode I), which, as mentioned above, is along direction 1 but also present in those aligned along
the most commonly activated in brittle materials [17], will direction 2. At a certain load, these maximum tensile stres-
ses will induce fracture along those joints, separating each
layer from its neighbors. When the tension is applied along
one of the rod axes (directions 1 or 2), the tensile stresses
are largely located on the vertical rods (Fig. 7b), with their
maxima in the free segments of those rods close to (though
not exactly at) the joints with the horizontal rods (see
inset). The orientation of these maximum tensile stresses
is largely along the rod axes, so that cracks originated by
them will tend to break the vertical rods transversally.
Fig. 8 shows analogous contour plots of rt at 250 N load
for the three different shear configurations selected (Fig.
4e–g). For the sake of clarity, the direction of the applied
load is included in the figures (recall that the bottom sur-
face is kept fixed). Although the three configurations are
significantly different, the stress contours generated exhibit
many similarities. For all three orientations, the tensile
stresses organize into bands of alternating intensity ori-
ented at 45° with respect to the load axis. The maxima of
these stresses, disregarding the near-contact field, are
located at the joints between the different rod layers, the
Fig. 5. Plot of maximum tensile stress in the structure (maximum of rt) vs. edges at the intersections acting as stress concentrators.
the applied stress (applied load normalized by initial cross-sectional area)
Therefore, for the three configurations, cracking will occur
for both HA and b-TCP scaffolds in the two compression directions
considered: perpendicular to the printing plane (direction 3, Fig. 4a) and at these locations and the different rod layers will tend to
along the rod axes (direction 1 or 2, Fig. 4b). Note that the results are detach from each other. However, the actual cracking pro-
independent of the material. cess will be slightly different in the case of the configuration
1720 P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724
Fig. 6. FEM-generated stress contours corresponding to the highest tensile stress, rt, generated when a compressive load of 250 N is applied (a)
orthogonal to the rods (direction 3) or (b) along a rod axis (direction 1, equivalent to direction 2).
in Fig. 8c, where adjacent layers tend to rotate relative to configurations considered, the stresses are higher when
each other instead of simply one displacing the others, as the load is applied perpendicular to the printing plane
in the cases of Fig. 8a and b. (direction 3), but this situation is reversed under tensile
Concerning the relative intensities of the maximum loads, which yield higher stresses when applied along the
stresses developed in each of the configurations analyzed, rod axes (direction 1 or 2). The most deleterious configura-
Fig. 9 shows the absolute maximum value of rt anywhere tion analyzed is that corresponding to shear load applied in
in the scaffold vs. the applied stress (i.e., applied load nor- planes perpendicular to the printing plane, along the direc-
malized by the corresponding scaffold’s external cross-sec- tion of the rod axes (Fig. 4g). The torsion-like relative dis-
tion) for each testing configuration. This representation placement of the rod layers generates a stress field around
and the conclusions that derive from it are independent the joint that is significantly greater than in the two other
of the material considered and therefore have a universal shear configurations, which nonetheless both give stress
validity for any scaffold with the same morphology. As levels more than twice as large as those of the tension con-
can be clearly appreciated, there are very large differences figurations. An immediate, though anticipated, conclusion
in the maximum tensile stress levels that develop in the from these results is that brittle porous scaffolds fabricated
scaffolds depending on the type of load applied. As by robocasting, regardless of the material they are made of,
expected, compressive loads are the mildest, with stresses will exhibit far superior strength properties under compres-
about three times lower than those that develop in the ten- sive loads; tensile and shear stresses acting on them should
sion tests and more than an order of magnitude lower than be minimized, if not altogether suppressed, to improve
those of the shear configurations. Of the two compression their mechanical performance.
P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724 1721
Fig. 7. FEM-generated stress contours corresponding to the greatest tensile stress, rt, generated when a tensile load of 250 N is applied (a) orthogonal to
the rods (direction 3) or (b) along a rod axis (direction 1, equivalent to direction 2).
From the plots in Fig. 9 it is simple to predict the three-point bending tests. As expected, the compressive
mechanical strength of a scaffold for each testing configura- strength values exhibited by both materials are significantly
tion by using a critical stress criterion, provided the intrin- higher than their corresponding tensile and, especially,
sic strength of the constituent material is known. Indeed, shear strengths. That the HA values are more than twice
these plots allow one to determine the applied stress at those corresponding to b-TCP is a simple reflection of their
which the maximum tensile stress in the scaffold will equal respective inert strengths.
the strength of the material, i.e., the structure strength Finally, Fig. 11 shows the comparison between the pre-
under that type of load. Fig. 10 shows the strength values dicted compressive strength and the experimental results
calculated following this procedure for the two materials obtained in the uniaxial compressive tests performed on
considered in this study, HA and b-TCP, using the values the actual scaffolds for the two testing orientations consid-
of the inert strength, rF, from Table 1. The error bars rep- ered (Fig. 4a and b). Again, the error bars represent the
resent standard deviations of the predictions due to uncer- standard deviation of the data. The FEM predictions are
tainties in the measurement of rF from the corresponding seen to agree with the experimental compressive strengths
1722 P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724
Fig. 8. FEM-generated stress contours corresponding to the greatest tensile stress, rt, generated when a shear load of 250 N is applied on the printing
plane along a rod axis (direction 1, equivalent to direction 2) (a), or on a plane orthogonal to a rod axis (direction 1, equivalent to direction 2) along the
printing direction (direction 3) (b) or along the perpendicular rod axis (direction 2, equivalent to direction 1) (c).
within uncertainties for both the HA and the b-TCP scaf- able for b-TCP, could be attributed to the use of an inert
folds. The apparently systematic slight (within the errors) strength value, while the actual experiments were per-
overestimate of the compressive strength, especially notice- formed at a low speed, where slow crack growth could be
P. Miranda et al. / Acta Biomaterialia 4 (2008) 1715–1724 1723
Fig. 9. Plot of maximum tensile stress in the scaffold (maximum of rt) vs. Fig. 11. Bar graph showing comparison between experimental measure-
the applied stress (applied load normalized by initial cross-sectional area) ments and FEM predictions of the compressive strength of HA and b-TCP
for all the testing configurations analyzed: compression (two solid lines), scaffolds for the two compression orientations considered. Error bars
tension (two dashed lines) and shear (three lines with dots). Legend represent standard deviations of the data.
notation is the same as used in Fig. 4.