Talzat Zacharia PDF
Talzat Zacharia PDF
Talzat Zacharia PDF
4411~426, 1998
© 1998ElsevierScienceLtd
~) Pergamon All rightsreserved.Printedin Great Britain
0020-7683/98 $19.00 + .00
PII : 80020-7683(97)00249-7
B. TALJAT, T. ZACHARIA
Oak Ridge National Laboratory, Metals and Ceramics Division, Oak Ridge, TN 37831-6140,
U.S.A.
and
F. KOSEL
University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, 1000 Ljubljana,
Slovenia
1. INTRODUCTION
A concept for relating parameters of a spherical indentation to the material true stress/
plastic-strain (at-ep) curve was introduced by Tabor (1951). He presented an important
correlation between ep and indentation parameters: ep = 0.2d/D, where d is the contact
diameter of a residual indentation and D is the indenter diameter. Tabor also found that
the corresponding stress value could be determined as one third of the mean pressure
beneath the indenter of a fully plastic indentation. Several other authors (Francis, 1976;
Johnson, 1970; Sinclair et al., 1985; Au et al., 1980) have also worked on development of
relations between a, and indentation parameters, always assuming Tabor's ep relationship
as a basis.
As a result, the spherical-indentation method can today be used as an experimental
procedure for obtaining flow properties of metals (Field and Swain, 1995). The spherical-
indentation technique is a unique method for determining material stress-strain behaviour
in a nondestructive and localized fashion. This is important in measuring local flow proper-
ties of welds, heat-affected zones, and metallic structures in operation.
Rapid advances in computational mechanics in the last decade have produced the
means to analyze the mechanical components under fully non-linear plastic conditions. As
an example, very sophisticated computer codes (ABAQUS, MSC-NASTRAN, DYNA 3D,
ANSYS) have been developed that enable numerical modeling of highly non-linear pro-
cesses such as metal forming. In these applications, ep often exceeds the value of 0.2, and
therefore the stress-strain relations should also apply to higher strain values. On the other
hand, the yield strength provides an important material information, as does the early
* Author to whom correspondence should be addressed.
4411
4412 B. Taljat et al.
portion of the stress-strain curve. Improving the data analysis of the spherical indentation
in order to accurately predict stress-strain curves at very low strains and to obtain its large
strain portion will empower the indentation technique and satisfy the above mentioned
requirements.
A finite element (FE) model of a spherical indentation process was developed. The
analysis was performed using ABAQUS (1996) computer code. Several simulations were
performed analyzing the response of the stress-strain field beneath the indenter and the
response of the load-depth curve to variations in different material properties. On the basis
of the FE results, a new set of equations relating the material flow properties to the
indentation parameters was developed. The advantage of the developed equations is a
significant extension of strain range over which the ~r,-ep curve can be determined. Also, a
continuous ~r,-ep curve is calculated during loading of the indenter, in contrast to the old
approach, where an unloading is required to determine a point on the a, e,p curve. Different
indentation diameter measures were considered and their influence on the calculated a,-ep
curve was analyzed. The significance of having the right measure of the indentation diameter
was emphasized.
Experimental compression and indentation testing was performed on five materials
with different mechanical properties (A533-B steel, SAE1010 steel, A1-Mg alloy, annealed
copper, and cold-worked copper). Compression tests were conducted to measure the uni-
axial ar-ep curves, which were used as an input to the FE analyses. Indentation experiments
were performed to verify the FE results.
2. OVERVIEW OF THEORY
True-stress is correlated to the mean pressure underneath the indenter, Pro, which is
the ratio between the applied force, F, and the projected indentation area, 7~d214 (Tabor,
1951):
P~
,~, ~ .
-- (1)
The parameter 4' is called the constraint factor and is, according to Francis (1976), rep-
resented as a function of the parameter • :
CI ~< 1
, = C I+ C 2 ln~ 1 < • ~< C3, (2)
4'< ~>C3
where
Equation (2) defines three mechanical stages : elastic, where after load removal the material
recovers elastically and 4' has a constant value of 1.11 ; transition, which is a combination
of elastic and plastic deformation and 4' is a linear function of ln(~) ; and fully plastic,
where the plastic deformation, which initiates in the material beneath the indenter reaches
the free material surface and 4' has a constant value 4'c. Different values can be obtained
for parameters C1, C2, C3, and 4'< from the literature (Tabor, 1951 ; Francis, 1976; Au et
al., 1980; Field and Swain, 1995). Since Francis (1976) worked on statistical analysis of
previously published experimental data, we may take his computed values for C~ = 1.11,
C2 = 0.534, C3 = 27.3, and 4'c _-_ 2.87 as the average values. The parameter 0 is the angle
between the tangent to the indenter at the indentation edge and the original surface. The
parameter • may be interpreted as a ratio of the strain imposed by the indenter to the
Stress-strain curve from spherical indentation data 4413
d/2
maximum strain which can be accommodated by the material before yielding (Johnson,
1970). For shallow indentations, where d << D, one can rewrite eqn (3a) as :
Ed
qb = a,D" (3b)
The parameter q~ for loaded indentations was defined in the current work as :
4Eh,
¢ - (3c)
dtat '
where E is Young's modulus of material, parameters h, and hps a r e defined in Fig. 1, and
the parameter d, is defined in eqn (8). Figure 1 is a schematic of the indentation cross-
section illustrating the main indentation parameters. Typical materials, where piling-up
appears include cold-worked materials. Sinking-in effects, on the other hand, are typical
for annealed materials.
Tabor (1951) defined the relationship between true-plastic strain and the indentation
parameter diD as:
ep = 0.2d/'D. (4)
The calculated a, and ep values correspond to stress and strain values at the indentation
edge, called also "representative stress and strain" (Tabor, 1951).
Equation (2) does not predict the constraint factor dependency on the strain hardening
exponent, n, defined by equation : a, = Ke~, where K is the strength coefficient. However,
the results of Matthews (1980) and Tirupataiah (1991) demonstrate such dependency.
Matthews has also presented an approximate equation for the mean contact pressure for a
work-hardening material from which ¢" can be written as a function of n :
Po,
- a, 2+n\9n] ' (5)
Equation (5) gives the value ¢ c = 3 for n = 0, and ¢c = 2.85 for n = 0.5. The exper-
imental results of Tirupataiah show the same trend: higher ¢c at lower n and vice versa.
His measured ¢c values vary between 2.4 and 3.1 for different materials.
Method presented by Au et al. (1980) uses the indentation diameter, d', calculated by
Hertz's equation (Hertz, 1896) :
(6)
a, = er '
where
4414 B. Taljat et al.
1 1 -- y2 1 -- y2
+ (7)
E r - El E
E is elastic modulus and v is Poissons's ratio for the material (E, v) and the indenter (El,
v/), he is the plastic indentation depth, and ht is the total indentation depth (see Fig. 1).
Several progressive loadings and unloadings were made in order to provide hp data, which
is necessary to calculate d' (Au et al., 1980). According to Tabor (1951), eqn (6) gives good
results for shallow indentations, but a large discrepancy appears for deeper indentations.
Also, according to the results of our FE calculations, the results obtained by eqn (6) are
almost identical (average difference is about 1% over the range of 0.1 ~< diD <~0.7 for an
E/ay = 500 material where ay is the yield stress) to the indentation for diameter, d,, cal-
culated as a diameter of the intersection between the indenter and the original material
surface under loading (see Fig. 1) :
4 = 2 x / h t ( D - h,). (8)
This is because eqn (6) was developed with the assumption of an ideal spherical depression
in which the diameter before and after load application is the same. This means that d as
calculated by eqn (6) is essentially identical to dr, if one neglects the change of indentation
diameter caused by the indenter's elastic deformation. This also means that neither eqn (6)
nor eqn (8) accounts for changes in the indentation diameter due to the piling-up or sinking-
in effect, which turns out to be very important for measuring the appropriate contact
surface. The importance of this effect was shown by Norbury and Samuel (1928), and
emphasized later by Hill et al. (1989) and Field and Swain (1995).
According to Norbury and Samuel (1928), raising (piling-up) or depressing the circle
of contact (sinking-in) relative to the original surface correlates with n (see Fig. 5). They
measured the contact diameter on recovered indentations and calculated its relative height
to the original surface. The result of their research is a relationship between the ratio of
s/hp and m, where s represents the height/depth of the pile-up/sink-in from the original
surface (s = hps-hp), m is the Meyer's index represented as F = kd", and k is a material
parameter. Furthermore, it was observed that the expression n = m - 2 is valid for a broad
range of materials (see Tabor, 1951). Hill et al. (1989) called the ratio hpjhp a numerical
invariant c2, which depends on n. From the geometry of a loaded shallow indentation
follows that hts and ht can be approximated by dZ/4D and ~ / 4 D respectively, which allows
the invariant c2 to be expressed as :
c2 hps ,~ ~ d2
hp and ~ (when d << D). (9)
d,
Different authors expressed it with different equations. Matthews (1980) proposed the
following equation as a fit to the Norbury c2 vs n data :
= \7-j • (lO)
c2 _5(2--n'~
= 2\4+nf (11)
with supportive eqn (10) or (11), represents an empirical approach, that tends to predict
the actual indentation diameter by accounting for the piling-up or sinking-in effect, in
contrast to eqn (6), which is an analytical solution of the contact problem and deals with
an ideal indentation shape.
3. FE A N A L Y S I S O F T H E I N D E N T A T I O N PROCESS
• T-' 4>
deformable indenter
' r
(a)
Z
It 4) indenter
rigid inde
! [11 I I I I I I I I I II..L_I t
/
(b)
Fig. 2. Finite element models.
4416 B. Taljat et al.
mesh shape and density, element type, and contact definition on the indentation load-depth
(F-h) results. The indenter diameter used in the analysis was in the range from 1.576 to 40
mm, dependent on the indentation depth analyzed. The indenter diameter of 1.576 mm was
used for model verification, matching the indenter diameter of the experimental system.
The cylindrical specimen of 8 mm in diameter and 8 mm in height was modeled by 1571
linear four-node elements and used in the majority of calculations (see Fig. 2a). Two other
models were developed to determine the modeling effects on the F-h results. A specimen of
8 mm in diameter and 4 mm in height was modeled by 588 linear four-node elements and
by quadratic eight-node elements (see Fig. 2b). For indentations made with larger indenters
a specimen of 30 mm in diameter and 20 mm in height, meshed by a similar, but expanded
mesh to that in Fig. 2a was used. A cylindrical coordinate system with radial coordinate, r,
and axial coordinate, z, was used in all models. As shown in Fig. 2, the bottom surface of
the specimen has the z displacement fixed, whereas a free movement was allowed in the r
direction. The appropriate boundary conditions to model the axisymmetric behavior were
applied along the centerline, and a free surface was modeled at the top and outside surface
of the specimen. The influence of friction at the interface between the indenter and the
specimen on the F-h curve was evaluated by varying the friction coefficient between 0.0 and
1.0. The friction coefficient of 0.2 was used in the verification analyses and in the analyses
that supported the new method development.
The elastic-plastic constitutive behavior was assumed for the materials used in the
study. A constant Young's modulus of 200 GPa and a constant yield stress of 400 MPa
was used in development of the new method, resulting in a n E/fly ratio of 500. However,
by using the appropriate non-dimensionalization the computed results can be generalized
for materials with different E and fly. The plastic constitutive behavior was represented by
fl, = K a~ power law curves and the Von Mises yield criterion was assumed. A set of uniaxial
true stress/plastic-strain (a,-~p) curves was defined by varying the strain hardening
exponent, n, in the range from 0.0 (elastic/ideal-plastic material) to 0.5. The strength
coefficient, K, for each of these curves was calculated with the assumption that the power
law constitutive behavior starts at the point defined by a 0.002 offset of the yield strain.
According to this assumption K was calculated as K = ay/(ay/E+O.O02) ~. The Poisson's
ratio of 0.3 was used in all analyses. An elastic material with E = 645 GPa and a Poisson's
ratio of 0.28 was used for the deformable indenter.
As a part of the new procedure development the effect of material properties on the
F-h data was studied. The unloading part of the F-h curve is primarily influenced by E, as
the unloading process is essentially an elastic process ; also shown by Taljat et al. (1997).
They also showed that the loading part of the F-h curve correlates with a,. and n. Due to
the fact that the computed indentation F-h data can be generalized by an appropriate
normalization to account for different Cry and E, the effect of parameter n was of primary
interest in this study. It was assumed that the results obtained for an E/fl, = 500 material
adequately represent indentation deformation behavior of basic engineering metals, limited
by A1-Mg alloy at one, and annealed copper at the other end. The behavior of materials
with extreme properties may be different (see Bolshkov et al., 1996).
The simulation was essentially used to calculate F-h curves for the defined range of
material properties and to analyze the stress-strain distributions in the material, which was
used as a basis for development of the new method. The obtained results also gave the
possibility to verify the existing equations which relate indentation data to the a, ep curve.
5oo
U.
i
"1000
500
Al-Mg alloy respectively. Poisson's ratio of 0.3 was used in all three cases. Fig. 3 shows a
comparison between the FE and experimental F-h results.
The effects of mesh density, mesh shape, element type, type of indenter model, and
contact definitions on the analysis results were investigated. The evaluation of various
effects was based on the computed F-h data. FE model verification provided the following
information :
1. A very small influence of different mesh shapes and element types on the analysis results
was found. G o o d results were obtained with a mesh density of about 100 elements in
contact with the indenter. Reducing the number of elements beneath the indenter to 50
slightly affected the F-h result.
2. The friction coefficient at the indenter-material contact was varied from 0.0 to 1.0. The
influence on the F h results was negligible when the friction coefficient equal or higher
than 0.2 was used, but an effect on the stress-strain distribution beneath the indenter
was observed. A large difference in the computed F-h data was obtained when a
frictionless contact or very low friction coefficients ( < 0.05) were assumed.
3. The indenter was modeled as a rigid and a deformable body. The penetration depth of
a node in contact at the indentation center was computed for both indenter types. In
comparing the penetration depth results a negligible difference was detected. On the
other hand, a large difference was detected when the calculated results were compared
with experimental results. It was proved that the difference is due to the indenter
compliance (Taljat et al., 1997), which cannot be ignored, even though the indenter's
elastic modulus is at least three times higher than the elastic modulus of the materials.
Indenter compliance was taken into account by modeling the indenter up to the point
where the depth sensor is mounted. Comparison of the experimentally measured and
calculated F-h results at this point shows a very good agreement (see Fig. 3). All the
subsequent indentation depth results are evaluated at the indenter-material contact,
because it represents the actual material behavior.
o.o
experimentally measured profiles for different materials and a good agreement was obtained
(see Fig. 4).
Figure 5 shows the experimental c2 vs n data of the current work, the data measured
by Norbury and Samuel (1928), and the FE data of loaded and unloaded indentations
computed for an E/ay = 500 material. A good comparison with the data of Norbury and
Samuel (1928) was obtained. Different c 2 data for loaded and unloaded indentations are due
to elastic recovery, which considerably affects the indentation depth and so the calculated d,,
whereas the change in actual contact diameter, d, during unloading is not significant (up to
2% for materials that exhibit sink-in; the change is even smaller for other materials;
calculated for did < 0.75). The results show that the pile-up and sink-in effect depends on
n. An E/ay = 500 material shows the pile-up behavior at low n (n < 0.2) and sink-in
behavior at higher n (n > 0.2). The FE results at Fig. 5 are reported at diD = 0.5.
1.4
• E / ay = 500 loaded
, • E / c = 500 unloaded
1.3 ~© O Experiment
t , . \ A Norbury and Samuel (1928)
1. 2
1.1 ~Z~,z~
%1.0
0.9
0.8
A
0.7
The functional relationships between n and c2, based on the computed FE data for an
E/ay = 500 material of loaded (c/) and unloaded indentations (c), can be represented by'
One can use the invariant c to calculate the actual diameter d from the measured d,
using eqn (9) only for shallow indentations (d << D) under loading. In some cases the
equation may give good results for deeper loaded indentations, whereas caution should be
exercised for unloaded indentations. A significant elastic recovery in some elastic-plastic
materials causes higher curvature of the residual indentation, which may cause large dis-
crepancies in the calculated values of d, and d.
3.5
3.0 ~ o o o ~ o
2.5 .~p~-~o~m
2.0 F " o
o
o
© n=O.O
1.5 o
o
o • n=0.1
1.0 [] n=0.3
)
)
• n=0.5
0.5 .... I .... i , ' ' i .... I' ' ' i ' ' ' ' l
0 25 50 75 1 O0 125 150
(I)
Fig. 6. Constraint factor calculated for the representative stress and strain values at the indentation
edge.
4420 B. Taljat et al.
1. By applying the empirical equations to the calculated F-h data the a,-e,p curve can be
calculated, which should be equal to the input tr, ep curve to the FE analysis.
2. Tabor's equation [see eqn (4)] can also be verified by the calculated strain distribution
in the material. This has already been done by Hill et al. (1989) and by Sinclair et al.
(1985). They compared octahedral strain at the "representative point" calculated by
their FE code to e,e calculated by eqn (4).
A similar comparison was made here. Using the FE method, the equivalent plastic
strain, e~, was calculated at the indentation edge for an E/ay = 500 material with three
different strain hardening exponents, n = 0.0, 0.25, and 0.5, and the results were compared
to ep calculated by eqn (4). The values of eeP~were computed at d i d = 0.5, which gives the
value of ~p equal to 0.1. The computed values of ePe~at the indentation edge are within 0.05
and 0.15. The gradient of e~q at the indentation edge is very high, which contributes to the
discrepancy in results evaluated exactly at the indentation edge. The e~q curves for all three
cases intersect the value of 0.1 in the region from 0.97d to 1.03d, which can be considered
a good agreement (see Fig. 7).
If one applies the indentation diameter calculated by eqns (8) or (6) to eqn (1) and
calculates parameters ff and ~, a common transition curve in the ff-~ plot is obtained for
materials with different n. A large variety in the ~c values for the fully plastic regime is
shown on the same plot (see Fig. 8). Materials with low n show a high ~,c value, and vice
versa. In this case ~ accounts for two effects of n: (1) a direct effect, as shown before (see
Fig. 6), when the actual indentation diameter was used (also shown by Matthews, 1980;
Tipataiah, 1991) ; and (2) an indirect effect, as a parameter governing piling-up and sinking-
in, which actually affects the indentation diameter. The indirect effect of n is in this case
more pronounced, since the ffc values are in the range from about 1.5 to almost 4.0 for
n = 0.5 and n = 0.0, respectively. This suggests that: (1) it is important to accurately
measure or calculate the actual contact diameter when one uses a ~ value around three
(approaches by Tabor, 1951 ; Francis, 1976; Matthews, 1980), or (2) one could calculate
the indentation diameter by eqn (8) and use the ~b value which accounts for the difference
between d, and the actual contact diameter; apply ~ as presented in Fig. 8.
0.6
0.5
0.4
¢o= 0.3
0.2
0.1
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
2rld
Fig. 7. Equivalent plastic strain at the indentation edge for n = 0.0, n = 0.25, and n = 0.5.
Stress strain curve from spherical indentation data 4421
4.0
3.5
3.0
2.5
2.0
1.5
p ~-- o n=O.O
0.3
1.0 0
o° • n=0.5
0.5 . . . . I . . . . I . . . . i . . . . I
0.0 50.0 100.0 150.0 200.0
Fig. 8. Constraint factor calculated for the representativestress and strain values at the indentation
edge using d,.
indentation center, which depends on the friction between the indenter and the specimen,
eeP~qreaches a maximum and then decreases with a smaller gradient to the indentation center.
This implies that by placing the point at which the stresses ad strains are related to the
indentation parameters to some other location beneath the indenter, the cr,-ep curve can be
predicted for ep higher than 0.2. The new selected point can be called the "reference point".
Stresses and strains at this point were related to the indentation parameters by a new set of
equations, different from those established for the stress and strain values at the indentation
edge.
In the present analysis, the reference point was selected at the maximum strain point,
which is at the radius of about d,/3 from the indentation center. A new relationship between
ep and the indentation parameters was established for this point. It was written as a function
of n to account for the piling-up or sinking-in effect :
ep..... =
( "44
0.5n+ ~ 0 . 1 1.6 ~' . (13)
The corresponding if value, if . . . . shows the same behavior as that computed at the
indentation edge using the actual contact diameter. After it reaches a certain value (i.e.,
total plastic state is reached), if max remains constant and independent of qb, if max = ifCax,
but on the other hand, is an obvious function of n, because the indentation diameter d, was
used. Before the if max curves reach their constant value if~ax, they follow the same path,
which happens to be a line in the ~maxvs ln(qb) plot (see Fig. 9), and could be approximated
by a function [see eqn (14)], which is very similar to eqn (2). Perhaps, the same explanation
about the transition regime, which is a combination of elastic and plastic deformation,
where if is a linear function of In(O), and the fully plastic regime, where the plastic
deformation reaches the free material surface and ~J has a constant value ~c can be
considered here. The form of the function is :
The if/max = - - 0 . 6 5 + In(O) line should be followed until it reaches the ffJCax value pertinent
4422 B. Taljat et al.
4.0 --
3.5
3.0
=0.1
• c
2.5
= 0.2
2.0 = 0.3
= 0.4
1.5 =0.5
o~
1.0
0.5 i I i
10 100 190
2
I~Cax = - - 0 . 8 1 + - - (15)
x/n+0.2
3.3.3. Minimum strain method. In contrast to the approach presented in the previous
section, we now look for the reference point that would give a reliable result up to the strain
value of about 0.05. Placing the point at a distance of about d,/lO outward the indentation
edge, ePZqhas the m a x i m u m value of 0.26 for n = 0.0. The correlation between ~q and dJD
at this point can be approximated by the following equation :
The corresponding I~min VS (I) plot, calculated at the radial distance of dt/lO from the
indentation edge, can be derived in the same way as described earlier in the m a x i m u m strain
approach. The results show that almost a constant ~bmi, (qJmin = 3.75) value can be accepted
for n < 0.3.
~ at the observed point depends on n and d,/D. The value of e~tq is higher
(0.12 < ~/q < 0.25) for materials with low n (n < 0.1) and lower (0.01 < e~q < 0.03) for
materials with higher n (n > 0.3). A favorable ~ range occurs at n = 0.2 where the ~
reaches the value of 0.06 at d,/D = 0.75. An efficient method would give results up to the
desired strain value, independently of the n value. To achieve this, one can use the ~q-dt/D
curve obtained for n = 0.2 for all n values. This means that the reference point moves to a
different location for materials with n different than 0.2. The corresponding ~/min c u r v e s can
be approximated as :
Stress strain curve from spherical indentation data 4423
3.18n2 + 2 . 8 9 n 3 )
~'mi. : (3.65 -4.30n2 +4.36n3)(-~'~ ~ (17)
\1.,/
The above equations provide a good fit to the computed data. They are rather sensitive
to n, which means that a good estimate of n is required. The equation for a, [eqn (1)]
remains the same as was used for the Tabor's approach, except for the parameter ~ which
changes to I]/ma x for the maximum strain approach, and to ~tmi n when the minimum strain
approach is considered.
3.3.4. Determ&in9 nfrom indentation data. A strong n influence introduced in the above
equations accounts for both, the difference between d, and the actual contact diameter, and
the direct effect of n on ~. Using d, of loaded indentations makes the procedure easier to
use, because no intermediate unloading is required, and also more comparable, because of
the unique way for determining the indentation diameter. A question arises at the point
when one wants to apply the equations to an experimental procedure, because the equations
are n dependent. Any approach to determine ~r,-~p curve by the indentation test requires
information of the indentation diameter. Since the piling-up or sinking-in phenomena
considerably influence the indentation diameter, and the phenomena depends on n, it is
obvious that the knowledge of n is mandatory. Therefore, it is essential to extract that
information from the indentation F-h measurement results. Several possibilities are dis-
cussed below:
1. Considering Meyer's law, F = kd% and the fact that n = m - 2 , one can estimate n from
the measured F-d data. It turns out that the method works well and that n can be
determined with reasonable accuracy for a broad range of materials. The difficulty of
this approach is that the information of the actual contact diameter, d, is required, and,
our analysis shows that n obtained using this approach, while substituting d with d, of a
loaded or unloaded indentation, does not always give accurate results (see also Field
and Swain, 1995).
2. The loading part of the F-h curve can provide additional information (Taljat, 1996). A
relationship among loading slope, ay, and n, proposed by Taljat (1996) can be solved if
one has information of a.v, in addition to the measured value of loading F-h slope.
3.5 i
m ~ FE
/I ~o~ e6
1.01.5 ~ i ~G~.~'~
~
0.5
0.0
0.000.020.040.060.080.100.120.140.16
.... I .... I .... i .... I .... I .... I .... I .... I
ht/D
Fig. 10. Loading F/cr,D2-h,/Dcurve shape as a function ofn.
4424 B. Taljat et al.
3. Another possibility is the shape of the loading F-h curve. Results from the FE analyses
and experimental work suggest that the relationship between F and h is almost linear
(except the first part) due to the combination of two inverse nonlinearities (nonlinear
indenter geometry and nonlinear material characteristic). The linear relationship is
usually obtained at a certain diD range (0.1 < d/D < 0.7). After the indenter is pushed
to more than a certain depth, the F-h plot becomes nonlinear and curved toward the h
axis. The explanation for this phenomenon could be the effect of material saturation,
which occurs after a critical amount of deformation is exceeded, or also, the nonlinearity
effect of the sphere geometry compared to the effect of material nonlinear characteristic
is becoming weaker as the penetration depth increases beyond a certain limit. It was
observed that the portion of F h loading curve which is expected to be linear is slightly
curved toward the F axis for materials with high n and curved toward the h axis for
materials with low n (see Fig. 10).
4. Combining the observed phenomenon with the method of measuring piling-up or sink-
ing-in of the material around the impression should give a reasonably good estimate of
the strain-hardening exponent, providing the E/a~ value of the material tested is known
[see Fig.4, Fig.5, and eqns (9) to (12)].
4. E X P E R I M E N T A L R E S U L T S A N D M E T H O D V E R I F I C A T I O N
5. C O N C L U S I O N S
400 1000
_ .L
r 800
30o
600
2oo [MPa]
[MPa] 400
100
,'~2~:t;:;test 2O0
! ; : : : ', : ',
, I i I = : 0
0.1 0.2 0,3 0.4 0.5 0.6 0.7 0.8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
£p £p
(a) (b)
700 ¸ 350 •
....
A ~ t"
60o i 300,
2O0 100,
100 50'
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0,1 0.2 0,3 0.4 0.5 0.6 0.7
Ep £p
(c) (d)
Fig. 11, Comparison between compression and indentation a,--~p curve: (a) cold-worked copper; (b)
A533-B steel; (c) A I - M g alloy; (d) annealed copper.
The importance of an accurate measure of the actual indentation contact diameter was
discussed and different equations for its prediction were evaluated in terms of their effect
on the subsequent procedure for the stress-strain curve calculation. Also, the effect of
piling-up and sinking-i n on the actual contact diameter was evaluated.
Equations for calculating the material cr,-ep curve from the indentation parameters [see
eqns (1) and (4)] relate the indentation parameters to the material stress-strain state at the
indentation edge. The maximum ep value calculated from eqn (4) is 0.2. By moving the
point for which the a,-ep relation is established, i.e., "the reference point," to a region of
maximum strain (approximately to a radius of d,/3 from the indentation center), the
material a, ep curve can be predicted at much higher strains (ez, up to 1.6). On the other
hand, by moving the reference point outward from the indentation edge, the material ~Tt-~. p
curve can be predicted for lower strains (yield strain < ep < 0.1). New relationships for ~.p
and ~ were proposed which enable continuous a,-ep curve calculation during the loading
process (no intermediate unloading is required) by the mentioned maximum and minimum
strain approaches. This was accomplished by using an apparent indention diameter, defined
4426 B. Taljat et al.
Acknowledgments--The authors would like to thank Prof. George M. Pharr and Dr Gorti Sarma for reviewing
the paper. The research was sponsored in part by an appointment to the Oak Ridge National Laboratory
Postdoctoral Research Associates Program administered jointly by the Oak Ridge Institute for Science and
Education and Oak Ridge National Laboratory. The research was also sponsored by the Division of Materials
Science, U.S. Department of Energy, under contract De-AC05-96OR22464 with Lockheed Martin Energy
Research Corp., and by the U.S. Navy, Office of Naval Research under interagency agreement DOE No. 1866-
EI26-A1, Navy No. N000014-92-F-0063 under U.S. Department of Energy contract De-AC05-96OR22464 with
Lockheed Martin Energy Research Corp.
REFERENCES
ABAQUS (1996), Users' Manual, Version 5.5. Hibbitt, Karlsson & Sorensen.
Au, P., Lucas, G. E., Scheckerd, J. W. and Odette, G. R. (1980) Flow property measurements from instrumented
hardness tests. In Nondestructive Evaluation in the Nuclear Industry, pp. 579~10, ASM International, Metals
Park, OH.
Bolshakov, A., Oliver, W. C. and Pharr, G. M. (1996) Finite element studies of the influence of pile-up on the
analysis of nanoindentation data, accepted for publication in Thin Films : Stresses and Mechanical Properties
VI--Materials Research Society Symposium Proceedings, 436, ed. W. W. Gerberich, H. Gao, J. E. Sundgren
and S. P. Baker, Materials Research Society, Pittsburgh, PA.
Field, J. S. and Swain, M. V. (1995) Determining the mechanical properties of small volumes of material from
submicrometer spherical indentations. J. Mater. Res. 10(1), 101- 112.
Francis, H. A. (1976) Phenomenological analysis of plastic spherical indentation, Trans. ASME, J. En. Mater.
Technol. 272-28 I.
George, R. A., Dinda, S. and Kasper, A. S. (1976) Estimating yield strength from hardness test. Met Progress,
3(~35.
Hertz, H. (1986) Miscellaneous Papers by H. Hertz, ed. Jones and Schott, Macmillan, London.
Hill, R., Storakers, B. and Zdunek, A. B. (1989) A theoretical study of the brinell hardness test. Proc. R. Soc.
London A423, 301.
Johnson, K. L. (1970) The correlation of indentation experiments. J. Mech. Phys. Solids 18, 115-126.
Matthews, J. R. (1980) Indentation hardness and hot pressing. Acta Metall. 28, 311-318.
Norbury, A. L. and Samuel, T. (1928) The recovery and sinking-in or piling-up of material in the brinell test, and
the effects of these factors on the correlation of the brinell with certain other hardness tests. J. Iron. Steel Inst.
117, 673.
Sinclair, G. B., Follansbee, P. S. and Johnson, K. L. (1985) Quasi-static normal indentation of an elasto-plastic
half-space by a rigid sphere--II, results. Int. J. Solids Structures. 21(8), 865-888.
Tabor, D. (1951) The Hardness of Metals. Clarendon Press, Oxford.
Taljat, B. (1996) Determinin 9 Stress-Strain Curves of Metals by Nondestructive Continuous Ball-Indentation
Methods. Ph.D. Thesis, University of Ljubljana, Slovenia.
Taljat, B., Zacharia, T., and Haggag, F. M. (1997) Analysis of ball-indentation load-depth data: part I--
determining elastic modulus. J. Mater. Res. 12(4), 965-974.
Tirupatahiah, Y. (1991) On the constraint factor associated with the indentation of work-hardening materials
with a spherical ball. Metall. Trans. A. 22A, 2375-2384.