You are on page 1of 6

An Enhanced Calibration Scheme for the EDM Hole-Drilling Strain Gage Method

for the Measurement of Residual Stress in Ferrous Materials


H. T. Lee, C. Liu
*
, F. C. Hsu
*
and J. M. Hsu
*
Department of Mechanical Engineering, National Cheng-Kung University, 701, Tainan, Taiwan, R. O. China
Recent studies revealed that EDM hole-drilling strain gage method is applicable for the measurement of residual stress in materials with
higher hardness and toughness. However, the metallurgical transformation layer formed on the wall of the hole induces an additional stress and
therefore generates a measurement error. Usually this error can be calibrated by estimating and reducing the hole-drilling induced extra stress,

IS
. However, the value of
IS
is highly sensitive to the EDM parameters. Accordingly, the current study aimed at lowering
IS
by optimizing the
working parameters of pulse current, pulse-on duration and pulse-o duration with experiments. An inverse power law relationship was
constructed to predict the magnitude of the hole-drilling induced stress where was referred. A convenient calibration method was proposed
where calibration factor can be easily determined. The calibrated results revealed good accuracy where deviation is no more than 10 MPa. The
accuracy of the proposed calibration scheme was conrmed in study. [doi:10.2320/matertrans.MER2008098]
(Received March 25, 2008; Accepted June 2, 2008; Published July 25, 2008)
Keywords: electrical discharge machining, hole-drilling strain gage method, residual stress, transformation layer
1. Introduction
In the measurement procedure of the hole-drilling strain
gage method, a strain gage rosette is attached to the specimen
surface, and a circular hole is drilled vertically into the
specimen at the center of the gage circle in order to partially
release the residual stress within the specimen.
1)
The
magnitude of the released strain is monitored continuously
during the drilling process and once the depth of the
measurement hole reaches the prescribed value, the strain
values are substituted into the stress-strain equation provided
within the ASTM E837 standard in order to derive the
residual stress value. However, the drilling operation inevi-
tably induces an additional stress within the component. This
additional stress is mixed with residual stress detected and
therefore leads to a corresponding error in computing the
residual stress.
2)
It has been reported that the additional stress
induced by High-Speed (HS) drilling techniques is relatively
lower than that generated by conventional hole-drilling
methods such as low-speed end milling technique and
abrasion jet machining technique, for example.
3)
Further-
more, HS drilling techniques have the advantages of simple
experimental setup, straightforward operation and improved
precision.
4,5)
Accordingly, the ASTM E837 standard explic-
itly recommends the use of HS drilling technique to
accomplish the hole-drilling strain gage measurement of
residual stress.
However, hole drilling with high speed leaded to a severe
wear on cutting tool when performed on specimens with high
hardness and high toughness. This tool wear increased not
only the induced stress and deviation,
6)
but also might prompt
a catastrophic failure. The completion of the measurement
process became questionable. Consequently, alternative
drilling method is required here. The electrical discharge
machining (EDM) technique has the advantage that its
applicability is not limited by the mechanical properties of
the specimen. It can be applied to all ferrous materials, even
by those with high hardness and high toughness.
Previous studies
79)
have shown that tensile residual
stresses as high as 400 MPa can be induced within the region
of the transformation layer extending from the machined
surface to a depth of around 40 micrometers. However, at
greater depths, the tensile stress gradually reduces toward
zero and may even transform to a small compressive stress.
By EDM hole-drilling strain-gage method, an extra strain
introduced by this transformation layer is inevitably included
within the released strain detected. Such measurement error
should be calibrated if the accuracy of the residual stress
measurement is aimed at. In previous studies by the current
group, it was shown that the hole-drilling induced stress (
IS
)
measured on stress-free specimens could be useful in com-
pensating the measurement error.
1012)
However, it was also
noted that the value of
IS
was highly sensitive to the working
parameters such as pulse current and pulse-on duration. Thus
it is required to specify the appropriate EDM working
conditions in order to ensure the validity of the calibration.
It is clear that the stress induced during the hole-drilling
process should be minimized. Accordingly, the current study
commences by nding suitable values of the pulse current,
pulse-on duration and pulse-o duration. Good combination
of those parameters can eectively suppress the occurrence
of secondary discharges and therefore result in a lower
induced stress. A series of experiments were performed using
the various combination of parameter settings to check the
hole-drilling induced stress generated. In this study, six
dierent ferrous specimens with varying thermal conductiv-
ities were used. A mathematical correlation between the
hole-drilling induced stress and the thermal conductivity
coecient of the specimen was derived on the base of the
experimental data. This correlation was employed further to
construct a calibration equation with which the measured
stress value obtained using the EDM hole-drilling strain-gage
method was compensated. Study results were further checked
and discussed.
2. Experimental Procedure
The residual stress measurements conducted in the present
*
Graduate Student, National Cheng-Kung University
Materials Transactions, Vol. 49, No. 8 (2008) pp. 1905 to 1910
#2008 The Japan Institute of Metals EXPRESS REGULAR ARTICLE
study were performed in accordance with the ASTM E837
standard using a TEA-06-062RE strain gage rosette and a
P-3500 strain indicator (both manufactured by the Vishay
Measurements Group Inc., USA). Figure 1 presents a
schematic illustration of the strain gage, in which D is the
diameter of the gage circle (5.13 mm in the present case) and
D
0
is the diameter of the drilled hole. According to ASTM
E837, the diameter of the measurement hole should be
between 0.3D and 0.5D, while its maximum depth should be
0.4D. During the drilling process, the released strain was
monitored continuously through the strain gage and indica-
tor. When the measurement hole reached its maximum depth,
the corresponding values of the strain, namely "
1
, "
2
, and "
3
(see Fig. 1) were read and substituted into eq. (1) shown
below to obtain the corresponding value of the residual stress.
Note that in eq. (1),
m,min
and
m,max
represent the minimum
and maximum principal residual stress.

(m,min),(m,max)
= ("
3
"
1
)=4A [("
3
"
1
)
2
("
3
"
1
2"
2
)
2
]
0:5
=4B:
(1)
A and B are material-dependent coecients, whose values
are derived directly from the tables presented in ASTM
E837 depends on Youngs modulus and Poissons ratio of
the specimen and the diameter of the measurement hole.
The experiments were performed using a CNC die-sinking
EDM machine (Yihawjet Inc., Taiwan) with an open circuit
voltage of 120 V. Kerosene dielectric with a temperature of
25

C and solid CuW electrode with an external diameter of


1.5 mm were used.
AISI 1045 mild steel was used to investigate the inuence
of the EDM parameters on the relative stability coecient of
the discharge duty ratio and the hole-drilling induced stress.
In our experiments, the specimens with a diameter of 35 mm
and a thickness of 6.2 mm were annealed to ensure a stress-
free state (as conrmed via X-ray diraction analysis). The
specimens were then drilled in accordance with the ASTM
E837 standard. Since the internal residual stress of a fully
annealed specimen prior to the drilling is nearly zero and can
be neglected. The values of the measured stress obtained
from eq. (1) should thus be the principal stress,
IS,min
and

IS,max
, induced by the hole-drilling operation.
1012)
Further step required was to explore the correlation
between the hole-drilling induced stress and the thermal
conductivity coecient of the sample. Derivation of a more
robust calibration scheme was aimed at. Six dierent ferrous
materials, namely AISI 1045, 4140, L6, H13, M2 and 304
were used in this experiment. They were deliberately chosen
so as to provide a wide range of thermal conductivity
coecients. The values correspondingly are 50.2, 41.8, 36.4,
28.6, 21.3 and 16.3 Wm
1
K
1
, respectively.
Finally, further checkup aimed at the validity and precision
of the proposed calibration approach. Annealed AISI H13
tool steel specimen was used here. Specimens were made and
pre-stressed with uniaxial loading equivalent to 20%, 35%,
50% and 65% of their yield strength, respectively, where the
presence of residual stress was simulated. Note that the
dimensions of the specimens and the grip used were designed
in conformance with the guidelines stated in ASTM E8M
tension testing standard.
13)
The magnitude of the pre-stress
was then measured using the EDM hole-drilling strain gage
method.
3. Results and Discussions
Figure 2 illustrated the variation of the measured stresses
in AISI 1045 stress-free specimens drilled with a solid CuW
electrode under dierent EDM conditions. It can be seen that
the maximum (
IS,max
) and minimum (
IS,min
) principal
stress induced by I
P
=
on
=
off
settings of 4A/23 ms/23 ms
and 12A/23 ms/23 ms were 157:9 MPa/136:4 MPa and
170:2 MPa/155:6 MPa, respectively. These stress values
were greater than 48% of the yield strength of AISI 1045
carbon steel (
ys
= 285 MPa) in magnitude, and thus these
parameter combinations were obviously inappropriate for
the EDM hole-drilling strain gage method. On the other
hand, the hole-drilling induced stresses obtained by 4A/
9 ms/9 ms and 12A/9 ms/9 ms were comparatively much
lower, where those corresponding values were 40:4 MPa/
30:2 MPa and 43:0 MPa/35:0 MPa, respectively. How-
ever, the discrepancy between the
IS,max
and
IS,min
was
found to be around 12.2 MPa and 8.0 MPa, respectively.
Theoretically, if the transformation layer formed on the hole
surface is homogenous and has uniform properties, then the
three strains ("
1
, "
2
, "
3
) detected by the strain gage should be
approximately equal, i.e.
IS,max
-
IS,min
. However, for the
Fig. 1 Schematic illustration of rosette strain gage.
Fig. 2 Hole-drilling induced stress generated under various combinations
of I
P
=
on
=
off
parameters with solid CuW electrode in AISI 1045 stress-
free specimens.
1906 H. T. Lee, C. Liu, F. C. Hsu and J. M. Hsu
parameter settings of 12A/9 ms/9 ms and 4A/9 ms/9 ms, the
discrepancies between
IS,max
and
IS,min
were more than
20% of the average of the two principal stresses in each case.
It means that the properties of the transformation layer were
far from uniform. However, by parameter setting of 12A/
6 ms/30 ms as shown in Fig. 2, the data for
IS,max
and
IS,min
were found to be 25:8 MPa and 23:7 MPa, respectively.
The discrepancy between these two values was reduced to
just 2.1 MPa. According to ASTM E837 regulation, the
resolution of the strain indicator used in the measurement
process requires less than 2 m". This resolution corresponds to
a stress of approximately 2.5 MPa. Comparing the discrep-
ancy of the principal induced stresses mentioned above
(2.1 MPa) with the strain indicator resolution (2.5 MPa), it
can be concluded that the discrepancy between the
IS,max
and
IS,min
under this working condition was suciently
small which can be ignored.
By EDM process, the electrical energy of each single
discharge (E) is contributed by the product of the pulse
current (I
P
), the voltage (V
g
) between the electrode and the
workpiece, and the pulse-on duration (
on
), i.e. E = I
P

V
g

on
. Theoretically, it seems reasonable to expect that a
lower discharge energy will also result in a lower hole-
drilling induced stress. However, this is found not to be the
case in practice. Under the ve EDM parameter settings as
shown in Fig. 2, V
g
voltage was restricted to the range 45
48 V. And thus ve sets of EDM conditions can be ranked
according to energy level as follows: 12A/23 ms/23 ms >
12A/9 ms/9 ms > 4A/23 ms/23 ms > 12A/6 ms/30 ms > 4A/
9 ms/9 ms, where the energy values were 12.4, 4.9, 4.1,
3.2 and 1.6 mJ, respectively. However, experimental results
showed that the magnitude of induced stresses was in the
order as: 12A/23 ms/23 ms > 4A/23 ms/23 ms > 12A/9 ms/
9 ms > 4A/9 ms/9 ms > 12A/6 ms/30 ms. The two ordered
sequences are clearly dierent, and thus it is apparent that
there must be some other factors aecting the magnitude of
the induced stress besides of I
P
and
on
. Numerous experi-
ments were performed in this study, we found that the hole-
drilling induced stress is basically aected by the occurrence
of the secondary discharge. Theoretically, if the dielectric
uid can be kept its good function as an eective insulation
layer between the electrode and the workpiece, the succes-
sive discharge sparks are randomly distributed over the
machined surface, as shown schematically in Fig. 3(a). The
craters shown in this schematic gure were pasted from SEM
micrograph symbolized spark results of a single discharge on
surface. If there is no secondary discharge happened during
discharging process, the morphology of EDMed surface
will take the form similar to that shown in Fig. 3(a) where
individual spark craters are separately dispersed on surface.
However, as the insulation property of the dielectric uid
after the discharge is not fully restored before the next
discharge event. The secondary discharges would occur
during the process, and some sparks would hit the same site
that caused overlapped craters formed on surface. Schematic
gure is shown in Fig. 3(b) where the third to the eighth
discharge hit the same place for example. Under these
circumstances, a signicant amount of discharge heat can be
accumulated on site of the place where secondary discharge
occurred, and thus an increased induced stress is generated.
Hence we assumed that the magnitude of the hole-drilling
induced stress was fundamentally related to the occurrence
of secondary discharge events during the drilling process.
However, it is practically quite dicult to detect the
occurrence of secondary discharges so far nowadays our
techniques attainable. As alternative parameter designated as
the relative stability coecient of the discharge duty ratio
(C
sta
) was introduced here in order to detect the occurrence
of the secondary discharge. Figure 4 presented a schematic
illustration showing several scope patterns of working
voltage in one discharge duty cycle. As shown, the pattern
of normal duty comprised delay duration (
d
), pulse-on
duration (
on
), and pulse-o duration (
off
).
14)
The ratio of the
pulse-on duration to the total duration of the cycle was
referred to as the discharge duty ratio, R
on
(see eq. (2a)).
R
on
=
on
=(
d

on

off
) (2a)
In the current study, an electronic monitoring system was
designed to detect the R
on
values 100 times every 350 ms.
The R
on
data were then used to compute its mean (R
on
) and
standard deviation (s
R
) in accordance with eqs. (2b) and (2c),
respectively. From a statistical perspective, the standard
deviations of data obtained from dierent experiments can
not directly be compared unless their means are close to
equal.
15)
Therefore, the s
R
should be converted into an
equivalent relative variability coecient, designated as CV
R
,
(a) (b)
Fig. 3 (a) Ten successive discharge sparks are randomly on machined surface; (b) occurrence of secondary discharge phenomenon during
third to eighth discharge events.
An Enhanced Calibration Scheme for the EDM Hole-Drilling Strain Gage Method for the Measurement of Residual Stress 1907
where CV
R
= standard deviation/mean (see eq. (2d)). The
relative stability coecient of the discharge duty ratio (C
sta
)
was then dened as 1 CV
R
(see eq. (2e)).
R
on
=
X
100
i=1
(R
on,i
=100) (2b)
s
R
=
X
100
i=1
(R
on,i
R
on
)
2
!
=(100 1)
" #
0:5
(2c)
CV
R
= s
R
=R
on
(2d)
C
sta
= 1 CV
R
(2e)
When the scope pattern of each discharge duty conforms
exactly to the EDM parameter settings, the C
sta
coecient
can be maintained at (or very close to) a value of 1. The
occurrence of secondary discharge events yields abnormal
scope patterns such as those labeled as B, C and D in
Fig. 4. The corresponding changes in
d
and
on
prompt a
signicant reduction in the value of C
sta
. Figure 5 illustrated
the C
sta
value when using the ve sets of EDM conditions
shown in Fig. 2. From the experimental results presented
in Figs. 2 and 5, it can be seen that the hole-drilling
induced stress decreased as the C
sta
value increased. When
the drilling operation was performed using parameter
setting of 12A/6 ms/30 ms, the value of C
sta
lain between
0.99 and 1. Accordingly, it was reasonable to infer that
the occurrence of secondary discharges was suppressed in
drilling process, and therefore, the hole-drilling induced
stress could be eciently decreased. Furthermore, Fig. 5
showed that when the hole-drilling process was performed
using EDM conditions of 4A/9 ms/9 ms, 12A/9 ms/9 ms,
4A/23 ms/23 ms and 12A/23 ms/23 ms, the corresponding
C
sta
coecient had a value of 0.93, 0.91, 0.62 and 0.56,
respectively. In other words, it was signicant to infer that
the secondary discharge events took place in these drilling
processes, which not only decreased the C
sta
value, but also
increased the induced stress.
Figure 6 presented a SEM micrograph showing the cross-
section of the transformation layer by AISI 1045 specimen.
Two distinguishable layers, namely the recast layer and the
heat aected zone, were readily identied via their distinc-
tive microstructures. Similar SEM images were obtained for
AISI 1045 specimens machined using each of the ve
parameter settings shown in Fig. 2 and were used to measure
the thickness of the recast layer in each case. The
corresponding results were presented in Fig. 7. Signicantly,
the results showed that the thickness of the recast layer was
not directly related to the discharge energy. For example, the
EDM condition of 12A/9 ms/9 ms had a higher value of
discharge energy than the condition of 4A/23 ms/23 ms,
however, the thickness of the corresponding recast layer was
Fig. 5 Relative stability coecient of discharge duty ratio for various
combinations of I
P
=
on
=
off
parameters. (material: AISI 1045).
Fig. 6 SEM micrograph of recast layer and heat aected zone. (material: AISI 1045; EDM condition: 4A/23 ms/23 ms).
Fig. 4 Scope patterns of discharge duty cycle: (A) prole of normal duty
cycle; (B), (C) and (D) proles of abnormal duty cycle.
1908 H. T. Lee, C. Liu, F. C. Hsu and J. M. Hsu
far less than that produced in the latter case. From the
experimental results presented in Figs. 2 and 7, it can be seen
that greater C
sta
(e.g. 0.99 by 12A/6 ms/30 ms setting) value
prompted a signicant reduction in the thickness of the recast
layer. Accordingly, it was reasonably conrmed that the
occurrence of secondary discharge was an important factor in
enhancing the accumulation of discharge heat. Because the
EDM condition of 12A/6 ms/30 ms ensured that the C
sta
coecient is kept within 0.991 throughout the drilling
operation, we suggest that this condition is optimal for EDM
hole-drilling strain gage method.
In previous studies
1012)
performed by the current group
demonstrated that the hole-drilling induced stress (
IS
) was
insensitive to the magnitude of the original residual stress
within the specimens. Therefore, the
IS
value obtained from
stress-free sample could be used to calibrate the value of
the measured stress. As shown in eq. (3), the calibration
procedure simply involves subtracting the
IS
value from the
measured value of the residual stress, i.e.

(cal,min),(cal,max)
=
(m,min),(m,max)

IS
: (3)
However, the application of this calibration scheme
requires the use of a extra measurement in order to
determine the value of calibration factor (
IS
). Accordingly,
the objective of the current study is to enhance this
correction scheme such that the calibration factor can be
predicted directly by the properties of the specimen without
the need for any auxiliary measurement. In a previous study
by the current group,
12)
it was shown that of the various
properties of a specimen, its thermal conductivity has the
greatest eect on the EDM hole-drilling induced stress. In
order to establish the relationship of the induced stress and
thermal conductivity, a series of measurement experiments
was performed using the optimal parameter setting of
12A/6 ms/30 ms in six dierent ferrous materials, namely
AISI 1045, 4140, L6, H13, M2 and 304, respectively.
From Fig. 8, it was found that this relationship, calculated
by regression, had the form
IS
= 325:9k
0:65
, where
R
2
= 0:998. Substituting this expression into eq. (3) yielded
the enhanced EDM hole-drilling strain gage calibration
equation shown in eq. (4).

(cal,min),(cal,max)
=
(m,min),(m,max)
325:9k
0:65
= [("
3
"
1
)=4A (("
3
"
1
)
2
("
3
"
1
2"
2
)
2
)
0:5
=4B]
325:9k
0:65
(4)
Since the thermal conductivity coecient of various ferrous
materials is readily available in the literature, and thus eq. (4)
provides a highly convenient means to calibrate the measure-
ment results obtained using the EDM hole-drilling strain
gage method.
In order to evaluate the accuracy of the calibration
equation given in eq. (4), various uniaxial pre-stress loadings
were applied to stress-free AISI H13 specimens to simulate
the presence of residual stress. In applying the pre-stress
loading, the minimum stress was specied as
P,min =

0,
while the maximum stress (
P,max
) was assigned values of
approximately 20%, 35%, 50% and 65%, respectively, of the
yield stress of AISI H13 tool steel (
ys
= 350 MPa). The
residual stress (i.e. the pre-stress) within each sample was
then evaluated by conducting an EDM hole-drilling measure-
ment process using the optimal parameter settings of 12A/
6 ms/30 ms. Figure 9 illustrated the un-calibrated and cali-
brated results of the measured stress. It can be seen that the
measured stress (
m,min
,
m,max
) obtained from eq. (1) was
lower than the original pre-stress. The thermal conductivity
coecient of AISI H13 tool steel is 28.6 Wm
1
K
1
.
Substituting this value into eq. (4), together with the strain
data measured in the corresponding EDM hole-drilling
process, yielded the calibrated residual stress values
(
cal,max
,
cal,min
). Plotting these calibrated values in Fig. 9,
it was found that the maximum discrepancy between the
calibrated results and the original pre-stress values was less
than 10 MPa in every case. This deviation was well within
the limit of 20 MPa recommended in the Handbook of
Measurement of Residual Stresses,
16)
and therefore con-
rmed the precision of the calibration scheme.
In this study, the ideal C
sta
value of 0.991 was obtained
only using parameter setting of 12A/6 ms/30 ms. In other
words, it can be inferred that when drilling employed this
EDM condition, pulse-o duration of 30 ms was sucient to
allow the insulating properties of the dielectric uid to be
restored. By contrast, in the other four parameter settings, the
Fig. 7 Recast layer thickness for various combinations of I
P
=
on
=
off
parameters. Note that C
sta
values for 12A/9 ms/9 ms and 4A/23 ms/23 ms
are 0.91 and 0.62, respectively. (material: AISI 1045).
Fig. 8 Regression correlation between hole-drilling induced stress and
thermal conductivity. (EDM conditions: 12A/6 ms/30 ms).
An Enhanced Calibration Scheme for the EDM Hole-Drilling Strain Gage Method for the Measurement of Residual Stress 1909
pulse-o duration was too short to restore the insulating
properties, and thus the secondary discharge phenomenon
occurred and yielded a lower C
sta
value.
In order to further investigate the inuence of pulse-o
duration on C
sta
coecient, a series of experiments was
performed where I
P
=
on
parameters were 12A/23 ms, 12A/
9 ms, 4A/23 ms, and 4A/9 ms, respectively. The experimental
results showed that when the pulse-o duration exceeded
45 ms, the value of C
sta
was generally close to 0.99 in every
case. However, it was found that on occasions, the value of
C
sta
suddenly dropped to a very low value and then oscillated
in an unstable state. The duration of each unstable state was
around 35 seconds. It was conjectured that this phenomenon
was a result of a poor removal of debris from the machining
area during the drilling process. Accordingly, in a future
study, the eects of the debris removal eciency on the
induced stress will be examined in order to enhance the
robustness of the proposed measurement of residual stress.
4. Conclusions
The experimental results presented in this study support
the following major conclusions:
(1) When the relative stability coecient of the discharge
duty ratio is kept within 0.991, the secondary dis-
charge phenomenon is eectively suppressed with the
result that the hole-drilling induced stress is reduced.
(2) For the ferrous materials employed in the present study,
the suggested optimal drilling parameters for the EDM
hole-drilling strain gage method are as follows: a pulse
current of 12A, a pulse-on duration of 6 ms, and a pulse-
o duration of 30 ms.
(3) When performing the hole-drilling operation using the
suggested parameter setting of 12A/6 ms/30 ms, the
value of the hole-drilling induced stress (
IS
) is related
to the thermal conductivity coecient (k) of the
specimen via the following inverse power law relation-
ship:
IS
= 325:9k
0:65
.
(4) When using EDM hole-drilling strain gage method to
measure the residual stress, the measurement error
caused by metallurgical transformation layer can be
calibrated by applying the following calibration
equation:

(cal,min),(cal,max)
= [("
3
"
1
)=4A (("
3
"
1
)
2
("
3
"
1
2"
2
)
2
)
0:5
=4B] 325:9k
0:65
:
(5) The calibrated values of the residual stress deviate from
the true values by no more than 10 MPa. Consequently,
the calibration scheme presented in this study provides
a highly precise and convenient means of obtaining a
signicant improvement in the accuracy of the EDM
hole-drilling strain gage method.
REFERENCES
1) ASTM Standard E837-95, Standard Test Method for Determining
Residual Stresses by Hole-Drilling Strain Gage Method, (ASTM,
Philadelphia, PA, USA, 1995).
2) H. T. Lee, F. C. Hsu and T. Y. Tai: Mater. Sci. Eng. A 364 (2004)
346356.
3) M. T. Flaman and J. A. Herring: Exp. Tech. 9 (1985) 3032.
4) M. Tootoonian and G. S. Schajer: Exp. Mech. 35 (1995) 124129.
5) G. S. Schajer: Transactions of the ASME 110 (1988) 338343.
6) M. T. Flaman and J. A. Herring: Exp. Tech. 10 (1986) 3435.
7) B. Ekmekci: Appl. Surf. Sci. 253 (2007) 92349240.
8) J. C. Rebelo, A. M. Dias, D. Kremer and J. L. Lebrun: J. Mater. Process.
Technol. 84 (1998) 9096.
9) B. Ekmekci, A. E. Tekkaya and A. Erden: Int. J. Mach. Tools Manuf.
46 (2006) 858868.
10) H. T. Lee and F. C. Hsu: Mater. Sci. Technol. 19 (2003) 12611265.
11) H. T. Lee, W. P. Rehbach, F. C. Hsu, T. Y. Tai and E. Hsu: J. Mater.
Process. Technol. 149 (2004) 8893.
12) H. T. Lee, J. Mayer and F. C. Hsu: Transactions of the ASME 128
(2006) 468475.
13) ASTM Standard E8M, Standard Test Methods for Tension Testing of
Metallic Material (metric), (ASTM, Philadelphia, PA, USA, 1999).
14) S. Kher and A. Dua: SICE Annual, Proceedings of 38th Annual
conference, (1999) pp. 10471052.
15) R. L. Ott and M. Longnecker: Statistical Methods and Data Analysis,
fth ed., (Duxbury, CA, USA, 2001) pp. 9396.
16) J. Lu: Handbook of Measurement of Residual Stresses, (Lilburn, GA,
Fairmont Press, Inc., 1996) p. 229.
Fig. 9 Calibrated (
cal,max
,
cal,min
) and un-calibrated (
m,min
,
m,max
)
results for principal measurement stress.
1910 H. T. Lee, C. Liu, F. C. Hsu and J. M. Hsu

You might also like