Beruflich Dokumente
Kultur Dokumente
Introduction
List of Symbols
The hole-drilling-method 1 5 is a well-established,
A, B, C = calibration constants popular technique for measuring residual stresses in a
c** = orthotropic strain relief compliances wide range of engineering materials. The method is
Ex, Ey = elastic moduli along x and y (elastic sym- easy to use, reliable in operation, and involves only
metry) axes limited damage to the specimen.
G~y = x - y shear modulus The conventional hole-drilling method can be used
m = orthotropic elastic modulus ratio [eq (10)] only with isotropic materials. However, many mod-
r~ = hole radius em materials, such as fiber-reinforced composites, have
r,, - mean radius of strain-gage rosette distinctly anisotropic elastic properties. Bert et al. 6'7
u, v = displacements in x and y directions and Prasad et al. 8 have generalized the computational
x, y - coordinates along elastic symmetry axes procedure for the hole-drilling method to extend the
W1, W2 = geometrical parameters [eqs (14) and (15)] use of the method to orthotropic materials. However,
XI, X2 = geometrical parameters [eqs (18) and (19)] this generalization is shown here not to be valid. This
Y1, I12 = geometrical parameters [eqs (20) and (21)] paper presents a different solution method that can be
a, 13 = orthotropic elastic material constants [eqs used for materials of any degree of elastic orthotropy.
(12) and (13)] An experimental example is presented to illustrate the
~,:y = x - y Cartesian shear strain use and applicability of the method.
8r measured relieved strain
=
• V((O" x __ 2
0",)/2) 2 -'1- "rxy (3)
and
1 [ 2"rxy ]
6 = - arctan
- 2 L~x - %J (4)
X
Orthotropic Material Properties
For the two-dimensional case, five elastic constants
are required to relate the Cartesian stresses and strains
in an orthotropic material, m'u When the x and y axes
lie along the principal elastic directions of the mate-
rial, Hooke's Law generalizes to
% = ~/6~ (5)
Only four of the five elastic constants are independent
When making residual-stress measurements, a cir- because of the elastic symmetry relationship
cular hole is drilled at the geometrical center of the
rosette to a depth slightly greater than the hole di- V~y/E~ = Vy~/Ey (6)
ameter. This hole locally relieves the stresses in the
surrounding material, and the associated strain reliefs In general, the shear modulus Gxy is independent of
are measured by the three strain gages. For an isotro- all other elastic constants. However, in the isotropic
pic material, the relieved strain measured by a strain
case, Ex = Ey = E, Vxy = Vyx = v, and G,~ = G =
gage whose axis is inclined at an angle 0 from the x
0.5E/(1 + v). There are then only two independent
direction is elastic constants and eqs (5) reduce to their more fa-
miliar forms.
er = A(~r~ + %) + B(% - %) cos 20 An interesting special case occurs when the shear
modulus of an orthoU-opic material happens to be re-
+ C %y sin 20 (1)
lated to the other elastic constants as follows.
stresses %, % and "rxydo not vary with depth from G,~ Ex Ey (7)
the specimen surface. The calibration constants A, B
and C depend on the material properties, the rosette In this particular case, the shear modulus is the same
geometry, the hole diameter and the hole depth. For in all directions. The material has isotropic shear be-
an isotropic material, C = 2B. 7 The calibration con- havior, but orthotropic axial behavior. An approxi-
stants can either be determined experimentally' or nu- mately opposite case occurs when Ex = Ey but G #
mericallyfl 0.5E/(1 + v). The latter material has an orthotropic
Equation (1) can be rewritten in matrix form to re- shear modulus, with equal (but not isotropic) princi-
late the three measured strains gl, 1~2and ~3 in Fig. 1 pal axial moduli. These two types of orthotropy are
to the Cartesian stresses ox, % and "rxy. considered in subsequent sections.
A ex = ~--~ A A
where
m = 4VEx/Ey (I0)
x c12c137E
j xEll
IC2, C22C231 (22)
C31 C32 C33J ,228 0 - . 6 5 9 ] (23)
Ex/ Ey Vxy Gxy/ Ey c,1 c,3 c2, c22 c23 c3, c33
E~/Ex ~'yx G~y/E~ c33 c31 c23 c22 c2~ c~3 c~1
1 .00 .10 -.591 .169 -.291 1.180 -,291 .169 -.591
1 .25 .10 -.583 .123 -.314 1.097 -,314 .123 -.583
1 .50 .10 -.575 .076 -.336 1.013 -,336 .076 -.575
1 .75 .10 -.568 .030 -.357 .930 -,357 .030 -.568
1 .00 .20 -.514 .193 -.188 .877 -,188 .193 -.514
1 .25 .20 -.503 .146 -.205 .788 -,205 .146 -.503
1 .50 .20 -,491 .098 -.221 .699 -,221 .098 --.491
1 .75 .20 -.478 .049 -.235 .609 -,235 .049 -.478
1 .00 .30 -.474 .205 -.146 .757 -.146 .205 -.474
1 .25 .30 -.460 .157 -.160 .665 -.160 .157 -.460
1 ,50 .30 -.445 .109 ".172 ,572 -.172 .109 -.445
1 .00 .40 -.450 .213 -.123 ,692 -.123 .213 -.450
1 .25 .40 -.433 .164 -.135 .598 -.135 .164 -.433
1 .00 .50 -.433 .218 -.108 .650 -.108 ,218 -.433
2 .00 .15 -.453 .156 - .226 1.156 -.345 .185 -,743
2 .25 .15 -,448 .122 -.244 1.097 -~357 .153 -.737
2 ,50 .15 -.443 .088 -.263 1.038 -.369 .122 -.730
2 .75 ,15 -.438 .053 -.281 .978 -.380 .090 -.723
2 .00 .30 -.403 .180 -,144 ,865 -.225 .207 -.631
2 .25 .30 -.396 .145 -.159 .802 -.233 .175 -.621
2 .50 .30 -.389 .110 -.174 .739 --.240 .143 --.611
2 .75 ,30 -.382 .074 -.188 .675 -.247 .110 -.600
2 .00 .45 -.377 .193 -.111 .751 -.177 .217 -.576
2 .25 .45 -.369 .158 -.124 .686 -.182 .185 -.563
2 .50 .45 -.360 .122 -.137 .620 -.187 .152 -.550
2 .75 .45 -.350 .085 -.149 .554 -.191 .119 -.536
2 .00 .60 -,361 .201 -.092 .689 -.150 .223 -.541
2 .25 .60 -.351 .165 --.105 .623 -.154 .191 -.527
4 .00 ,20 -.350 ,137 -.186 1,212 -.435 .195 -.949
4 .25 .20 -.347 .112 -.201 1.171 -.441 .174 -.944
4 .50 .20 -.344 .088 -.216 1.129 -.447 .152 -.938
4 .75 .20 -.341 .063 -.232 1.087 -.453 .131 -.932
4 .00 .40 -.318 .162 -.116 .910 -.288 .215 -.787
4 ,25 .40 -.314 .136 -.130 .866 -.291 .193 -.778
4 .50 .40 -.310 .110 -.143 .821 -.294 .171 -.770
4 .75 .40 -.306 .084 -,156 .776 -.296 .149 -.761
4 .00 .60 -.301 .175 -.088 .792 -.229 ,223 -.707
4 .25 .60 -.296 .149 -.101 .746 -.230 .201 -.696
4 .50 .60 -.291 .122 -.112 .700 -.231 .179 -,685
4 .75 .60 -.286 .096 -.124 .653 -.231 .157 -.674
4 .00 ,80 -.291 .183 -.073 ,728 --.196 ,228 --.659
4 .25 .80 -.285 .157 -.084 .681 -.196 .206 -.646
4 .50 .80 -.279 .130 -.095 .634 -.196 .184 -.633
4 .00 1.00 -.283 .189 -.063 .687 -.175 .231 -.626
8 .00 .30 -,263 .123 -.138 1.224 -.508 .207 -1.160
8 .25 .30 -.262 .105 -.150 1.194 -.510 .192 -1.155
8 .50 .30 -.260 .087 -.162 1.164 -.513 .177 -1.149
8 ,75 .30 -.258 .068 --.174 1.135 -.515 ,163 -1.144
8 .00 .60 -.244 .146 -.084 .935 -.343 .222 -.939
8 .25 .60 -.242 .127 -.095 .903 -.343 .207 -.931
8 .50 .60 -.239 .108 -.105 .871 -.342 .192 -.923
8 .75 .60 -.236 .089 -.116 .839 -.342 .177 -,914
8 .00 .90 -.234 .158 -.062 .823 -.278 .228 -.834
8 .25 .90 -.231 .139 -.072 .790 -.276 .213 -.824
8 .50 .90 -.228 .120 -.082 .757 --.275 .198 --.814
8 .75 .90 -.224 .100 -,092 .724 -.273 .183 -.803
8 .00 1.20 -,227 .166 -,050 .763 -.242 .231 -.771
8 .25 1.20 -.224 .147 -.060 .729 -.240 ,216 -.760
8 .50 1.20 -.220 .127 -,069 .696 -.237 ,201 -.748
16 .00 .40 -.199 .105 -.109 1.323 -.634 .213 -1.450
16 .25 ,40 -.198 .092 -.118 1.301 -.634 .203 -1.446
16 .50 .40 -.196 .079 -.127 1.280 -.634 ,193 -1.441
16 .75 .40 -.195 .065 -.136 1.259 -.634 .182 -1.437
The strain responses in column 5 of Table 2 are expected from c~z in eqs (2) and (22). A numerical
particularly sensitive to measurement errors because error is suspected because their calculated C values
they derive from the difference between the much larger do not approach 2B in the two nearly isotropic ex-
values in columns 3 and 4. In most cases, the values amples that they examine. The results in columns 5
in column 5 are less than half of those in column 3. and 6 are believed to be reliable because the calcu-
Thus, any errors in columns 3 and 4 have magnified lation method gives C = 2B for an isotropic material
relative effects on column 5. and also gives displacement fields identical to those
Rosette angular misalignment was found to be an reported by Schimke e t al. 11
additional source of error in Table 2. For the 0-deg Column 6 of Table 4 lists the predicted strain re-
and 90-deg specimens before hole drilling, the read- sponse when the finite areas of the hole-drilling ro-
ings from gage 2 are expected to be the average of sette gages are taken into account. The calculation
the readings from gages 1 and 3. This is true for the method is the same as used for column 6 of Table 2.
0-deg specimen but not for the 90-deg specimen. The In some cases, the difference between the 'point' and
discrepancy was traced to a 1.5-deg error in the ro- finite-area gage calculations are quite significant. In
sette alignment. This alignment error was taken into all cases, the relieved strain responses listed in col-
account when computing the values in column 6. umn 6 of Table 4 correspond more closely with the
measured values in column 3 than do the 'point' gage
values in columns 4 and 5. These results demonstrate
Comparisons with Other Published Data the significance of using the finite-area strain-gage
Prasad e t al. 8 did a series of hole-drilling measure- calculation.
ments, similar to those reported here, using a graph-
ite-polyimide laminate. They also made calculations
Residual-stress Measurement Accuracy
of the strain responses expected during their experi-
mental measurements. Their mathematical method The discussion so far has focused on how well the
differs from that presented here, notably by their choice proposed calculation method predicts the measured
of working in terms of strains. To simplify their strain- strain responses in calibration tests. This assessment
based computations, Prasad e t al. approximated each provides an important measure of the theoretical
strain gage in the hole-drilling rosette as being con- method. However, in practice, the question of interest
centrated at a point. In the present study, displace- is "What level of accuracy can be expected when ma-
ment-based calculations were chosen because this ap- trix eq (22) and theoretical compliance values are used
proach lends itself very conveniently to computin~ the to evaluate residual stresses from experimental strain
response of practical strain gages of finite area. measurements?" A related question is "What is the
Table 4 lists the measured strain responses of Pra- consequence of using the trigonometric assumption,
sad e t al. and the corresponding theoretical values, eq (1), rather than the matrix eq (22)?" These ques-
calculated in three different ways. Column 3 lists the tions are examined here.
measured strain responses, and column 4 lists Pra- Table 5 lists the residual stresses that would be cal-
sad's corresponding calculated values. Column 5 lists culated for the five tensile specimens whose measured
the strain responses calculated from eqs (8)-(21) as- strain data are reported in Tables 2 and 4. Each spec-
suming a strain gage concentrated at a point. These imen supports a purely longitudinal stress, which for
results are mostly the same as those of Prasad et al. simplicity of comparison has been normalized to 1.00
The differences between columns 4 and 5 for the 45- MPa. The table lists the stresses that would be cal-
deg specimen are due to a suspected numerical error culated from the measured strains in column 5 of Ta-
by Prasad e t al. in calculating shear strain response. ble 2 and column 3 of Table 4, using the finite-area
In their study, they worked in terms of constants A, compliance values in columns 6. The actual applied
B and C in eq (1). Their value of C is smaller than stresses are shown in parentheses. In general, the cal-
TABLE 5--COMPUTED STRESSES FOR A 1.00 MPa APPLIED LONGITUDINAL STRESS APPLIED STRESSES
IN PARENTHESES)
Computed Stress, MPa
Specimen cr,~ "rxy %, ~rr,~ ~rm,n s
0 deg 1.09 .04 .11 1.09 .11 3 deg
(1.00) (.00) (.00) (1.00) (.00) (0 deg)
45 deg .57 -.58 .61 1.17 .01 46 deg
(.50) (-.50) (.50) (1.00) (.00) (45 deg)
90 deg .06 -.07 1.00 1.00 .05 96 deg
(.00) (.00) (1.00) (1.00) (.00) (90 deg)
Tension 1.12 -.04 .03 1.12 .03 - 2 deg
aef. 8 (1.00) (.00) (.00) (1.00) (.00) (0 deg)
Shear - . 19 1.12 .04 1.04 - 1.20 48 deg
aef. 8 (.00) (1.00) (.00) (1.00) (-1.00) (45 deg)
culated stresses are somewhat higher than the actually centage increase corresponds well with the results in
applied stresses, up to 20 percent higher in extreme Table 5.
cases. This over-estimation of the actual stresses is Table 6 compares the calculated stresses for the 0-
believed to be a consequence of the laminar structure deg and 90-deg specimens used in this study, deter-
of the test material. mined using matrix eq (22) and also using the trigo-
When using eqs (8)-(21) to calculate the strain re- nometric assumption, eq (1). Column 2 of Table 6
sponses in Tables 2 and 4, the assumption is made lists the A, B and C values that would be used for
that the graphite-epoxy laminate is a homogeneous each specimen, determined from longitudinal tension
continuum. However, in reality, the test material con- calibration tests. Column 3 lists the corresponding A,
sists of 24 discrete layers, each about 0.15-mm thick. B and C values from hypothetical transverse tension
In such a case, the continuum assumption is reason- calibration tests. The values in columns 2 and 3 sig-
able only for macroscopic features that extend over nificantly differ. The longitudinal A, B and C values
regions significantly larger than one layer thickness. for one specimen should normally equal the trans-
For the tests done in this study, the hole is 4.8 mm verse values for the other specimen. They are not ex-
in diameter, which is much larger than the 0.15-mm actly equal here because of the 10-percent difference
layer thickness. Thus, the hole can be expected to be in elastic properties of the two specimens.
a 'macroscopic' feature. However, St. Venant's prin- Column 4 of Table 6 lists the stresses calculated
ciple suggests that the laminar structure is likely to using eq (22) for a nominal longitudinal tension of
disturb the continuum assumption within about one 1.00 MPa. These values are the same as those listed
layer thickness from the hole boundary. Thus, the ef- in Table 5. Column 5 lists the stresses calculated us-
fective hole diameter could be expected to be slightly ing eq (2) with the 'longitudinal' A, B and C values
larger than the actual hole diameter. If the enlarge- from column 2. These calculated stresses are similar
ment in effective hole radius is one layer thickness, to those from eq (22) in column 4. Column 6 lists the
then the expected increase in strain response is about stresses calculated using eq (2) with the 'transverse'
12 percent. (The strain response is approximately pro- A, B and C values from column 3. These calculated
portional to the square of the hole diameter). This per- stresses greatly differ from the expected values, and