Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1007/s11340-015-0104-3
Received: 31 July 2015 / Accepted: 12 October 2015 / Published online: 28 October 2015
# Society for Experimental Mechanics 2015
* S. Schuster
simone.schuster@kit.edu
1
Introduction
Crystallographic textures in technical components might result for instance from process chains like sheet metal forming
(e.g., warm/cold rolling deep drawing) due to the fact that
limited possibilities of gliding exist with respect to the direction of mechanical loading [1]. The extreme case of a crystallographic texture is a single crystal, which possesses only one
distinct crystal orientation. Beside a crystallographic texture,
characteristic residual stress distributions might be induced by
the forming process and most manufacturing processes, respectively. Concerning the trend to light weight constructions
especially in the automotive industry, the knowledge of the
residual stress state is very important for components design,
not least for the validation of finite element simulations.
A multitude of residual stress analysis methods is developed like for instance diffraction methods, mechanical
methods, ultrasonic methods or magnetic methods. However
all of the established methods have their limitations/problems
dealing with anisotropic material behavior. From the vast of
various analysis techniques, X-ray diffraction and the incremental hole-drilling method are most widely used in industry
and scientific applications. The standard procedure of X-ray
diffraction, in particular the sin2-method [2], is not applicable in case of textured materials, since the sin2-distribution is
no longer linear [3, 4]. Special measuring strategies for an
application on textured materials as for instance the crystallite
group method [5, 6] or stress factors F ij [7] are well
established, but they are very elaborate and time consuming
and in addition knowledge of the stress free lattice parameter
d0 for the stress calculation is required. In many cases, in
370
specific calibration functions is analyzed. Finally, the presented approach is used for experimental stress analysis on an
Inconel single crystal with cube orientation.
Theory
Fundamentals
The incremental hole-drilling method is one of the widest used
residual stress analysis techniques and the fundamental approach was first developed by Mathar [14] in 1933. The incremental hole-drilling method is based on the elastic strain
relaxation due to stepwise removal of stressed material in the
region of the hole. Conventionally the released strain for each
drilling increment is measured by means of a dedicated strain
gage rosette. Afterwards, the residual stress is calculated from
the measured strain using generalized Hookes law and calibration data. An alternative to the strain gage rosettes are
optical strain measurement techniques like digital image correlation [15, 16], moir interferometry [1719] or holography
[20, 21]. However, these optical techniques are in general not
as accurate as strain gages and more sensitive to external influences like vibrations or temperature variations [22, 23].
Also the effort for stress computation is often relatively large
due to the huge amount of data. For this reason conventional
hole drilling strain gage rosettes are used for strain determination in this study.
Basically, mainly two evaluation methods are used to calculate residual stresses from the depth dependent strain relaxations, i.e., the integral method [24, 25] and the differential
method [26, 27]. Both evaluation approaches are combined in
the necessity of calibration data regarding that only a part of
the residual stress is released due to the drilling of a hole.
Calibration data can only be determined by means of FEsimulations in case of the integral method. Here, the stress
calculation is based on the trigonometric relation of the released strain around the drilled hole, which is not valid for
anisotropic materials. The calibration constants A and B are
calculated for normal stress and shear stress, respectively, and
are not directly assigned to any directions.
The differential methods allows both, an experimental calibration (tension/bending test) [28, 29] as well as a numerical
determination of the calibration data. This is a big plus regarding textured materials, since the knowledge of the texture by
means of the orientation distribution function and more precisely the effective elasticity tensor is not mandatory. Contrary
to the integral method, the measured strain relaxation is only
assigned to the stress within the actual drilled hole increment
for the differential approach. Residual stresses of previously
drilling steps are not considered for stress calculation; hence
steep stress gradients might be problematic. Thus, for larger
drilling depths the differential method wont be as reliable as
371
dy
dx
C;x
C;y
d
d
h
i
K y
v 2
C;x 2C;y
E
d0-
d90-
vK
x
y
2
d
d
K x 2 v2 K y
E
3
45-
E
K x
v2
d45d0- d90- d45-
vK y
2 K x
d
d
d
d
K y
4
90-
d90-
d0-
vK
x
y
2
d
d
K x 2 v2 K y
E
5
Figure (1) gives an overview of the coordinate system
and the position of the strain gages.
372
q
0- 45- 2 90- 45- 2
6
1
245- 0- 90-
arctan
2
0- 90-
10
11
Consequently, two simulations or experiments are necessary to determine the four case-specific calibration
functions. A graphical overview of the whole residual
stress analysis procedure applying this new evaluation
approach is given in Fig. (2). Due to the establishment
of four calibration functions a modification of the stress
calculation scheme is necessary according to
0-
Ex
d0-
d90-
vxy K xy
K yy
d
d
K xx K yy vxy vyx K xy K yx
12
90-
Ey
d90-
d0-
vyx K yx
K xx
d
d
K xx K yy vxy vyx K xy K yx
13
Basically, the Youngs modulus and the Poisson ratio
can be removed of equations (8, 9, 10, 11, 12, 13),
because it will be canceled down by applying the
case-specifically determined calibration functions during
stress calculation. It comes out that using this new approach a determination of residual stress is possible in
the two orthogonal directions of the strain gauge rosette
only.
373
Fig 2 Scheme of the new evaluation approach using multiple case-specific calibration functions
If the elastic constants are identical in these two measuring directions as it is for example for a cube texture,
the case-specific calibration functions reduce to the two
functions in equation (1 and 2). This way a biaxial
calibration stress state is also applicable to determine
the case-specific calibration functions. Hence, the evaluation algorithms for the calculation of the stresses in
measuring direction 0- and 90- simplify to the conventional equations (3 and 5). In order to proof the potential of our proposed evaluation strategy using four casespecific calibration functions, the stress is calculated
with three different approaches and the stress deviation
serves as evaluation criteria, using:
&
&
&
FE-Simulations
In this study case-specific calibration functions are numerically determined to evaluate the simulated strain relaxations
374
Fig. 3 FE-model for the simulation of the strain relaxations during hole
drilling, applied stress and position of strain gages
x 300MPa; x 150MPa
x 2y
x 300MPa; x 300MPa
x y
&
evaluation nominal
100%
nominal
14
Fig. 4 Youngs modulus distribution within the measuring plane for the investigated texture orientations of a nickel single crystal
375
Fig. 5 Relieved strain of a nickel single crystal in cube orientation with an applied stress of x =2y (a) and the corresponding stress distribution
evaluated conventionally (isotropic) and case-specifically using multiple case specific calibration functions (b)
Fig. 6 Comparison of the stress deviation of a Cube orientation using conventional calibration functions or multiple case-specific calibration functions
for a stress state x =2y (a) and x =y (b)
376
specific calibration functions results in a considerable reduction of stress deviations due to the directional consideration of
material stiffness and strain reaction. All stress distributions
evaluated by the new approach are in good accordance with
the nominal applied stress, since the maximum stress deviations are smaller than 5 %. If a nominal stress state of x =2y
is applied on a nickel single crystal in Goss orientation the
maximum stress deviation for the stress component in direction of 0 is reduced from approx. 32 % to 2 % and in a
direction of 90 from 38 to 5 %. For an axi-symmetric stress
state the 0-component decreases from 34 to 2 % and
the 90-component decreases from 37 to 5 %.
In case of the brass texture (see Fig. (9)), the conventionally
evaluated stress deviates about 8 % and 19 %, respectively, for
the components in 0 and 90 direction, when a stress state of
x =2y is applied. If the stress is evaluated using the proposed
evaluation approach the maximum stress deviation reduces to
1 % in 0 direction and 2 % in 90 direction. A conventional
evaluation for axi-symmetric stress state results in a maximum
Fig. 8 Comparison of the stress deviation of a Goss orientation using conventional calibration functions or multiple case-specific calibration functions
for a stress state x =2y (a) and x =y (b)
377
Fig. 9 Comparison of the stress deviation of a Brass orientation using conventional calibration functions or multiple case-specific calibration functions
for a stress state x =2y (a) and x =y (b)
378
Experimental Validation
Up to this point we only focused on fundamental investigations using finite element simulations, whilst measurement
uncertainties (scatter) were not considered. For this reason
experimental validation of the proposed approach is carried
out for a CMSX-4 single crystal sample, which was subjected
to a defined bending stress distribution. Figure (11) schematically shows the experimental setup as well as the orientation
of the sample by means of the {100} pole figure. The single
crystal obtains a cube orientation with its edges parallel to the
samples edges. Single crystal elastic constants of nickel were
assumed (see Chapter 3). Thus, the same Youngs modulus
distribution for a cube orientation as in Fig. (4) results for the
CMSX-4 single crystal.
The bent beam sample with dimension of 70 mm x 18 mm
x 5.3 mm (L x W x T) was elastically loaded in a four point
bending fixture to a defined surface strain of 3106 m/m. This
load corresponds to a surface stress of about 407 MPa. Prior to
load stress analysis the initial residual stress state was analyzed by means of the incremental hole drilling and slitting
method. The results show that only at the very surface
(<0.1 mm) small amount of grinding residual stresses exist.
For larger drilling depth the bending bar can be assumed to be
stress free (RS <10 MPa). For the hole-drilling stress analysis
a hole-drilling apparatus with an electrical turbine from Newport was used. The rotational speed was 25,000 rpm and the
feed rate was set to 0.02 mm/s. The released strain is measured
by a strain gage rosette of type EA-06-031RE-120 from
Vishay Measurements Group. A TiN coated 2-flute drill bit
with a diameter of 0.795 mm led to a final hole diameter D0 of
Fig. 12 Stress depth profile for the CMSX-4 single crystal during loading stress experiment using a four point bending setup
Conclusion
A new evaluation approach for hole-drilling residual stress
analysis on strongly textured materials was presented. The
proposed evaluation strategy, which bases on the differential
method, takes into account the anisotropic material behavior
during the stress calculation by means of multiple casespecific calibration functions.
From the systematic studies for different ideal orientations
(cube-, Goss-, brass-texture) of a nickel-based single crystal
for different stress states the following conclusions can be
drawn:
&
&
&
&
&
A conventional stress evaluation assuming isotropic materials behavior leads to significant stress deviations, in the
cases studied here up to 49 %, and thus, a reliable stress
analysis is not feasible (see also [13]).
An evaluation using conventional case-specific calibration
functions (i.e., two calibration functions are used), basically leads to no improvement in stress evaluation on
strongly textured materials. Only for cube orientations,
where the effective Youngs modulus distribution is symmetric with respect to the strain gage rosette orientations
the conventional case-specific approach results in a considerable advancement.
The application of the proposed evaluation approach using
multiple case-specific calibration functions (i.e., four casespecific calibration functions are considered) reduces the
stress deviation (referring to conventional calibration
based on isotropic materials behavior) to 15 % and the
evaluated stress is in good accordance to the nominal applied stress.
For the implementation of the new approach a higher simulation effort must be invested, which is unquestionable
worthy in case of strongly textured materials, where all
other residual stress analysis approaches have their limitations. Further the ODF must be known beforehand.
The usage of multiple case-specific calibration functions
is still applicable, when the rosette orientation does not
perfectly match between the hole-drilling measurement
379
&
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
380
14.