Professional Documents
Culture Documents
On The Nature of The Yield Point Phenomenon
On The Nature of The Yield Point Phenomenon
On The Nature of The Yield Point Phenomenon
com
Received 25 September 2012; received in revised form 2 December 2012; accepted 3 December 2012
Available online 25 December 2012
Abstract
Materials in which the yield point phenomenon occurs exhibit a characteristic drop in the stress–strain curve at the end of the elastic
(or mainly elastic) region. In most cases the subsequent plastic deformation takes place locally in deformation bands, often termed
Lüders bands. Although first described more than 150 years ago, the yield point phenomenon is not yet fully understood. In this paper,
a new and simple macromechanical model is presented to explain the yield point phenomenon. It is based on three major points: (i) a
realistic or true upper yield point, (ii) typical strain-hardening behaviour common to many materials and (iii) the triaxiality of the stress
state that necessarily has to develop at the Lüders front and that determines the stress level at the lower yield strength. The model is
evaluated by experimental, analytical and numerical evidence. Combining all evidence, the model basically describes the yield point phe-
nomenon correctly. It may also be applied to related strain-softening phenomena and it explains the comparatively low ductility of some
nanocrystalline materials.
Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Yield strength; Yield phenomenon; Yield behaviour; Steel; Lüders effect
1359-6454/$36.00 Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.actamat.2012.12.003
R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808 1799
2. The model
Fig. 2. True and observed stress–strain behaviour of a typical structural steel in the tensile test.
the specimen and the full Lüders strain is observed [46]. stresses in the elastic region (r2e and r3e), and to tensile lat-
Sometimes single Lüders bands appear, sometimes multiple eral stresses in the plastic region (r2p and r3p).
bands, and in many cases also complex bands [1]. Lüders Some distance away from the Lüders front, the stress
fronts in tensile tests are mostly flat, in the case of complex state is uniaxial, as normally to be expected in the standard
bands the fronts may be curved. An essential point is that tensile test. So the upper small grey volume element (marked
the plastic deformation observed near well-developed with an asterisk in Fig. 5) within the elastic region will expe-
Lüders fronts in steel is not a pure shear deformation, rience a uniaxial stress state at the stress level of the observed
but a combination of shear deformation and a deformation lower yield strength ReL(obs). Almost the same stress acts in
perpendicular to the Lüders front [43,45,47]. the grey volume element in the plastic region, also marked
To conclude, experimental results strongly support the with an asterisk, if surface effects and the smaller cross-sec-
first two points of our model; some give a hint for the third tional area are neglected. The plastic region some distance
point. Now, if this is accepted for the moment, why is the away from the Lüders front has strain-hardened to the stress
true material behaviour not observed in the tensile test? level of the observed lower yield strength ReL(obs).
The most important part of our model certainly is the To further explain our model, the stress states will be
occurrence of a triaxial stress state that necessarily has to represented with the help of Mohr’s stress circles in s–r
develop at a Lüders front and that hides the true material diagrams, showing shear and normal stresses in differently
behaviour. To show this, some analytical calculations are inclined planes. The state of stress in the two grey volume
presented in the next section. elements is shown by the grey circle in the s–r diagram on
the right side of Fig. 5. Directly at the Lüders front, two
4. Analytical evidence material states – an elastic and a plastic one – must coexist.
Coexistence means that the equilibrium of forces and geo-
To begin with, a Lüders front at an angle u of 90° to metrical compatibility have to be fulfilled. The stress states
the tensile axis is considered, as shown in Fig. 5. This is on the “still-elastic” and “just-yielded” sides of the Lüders
not the typical inclination that in most cases will develop front are shown with the red and blue volume elements
in reality, but it is shown here to explain the principles of respectively; the corresponding stress circles are shown
the phenomenon. Typical strain-hardening behaviour with the same colours, additionally marked with E and P.
in the plastic region and fine-grained material with Due to specimen symmetry no shear stresses act on the
homogenous and quasi-isotropic material behaviour are red and blue volume elements. All normal stresses, r1e,
assumed. r2e, r3e, r1p, r2p and r3p, are principal stresses.
Four main ideas are used to explain the observed lower
4.1. Lüders front perpendicular to tensile axis yield strength ReL(obs).
On the left side of Fig. 5 a tensile specimen with a 4.1.1. Position of the stress circles relative to each other
Lüders front about halfway through the cylindrical part The red and blue volume elements are positioned next to
of the specimen can be seen. The middle part of Fig. 5 each other. Therefore the normal stresses r1e and r1p have
shows an enlargement of the Lüders front region, the spec- to be equal:
imen deformation being exaggerated. Clearly there is a r1e ¼ r1p ð1Þ
large difference between the lateral contraction in the elas-
tic and in the plastic region of the specimen. Directly at the This fact relates the position of the elastic and plastic
Lüders front this invariably leads to compressive lateral stress circles to each other.
Fig. 5. Stress state at a Lüders front with an inclination of 90° to the tensile axis in a tensile test.
1802 R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808
4.1.2. Middle principal stresses small region near the surface the described deviations
Due to specimen symmetry, and due to symmetrical lat- exist. This is well supported by finite element calculations.
eral contraction, in a cylindrical specimen r2e and r3e have To conclude, Eqs. (1)–(6) should be a first-order
to have the same magnitude: approximation.
r2e ¼ r3e ð2Þ
4.1.5. Results
Similar considerations can be made on the plastic side of Solving the six equations for the six variables leads to
the Lüders front, where freshly yielded material with the obvious solution
hardly any strain-hardening exists:
ðReHðtrÞ þ ReLðtrÞ Þ
r2p ¼ r3p ð3Þ ReLðobsÞ ¼ ð7Þ
2
Thus Eqs. (2) and (3) determine the level of the middle So Eq. (7) gives the solution for the observed lower yield
principal stresses relative to the minimum principal strength ReL(obs) with the Lüders front perpendicular to the
stresses. tensile axis. During the tensile test a small volume element,
which is still in the elastic region of the specimen, would
4.1.3. Yield criteria experience lateral compressive stresses with a Lüders front
The equivalent stresses for plastic yielding on the elastic approaching. When the Lüders front arrives at the volume
and on the plastic side of the Lüders front have to satisfy element, the von Mises yield criterion for the true upper
the true material model. So, on the elastic side of the yield strength ReH(tr) will be reached at nominal stress level
Lüders front, the equivalent stress in the red volume ele- ReL(obs). Plastic yielding then takes place with a large drop
ment has to be equal to the true yield strength ReH(tr). of equivalent stress down to the true lower yield strength
Using the von Mises criterion for plastic yielding and Eq. ReL(tr). This is accompanied with lateral tensile stresses,
(2) leads to causing the nominal stress r to rise to the observed lower
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi yield strength ReL(obs). Shortly afterwards the material
1 2 2 2
ReHðtrÞ ¼ pffiffiffi ðr1e r2e Þ þ ðr2e r3e Þ þ ðr3e r1e Þ strain hardens to the nominal stress level of the observed
2
lower yield strength ReL(obs).
¼ r1e r3e ð4Þ
Hence there is always a mechanical equilibrium between
This also means that the diameter of Mohr’s stress circle the stress level at the Lüders front and the strain-hardened
on the elastic side of the Lüders front equals the true upper plastic region. If a small amount of further strain-harden-
yield strength, illustrated in Fig. 5. Similarly the von Mises ing occurs in the plastic region, the Lüders front will carry
criterion on the freshly yielded plastic side of the Lüders on moving towards the still elastic region, until the whole
front and Eq. (3) lead to uniform specimen length has yielded.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Lüders fronts perpendicular to the tensile axis are not
1
ReLðtrÞ ¼ pffiffiffi ðr1p r2p Þ2 þ ðr2p r3p Þ2 þ ðr3p r1p Þ2 observed in most cases. The main reason for this is the dif-
2 ficulty of the material to deform plastically in the lateral
¼ r1p r3p ð5Þ direction along an initially narrow band, especially at the
initiation stage. Therefore a Lüders front with a typical
4.1.4. Absolute values of the principal stresses inclination is considered in the next step.
Eqs. (1)–(5) only relate the principal stresses to each
other; no absolute values are obtained. Therefore, one 4.2. Lüders front with typical inclination to tensile axis
more equation has to be set up:
Again, a Lüders front partly through the parallel length
r3e ¼ r3p ð6Þ
of the specimen is regarded (see Fig. 6). An essential part of
At first glance, Eq. (6) seems to be logical; it is based on our model is the experimental fact that plastic deformation
the equilibrium of forces. Nevertheless, it is only approxi- close to the Lüders front is not of a pure shear type, but is a
mately valid and should rather be regarded as a reasonable combination of shear deformation and a deformation per-
guess. This fact, however, does not affect our model funda- pendicular to the Lüders front [43,45,47]. The typical spec-
mentally, only qualitatively. Moreover, the stress state imen shape is shown exaggeratedly in the middle part of
along the Lüders front is not homogeneous. The lateral Fig. 6. It is worth mentioning that this state only applies
stresses r3e and r3p must be equal to zero at the specimen for a well-developed broad single Lüders band, and not
surface; shear stresses in the Lüders front are only equal for a narrow band in the initiation stage.
to zero at the specimen surface and in the central region. In an analogous way to Section 4.1, small volume ele-
On the other hand, the lateral stresses near the Lüders ments show the states of stress in different places. Far away
front only exist in a comparatively narrow region com- from the Lüders front, the stress state is mainly uniaxial. So
pared with the specimen diameter. As a consequence, the the upper grey volume element in the elastic region, marked
stress state along the Lüders front will be almost homoge- with an asterisk, experiences a uniaxial stress state at stress
neous for a wide central area, and only in a comparatively level ReL(obs). The same applies for the lower grey element
R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808 1803
Fig. 6. Stress state at a Lüders front with typical inclination in a tensile test.
in the plastic region, also marked with an asterisk, if again sides of the Lüders front. Again, the lateral contractions
surface effects and the smaller cross-sectional area are are assumed to be symmetric, but now certain problems
neglected. The state of stress in these two elements is repre- arise to determine the middle principal stresses logically.
sented by the grey stress circle on the right side of Fig. 6. The A reasonable guess, based on the equilibrium of forces, is
stress states on the “still-elastic” and “just-yielded” sides of to set r2e equal to r3e and r2p equal to –r2e (see Fig. 6):
the Lüders front are shown with the red and blue volume r2e ¼ r3e ð9Þ
elements, respectively; the corresponding stress circles are
drawn with the same colours and marked with E and P. r2p ¼ r2e ð10Þ
Now, in a similar way to Section 4.1, six equations can Instead of Eqs. (9) and (10), related ideas may be
be set up. r1e, r2e and r3e are the principal stresses in the applied; in most cases, the results do not deviate much,
red volume element on the elastic side of the Lüders front. since the influence of the middle principal stresses on yield-
r1p, r2p and r3p are the principal stresses on the plastic side. ing is limited.
It is essential that both the inclination u as well as the
observed lower yield strength ReL(obs) are not predeter- 4.2.3. Yield criteria
mined. They will be the result of the calculation. Similar to Eqs. (4) and (5), the equivalent stresses on
The main ideas are fourfold again. both sides of the Lüders front have to reach the true upper
respectively q lower yield strength. Using Eq. (9) leads to
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
4.2.1. Position of the stress circles relative to each other ReHðtrÞ ¼ pffiffiffi ðr1e r2e Þ2 þ ðr2e r3e Þ2 þ ðr3e r1e Þ2
At this stage a significant hypothesis has to be made: 2
¼ r1e r3e ð11Þ
The Lüders front plane is the plane, where the maximum qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2 2 2
shear stress acts on the plastic side. Or in other words: ReLðtrÞ ¼ pffiffiffi ðr1p r2p Þ þ ðr2p r3p Þ þ ðr3p r1p Þ
the maximum respectively minimum points of the 2
blue circle (P) are the intersection points with the red ð12Þ
circle (E).
4.2.4. Absolute values of the principal stresses
This hypothesis appears to be reasonable; it expresses To obtain the absolute values of the principal stresses, a
the fact that plastic deformation under these geometrical similar assumption to Section 4.1.4 has to be made. Here it
restrictions is easiest in a plane of maximum shear stress. means that the grey stress circle intersects the dashed black
In analogy to Eq. (1), the position of the stress circles rel- horizontal line in the centre of the dashed line (grey point
ative to each other is determined by their intersection 1). Using the theorem of Pythagoras, this is expressed by
points. Using the theorem of Pythagoras leads to r þ r r1e þ r3e 2
1p 3p 2
r r 2 r þ r þ ¼ ðr1p r3p Þ
1e 3e 1p 3p r1e þ r3e 2 r1p r3p 2 2 2
¼ þ r þ r
2 2 2 2 1p 3p r1e þ r3e 2
þ
or simplified: 2 2
2 2 2 or simplified:
ðr1e r3e Þ ¼ ðr1p þ r3p r1e r3e Þ þ ðr1p r3p Þ ð8Þ
2
ðr1p þ r3p Þðr1e þ r3e Þ ¼ ðr1p r3p Þ ð13Þ
4.2.2. Middle principal stresses Again, this assumption is a reasonable guess; again it
The normal stresses rne and rnp are responsible for the only affects the results qualitatively and does not contradict
lateral contractions, respectively compressions on both the fundamental idea of our model.
1804 R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808
4.2.5. Results So the observed lower yield strength ReL(obs) and the
Solving Eqs. (8)–(13) gives the following results, using inclination u of the Lüders front are only a function of
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R ¼ R2eHðtrÞ þ 8R2eLðtrÞ as an abbreviation: the true upper and true lower yield strength. Fig. 7 shows
the analytical results graphically in a normalized form.
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
2 2 The ReL(obs)/ReL(tr) ratio as well as u are plotted vs. ReH(tr)/
1 14R eHðtrÞ 2R eHðtrÞ R 8R eLðtrÞ
r1e ¼ @5ReHðtrÞ þ RA ReL(tr). For comparison, the results for a Lüders front at
8 3 90° to the tensile axis (Section 4.1) are included.
According to the analytical results, a Lüders front incli-
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
nation u of 45° has to be expected at an ReH(tr)/ReL(tr) ratio
1@ 14R2eHðtrÞ 2ReHðtrÞ R 8R2eLðtrÞ
r2e ¼ 3ReHðtrÞ þ RA of 1. The larger the ReH(tr)/ReL(tr) ratio, the more u
8 3 approaches 90°. The result ReL(obs)/ReL(tr) for free inclina-
tion of the Lüders front is always lower than for predeter-
r3e ¼ r2e mined inclination of 90° to the tensile axis.
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi It would be most interesting to compare these results
1@ 14R2eHðtrÞ 2ReHðtrÞ R 8R2eLðtrÞ with experimental data. Unfortunately both ReH(tr) and
r1p ¼ ReHðtrÞ þ
8 3 ReL(tr) are very difficult to measure by experiments. But
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 taking Eq. (14), solving it for ReL(tr) and making use of
8R2eHðtrÞ þ 8ReHðtrÞ R þ 32R2eLðtrÞ Hutchinson’s data, the true lower yield strength of Armco
þ þ RA iron can be back-calculated (see Fig. 4). Most of the back-
3
calculated values of ReL(tr), in Fig. 4 termed ReL(tr, calculated),
r2p ¼ r2e appear reasonable. The data obtained for room tempera-
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ture (293 K) are not easily explained at the moment.
1@ 14R2eHðtrÞ 2ReHðtrÞ R 8R2eLðtrÞ
r3p ¼ ReHðtrÞ þ 5. Numerical evidence
8 3
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
Several hundred finite element calculations have been
8R2eHðtrÞ þ 8ReHðtrÞ R þ 32R2eLðtrÞ
þ RA carried out to evaluate our material model using the Aba-
3 qus and CalculiX code. The general problems and restric-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tions of numerical calculations are well known and need
ReHðtrÞ þ R ReHðtrÞ þ R2eHðtrÞ þ 8R2eLðtrÞ not be discussed here. One problem, however, deserves spe-
ReLðobsÞ ¼ ¼ ð14Þ cial attention, since it is of vital importance. It is the extre-
4 4
0 v ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
u 2 mely sharp and large drop of stress at the yield point and
p 1 u 4ReLðtrÞ R2eHðtrÞ þ ReHðtrÞ R the resulting sharp spatial stress and strain gradients near
u ¼ arctan @2t 2 A ð15Þ the Lüders front. That makes it particularly difficult for
2 2 7ReHðtrÞ 4R2eLðtrÞ ReHðtrÞ R
finite element programmes to come to stable and good
solutions. Further attendance has to be given to the fact
that the numerically calculated Lüders fronts tend to form
along the element edges.
In consideration of this, typical parameter studies were
carried out, varying
shapes, sometimes no Lüders bands are calculated, since all as well, but the problem of the numerical calculation to
elements behave completely symmetrically. In these cases a cope with the resulting large spatial stress and strain gradi-
slight asymmetry or a weak trigger point has to be included ents remains. Furthermore, dynamic analysis needs a cer-
in the modelling. tain amount of damping, a point that still has to be
As an example, the results of a finite element simulation investigated.
of a typical tensile test of a structural steel will be pre- The smoothing out of the yield point drop has several
sented. The simulation was carried out using the Abaqus consequences. On the one hand, it enables stability and
Standard code, large deformation mode, static three- probably leads to better numerical results, because the
dimensional calculation with default stabilizing option stress gradients are not so extreme. On the other hand,
and 15,000 hexahedral elements, type C3D20. Fig. 8 the flow curve is altered considerably. Nevertheless, we
shows the specimen shape, the elements, the main bound- would like to present this static analysis, since there were
ary conditions and the material model. For the material no further parameter adjustments necessary.
model a realistic set of data was chosen with ReH(tr) = 500 The results of the finite element calculations are shown
N mm2, ReL(tr) = 150 N mm2 and a typical flow curve. in Fig. 9. In the upper left hand corner the resulting
The simulated specimen was loaded in a similar way to a stress–strain diagram is displayed. For easy and realistic
real tensile test. The upper ring area at the upper specimen comparison with real tensile tests, the nominal stress is
shoulder was held in constant vertical position; the lower plotted vs. nominal strain of the actual cylindrical test
ring area at the lower specimen shoulder was moved down- region. The nominal strain is extensometer based; it is
wards to simulate a strain controlled tensile test. A trigger determined by e = (L1 + L2)/2L0 (see Fig. 8), with
element with a slightly lower strength (ReH(tr) = 490 L0 = 50 mm. Since symmetrical loading and homogeneous
N mm2) than all the other elements served as starting as well as isotropic material behaviour were used, the true
point for the Lüders band. upper yield strength ReH(tr) of 500 N mm2 is obtained.
Several tests using static analysis showed that the steep The very steep yield drop originates partly from the fact
fall of the stress–strain curve at the yield point leads to that the Lüders band starts outside the gauge length.
instability in the calculation, a problem that has to be The finite element result of the lower yield strength
expected. As a consequence, the vertical drop at the yield ReL(FEM) of 290 N mm2 compares well with the analyt-
point was smoothed more and more in the calculations, ical prediction. Inserting ReH(tr) = 500 N mm2 and
until stability was achieved. A true plastic strain of ReL(tr) = 150 N mm2 into Eq. (14) leads to
0.0075 at the lower yield point (see Fig. 8) proved to be suf-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ficient under the given conditions. Other simulations using
ReHðtrÞ þ R2eHðtrÞ þ 8R2eLðtrÞ
the implicit dynamic mode and sharp drops in stress at the ReLðobsÞ ¼ ¼ 289 N mm2
yield point proved to come to good and very similar results 4
Fig. 8. Specimen shape, elements and material model used for finite element calculations.
1806 R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808
Fig. 9. Results of a finite element simulation; the equivalent plastic strain distribution shown is the strain at the specimen surface.
Although other finite element calculations also show points, the von Mises equivalent stresses reach ReH(tr) and
good correlation with the analytical results, the very good ReL(tr) at the Lüders front, as to be expected.
concurrence in this case may rather be regarded as coinci- Further preliminary finite element calculation results
dence. But on the whole, the stress–strain diagram com- may briefly be summarized as follows.
pares very well with a real tensile test.
In the upper right hand side of Fig. 9 the distribution of The calculated lower yield strength ReL(FEM) generally
the equivalent plastic strain at the specimen surface is pre- follows the trend given by Eq. (14).
sented. Since the Lüders front is almost planar, the plastic The angle of inclination u is generally lower in the finite
strain distribution shown in this side view resembles the element simulations than predicted by Eq. (15), but an
distribution along a longitudinal section through the spec- increase of u with increasing ReH(tr)/ReL(tr) ratio accord-
imen axis. The angle of inclination u of the Lüders band is ing to Fig. 7 is calculated, though to a much smaller
55° at the initiation point and increases to 62° during the amount.
simulated tensile test. In this case the analytically predicted Slightly asymmetrical loading in the tensile test leads to
angle of 72° (by Eq. (15)) differs substantially. The exagger- a remarkable suppression of the observed upper yield
ated deformation shown in the lower right-hand corner of strength.
Fig. 9 seems to be in accord with the main ideas of Simulations of bending and torsion tests match up well
Section 4.2. with experimental data.
The stress–strain diagram in the lower left hand corner Simulations of deep drawing processes show typical
of Fig. 9 shows the stress history of an integration point stretcher strains that are observed in reality.
almost in the centre of the simulated specimen. In order Specimens with homogeneous materials predominantly
to compare these results with the stress–strain diagram show single Lüders bands. The more inhomogeneous
above, the stresses are plotted vs. nominal strain as well. the material, the more complex the simulated Lüders
The main ideas of the analytical calculations in Sections bands.
4.1 and 4.2 are confirmed: with approaching Lüders front, Tensile and compressive loading of more complicated
the von Mises stress rises to almost ReH(tr), falls down to prismatic shapes show Lüders bands as well.
almost ReL(tr) then approximately recovers the level of
ReL(FEM). The curves of the principal stresses r1, r2 and
r3 compare reasonably well with the prediction of our the- 6. Discussion
ory (open circles). It must be emphasized that these plots
vary a lot between different integration points, but the gen- All of the presented evidence have specific advantages
eral tendency is consistent. In some of the other integration and drawbacks. Experimental results mainly suffer from
R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808 1807
asymmetrical loading of specimens, the difficulty of mea- Since our material model is valid for any steep drop in
suring the true flow curve within a Lüders band, internal the flow stress, other strain softening mechanisms may be
stresses in specimens, inhomogeneous as well as anisotropic explained as well, such as the Portevin–Le Chatelier effect,
materials and the problem of measuring the strain and the yielding behaviour of some polymers or materials with
stress distribution inside the specimens. The analytical cal- certain phase transformations. Of special interest are extre-
culation has to make several assumptions to come to rea- mely fine-grained materials and also some metallic glasses,
sonable solutions. The numerical simulation has to deal where the true upper yield strength (or a substitute for it) is
with the sharp stress and strain gradients near the Lüders very high. In this case even the observed lower yield
front. strength exceeds the tensile strength. The strain-hardening
Nevertheless, taking all evidence and problems into con- ability of these materials is not sufficient to meet the stress
sideration, our model generally is able to explain material level of the lower yield strength. And this in turn does not
behaviour with a yield phenomenon. A critical point cer- lead to stable spreading out of Lüders bands, but instead to
tainly is the assumption of a true and well-defined lower a strain concentration in a freshly created Lüders band
yield strength ReL(tr). This could also be the lowest point (then called shear band) with shear type fracture and poor
in the flow curve of the dislocation multiplication model ductility in the tensile test.
for steels, but the idea of most of the locked dislocations
in steels being unlocked and free at yielding seems to be 7. Conclusions
possible. Moreover our model only applies for well-defined
broad Lüders bands. Thin-walled specimens might show a In this paper a simple macromechanical model to
smaller lower yield strength than predicted analytically, explain the nature of the yield phenomenon in steels is pre-
since the influence of the surface is more pronounced. sented. The model is based on three major points:
The finite element simulation presented here predicts an
angle of inclination u of the Lüders band of 55° at the ini- The true upper yield strength is much higher than the
tiation stage and 62°, when the Lüders band is compara- upper yield strength commonly observed. The main
tively broad. So in the simulation the inclination of the cause for this is that even slight misalignment of speci-
Lüders front changes in the course of the tensile test. In mens in the tensile test, notches and residual stresses
experiments on steels close to the simulated one, the lead to stress concentrations with earlier yielding. There
observed inclination is 51–55° and does not seem to is enough reliable experimental and numerical evidence
change a lot during the test. Rather than a change in incli- to confirm this point without doubt.
nation, sometimes new Lüders bands appear in different The true material behaviour after first yielding is not the
directions. This was also observed in finite element simula- observed one, but instead the normal one with normal
tions of slightly inhomogeneous materials. An interpreta- strain-hardening. Experimental and numerical evidence
tion of this fact could be that, once a Lüders band has strongly support this point; certain variations to this
originated in reality, it tends to keep its inclination, since point are not excluded.
plastic deformation is focussed on a region close to the The stress level at the observed lower yield strength is
Lüders front. In contrast, the finite element Lüders front predominantly a result of the triaxial stress state that
is able to change its inclination to some extent towards develops at the Lüders front. This is strongly supported
the analytically predicted one. by experimental, analytical and numerical evidence.
Another important point is the dependence of the upper
and lower yield strength of plain carbon steels on the strain Preliminary numerical simulations show that the model
rate in real tensile tests. In most cases the upper yield also seems to predict the yield behaviour of steels in more
strength shows a comparatively large sensitivity to strain complex structures, in compression, bending and torsion.
rate (sometimes described as delayed yielding). The lower Although primarily intended to describe the yield phenom-
yield strength shows a moderate one and the flow curve a enon in plain carbon steels, the model may also be applied
comparatively small sensitivity [1,38,39]. This experimental for similar materials and related strain softening effects. It
fact qualitatively agrees with our model, because according even explains the low ductility of some nanocrystalline
to Eq. (14) ReL(obs) is a function of ReH(tr) (highest sensitiv- materials.
ity) and ReL(tr) (part of the flow curve, lowest sensitivity).
Or in other words: with increasing strain rate, ReH(tr) References
increases substantially, ReL(tr) only a little, and ReL(obs)
moderately, according to Eq. (14). [1] Hall EO. Yield point phenomena in metals and alloys. London: Mac-
The next point is the different sensitivity of the upper millan; 1970.
[2] Gaal I, Tóth AL, Uray L, Harmat P. Int J Refract Met H
yield strength, the lower yield strength and the flow stress 2006;24:325.
on grain size and temperature [1,48]. Here again the idea [3] Shibkov AA, Zolotov AE, Zheltov MA, Denisov AA. Phys Solid
of ReL(obs) being a function of ReH(tr) and ReL(tr) seems to State 2011;53:887.
agree with experimental data. [4] Vöhringer O, Macherauch E. Phys Status Solidi B 1967;19:793.
1808 R. Schwab, V. Ruff / Acta Materialia 61 (2013) 1798–1808
[5] González-Doncel G, Adeva P, Cristina MC, Ibáñez J. Acta Metall [27] Van Rooyen GT. Mater Sci Eng 1968;3:105.
Mater 1995;43:4281. [28] Van Rooyen GT. Mater Sci Eng 1971;7:37.
[6] Malygin GA. Phys Solid State 2005;47:656. [29] Shaw JA, Kyriakides S. Int J Plasticity 1997;13:837.
[7] Tsuji N, Ito Y, Saito Y, Minamino Y. Scripta Mater 2002;47:893. [30] Tsukahara H, Iung T. Mater Sci Eng A – Struct 1998;248:304.
[8] Han BQ, Lavernia EJ. Adv Eng Mater 2005;7:457. [31] Kyriakides S, Miller JE. J Appl Mech 2000;67:645.
[9] Tsuchida N, Masuda H, Harada Y, Fukaura K, Tomota Y, Nagai K. [32] Childs THC. Int J Mech Sci 2009;51:402.
Mater Sci Eng A – Struct 2008;488:446. [33] Zhang YT, Qiao JL, Ao T. Model Simul Mater Sci 2007;15:147.
[10] Zhang ZF, He G, Eckert J, Schultz L. Phys Rev Lett 2003;91:045505. [34] Zhang YT, Ao T, Jiao W, Cui YH. Comp Mater Sci 2008;41:547.
[11] Schuh CA, Hufnagel TC, Ramamurty U. Acta Mater 2007;55:4067. [35] Marais A, Mazière M, Forest S, Parrot A, Le Delliou P. Philos Mag
[12] Johnston WG, Gilman JJ. J Appl Phys 1959;30:129. 2012:1.
[13] Brown N, Ward IM. J Polym Sci A2 1968;6:607. [36] Hutchinson MM. Philos Mag 1963;8:121.
[14] Lu J, Ravi-Chandar K. Int J Solids Struct 1999;36:391. [37] Petch JN. Acta Metall Mater 1964;12:59.
[15] Friedman M, Logan JM. Geol Soc Am Bull 1973;84:1465. [38] Krafft JM, Sullivan AM. Ship structure committee 1960 SSC-127.
[16] Burg JP, Harris LB. Tectonophysics 1982;83:347. <http://66.180.169.221/pdf/127.pdf>.
[17] Watterson J. J Struct Geol 1999;21:939. [39] Krafft JM. Acta Metall Mater 1962;10:85. <http://www.shipstruc-
[18] Emadoddin E, Akbarzadeh A, Daneshi GH. Mater Sci Eng A – ture.org/pdf/123.pdf>.
Struct 2007;447:174. [40] Bullen FP, Henderson F, Hutchinson MM, Wain HL. Philos Mag
[19] Lee S, Lee SJ, Santhosh Kumar S, Lee K, De Cooman BC. Metall 1964;9:285.
Mater Trans A 2011;42A:3638. [41] Oates G, Wilson DV. Acta Metall Mater 1964;12:21.
[20] Young ML, Wagner MFX, Frenzel J, Schmahl WW, Eggeler G. Acta [42] Shioya T, Shioiri J. J Mech Phys Solids 1976;24:187.
Mater 2010;58:2344. [43] Butler FJ. J Mech Phys Solids 1962;10:313.
[21] Shaw MC, Sata T. Int J Mech Sci 1966;8:469. [44] Delwiche DE, Moon DW. Mater Sci Eng 1971;7:203.
[22] Gajo A, Bigoni D, Muir Wood D. J Mech Phys Solids 2004;52:2683. [45] Ananthan VS, Hall EO. Acta Metall Mater 1991;39:3153.
[23] Piobert G, Morin AJ, Didion I. Mem l’artillerie 1842;5:501. [46] Moon DW. Mater Sci Eng 1971;8:235.
[24] Lüders W. Dinglers Polytech J 1860;155:18. [47] Lomer WM. J Mech Phys Solids 1952;1:64.
[25] Cottrell AH, Bilby BA. Proc Phys Soc A 1949;62:49. [48] Armstrong R, Codd I, Douthwaite RM, Petch NJ. Philos Mag
[26] Hahn GT. Acta Metall Mater 1962;10:727. 1962;7:45.