Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 13 (2014) 282 287

2nd CIRP Conference on Surface Integrity (CSI)

Finite Element Simulation of Residual Stresses in Cryogenic

Machining of AZ31B Mg Alloy
Z. Pua, D. Umbrellob*, O.W. Dillon Jr.a, I.S. Jawahira
a
University of Kentucky, Institute for Sustainable Manufacturing (ISM), Lexington, KY 40506, USA
University of Calabria, Department of Mechanical, Energy and Management Engineering, 87036 Rende (CS), Italy
* Corresponding author. Tel.: +39-0984-494820; fax: +39-0984-494673. E-mail address: d.umbrello@unical.it.
b

Abstract
Magnesium alloys are lightweight materials primarily used in transportation industry, and are also emerging as a potential material
for biodegradable fixation implants. However, unsatisfactory corrosion resistance largely limits the application of these materials.
Residual stresses were reported to have significant influence on corrosion resistance of Mg alloys. In this study, a finite element
model was developed to simulate the residual stresses in cryogenic machining of AZ31B Mg alloy. After calibration using
experimental data, numerical simulations were conducted to study the influence of cutting edge radius and cooling method (dry vs.
cryogenic) on residual stresses. The model can be used to establish proper cutting conditions to induce compressive residual
stresses to enhance the corrosion resistance of Mg alloys.
2014
2014 The

The Authors.
Authors. Published
Publishedby
byElsevier
ElsevierB.V.
B.V.Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of The International Scientific Committee of the 2nd Conference
on Surface Integrity
Selection and/or peer-review under responsibility of The International Scientific Committee of the 2nd CIRP
Conference on
in the person of the Conference Chair Prof Dragos Axinte dragos.axinte@nottingham.ac.uk
Surface Integrity in the person of the Conference Chair Prof. Dragos Axinte dragos.axinte@nottingham.ac.uk
Keywords: cryogenic machining; finite element modeling; biodegradable implants; residual stresses

1. Introduction
Magnesium alloys are lightweight materials with
increasing applications in transportation industry, and
are also emerging as a potential material for
biodegradable fixation implants [1]. These alloys can be
safely absorbed by human body after the fractured bone
heals itself [1]. However, unsatisfactory corrosion
resistance limits the application of these materials to a
large extent. Surface integrity induced by manufacturing
processes has been found to have significant influences
on the corrosion resistance of Mg alloys [2]. Residual
stress is one of the most critical influencing factors on
the performance of components made from materials.
The high compressive residual stress generated in the
subsurface via a deep rolling process was claimed to
reduce the corrosion rate of a biphasic magnesiumcalcium alloy by a factor of approximately 100 [3].
Compressive residual stresses induced by cryogenic
machining [2], cryogenic burnishing [4] and laser shock
peening [5] were also claimed to improve the corrosion
resistance of magnesium alloys.

Due to their remarkable influence on corrosion

resistance, residual stresses generated in the machined
workpieces need to be controlled. In addition to
experimental studies, there is a great need for developing
predictive models to simulate the distribution of residual
stresses on machined surface and sub-surface so that
proper machining conditions can be selected. Finite
element (FE) modeling has been used successfully to
predict residual stresses induced by machining [6, 7].
However, there are limited studies on simulating residual
stresses in machining magnesium alloys. The influence
of cryogenic cooling on residual stress distribution has
not been simulated by researchers. It is the aim of this
paper to stress these issues. The influence of cutting
edge radius will also be simulated.
2. Experimental Work
2.1. Experimental procedure
In order to calibrate and validate the FE model,
orthogonal machining tests were conducted on AZ31B

2212-8271 2014 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of The International Scientific Committee of the 2nd Conference on Surface Integrity in the person
of the Conference Chair Prof Dragos Axinte dragos.axinte@nottingham.ac.uk
doi:10.1016/j.procir.2014.04.048

283

Z. Pu et al. / Procedia CIRP 13 (2014) 282 287

Mg alloy under both dry and cryogenic conditions. The

composition of the workpiece material is shown in Table
1. The initial hardness is 52.3 HV. The details of the
experimental study were reported elsewhere [2]. The
machining tests were conducted on a Mazak Quick
Turn-10 Turning Center using uncoated carbide tools
with TNMG 432 geometry. For cryogenic machining, an
Air Products ICEFLY system was used to spray liquid
nitrogen to the machined surface from the clearance side
of the cutting tool as shown in Figure 1. Forces and
temperatures were recorded during the machining tests.
The machining conditions tested are shown in Table 2.

Table 3. X-ray diffraction parameters for residual stress measurement

of AZ31B Mg alloy
Radiation
Voltage and current
Collimator diameter (mm)
X-ray elastic constants
(MPa-1)
Bragg angle 2 ()
Number of angles

Mn-K
20kV, 4 mA
5
()S2 = 29.32 10-6, S1 = -6.59 106

151.06
(hkl) = (203)
30

3. FE Analysis
3.1. FE Model

Fig. 1. Experimental setup for cryogenic machining [2].

Table 1. Nominal composition of AZ31B magnesium alloy (wt. %)
Al

Zn

Mn

Mg

2.5 - 3.5 0.7 - 1.3 0.2 1.0 balance

Table 2. Experiment matrix (V = 100 m/min; f = 0.1 mm/rev) [2].
Test

Cooling Method

Cutting Edge Radius (m)

Dry

30

Cryogenic

30

Dry

70

Cryogenic

70

The proposed numerical procedure developed in this

paper employs a FE based thermo-mechanical model
formulation of the orthogonal hard turning process. In
particular, 2D plane-strain simulation was carried out
using SFTC Deform 2D, and it was based on the
following assumptions:
x rigid cutting tool (divided into 3500 elements);
x isotropic hardening for workpiece material, modelled
as elastic-plastic and divided into 8500 elements;
x high mesh density was defined on the workpiece. In
particular, the elements located around the cutting
edge, along the machined surface and, for about 1
mm, below it were fifty times as dense as the other
ones (average element edge length 5-6 Pm);
x nonisothermal
elastic-visco-plastic
material
governed by incremental theory of plasticity and Von
Mises yield condition. A modified Johnson-Cook [9]
flow stress rule [10, 11] was implemented in the FE
code as material constitutive law:
V

0.337

A 0.0001  e100H 

2.2. Residual stresses

The experimental residual stress state in machined
AZ31B Mg samples was analyzed by X-ray diffraction
technique using the sin2 method [8]. The equipment
used was a PROTO machine (model iXRD).The
parameters used in the X-ray analysis are shown in
Table 3. To determine the in-depth residual stress
profiles, successive layers of material were removed by
electropolishing to avoid the modification of machininginduced stresses. It is noted that in the present study, the
total penetration depth of the X-ray beam for Mg is
about 25 m. Since the strength of the X-ray beam
decreases exponentially with the distance from the
surface, it is estimated that the measurement for the
nominal depth is an average value of a layer about 15
m thick below the surface.

>200  400 H

T T
@ >1  0.016 ln H  A@ 1  618
T

room

1.829

room

(1)

The detailed procedure and the explanation for the

calibration of above equation can be found in [10];
Cockroft and Lathams fracture criterion was used to
predict the effect of the stress on the chip
segmentation during cutting. A modified material
flow stress model incorporating flow softening
effects was successfully used in Ti6Al4V to
simulate chip serration [12]. It is noted that the
authors prefer to use the modified flow stress
approach. However, due to the limitation of the
existing material testing data on AZ31B Mg alloy, it
is not possible at this time to develop a modified
material constitutive model incorporating flow
softening effects and this will be the focus of future
research. The critical damage value, Dcritical, was set
equal to 35 MPa, as found by a FE calibration in a
prevoius work [10]. More details on the changes of
Dcritical during a chip segmentation cycle were

284

Z. Pu et al. / Procedia CIRP 13 (2014) 282 287

reported elsewhere [10, 11];

x constant shear model W mW) was implemented in
the FE model, setting P= 0.7 (estimated in [10, 11]).
In the mentioned equation, W is the shear stress and W0
the flow shear stress obtained as 0 0 3 ;
x workpiece and cutting tool were allowed to exchange
heat with the environment; the convection coefficient
was selected at 20 W/(m2K), while the global heat
transfer coefficient at the tool-chip workpiece
interfaces, hint, was set equal to 5000 kW/(m2K) as
found during a previous FE calibration [10] taking
the temperature into account;
x to simulate the effect of cryogenic cooling, an
environmental window for heat exchange was
defined as shown in Figure 2, and the width of the
windows is 1 mm. The window is fixed in its position
and does not move with the workpiece. The local
convection coefficient in the window can be adjusted
to simulate the cryogenic cooling effect, and is one of
the major parameters to be calibrated for modeling
cryogenic machining. This local convection
coefficient, hcryo, replaces the default air convection
coefficient for the workpiece covered by the window;
the rest of the workpiece is still subject to air
convection. As demonstrated in a previous work [10],
a good agreement with measured temperature was
achieved when the convection coefficient was set to
5000 kW/(m2K).

Fig. 2. Mesh and boundary conditions for the FE model [10].

a user subroutine was developed and implemented

in the proposed FE model for predicting the grain
size evolution and the hardness modification on the
machined surface and the sub-surface. Details,
including the FE calibration and its validation are
reported and extensively discussed in [10].

3.2. Method for extracting in-depth residual stress

profiles
The proposed FE strategy in this study is to predict
the residual stress evolution in the workpiece after
machining. The FE model is the same as described in
Section 3.1. A manual routine is used since DEFORM2DTM cannot directly obtain the simulated residual
stresses. The follow procedure was used:
1. consider tool flank wear in the simulations;

2. include both the circumferential and axial stresses

due to prior operations, and load them as initial
residual stress condition. Figure 3 shows the
circumferential residual stress due to prior operation
and as it is load as initial stress in the workpiece;
3. perform at least two cutting passes before extracting
the residual stresses from the model [13 - 15];
4. for several time steps, the tool was released from the
machined surface (unloading phase) and the
workpiece was cooled down to the room temperature;
5. residual stress profiles at several locations of the
machined surface were collected and the average
values were calculated, as described in [13].

Fig. 3. Circumferential residual stress due to prior operation, loaded as

initial stress in the workpiece before machining.

Figures 4, 5 and 6 report the sequences to be carried

out in order to take into account Points 3, 4 and 5 of the
described methodology for extracting in-depth residual
stress profiles. In particular, Figure 4 shows the
predicted circumferential residual stress at the end of the
first cut, after the unloading phase and when the
workpiece was cooled down to the room temperature.
Furthermore, Figure 4 highlights the region used in the
residual stress analysis in accordance with the findings
of Outeiro et al. [13] and the comparison between the
FEM-predicted and the XRD-measured circumferential
residual stresses. As can be seen from Figure 4, the
predicted circumferential residual stresses are far from a
good agreement with the experimental values,
confirming that at least two cutting passes are needed
before extracting the residual stresses from the model.
Similar trends were also observed with the prediction of
the axial residual stresses.
A second cutting pass must be conducted in order to
take into account the residual stress induced by the first
cut. Figure 5 shows the circumferential residual stress
due to the first, and loaded as initial stress in the
workpiece before running the second cut. After
the
second
pass,
the
predicted
conducting
circumferential residual stresses are in a good agreement
with the experimental values (Figure 6). Similar trends

Z. Pu et al. / Procedia CIRP 13 (2014) 282 287

285

smaller than 25 Pm could not be measured by the

method used in this study due to the large penetration
depth of X-ray in magnesium alloys. It was expected that
there might be larger differences between measured and
simulated residual stresses in the first 25Pm since
complex microstructural changes occurred in this zone
including twinning and significant grain refinement
which may influence residual stresses and are not
considered by the FE model.
Numerical prediction of axial residual stresses is also
satisfactory even if it is not as accurate as those along
the circumferential direction, especially in dry
machining. With a 30 m edge radius tool, both the

Fig. 4. Circumferential residual stress at the end of the first cut

(Dry machining; edge radius: 30 m; cutting speed: 100 m/min;
feed rate: 0.1 mm/rev).

were also observed with the prediction of the axial

residual stress.

Fig. 6. Circumferential residual stress at the end of the second cut

(Dry machining; edge radius: 30 m; cutting speed: 100 m/min;
feed rate: 0.1 mm/rev).

Fig. 5. Circumferential residual stress due to first cut, loaded as initial

stress in the workpiece before the second cut.

4. Results and Discussion

Figure 7 shows the measured and the predicted
residual stresses in the circumferential and axial
directions after machining using cutting tools with 30
m as edge radius under both dry and cryogenic
conditions. First of all, prediction of circumferential
residual stresses under both dry and cryogenic
machining can be considered very good when the depth
is larger than 25 Pm. The residual stresses at depth

experimental
and
the
predicted
compressive
circumferential residual stresses are induced up to 140
m under dry conditions and up to 110-140 m under
cryogenic condition below the machined surface. The
peak compressive stress for both dry and cryogenic
machining along the circumferential direction was about
-40 MPa at a depth of 30-40 m below the surface. The
predicted peak in compressive stress along the
circumferential direction show similar data. Much larger
differences were found in the axial direction between dry
and cryogenic machining with the same edge radius of
30 m. Tensile residual stress of 37 MPa was generated
in the axial direction while cryogenic machining led to
compressive residual stress of -39 MPa in that direction.
Also the FE model is capable of simulating similar

286

Z. Pu et al. / Procedia CIRP 13 (2014) 282 287

trends. In fact, a tensile residual stress of 30 MPa was

generated in the axial direction on the machined surface
obtained in dry machining, while a compressive residual
stress of about -40 MPa was predicted in that direction
under cryogenic condition.
Figure 8 shows the measured and the predicted
residual stresses in the circumferential and axial
directions after machining using cutting tools with 70
m as edge radius under both dry and cryogenic
conditions. When the cutting edge radius was increased
to 70 m, the penetration depth of the experimental
compressive residual stresses was extended to about 200
m in the circumferential direction under cryogenic
conditions compared with 150 m under dry conditions,
a 33% increase. The numerical simulation shows a
similar trend. Moreover, the comparison between the
predicted and experimental circumferential residual
stresses is good although no experimental data are
available in the first 25 Pm. The residual stresses in the

axial direction were more compressive and extended to

greater depths during cryogenic machining than those
created during dry machining. For dry machining, the
peak compressive stresses in both directions were
reduced by about 16% when the cutting edge radius was
increased from 30 m to 70 m due to the increased
temperature resulting from more severe ploughing
effects. For cryogenic machining, the peak compressive
residual stress was reduced slightly in the
circumferential direction and increased by 12% in the
axial direction. The FE prediction shows satisfactory
results especially under cryogenic condition for both
circumferential and axial directions.
It is noted that phase transformation did not occur
during machining of this magnesium alloy which
contributes to the good agreement between simulated
and experimental residual stresses.

Fig. 7. Experimental [2] and numerical residual stresses after machining using cutting tools with an edge radius of 30 Pm and under (a) dry and (b)
cryogenic conditions (V =100 m/min, f = 0.1 mm/rev).

Fig. 8. Experimental [2] and numerical residual stresses after machining using cutting tools with an edge radius of 70 Pm and under (a) dry and (b)
cryogenic conditions (V =100 m/min, f = 0.1 mm/rev).

Z. Pu et al. / Procedia CIRP 13 (2014) 282 287

[6]

5. Conclusions
In this paper a calibrated FE model [10] was used for
studying the effect of the cooling conditions and tool
edge radius on residual stresses distribution. The
proposed FE procedure and the methodology for
predicting the residual stress state were respectively
validated and demonstrated by using experimental
observations [2]. In particular, numerical prediction of
both circumferential and axial residual stresses was
found to be consistent with those experimentally
measured.
Also, it was found that sequential cuts deeply affect
the residual stress distribution and they have to be
considered during the FE modeling. The present
investigation suggests that a better agreement between
the experimental and predicted residual stress state in
machining of AZ31B Mg alloy can be achieved if the FE
model takes the initial residual stress into account. In
summary, the proposed FEM procedure could be used
for intelligent process parameter selection to achieve a
more desirable final surface integrity state.
Acknowledgements
The authors would like to thank Air Products and
Chemicals for providing the ICEFLY liquid nitrogen
delivery system to the Institute of Sustainable
Manufacturing (ISM) at the University of Kentucky
which enabled the experimental work. The help from
Tao Lu (PhD candidate in Mechanical Engineering,
University of Kentucky) on machining experiments and
partial financial support from ISM are also sincerely
acknowledged.
References
[1]
[2]

[3]

[4]

[5]

Witte, F., 2010, The history of biodegradable magnesium implants:

A review, Acta Biomaterials, 6(5), pp. 1680-1692.
Pu, Z., Outeiro, J. C., Batista, A. C., Dillon Jr, O. W., Puleo, D. A.,
Jawahir, I. S., 2012, Enhanced surface integrity of AZ31B Mg
alloy by cryogenic machining towards improved functional
performance of machined components, International Journal of
Machine Tools and Manufacture, 56, pp. 17-27.
Denkena, B., Lucas, A., 2007, Biocompatible magnesium alloys as
absorbable implant materials Adjusted surface and subsurface
properties by machining processes, CIRP Annals - Manufacturing
Technology, 56(1), pp. 113-116.
Pu, Z., Song, G. L., Yang, S., Outeiro, J. C., Dillon Jr, O. W.,
Puleo, D. A., Jawahir, I. S., 2012, Grain refined and basal textured
surface produced by burnishing for improved corrosion
performance of AZ31B Mg alloy, Corrosion Science, 57, pp. 192201.
Guo, Y., Sealy, M. P., Guo, C., 2012, Significant improvement of
corrosion resistance of biodegradable metallic implants processed
by laser shock peening, CIRP Annals - Manufacturing Technology,
61(1), pp. 583-586.

Umbrello, D., Outeiro, J. C., M'Saoubi, R., Jayal, A. D., Jawahir, I.

S., 2010, A numerical model incorporating the microstructure
alteration for predicting residual stresses in hard machining of
AISI 52100 steel, CIRP Annals - Manufacturing Technology,
59(1), pp. 113-116.
[7] Outeiro, J. C., Umbrello, D., M'Saoubi, R., 2006, Experimental
and numerical modelling of the residual stresses induced in
orthogonal cutting of AISI 316L steel, International Journal of
Machine Tools and Manufacture, 46(14), pp. 1786-1794.
[8] Noyan, I. C., Cohen, J. B., 1987, Residual stress: measurement by
diffraction and interpretation, Springer, New York, N.Y.
[9] Cockcroft, M. G., Latham, D. J., 1968, Ductility and workability
of metals, Journal Institute of Metals, 96, pp. 33-39.
[10] Pu, Z., Umbrello, D., Puleo, D. A., Dillon Jr, O. W., Jawahir, I. S.,
2013, Finite Element Modeling of Microstructural Changes in Dry
and Cryogenic Machining AZ31B Magnesium Alloy for Enhanced
Corrosion Resistance, Proceedings of NAMRI/SME, Vol. 41.
[11] Pu, Z., 2012, Cryogenic Machining and Burnishing of AZ31B
Magnesium Alloy for Enhanced Surface Integrity and Functional
Performance, PhD Dissertation, University of Kentucky.
[12] Sima, M. and zel, T., 2010, Modified material constitutive
models for serrated chip formation simulations and experimental
validation in machining of titanium alloy Ti6Al4V. International
Journal of Machine Tools and Manufacture, 50(11), 943-960.
[13] Outeiro, J. C., Ee, K. C., Dillon Jr., O. W., Wanigarathne, P. C.,
Jawahir, I. S. 2006, Some observations on comparing the modelled
and measured residual stresses on the machined surface Induced
by orthogonal cutting of AISI 316L steel, Proceedings of the 9th
CIRP International Workshop on Modeling of Machining
Operations. Bled, Slovenia, 1112 May, pp 475-481
[14] Outeiro, J. C., Umbrello, D., M'Saoubi, R. (2006) Experimental
and FEM Analysis of Cutting Sequence on Residual Stresses in
Machined Layers of AISI 316L Steel, Materials Science Forum,
524-525, pp. 179-184.
[15] Ee, K.C., O.W. Dillon, Jr. and I.S. Jawahir, 2005, Finite Element
Modeling of Residual Stresses in Machining Induced by Cutting
Tool with a Finite Edge Radius, Int. J. Mechanical Sciences, Vol.
47, pp. 1611-1628.

287