Keywords: laser assisted milling (LAMill), silicon nitride, thermal model, finite element
analysis (FEA), heat generation
Journal of Manufacturing Science and Engineering OCTOBER 2009, Vol. 131 / 051007-1
Copyright © 2009 by ASME
heat transfer model using FEA. With this model, a parametric and cutting speed. Hence, through thermal analysis, guidelines for
study was conducted involving parameters of laser power, laser- the parameter selection can be made and thus the thermal gradi-
beam diameter, laser moving speed, and preheat time. After this, ents over the cutting zone can be reduced. So far no literature has
Shen et al. 关18兴 improved that model 关17兴 so that it could be used reported on the thermal analysis of the cutting zone in detail for
to predict the potential surface/subsurface cracks through thermal LAMill. In addition, previous researchers usually neglected heat
stress analysis. However, neither the material removal nor model generation associated with machining in their thermal models, be-
validation was performed for both models. Later on, Yang et al. cause they argued that the value was very small compared with
关19兴 conducted a number of experiments of laser assisted face the energy from the laser source, while this study conducts simu-
milling, but no thermal model was built for this operation. They lations to explore its effect on the temperature at the cutting zone.
indicated that as operating temperature increases, both the cutting In the following, the FEA model is first introduced in detail in
force and specific cutting energy decrease. Yang and Lei 关20兴 Sec. 2 and an example of temperature distribution is illustrated in
further pointed out that once the operating temperature is higher Sec. 4. Then, model validation is conducted in Sec. 5 via varying
than 1000° C, entry edge chipping was not observed. Exit edge the parameters of laser power, feed, and cutting speed. Finally, in
chipping decreases considerably as the temperature increases. Sec. 6, some guidelines on parameter selection is provided for
Tian et al. 关21,22兴 also did some experiments of laser assisted future laser assisted milling operations.
milling of silicon nitride. The TiAlN coated carbide end mills
were used. They indicated that there was a narrow range of ma-
terial removal temperature 共about 1200– 1300° C兲 with the TiAlN 2 Transient Three-Dimensional FEA Model
coated carbide tools. Also, a transient three-dimensional thermal
model was developed using FVM. However, the model only fo- 2.1 Mathematical Description of the Model. Figure 1 illus-
cused on laser heating without considering the material removal, trates the schematic of laser assisted face milling of silicon nitride
and only the effect of laser power was discussed. ceramics in this study. The laser is positioned in front of the mill-
This study improves our previous FEA thermal model 关18兴 by ing cutter. Between the laser spot and the cutter, a small distance,
considering material removal, and the new model is validated laser-cutter allowance, is maintained in case the laser spot over-
through a series of experiments of laser assisted face milling of laps and damages the cutter. The workpiece is fed into the milling
silicon nitride ceramics. Temperature simulation reveals that ther- cutter along the z direction, which is clamped by an insulated vise
mal gradients exist inside the workpiece in LAMill due to local- so as to prevent the heat dissipation from the workpiece. The
ized laser heating. Basically, for a successful LAMill, small ther- clamped areas 共S1兲 are shown in gray in Fig. 1. The remaining
mal gradients in the workpiece are expected, especially over the areas on the workpiece surfaces are exposed to the surroundings.
cutting zone. They are strongly affected by the parameters of The heat transfer model can be described by the following gov-
LAMill such as laser power, laser-beam diameter, feed rate, feed, erning equation integrating over the volume of the workpiece:
共1兲
where S2 is the area covered by the laser beam 共laser spot兲, S3 is In Eq. 共1兲, the first, second, and third terms on the left-hand side
the area where the cutter is in contact with the workpiece in one represent the change in the internal energy, energy convection,
cut, and S4 are the areas including all the workpiece surfaces open and radiation, respectively. The first term on the right-hand side
to the surroundings except S2 and S3. represents input heat flux from the laser source. The second term
冏 冏
that temperature for silicon nitride ceramics.
T 共2兲 In this study, the top-hat distribution of laser intensity re-
=0 共3兲
n S1 places the Gaussian distribution in our previous work based
on the laser profile measurement from the manufacturer
In the area of S2, in addition to free convection and thermal 共Visotek Inc.兲. As a result, an approximately uniform heat
radiation, there is also a heat flux from the laser source. Thus the flux input, ql⬙, yields, assuming that the top-hat distribution
boundary condition can be defined as is not altered for the inclination angle of 70 deg between
k 冏 冏
T
n S2,t⬎0
= ql⬙ − q⬙con − qrad
⬙ 共4兲
the laser beam and the top face of the workpiece.
In order to apply the heat flux to the elements within the
elliptic laser spot, it is roughly divided into 30x ⫻ 32z square
where ql⬙ is heat flux from the laser, q⬙con is convective heat flux, divisions due to the laser inclination. The division size 共in
q⬙con = h共Ts − T⬁兲, and qrad
⬙ is radiation heat flux, qrad
⬙ = 共Ts4 − T4⬁兲. the x and z directions兲, ␦, is the same as the element size 共in
For the area of S3, besides free convection and thermal radia- the x and z directions兲. The relative error between the
tion, there exists generated heat due to plastic deformation and power input for the simulation and the experimental value
tool-workpiece friction in machining. Thus the boundary condi- is 0.5%. Each division is specific to a corresponding ele-
tion can be represented by ment. As the laser spot moves, the location of each division
冏 冏
is changed and the division corresponds to another element.
T In each moving step, the heat flux is applied to every laser-
k ⬙ − q⬙con − qrad
= qgen ⬙ 共5兲
n S3,t⬎t p
beam covered element and removed from every newly un-
covered one. Note that the distance that the laser spot
It should be noted that the surface of S3 moves with the laser moves in each step is equal to the element size.
beam. 共3兲 Material removal in ANSYS is achieved through “killing”
For the areas of S4, only free convection and thermal radiation elements, which are deactivated by multiplying their stiff-
exist and thereby the boundary condition can be expressed as ness 共or conductivity, or other analogous quantity兲 by a
k 冏 冏
T
n S4
= − q⬙con − qrad
⬙ 共6兲
severe reduction factor 共say, 1.0⫻ 10−8兲. All the effects as-
sociated with deactivated elements are set to zero, such as
element loads, mass, damping, specific heat, and others.
But, in the actual material removal process, the milling cut-
2.2 Model Development Using FEA. In this study, the ther- ter contacts the workpiece with only one part of its nose
mal model is developed using the finite element package ANSYS edge 共Fig. 3兲, because the nose radius, r, of the cutter
11.0 关23兴. The dimensions of the model are the same as those of the 共0.787 mm兲 is much larger than the depth of cut of 0.2 mm.
actual workpiece with 4.3 mm in width 共x direction兲, 5.3 mm in It is also noted that the cutter diameter of 76.2 mm is con-
height 共y direction兲, and 48 mm in length 共z direction兲. Because of siderably larger than the workpiece width of 4.3 mm.
symmetry considerations only one-half of the workpiece is used in Therefore, the cutting path on the workpiece can be ap-
the model. As shown in Fig. 2, the z direction is the laser moving proximated as a straight line, and the cross section of the
direction. The region to be covered with a laser beam is a long cutting zone in any y-z plane can be seen in Fig. 3.
quasirectangular area located on the top face of the workpiece. The FEA mesh adopted in this study for the cutting zone
Journal of Manufacturing Science and Engineering OCTOBER 2009, Vol. 131 / 051007-3
Table 1 Thermal properties of silicon nitride. The values of thermal expansion coefficient,
thermal conductivity and specific heat are provided by Ceradyne Inc.
Thermal Thermal
Temperature expansion conductivity, k Specific heat, c
共5兲
共K兲 Emissivity, coefficient,  共W / m K兲 共J / kg K兲
4 Simulation Results
4.1 Temperature Distributions of the Workpiece. Simula-
Table 2 Properties of silicon nitride „25° C… „Ceradyne Inc.…
tions are performed with different operating conditions. One typi-
Property Value cal simulation result is illustrated with temperature distributions in
Fig. 7. The cutting length, Lc, is 7.0 mm, and the operating con-
Density 共kg/ m3兲 3200 ditions are Pl = 410 W, Vc = 1.0 m / s, f = 0.024 mm/ tooth/ rev,
Flexural strength 共MPa兲 800 V f = 6 mm/ min, t p = 12 s, Dl = 3.6 mm, DOC= 0.2, and Ll
Elastic modulus 共GPa兲 310 = 0.5 mm.
Poisson ratio 0.27 In Fig. 7, the MN and MX symbols identify the locations of the
Fracture toughness 共MPa m1/2兲 6.0 minimum and maximum temperatures 共the left face of the work-
piece and the laser spot center兲, respectively. Note that the uni-
form interval is used in the contour display. The contour interval,
Fig. 7 Temperature contours of the workpiece obtained from simulation „Pl = 410 W, Vc = 1.0 m / s,
f = 0.024 mm/ tooth/ rev, Vf = 6 mm/ min, tp = 12 s, and Lc = 7.0 mm…
Journal of Manufacturing Science and Engineering OCTOBER 2009, Vol. 131 / 051007-5
defined as the difference between two adjacent contour lines, is ence over the cutting zone, is 120° C.
around 140° C. Obviously, the region near the laser spot has Figure 9 illustrates the temperature contours of the back face
higher temperatures 共over 1000° C兲, which is verified by experi- obtained from simulation. Point F is the MTP with a temperature
mental observations. In the experiments, the color of the material of 1165° C. Assuming that another cutting is conducted, all the
in that region turns into red and fades away to the surroundings. operating conditions are the same except that the depth of cut is
The approaching portion of the laser spot continuously absorbs set to be the nose radius of 0.787 mm. Then Point G is the MTP
energy from heat conduction and, as a result, temperatures in- of the new cutting with a temperature of 1147° C. It means that,
crease quickly, while the newly uncovered region is gradually from the thermal aspects, a larger DOC is feasible under the cur-
cooling down due to free convection and thermal radiation. The rent operating conditions. However, from the machining aspects, a
geometrical widths of the contours in the x, y, and z directions are very large DOC may cause severe surface/subsurface damage,
not uniform and instead tend to increase from the laser spot to the which means that too much energy is deposited in the region not
surrounding portion. That is, along the x, y, and z directions, the to be machined. The approach to prevent it is either to increase the
thermal gradients are becoming smaller. Contour display reveals laser power or feed rate or both.
that the laser heat source has a strong influence on the region
around the laser spot and a weaker influence on the region be- 4.2 Effect of Heat Generation on Machining. As mentioned
yond. Section A-A in Fig. 7 passes through the cutting zone at the in Sec. 2.1, another factor that affects the thermal model is heat
top face of the workpiece. generation due to plastic deformation and tool-workpiece friction,
The minimum temperature point 共MTP兲, as the term suggests, which is associated with machining. In this model, heat generation
is defined as the point with the minimum temperature at the cut- is being treated. Although many studies indicate that the value is
ting zone. Usually, it is farthest from the laser spot center among very small, a few of them calculate it in detail. Since the ratio of
the points in the cutting zone. It depends on the relative positions the cutting energy converting to heat, h, and the ratios of the
of the workpiece and the cutter, and it moves with the cutting generated heat transferring to the cutter, the chips, and the work-
zone. The temperatures at the MTPs should be kept higher than piece, respectively, are not well known, in order to assess the
the material softening temperature in the milling process. maximum possible effect of generated heat on the workpiece tem-
Figures 8共a兲 and 8共b兲 illustrate the cutting zone and its tempera- peratures, the model assumes that all the cutting energy is con-
ture contours obtained from simulations, respectively. Due to the verted to heat, which dissipates into the workpiece 共hw = 1兲.
very small feed, the surface area OCDE is considered here to Equation 共7兲 indicates that the cutting energy is proportional to
represent the temperature distribution for the cutting zone. Point the main cutting force, Fc. Our experimental results show that the
O V is the location of the MTP and Point D is the maximum average main cutting forces with depth of cut of 0.2 mm and
temperature point over the cutting zone. The temperature differ- cutting length of 1 mm are 8.6 N, 6.4 N, and 2.2 N for operating
ence between them, as well as the maximum temperature differ- temperatures of 1140° C, 1320° C, and 1400° C, respectively. As
300 15
Group I 410 0.024 1.0 6 12 3.6 0.2 0.5
470 8
0.012 3
Group II 410 0.024 1.0 6 12 3.6 0.2 0.5
0.048 12
0.5 3
Group III 410 0.024 1.0 6 12 3.6 0.2 0.5
2.0 12
expected, the average cutting force decreases as the operating Vl = V f = Zf · RPM 共11兲
temperature increases. In order to evaluate the effect of heat gen-
eration, the average cutting force of 9.8 N at the operating tem- Because of the relative motion between the laser and the work-
perature of 1400° C is used as an example in this section. The heat piece, Vl is equal to feed rate, V f . For convenience, in this section,
feed rate, V f , will be used rather than the laser moving speed, Vl.
flux q⬙gen thereby yields 2.0⫻ 105 W / m2 according to Eq. 共10兲.
In this study, the parameters of laser power, feed, and cutting
However, comparing the results from considering heat generation
speed are the most important. The parameters of laser-beam diam-
with those not, the temperature discrepancies at the MTP and
eter, depth of cut, tooth number, and laser-cutter allowance are
maximum temperature point over the cutting zone are about
fixed. The measurement points are shown in Fig. 6. The tempera-
0.3° C and 2 ° C, respectively. That is, q⬙gen indeed can be ne- tures at the MTP and maximum temperature point over the cutting
glected. In fact, the heat flux input from the laser source, ql⬙, is zone are also predicted.
about 150 times of q⬙gen under these operating conditions, even Figures 10–14 show the temperature histories under the operat-
with all the cutting energy being assumed to dissipate into the ing conditions in Groups I and II 共Table 3兲. Each of them includes
workpiece. four curves: measurement, simulation, Simu_MTP, and Simu-
4.3 Uncertainty Analysis. There are some uncertainties in- _Max, which represent experimental data, simulation data, pre-
volved in this study. All the uncertainty values are for the reported dicted minimum, and maximum data at the cutting zone, respec-
quantities. These uncertainties originate from both measurements tively. For uncertainty considerations, the curves of measurement
and simulations. For measurements, the uncertainty is mainly and simulation are shown with the error bars. The curves of Si-
from pyrometer system error, ⫾0.5%. For simulations, the uncer- mu_MTP and Simu_Max are not shown with the error bars be-
tainties are mainly from 共1兲 laser-beam diameter determination, cause the difference between Simu_MTP and Simu_Max, as well
⫾11.1%, which comes from the actual measurements of laser- as the maximum temperature difference over the cutting zone, is
beam diameter; 共2兲 the absorptivity of the workpiece 关33兴, ⫾3.2%, the most important. The curves of measurement and simulation
which is due to temperature variations; 共3兲 meshing and discreti- show that the temperatures quickly rise to a peak value and then
zation of the model, 1.2%, which is from model meshing and gradually go down. This is because during preheating, the position
calculations; 共4兲 average convection coefficient 关29兴, ⫾11.8%, of the laser beam is fixed. After that time the measurement spots
which arises from nonuniform temperatures and different convec- start to move with the workpiece. Furthermore, the maximum
tion surfaces of the workpiece; and 共5兲 laser power input, 0.5%, temperature differences over the cutting zone are found to in-
which is due to power difference between the simulation and the crease gradually as the workpiece moves. The reason is that the
experiment 共Sec. 2.2 共2兲兲. The contributions of these uncertainties insulating materials are not ideally adiabatic. The simplified
to the simulation error also rank as above. For example, under the boundary conditions in the model actually neglect the heat trans-
operating conditions Pl = 410 W, Vc = 1.0 m / s, f fer between the insulating materials and the vise, thus causing the
= 0.024 mm/ tooth/ rev, V f = 6 mm/ min, t p = 12 s, Dl = 3.6 mm, predicted temperatures higher than the measured ones. As the tem-
DOC= 0.2, and Lc = 7.0 mm, the temperature at the MTP is
1165° C and the corresponding simulation error has upper and
lower limits of 4.2% and 4.6%, respectively, which are deter-
mined via several simulations using all the possible combinations
of the limit values of the above factors that can result in the
uncertainties in LAMill and then finding the upper and lower lim-
its. In addition, there are some uncertainties neglected in this
model such as inhomogeneous properties of the material, surface
oxidation, the air jet, and the insulating material.
5 Model Validation
To develop confidence in our FEA model, a series of experi-
ments was conducted for validation purposes. Three groups of
them are listed in Table 3. The operating conditions include the
following parameters: laser power, Pl, laser-beam diameter, Dl,
laser moving speed, Vl, preheat time, t p, depth of cut, DOC, feed,
f, cutting speed, Vc, tooth number, Z, and laser-cutter allowance, Fig. 10 Temperature histories with Pl = 300 W „Vc = 1.0 m / s, f
Ll. Vl and V f are linked through the following equation: = 0.024 mm/ tooth/ rev, Vf = 6 mm/ min, and tp = 15 s…
Journal of Manufacturing Science and Engineering OCTOBER 2009, Vol. 131 / 051007-7
Journal of Manufacturing Science and Engineering OCTOBER 2009, Vol. 131 / 051007-9