Вы находитесь на странице: 1из 8

The Prediction of Case Depth in

Laser Transformation Hardening


H.R. SHERCLIFF and M.F. A S H B Y

An approximate heat flow model is developed to predict the case depth in laser transformation
hardening of steel surfaces. The model exploits the dimensional relationships between the pro-
cess variables to give master diagrams for the hardened depth using Gauss[an and uniform,
rectangular sources. Critical values of dimensionless parameters are identified which predict the
conditions for first hardening and the onset of surface melting. Good agreement is demonstrated
with a wide range of experimental data, and comparisons are made with previous modeling
methods and process diagrams.

I. INTRODUCTION In any multivariable problem, it usually pays to use


dimensionless groups to simplify computation and to make
H I G H - p o w e r lasers are now established as industrial the results general to all materials. Numerous examples
tools. Recent reviews of the principles and applications can be found in the analysis of welding, [18-22~and in laser
of lasers describe the use of lasers as a controlled heat surface treatment, Kou [141 has suggested appropriate
source for transformation hardening, surface alloying, groups. In this article, we rework Ashby and Easterling's
welding, and cutting. [1-5~Heat flow problems such as these model t~71 in dimensionless terms and extend its appli-
are complicated by the number of variables involved, so cation to a wide range of treatments. The notation is given
there is a need for process models to provide predictions in Table I.
of the effect of the process variables and for process con-
trol. In this article, we describe a process model for laser
transformation hardening of steel surfaces. II. MATHEMATICAL MODELING
The hardening depth depends on two sets of process In common with other workers in the field, we make
variables. The first is the properties of the steel: its trans- the following assumptions:
formation temperatures, thermal conductivity, and vol-
umetric specific heat. The second set describes the (1) the surface absorptivity, A, is constant;
(2) the latent heat of the a - y transformation is negligible
performance of the laser: its power, beam size, scan speed,
compared to the other terms in the heat-flow equations;
surface absorptivity, and mode (the energy distribution
(3) the thermal conductivity, A, and diffusivity, a, (and
in the beam).
hence, specific heat, pc) are constant;
(4) the eutectoid temperature, A1, is as given by the phase
Modeling Methods
diagram; and
The classical approach to modeling the heat flow in- (5) the radius, rB, of a Gauss[an beam is the distance
duced by a distributed heat source moving over the sur- from the beam center to the position at which the inten-
face of a semi-infinite solid starts with the solution for sity has fallen to 1/e times the peak value.
a point source [6] and integrates it over the area of the The coordinate system has the origin at the beam cen-
beam. This widely used method [7-131 requires numerical ter. The laser, of total power q, moves in the x-direction
procedures for its evaluation, as do the finite-difference with speed v, with the y-axis across the track and z the
solutions of Kou and co-workers, t~4,151 While these so- distance below the surface.
lutions are rigorous, they have the drawback that the
computations are complex and the results hard to use. A. Modeling of Laser Hardening with a
An alternative approach is that of Bass, [16[ who presents Gauss[an Beam
temperature field equations for various beam geome-
tries, and by taking limits, extracts analytical results which 1. The general solution for the temperature field
provide a good overview of the response of different ma- The temperature field equation from Ashby and
terials. Ashby and Easterling t~71 carry the analytical ap- Easterling [17[is an adapted form of the Gauss[an line source
proach further, developing an approximate solution for solution due to Rykalin et al. :[23]
the entire temperature field. Comparison of the analyt- Aq
ical results with more exact numerical calculations shows T-To =
that this approach is entirely adequate for the description 27rAv [t(t + to)] 1/2
of laser hardening, in which variability of laser output
and material properties always introduces some scatter. 9e x p - - - - - + [11
4a t (t
H.R. SHERCLIFF, Research Associate, and M.F. ASHBY, The equation contains two reference parameters. The first,
Professor, are with the Engineering Department (Materials Group),
Cambridge University, Trumpington St., Cambridge CB2 1PZ, United to, is a characteristic heat-transfer time defined by to =
Kingdom. r2n/4a. The second, z0, is a characteristic length; its func-
Manuscript submitted November 9, 1990. tion is to limit the surface temperature to a finite v a l u e - -

METALLURGICAL TRANSACTIONS A VOLUME 22A, OCTOBER 1991--2459


Table I. Symbols T* (2/70 (q*/v*) exp - [(z* + z~) 2 y.2 ]
A surface absorptivity ( - ) = [t* (t* + 1)] w2 _ t* + (t* + 1)J
AI A I temperature of the steel (~ or K)
R aspect ratio of a rectangular source ( - ) [3]
I0 intensity of laser (power/area) (W m -z) Means for determining z0* (a function of v* only) are
T temperature (~ or K)
discussed below. The time to peak temperature, tp*, at a
Tm melting temperature of the steel (~ or K)
ro, Tp initial and peak temperatures (~ or K) given position (y*, z*) is found by differentiating with
T , Tp normalized temperature and peak temperature respect to time. ~241The m a x i m u m case depth and onset
(-) of melt occur at the midtrack position, y* = 0. In this
thermal diffusivity of the steel (mZs-1) instance, the derivative is a quadratic in t* which may
length of rectangular source in the travel be solved directly:
direction (m)
q beam power (W) 1
q*, q* normalized beam power, Gaussian and tp* = 4 [2(z* + z~) 2 - 1
non-Gaussian beams ( - )
rn radius of a Gaussian beam (m) + [4(z* + z*) 4 + 12(z* + z*) 2 + 1] 1/2] [41
t time (s)
to heat flow time constant (s) Substitution of t* into Eq. [3] yields the peak normalized
tp time to reach peak temperature (s) temperature, T f .
t* normalized time ( - ) 2. Surface temperature calibration
v beam speed (ms -1) Ashby and Easterling tm used a simple solution for the
V*, VR* normalized beam speed, Gaussian and surface temperature (z = z* = 0) due to Bass t161 to cal-
non-Gaussian beams ( - )
culate the characteristic lengths for two limits of the tem-
W width of rectangular source normal to travel
direction (m) perature field: the "near field" and "far field," We use
x, y, z Cartesian coordinates with origin at the beam the same method, but now find a single length Zo valid
center (m) for the whole temperature field. The result they used is
x*, y*, z* normalized x, y, z coordinates ( - ) valid if the beam radius is large compared to the distance
Zc case depth (m) over which heat flows during the interaction time, ~-, i.e.,
ZO characteristic length (m) rB ~> 2(at) I/z. At midtrack, -r = 2rB/v, so this condition
A thermal conductivity of the steel (J m-~s-~K-1) may be written v* >> 8. In this limit, a stationary beam
pc volumetric specific heat (J m-3K -~) of uniform intensity 10 applied for a time equal to the
T interaction time (s) interaction time produces a peak surface temperature given
by (Bass, El61 Eq. [10])

2A I0 (a~.)1/2 [5a]
Tp - To = .n.1/2 h
the true surface is taken to be a distance z0 below the
"model" surface. It is evaluated by matching our solu- The average intensity of a Gaussian beam is Io =
tion at z = z0 to known solutions for the peak surface q/(~r r2); hence,
temperature. Strictly speaking, Rykalin's solution is for
a high-speed source, but the use of z0 permits the use of 2Aq (a.01/2 [5b]
the solution at all beam speeds. Tp - To = r r~------~A
We define the following dimensionless groups:*
or in dimensionless form, with r = 2rs/v:
*In the context of laser hardening of plain carbon steels and cast
irons, it is sensible to normalize the temperature rise and power (Tp)z,=O = ( 2 / 7 ) 3/2 q * / ( v * ) 1/2 [6a]
using the A~ temperature, as this does not depend on the material. To
make the solution more generally applicable to laser processing of The conditions for first hardening (Tp = A1) or the onset
materials, the melt temperature would be more suitable.
of melt (Tp = T,,) are thus defined by a constant value
of a single dimensionless group containing all of the pro-
T* = ( T - To)/(A1 - To) cess variables:
(dimensionless temperature rise) [2a] q*/-r*r. ,~1/2 (7/./2)3/2 constant [6b]
q* = A q / r s A (Al -- To) with T* taking the value appropriate to the peak tem-
perature of interest (A~ or T,,).
(dimensionless beam power) [2b] Data for Gaussian treatments typically lie in the range
v* = vrB/a (dimensionless beam speed) [2c] 0.5 < v* < 500, so this high v* solution is not good
enough. Bass I161 gives a more general solution (in his
t* = t/to (dimensionless time) [2d] Eq. [23]) for the peak surface temperature induced by a
(x*, y*, z*) -- (x/rB, y/rB, z/rs) stationary Ganssian beam acting for a time r as before,
which in dimensionless form is
(dimensionless x, y, z coordinates) [2e]
:1r
(Tp)z,= 0 ---- ( 1 / T r ) 3/2 q* t a n -1 ( 8 / / v * ) 1/2 [7a]
The distance z0 is similarly normalized, z* = zo/re. The
temperature field equation is then Once again, a constant value of a single dimensionless

2460--VOLUME 22A, OCTOBER 1991 METALLURGICAL TRANSACTIONS A


parameter defines first hardening and the onset of melt, Table II. Thermal and Process
now valid for all v*: C o n s t a n t s - - M e d i u m Carbon Steel

(q*/T*) tan -1 ( 8 / 1 2 " ) 1/2 = ,/./.3/2 = constant [7b] Absorptivity, A 0.7


Thermal conductivity, A (780 0(2) 25.6 J s-lm-lK -l
Note that in the limit of a stationary beam, (v* ~ 0) Thermal diffusivity, a (780 ~ 3.6 • 1 0 - 6 m 2 s -1
tan -1 (8/v*)1/2---~ 7r/2, so there is a minimum value of Specific heat, c (780 ~ 925 J kg-lK -l
(q*/T*) equal to 2~r-~ or approximately 3.5. This iden- Density, p 7.64 x 1 0 3 k g m - 3
tifies the minimum q* required to reach a given tem- Initial temperature, To 300 K
perature: surface hardening, for example, occurs when A 1 temperature, A~ 996 K
Tp = A1 or T* = 1, so q'in = 3.5; surface melting occurs Melting temperature, Tm (0.4 pct C) 1774 K
Melting temperature, Tm (0.6 pet C) 1756 K
when Tp = Tm or (taking values for 0.4 pct carbon steel
from Table II) T* --~ 2.1 2, and thus, qmin
* = 7.5.
The question arises as to whether a solution for a sta-
tionary beam is a valid approximation to a moving beam. quired track width is usually fixed, and the length in the
Shercliff and Ashby [24] compared Bass' solutions tl6] with travel direction is varied, so it is sensible to normalize
a numerical one due to Pittaway [25] for the surface tem- the process variables using the beam width:
perature due to a moving Gaussian heat source. The
agreement was good and justifies the use of this approach.
q* = Aq/wA(A1 - - To) [8a]
To solve for z*, we must increment z* until the peak v* = v w / a [8b]
surface temperature solutions match, as z* cannot be ex-
pressed explicitly as a function of v*. For a given value z~ = z / w [8c]
of z*, we find l* at z* = 0 from Eq. [4], substitute for with T* as before and the case depth z'c; the subscript
t* in Eq. [3] to give T*, and compare the result with R refers to a rectangular, as opposed to Gaussian, beam.
Eq. [7a]; z* is incremented and the calculation repeated There is wide scope for varying the number of sources,
until the values of T* are equal. Note that as T* ~ q* their radius and location, and the share of the energy in
in both Eqs. [3] and [7a], the value of z* depends only each. We use a single Gaussian radius in a given case
on v* and not on q*. (defined by the ratio rB/w) and a symmetrical array of
3. Choice of "master diagram" sources. Figure 2 shows isometric views of the model
Four dimensionless groups define laser hardening with and true beam shapes for R = 1.0: such plots were used
a Gaussian beam. The aim is to produce a diagram from to optimize the model distribution of Gaussian sources.
which process variables may be readily selected. A con- The mismatch between the true and composite sources
venient plot is the dependent variable, Zc*, against v* with rapidly loses significance with increasing distance below
contours of constant q* (and with T* = 1 to give Tp = the surface. Note that in general, the fit across the track
AI), as shown in Figure 1. As surface melting is gen- is more significant than that along the track, as at high
erally undesirable, the contours are truncated if the sur- speeds, all sources resemble line sources.
face melts, that is, when T* = (Tin - To)/(A, - To) at 2. Surface temperature calibration
z* = 0 (Figure 1 shows the depth at which surface melt To calibrate the temperature field, we use Bass '[161 high
commences for a 0.4 pct carbon steel, for which the melt v* solution for a stationary uniform beam of intensity I0
T* --~ 2.12). Figure 1 also illustrates that there is a min-
imum value of q* (evaluated earlier) below which hard-
ening is not possible: case depth contours cannot be plotted 10
for values of q* below qmi,* -- 3.5. Similarly, for 0.4 pct
carbon steel, surface melt only occurs for contours above
*
qrnin 7.5.
I
B. Modeling of Laser Hardening with a Zc~' ~. ONSET OF
Non-Gaussian Beam .~ ~ J SURFACE
~.-/" MELTING
1. Superposition of Gaussian sources to model 10-I " ~
non-Gaussian mode structures
A non-Gaussian source may be simulated by super-
posing a number of Gaussian sources. The total power
is shared between the sources, and their spatial arrange- 10-2
q* = 4 8 15 30
ment adjusted, to give the best fit between the theoretical
and true energy profiles. For any position and time, the
temperature rises due to each source are summed. This
10-3 i
is valid if the thermal properties of the material are as- 10-2 10-1 1' J
10 1~0 2 10 3
sumed constant, as the differential heat flow equation is -Lv ~
then linear. This technique will be demonstrated for rect-
angular, uniform sources, defined by two dimensions: 1, Fig. 1 - - T h e proposed dimensionless master plot for Gaussian treat-
ments: case depth z* against beam speed v* with contours of constant
the length in the travel (x) direction, and w, the width beam power q* (for peak temperature T* = 1.0, corresponding to
in the cross-track (y) direction. A beam aspect ratio, R, Tp = A0; the contours are truncated if the surface reaches the melt
is defined by R = l/w. In practical treatments, the re- temperature for 0.4 pct carbon steel (T* = 2.12 at z* = 0).

METALLURGICAL TRANSACTIONS A VOLUME 22A, OCTOBER 1991--2461


(a) (b)

ra
11.5 11 11.5
3-
11 10 11
11.5 11 11.5

X = W

(c)
Fig. 2 - - A r r a n g e m e n t of nine Gaussian sources to model a rectangular source of aspect ratio R = 1.0: (a) isometric view of the composite
Gaussian source; (b) isometric view of the equivalent uniform source; and (c) plan view of the sources; the numbers show the percentage of the
total power in each source.

(Eq. [5a]), noting that I0 = q / l w and time ~- = l/v. The with constant speed in the x-direction (x = vt). As to =
peak temperature is then r2/4a and t* = t/to, x = vt is equivalent to x* = v't*~4.
Thus, the contribution of a source offset by a distance
2Aq (al/v)W2 [9] x* is calculated by introducing a time shift equal to 4x*/v*.
Tp - To - 7r~/2Alw The distance z* is found by summing the contributions
due to each Gaussian at z* = 0, and the net T* set equal
Normalizing with the parameters in Eqs. [8]: to Bass' solution [16J (Eq. [10a]).
T* = (2/77/2) q*/(R v*) ~/2 [10a]
3. Choice of master diagram
This is valid if the beam is large compared with the dis- For rectangular beams, there are five dimensionless
tance over which heat flows during the interaction time, groups: T*, q*, v*, Z'c, and R. The master plot for rect-
i.e., w / 2 >> 2(at) 1/2, or in dimensionless terms, v* >> angular beams, which is the parallel of that for the
16R. Note that a single value of a dimensionless param- Gaussian case, is z*c against v*, with contours of con-
eter containing all the process variables governs the po- stant q*, for a given aspect ratio R, and T* = 1 (cor-
sition of first hardening or the onset of melting, as for responding to Tp = A0; examples are given later to
Gaussian beams: compare experimental data with theory.
qh/Tp (R VR~) 1/2 = 7 r l / 2 / 2 = constant [10b] An alternative plot is more suitable for process dia-
grams. The available power is set by the laser, and so
Critical values of q* for melting or first hardening cannot the parameter q* is fixed first. We note from Eq. [10b]
be identified, as this solution becomes invalid at low that for a given q* and T*, the value of Rv* is constant;
v* (qR/Tp
* * ~ oo as v*--~ 0). i.e., first hardening or the onset of melt occur at fixed
The net temperature rise at any position and time is values of Rv* for all R. The best dimensionless process
found by summing the contribution of the separate sources diagram is therefore z*c against Rv*, with contours of
using Eq. [3], and the peak condition found by incre- constant R for given values of q*. Figure 3 shows this
menting t* (details in Reference 24). The temperature at plot for two values of q* (four parameters can appear on
a position relative to a given source in the y-direction is a single diagram, as the regions covered by two suffi-
given directly by the equations. Relative positions in the ciently different values of q* do not overlap). The con-
x-direction correspond to intervals in t, as the beam moves tours for a given q* lie in an "operating window" between

2462--VOLUME 22A, OCTOBER 1991 METALLURGICAL TRANSACTIONS A


i i 1 i I 1
ONSET OF SURFACE MELTING 0NETOFSURFACE
ME,NG
I
2.0 v
1.0
10-1 2.0
t.)
1.0 I1,1
Z*Rc 0.2 a:-
t--
t O
R
0.2
t.)
<
127
I--
"1-
I--
163 13_ 10-~
ILl a/w = 5c 200 800 1
t23
qR = 5 qR = 25 laJ
<
L.)
(Wlmm) F,RST
DEN,NO_
. j)
FIRST HARDENING
I() l i r
10-4 I I I
10 102 103 lO z' I 10 102 103 104

SPEED x SPOT LENGTH, 9 ( ( m r n 2 / s )


Fig. 3- A dimensionless master plot for rectangular, uniform beams: Fig. 4 - - A dimensional process diagram for medium carbon steel; the
case depth, z'c, a g a i n s t a m o d i f i e d b e a m s p e e d , Rv*, w i t h c o n t o u r s values of the material properties have been substituted into the di-
of constant a s p e c t ratio, R, f o r a g i v e n v a l u e o f q*; the c u r v e s f o r mensionless groups to give a plot of zc/w against vl with contours of
q* = 5 a n d q* = 25 are s u p e r p o s e d o n the s i n g l e p l o t as the r e g i o n s constant l/w for values of q/w = 50, 200, and 800 W/mm. The
covered by e a c h v a l u e d o not o v e r l a p . position of first hardening and the onset of surface melting are readily
identified.

the values of Rv* corresponding to surface melt and


first hardening. The ratio of these values is constant
(=(Tin - To)Z/(A1 To)z) for all q*, so on a log scale,
-
numerical solutions lead first to plots of Z*(V*)1/2 against
the "window" is of constant width. 7r3/Z(v*)]/Z/q* with contours of constant v*. A set of sim-
For convenience of use, the parameters of a process plified master equations are used to plot a diagram por-
diagram should not contain more than two variable quan- traying q*, v*, and zc* (exactly as we do), but they chose
tities. In addition, it is useful to present the diagrams normalized power and speed as axes, with contours of
with dimensional parameters by substituting values for constant case depth. Our model agrees reasonably well
the constants in all of the groups. The optimum process with all of their numerical results (a maximum disparity
diagram is therefore a dimensional form of Figure 3: a of roughly a factor of 2 in q* at high v* and z* but usu-
plot o f z J w against vl for given q / w and aspect ratio ally much less). The drawbacks of the scheme chosen
l/w. Figure 4 shows a process diagram for En8 steel by Davis et al. are that (a) the range of data extends to
(using the material constants in Table II). values of v* which are almost 20 times greater than the
maximum value of their master plot; (b) the deviation
between theory and experiment is unclear; and (c) there
III. COMPARISON WITH OTHER MODELS
is no assessment of the onset of surface melting.
The model presented above has been compared with
the parameters, equations, and diagrams used by other
workers. Some of the results are given below; more can C. The Method of Kou and Co-Workers
be found in Reference 24. Kou and co-workers (14"15] used a three-dimensional fi-
nite difference procedure to calculate the temperature field
A. The Plot of Megaw and Kaye induced by a moving rectangular, uniform heat source.
Megaw and Kaye (~21 plotted case depth against the re- Their dimensionless plot for a square beam is (in our
ciprocal of the beam interaction time (1/~- = v/l), with notation): normalized peak temperature Tp* /(qi~
* v*) against
contours of constant power density (power/source area). depth z* with contours of constant v*. This plot is shown
in Figure 5 with our solution for a uniform square beam
This plot agrees with our Gaussian model at large radii
superposed; the agreement is good. Kou and co-workers'
(when the heat flow is one-dimensional) but becomes
inaccurate at small radii, when the heat flow becomes analysis uses more sophisticated techniques based on re-
lated dimensionless parameters and provides a test of the
two-dimensional.
accuracy of the analytical model developed here. How-
ever, Kou and co-workers' plot is not ideal for selecting
B. The Model of Davis et al. operating conditions, and their data covered too small a
Davis et al. It31 model a Gaussian source and define range of the process variables to give a critical test of
groups which are essentially identical with our own. Their the model.

METALLURGICAL TRANSACTIONS A VOLUME 22A, OCTOBER 1 9 9 1 - - 2 4 6 3


15 I zx q"=7.4]
~ KOU (1983) "~'" • t3 8.6~'Li (19841
x 17.9,1
PRESENT WORK ~"~

*"~ ~,,~ ONSETOF


k ~''"% ~ T'FANCE

10
102 q*~= 5 7
12 20 40
=20

x
% 10-30-1 1 10 13 ~ 102 103

Fig. 6 - - T h e master plot for Gaussian laser hardening with data due
~ to Li t26] for a 0.6 pct carbon steel.
R=/+O

.4 , . ~ - . . . "~ I+ R-- /STEEN AND


~;; Ic0u.T.EY,9.9,
L _ ~ rB =l.02mm

0 0.25 0.5
~ N ~ ~ ~ ONSETOF

Fig. 5--ThedimensionlessplotofKouetal.:U']peaktemperature,
T*/(q* v*), against depth, z*, with contours o f constant b e a m speed,
v*; our solution for a u n i f o r m b e a m o f R = 1.0 is s h o w n superposed.
Z, ~ ~ ~~ ~
10-2
q~ = 5 7 12 20 40

IV. C O M P A R I S O N
WITH EXPERIMENTAL DATA
I I I
Data to validate the model have been obtained from ]0-310-1 I 10 102 103
our o w n experiments [24] and from those of other work-
ers. [26'27'281 The values o f the thermal and process con- (a)
stants contained in the dimensionless groups are listed
in Table II.
~.~ , , lu q. =19.81
~ - - ~ I x+ 23"2/ STEENAND
A. Data f o r Laser Hardening with a Gaussian Beam
Figure 6 shows the data due to L i [261 for laser hard-
ening of a 0.6 pct carbon steel. The agreement with the
10-!
master plot is good. The data for the highest value of q*
lie parallel to the melt line, probably because melting
Zc
occurred: the hardened depth is thus limited by the heat
absorbed by latent heat of melting and reduced absorp-
tivity. Figures 7(a) and (b) show some o f the data for
10-~
0.4 pct carbon steel due to Steen and Courtney. [271 At q* = 5 7 12 20 40
the small beam radius (Figure 7(a)), all of the data lie
above and parallel to the melt line, and there is distinct
bunching of the data over a wide range of q* values.
This again suggests that melting has occurred. For the 0-~ i i I
larger radius (Figure 7(b)), bunching of the data above 1 0~ 1 10 102 103
the melt line is still apparent, but the data points which
lie below it show numerical agreement which is gener- (b)
ally within the accumulated experimental error. Similar Fig. 7 - - T h e master plots for Gaussian laser hardening with data due
trends were observed at three other radii from the same to Steen and Courtney t27] for a 0.4 pct carbon steel, (a) with b e a m
data source. [24] The discrepancy between theoretical and radius r8 = 1.02 m m and (b) with b e a m radius rB = 2.38 m m .

2464-- VOLUME 22A, OCTOBER 1991 METALLURGICAL TRANSACTIONS A


experimental q* increased with radius to a m a x i m u m of Bransden. E2sl Master diagrams for two aspect ratios are
about 30 pct. shown in Figures 8(a) and (b); data for a small range of
aspect ratios are plotted together, as the master diagram
B. Data f o r Laser Hardening with a is essentially unchanged. The agreement between theory
Non-Gaussian B e a m and experiment is good, with the discrepancy in q* in-
creasing toward the surface. Similar data for R = 0.35,
Case depth data for uniform rectangular beams have 0.42, and 1.71 showed the same pattern. [24] We conclude
been obtained from our own work t24} and from that of that the superposition technique gives predictions for non-
Gaussian treatments which are at least as good as those
made for single Gaussian sources.
qR* R

x[] 3.72'9 0.220"22}


BRANSDEN(1986
+ 7.9 0.2 PRESENT WORK V. CONCLUSIONS
I. The approximate heat flow model of Ashby and
Easterling t171 for laser hardening has been developed
to describe both Gaussian and non-Gaussian sources
10-1 +'h- over a wide range of process variables. The main ad-
ONSET OF vances on the previous work are
su%Ag a. a simplification by judicious choice of dimen-
sionless parameters;
10-2 b. a new surface temperature calibration, extending
the approximate Gaussian solution to all practical
beam speeds; a high-speed solution was found to
be acceptable for rectangular, uniform sources;
c. a general Gaussian solution with a single calibra-
tion distance, enabling extension of the model to
non-Gaussian sources;
qR = 2.5 5 10 20 d. identification of constant dimensionless parame-
ters containing all of the process variables for both
i i i Gaussian and non-Gaussian sources, which deter-
lo 10 102 103 104 mine the position of first hardening or the onset
VR of melting, thus identifying the "operating win-
(a) dow" on the master diagrams. The equations for
peak surface temperature are
%* R Gaussian (all v*):
+ 4.7 0.89 BRANSDEN(1986)
[] 7.9085'1 PRESENTWORK ( q * / T * ) tan -1 (8/v*) 1/2
x 11.2 0186 f
Aq r 8 a ] 1/2
- tan-1 _ _
x
reA (Tp - To) kvrBj
ONSET OF
10-1 = 71-3/2

ZRc Gaussian (high v*):
Aqa 1/2
10-2 :'Ir :tr 9 * x l / 2 = (~/2) 3/2
q /Tp(v ) vl/Er3/EA(Tp_To)

Rectangular, uniform:
A q a 1/2
16 3 :~ ~ $ 1/2
qi~/Tp(RVR) v 1/2 11/2 wA (Tp - To)
qR :5 10 20 40
= rr112/2
4 I [ I 2. The models compare favorably with previous work
10-1 I0 102 103 I04 in the literature which require more elaborate nu-
~YR merical computation and with a wide range of ex-
(b) perimental data. The accuracy of the models for both
Gaussian and non-Gaussian sources is good enough
Fig. 8 - - M a s t e r plots for n o n - O a u s s i a n l a s e r h a r d e n i n g (z*c a g a i n s t
v*, w i t h contours of constant qR*, for 7~*p= 1.0) for a s p e c t ratios
for selecting trial operating conditions.
(a) R = 0 . 2 to 0 . 2 2 and (b) R = 0 . 8 5 to 0 . 8 9 , w i t h data f r o m our 3. The process diagrams for rectangular sources pre-
own e x p e r i m e n t s t241 and due to B r a n s d e n . t28~ sented in this work give straightforward means for

METALLURGICAL TRANSACTIONS A VOLUME 22A, OCTOBER 1991--2465


selecting the process variables in the most useful se- 11. Y. Arata, H. Maruo, and I. Miyamoto: I/W Documents IV-241-78
quence: first the beam power, then the track width, and 212-436-78, 1978.
12. J.H.P.C. Megaw and A.S. Kaye: Proc. 4th European Electro-
and finally, the scan speed and spot length.
Optics Conf., Utrecht, The Netherlands, Oct. 1978, Soc.
Photo-Optical Instrumentation Engineers 164, 1979, pp. 241-51.
13. M. Davis, P. Kapadia, J. Dowden, W.M. Steen, and C.H.G.
ACKNOWLEDGMENTS Courtney: J. Phys. D: Appl. Phys., 1986, vol. 19, pp. 1981-97.
14. S. Kou: Metall. Trans. A, 1982, vol. 13A (3), pp. 363-71.
The use of the lasers and assistance of Dr. J.H.P.C. 15. S. Kou, D.K. Sun, and Y.P. Le: MetaU. Trans. A, 1983,
Megaw and Mr. A.S. Bransden of the Culham vol. 14A (4), pp. 643-53.
Laboratory Laser Applications Group (U.K.A.E.A.) are 16. M. Bass: in Physical Processes in Laser-Materials Interactions,
gratefully acknowledged. Mr. A.S. Bransden, Dr. W.B. M. Bertolotti, ed., Plenum Press, New York, NY, 1983, ch. 3.
17. M.F. Ashby and K.E. Easterling: Acta Metall., 1984,
Li (University of Lulea, Sweden), and Professor W.M. vol. 32 (11), pp. 1935-48.
Steen (University of Liverpool) all readily provided ex- 18. N. Christensen, V. de L. Davies, and K. Gjermundsen: Br. Weld.
perimental data. The financial assistance of the Science J., 1965, Feb., pp. 54-75.
and Engineering Research Council in the form of a 19. T.W. Eager and N.S. Tsai: U.S. Welding Res. Supp., 1983, Dec.,
Research Studentship is acknowledged. pp. 346s-355s.
20. J. Mazumder and W.M. Steen: J. Appl. Phys., 1980, vol. 51 (2),
pp. 941-47.
REFERENCES 21. P. Henry, T. Chande, K. Lipscombe, J. Mazumder, and W.M.
Steen: Proc. Conf. 1CALEO '82, 1982, paper 4B-2.
1. P.J. Oakley: Weld. Inst. Res. Bull., 1981, Jan., pp. 4-11. 22. O.R. Myhr and O. Grong: Acta MetaU. Mater., 1990,
2. A.J. Hick: Heat Treat. Met., 1983, vol. 1, pp. 3-11. vol. 38 (3), pp. 449-60.
3. J. Mazumder: J. Met., 1983, May, pp. 18-26. 23. N.N. Rykalin, A.A. Uglov, and A. Kokora: Laser Machining
4. J.H.P.C. Megaw: Laser Welding, Cutting and Surface Treat- and Welding, Pergamon Press, Oxford, United Kingdom, 1978,
ment, Welding Institute, Cambridge, United Kingdom, 1984, ch. 3.
ch. 6. 24. H.R. Shercliff and M.F. Ashby: Cambridge University
5. P.A. Molian: Surf. Eng., 1986, vol. 2 (1), pp. 19-28. Engineering Department Report, CUED/C-MAT/TR169,
6. H.C. Carslaw and J.C. Jaegar: Conduction of Heat in Solids, 2nd Cambridge, United Kingdom, Sept. 1989.
ed., Oxford University Press, Oxford, United Kingdom, 1959. 25. L.G. Pittaway: Br. J. Appl. Phys., 1964, vol. 15, pp. 967-82.
7. H.E. Cline and T.R. Anthony: J. Appl. Phys., 1977, vol. 48 (9), 26. W.-B. Li: Doctoral Thesis, University of Lulea, Sweden, 1984.
pp. 3895-3900. 27. W.M. Steen and C.H.G. Courtney: Met. Technol., 1979, Dec.,
8. M. Lax: J. Appl. Phys., 1977, vol. 48 (9), pp. 3919-24. pp. 456-62.
9. N.N. Rykalin, A.A. Uglov, and M.M. Nizametdinov: Sov. 28. A.S. Bransden: Culham Laboratory, UKAEA, Abingdon,
J. Quantum Electron, 1977, vol. 7 (7), pp. 853-56. Oxfordshire, United Kingdom, personal communication, 1986.
10. Y.I. Nissim, A. Lietoila, R.B. Gold, and J.F. Gibbons: J. Appl.
Phys., 1980, vol. 51 (1), pp. 274-79.

2466-- VOLUME 22A, OCTOBER 1991 METALLURGICAL TRANSACTIONS A

Вам также может понравиться