Professional Documents
Culture Documents
(Basic: Monitoring Assisted by Image Technique'
(Basic: Monitoring Assisted by Image Technique'
374
a laser speckle technique, This method isbased on observations of the change of the
laserspeckle patterndependingon the surface roughness and texture change of the
material caused by slip bands due to plasticdeformation, The laser speckle pattern
was analyzed automatically and quantitativelyusing an image processing system in
thisexperiment. A new parameter, ck, was derivedto statistically express the features
of the speckle pattern.Itwas foundthat thisparameter can express the features of the
speckle pattern accurately and has excellent reproducibility. Also,itallows an accu-
rate measurement of plastic strain. The experiment was carried out for cases of both
static tensileand fatigue loading using aluminum alloy as the material,
with the change of surface roughness and profile.If tribution of the lightintensity of the laser speckle
the relationship betweenthe laserspeckle pattern and variesdrastically, we considered a new statistical
the amount of plastic deformationor fatigue damage parameterto express the features of the laser speckle
is known in advance, we can estimate the damage of pattern. The reproducibility of this parameter was
metal materials based on the speckle pattern. examined and the accuracy of the measurement was
37S
of a testing machine and the possibilityof fatigue expressed by value from O to 255 in the
a numerical
monitoring by the laser speckle technique was inves- image memory. The digitalimage data were transfer-
tigated. red to the main memory of a minicomputer (VAX 111
730)and then stored in itsmagnetic tape. Softwareto
2. Experimental Procedure
analyze the speckle image was developed on the
2.1 Experimentalsystem mlnlcomputer.
tional, the speckle pattern is directional and not plasticstrain caused by nniform tension. These figures
axisymmetrical. AIso,ifthe initial surface roughness show thatthe speckle patternchanges clearly with the
islarge,speckle pattem change is small in the range magnitude of plastic strain, The peak gray levelat the
of small plasticstrain. Therefore, if it is required to center of the speckle image decreasesand the gray
detect a small change of speckle pattern, the specimen level distribution spreads out with the increaseof
surface should initially be smooth and nondirectional. plastic strain.
In thisexperiment, the specimens were polished finally Figure3 shows the relationship between the ratio
with aluminum dioxidepowder in random directions of vamtldrnax and plastic strain. Here, M.dn and nlnax
so that the surface texture became macroscopically are the minimum and maximum values, respectively,
isotropic and uniform. The initial surface roughness among second-order moments of inertia of the gray
was about O,1 Fm Ra forevery specimen, leveldistribution around the fourdifferent axes shown
Figure 1 shows the layout of the experimental in the chart of Fig. 3, If the ratio n(tttnvamax isclose to
systern. He-Ne laserwas illuminated on the specimen 1, itmeans that the second-order moments of inertia
surface and the laser speckle pattern was formed on around the four axes take almost the sarne value;
the ground glassplacedin frontof the specimen. The that is, the gray level distribution of the speckle
diameter of the laser beam was about 1mm. The pattern isround and axisymmetrical. From the figure,
image of the speckle pattern was inputintothe image thisratio issmaller than 1 when p]asticstrain issmall,
processing system (HamamatsuCIOOe)through a TV but itapproaches 1 with the increaseof plasticstrain.
camera and then AID converted, Resolutionof the This means that the gray leveldistribution is slightly
image memory was 256× 256and the gray scale of one directionalinitially, butitbecomes nondirectional and
pixel was 256. The brightness of the laser speckle was axisymmetrical around the centroid of the gray level
distribution when plastic deformationbecomes large.
Hence we considered the followingparameter to
express the spread of the gray leveldistribution
around the centroid of the distribution.
First,the image of the speckle pattern was
smoothed to reduce the noise, We used a moving
averaging method of 3 × 3 pixels for smoothing. Then
the centroid, O(xo,yo), of the gray leveldistribution
was obtained (Fig. 4(a)), and the average gray level,
g(r), of pixels on a circle c with the center O and the
radius of r was obtained <Fig.4(b)).
The average
Zg(x,y)
g(r)-C , (1)
nc
where g(x,y) isthe gray levelat a point (x,y) on
JSME Internatlonat
Jotcrnal SeriesI,Vol.34, No. 3, 1991
376
m mo
mo pt
mo m::
;
-: wo ioo-.p:
v tse
: lco
Loo IOb
se N
oso ioo iso ISO m 4oe 1se ooo sso oco too tSO 1co Jco su -o soo abo
IInitia]soo #ep=5.20%-oe
1,k' m
}sc::
mo:t
reo-e; )o'o'tt
Lco LN
lco too
co
se
oes tco tso aco(b)tSO JOO "O mo soe sso o se ico lm 2oo 2sc sce sw too eo soo cao
lEp=1.55%4oo
r
(e) ep=6.60%
ma
}w-tt
isc
lco
co
o,L
S O・6X
O.4
y
O.2
e
O123456 7
x
PlastiestrsinS %
(b) Imageplane
Fig. 3 Relationshipbetween Mminuaax and plastic
strain Fig.4 Procedurefor obtaining g(r)
377
used the followinglinearmapping functionfor nor- where M isthelimitof r, which was taken as 110 in
malization: this analysis.
g,(.)=255
r- tn.
(2) Next, we examined the influenceof brightness of
gnal- gVndn
the laserspeckle on the ch parameter, Figure7shows
This functionmakes the maximum gray levelin g(r), the relationship between ck and gmni when thespeckle
to be 255 and
ginai, the minimum value, thdn,to be O,
pattern was obtained at the same point of a specimen
The normalized gray leveldistributionis shown in with different brightness of the laserspeckle, The
Fig.5(b). Figure 6 shows the difference in the nor-
speckle brightness was changed by two methods. One
malized gray level distributiondependingon the plas-
is that the aperture of the lensof the TV camera was
tic strain, The gray leveldistribution, g'(r),is found changed under a constant intensity of the laser light
to spread with the increaseof plasticstrain. Hence,
and the other is thatthe intensity of the laserlight
we considered the followingparameter that can
was changed with an ND filterunder a constant
express the feature of the gray level distribution:
aperture of the lens. This figureshows that the ch
parameter isalmost constant inthe wide range of thnax
m (from20 to 160)and takes almost the sarne value in
10T
noV.
both cases, This means that the ci parameter is not
NO by the brightness
significantly influenced of the laser
tso:
ep
under four differenttesting conditions of speckle
to brightness. The first one was thatthe aperture of the
o
orc -o eo eo tco tmo le tsc TV camera was constant butthe light intensityof the
r
laser lightwas adjusted with an ND filter so that the
(b) g'(r)(Normalized)
Fig,5 Normalization
of g(r)
2000
i-:
1800
Ck
utLV-
-w l600
lst;
ten:n
t-/'e 1400
tie;
Lco'
co 1200
co
to
N 1OOOo
o
100 200
o ee va sc eo too tN l" tse
r
gmar
rsME International
JourTtal SeriesI,Vol. 34, No, 3, 1991
NII-Electronic
NII-Electronic Library Service
The JapanSociety
The Japan Society of Mechanical
of Mechanical Engineers
Engineers
378
- (5)
average gray level of the speckle image became Ep=c(cA)`.
approximately constant. The others were that the Here, log(c)=-a/b and d=1/b. The values of c and
aperture of the lensof the TV carnera was fixedat d were obtained by the leastsquares method ; c=3.99
three different
values, i,e., "32,"16 and rr8, under a × 10'` and d==O.74. Therefore, we can estimate the
constant light
intensity, It isfound that the values of magnitude of plasticstrain from the value of the ch
ch obtained under differentconditions of speckle parameter using this equation, It was found thatthis
brightness coincide very well and the ck parameter is parameter ch is not infiuenced signhicantly by the
not influenced by speckle brightnessin all cases of brightness of the laserspeckle and thatithas a very
plastic strain, The parameter ch increases with Ep and good reproducibility. It can be concluded that this
has a tendency similar to the surface roughness shown parameter will allow us to make an accurate measure-
specimens coincide very well, This shows that the ch in the figure. The testing machine used was a servo-
parameter has a very good reproducibility and that hydraulicfatiguetestingmachine (Instron) .The spec-
there is a fixed relationship between ch and plastic imens were subjected to uniaxial tensilefatigueload,
strain Ep. Figure 10 shows that there isa linearrela- The specimen was set in the testing machine and the
tionshipbetweenlog(ck) and log(Ep), Thisrelation can laser light was illuminated on the specimen surface,
be expressed by the following equation : The experimental layoutisalmost the same as in Fig,
log(ck)=a+blog(Ep). (4) 1 except that we used a high-speedcamera (SP2000,
Alternatively, itcan be expressed by Kodak) to obtain the speckle image and a personal
computer (IBM PS!2) to analyze the image data,The
image of the speckle pattern was taken using the high-
O.6 speed carnera during fatigue loading without stopping
e.3y98
O.28:en
1Ooo
O.1
Ck
O 1234567
PSasticstrain ep %
1OO
Fig. 8 Relationchipbetween surface roughness and ep
2000 10
o.eol o.ol o.1
Plasticstrain
%
15oo
500
O 2 4 6 8 10 12
Plasticstrain% %
Fig. 9 Relationship
betweenc- and Ep Fig. 11 Test specimen for fatiguetest
3mp
radius of 6.4mm, As the notch radius is rather large, derivedin this study expresses the featuresof the
the stress concentration factor dueto the notch can be laser speckle pattern numerically and hasa very good
considered to be 1.The conditions of fatiguestress at reproducibility. The relation between ch and plastic
the notch root were as follows:the mean stress was strain isexpressed by a simple equation in the range
fixed to am=217MPa and the stress amplitude was of plastic strain (Ep<le%) where we made experi-
changed oa=155, 167, 186, 205 and 217MPa. Figure
as ments. Then, plastic strain can be estimated easily
12 the relationship between stress arnplitude
shows using this equation. It isconcluded that the magnitude
and fatigue life,IVf.Figure 13 shows the relationship of plastic strain can be estimated accurately by the
betweenthe ck pararneter of the laserspeckle pattern laserspeckle technique.
and AVNf. Here,N isthe number of stress cycles. Itis The patterndid not change so dramati-
speckle
normalized M, forthe
by fatiguelife, corresponding cally with fatigueloading,and the change of the cA
stress amplitude shown in Fig,12,The value of ch is parameterwas very small, Also, the fluctuationof the
very small compared to that in the case of static ch parameter was not as distinctive
as that of static
plastic deforrnation,
especially when aa issmall, Also, tensile test.Therefore,detectionof fatigue damage
the change of the ch value is very small when oa is will not be as accurate. However, the ck parameter
smaller than rs6MPa, However, the ck value in- changed appreciably when the stress amplitude was
creases monotonously with IVIM, The ck value be- rather large. It be concluded
can that there is a
comes largewith the increase of oh, and the change of possibility of detectingfatigue damage by this
the ch parameter IVIACf becomes perceptible.Itis
with method,
Y. 200='a:zaees1OO1 100
10 o
O,Ol O,1 1
Fatiguelife IV7 xlO` NtA}
Fig. 12 Relationshipbetween stress amplitude and Fig. 13 Relationshipbetween ch and IVINfunder different
fatigue
life stress amplitudes
JSME Internationat
Jburnat SeriesI,Vol.34, No. 3, 1991
NII-Electronic
NII-Electronic Library Service
The JapanSociety
The Japan Society of Mechanical Engineers
ofMechanical Engineers
380