IngenleurArchiv
t]bersieht: Das elastischplastische Biegen einer einfach gelagerten Ringplatte wird theoretisch untersucht, wobei der Einfluf~ der Membrankr/ifte beriicksichtigt wird. Der inhere Rand der Platte ist frei gelagert und tr/igt eine stetig verteilte Querlast. Es wird ein homogenes, isotropes und ideal elastisehplastisches :Material vorausgesetzt, das der v. Misesschen Fliel~bedingung und dem Henkyschen SpannungsDehnungsGesetz geniigt. Die Rechenergebnisse werden diskutiert und mit entsprechenden Versuchsergebnissen verglichen, wobei besonderen Wert auf die Beziehung zwischen dem wirklichen Materialverhalten und den iiblicherweise in der Theorie verwendeten Annahmen gelegt wird. Die Ergebnisse werden auBerdem mit entsprechenden Berechnungsergebnissen verglichen, bei denen die Membrankr/ifte vernachl/issigt wurden, so dab der EinfluB dieser Kr/ifte deutlich wird.
1. Introduetion. Recently, the authors analysed a few problems of elastoplastic bending of circular plates at large deflection assuming the Hencky's stressstrain relation, the Mises' yield condition and the elastic perfectly plastic material 1,2. They further examined experimentally the validity of the procedure by using mild steel specimens. The results of the above analyses were found to agree quite well with the corresponding experimental results in the range of calculation (i.e., w/h ~ 1 for b/h = 25, where w: maximum deflection, b: outer radius, 2 h: thickness). Hitherto, on the other hand, the analyses of annular plates at large deflection subjected to lateral load have been restricted to either those of elastic bending or those using the limit analysis. Though the elastoplastic bending of annular plates was analysed by the present authors 3, the applicable range of the analysis which did not take into account the membrane force is limited within relatively small deflection. The problem of elastoplastic bending at large deflection generally becomes very complicated because of the effect of membrane force, and the investigation concerning the large deflection of elastoplastic bending of the annular plate has been quite lacking. However, the aforementioned analytical method of circular plates1. 2 m a y be readily applied also to the case of annular plates. The first purpose of the present paper is to analyse by t h a t method the large elastoplastic deformation of annular plates under lateral load and to compare the numerical results with those of the theory neglecting membrane force z. The second purpose is to examine by performing the corresponding experiment whether the above mentioned procedure which furnished satisfactory agreement with the corresponding experiment in the case of circular plate m a y be really applied to the present problem of annular plates in which the states of stress distribution are supposed to be considerably different from those of circular plates, and to examine the extent to which the procedure can be applied with satisfactory accuracy. 2. Basic Relations. With some care for loading and boundary conditions, the fundamental relations hold almost the same as those of circular plate, derivation of which was discussed already in the previous papers 1,2. I n the following, the derivation of basic relations is outlined briefly. Cylindrical coordinates r, 0, z, where z = 0 plane lies in the middle plane of the plate, are taken as in Fig. 1. An annular plate of outer radius b, inner radius a and thickness 2 h (h ~ b) is subjected to a laterally distributed load p(r). Let T,, To and MT, Mo be the components of membrane force and bending moment per unit width of the plate, u and w the radial displacement
1 y . Ohashi and S. Murakami; Proc. 11th Internat. Cong. Appl. Mech. Springer, Berlin, Heidelberg, 1N~ew York (1966) p. 212. Y. Ohashi and S. Murakami, Trans. AS:ME, ser. E, J. Appl. Mech. Vol. 33. No. 4. (1966). 3 y . Ohashi and S. Murakami, Bull. Japan Society of :Mech. Eng. Vol. 9. 1~o. 34. (1966) p. 271.
XXXV. Band
341
and the deflection of the middle plane, respectively. Then the equations of equilihrium of an annular plate are expressed as follows:
dr + 7  (T~   To) = O,
(1)
dr q
(M~   M e ) q T ~ dw dr  r ~]
a
r p ( r ) dr = 0 .
(2)
ZS
Fig, 1. Annular plate with nomenclatures.
The components of unit elongation er, eo and the principaI c u r v a t u r e %, a0 are given as follows :
du 1 [d,v~ a u d2w 1 dw
e,. = aTr + ~ \ dr }
e0 
(3)
~r =  a r ~ '
~0 
r dr .
(4)
1 r a'~ = 0
'
(5)
dao 
, 1
& ~ T (o:0  %) = 0 .
(6)
Components of strain e~, e0 in the plate are expressed in terms of the unit elongation and the curvature : e r = e~ @ z a ~ , e0 = e0 @ z a 0. (7) According to the S o k o l o v s k y ' s representation 4 for the principal curvatures c~ ~ = T Ao sin (wo ~ ~) ~0 J cos/~ and the analogous representation for the components of unit elongation e~ e0 = 7 A1 sin (~1 T ,~) , cos ke (9) (8)
the effective stress a ~ (@   a, aa @ ff~)U2 in the plane stress state and t h e effective s t r a i n e ~ a/3 G in the elastic range are expressed as ~ = 2 I/~G {z~ A~ + 2 ~ Ao A~ cos (~o~   %) + A~}'/~
= (9/t/3) (Z 2 m~ u ~ ~ A0 z<~l cos (~o1  
00)
~/)o) ~ 412}l/2, /
I n these relations, G denotes the modulus of rigidity, and # is related to the P o i s s o n ' s ratio v b y the relation # = tan a {~/3(1   v)/(1 @ r ) } . (11) a) R a n g e of E l a s t o  p l a s t i c B e n d i n g (a ~< r ~< rb). According to the ~ I i s e s ~ yield condition = as (a, is yield stress of the m a t e r i a l under uniaxial tension), the distances from the middle plane to the elastoplastic interfaces in the meridian section Z* h and Zs h (Z1 ~ Z2) are given by (10) as the roots of the q u a d r a t i c equations: z~ + 2 z ~ cos (w~   ,po) + ~ = (~,/2 V ~ c h Ao)~,
whel'e
a = A1/h A o 9
02)
03) 26
342
IngenieurArchiv
(14)
(]5)
Assuming the incompressibility (v = 1/2,/~ :~/6) of the material in the plastic state, the relations (8) and (9), and the second of (10) turn to the following:
s0 /
Assuming the continuity of ~ at the elastoplastic interface, the next relation holds on the plastic side of the interface: ~. = 2 / ~ o h A'0 { # + 2 z ~, cos (~o~ ~Oo) + ~}~/', where (19) .. = A;/h A ; . Then A~, A~ in (16), (17) are obtained as A; = %/2 (,3G h F ~ ) ,
where
(20) (21)
Representations (8), (9) and (16), (17) give the following relationships between W0, 10s a and tan V2o~ V'3tan~ tano9 o , tan~ft  ]/3 tan/~ tanco 1 , = a, sin (V'0 /z) sin (o9t  ~r/6)/sin (Ft  #) sin ({oo  ~z/6). (22)
The strain energy W~ in the elastic layer consists of the energy due to dilatation as well as distorsion, and is expressed as follows:
z~h
(23)
where (r denotes mean stress. In the plastic layer, however, as the energy due to dilatation remains constant, and neglecting strain hardening of the material, the strain energy We can be expressed as follows under the assumption of H e n c k y ' s stressstrain relation:
z~h  h z1h h
We = f ((r, ~ @ const.) dz @ f ((Ys ~ @ const.) d z . Then, T~, To and Mr, Mo are given by the relations
aW T. = Oe~ ' ~W To   o ~ '
(24)
(25)
OW Mr. = ~ ,
~W Mo  gs ,
(26)
where W denotes the sum of W~ and We. Substitution of (23), (24) into (25), (26) yields
@ ~)
@ 20~(~1k
' 2 ) c o s ( ~)1 @ 6 ) }
+,

{,(.
+ (o.o+
,,(oo _+6)
I
,...,.(..
F~(Z,) @ F(ZI) E (  ] )
",')+
(2.)
a~ {F(1) F(X,)
xxxv. Band 1967 Y. Ohashi et al: Elastoplastic Bending of an Annular Plate at Large Deflection
343
~
0r 1




x In
E(Zl) + F(Zl)
E(~) + FO)
(28)
(29)
where
E(Z) = Z + al cos (col  (~
The radial displacement, the deflection and the strain components can be calculated by the relations
U
(30) (31)
dw _ dr eo
e,
= T
(32)
which are derived from (3), (4), (8), (9) and (14). Substituting (27), (28), (31) into (1), (2), and (8), (9), (14) into (5), (6) together with (13), (22), four simultaneous nonlinear ordinary differential equations with respect to Z1, al, COo,COl are obtained as the fundamental equations for this range 1. The details of these equations are, however, omitted here for brevities sake. The relations for the range ra < r < r b are obtained b y putting Z2 =  1 in the above relations. b) R a n g e of P u r e l y E l a s t i c B e n d i n g (r6 < r ~ b). Under the assumption of M i s e s ' yield condition, the bending state changes from purely elastic to elastoplastic when the effective stress on the surface of plate (~)h attains to as. Then we introduce a new variable s (0 ~< s <~ 1) which is given by the relation (~)t, ~ s ~,. Thus, in place of (14), A 0 and A 1 are expressed as follows : A 0 = s a,/2 where
~ ( 1 ) = {1 + 2 ~ cos ( ~ ~o) + ~ ' } ' / ' 9
I/3G h F ( 1 ) ,
A~ ~ s a~ a12 ~ 3 G F ( 1 ) , ]
(33)
Wo = f {3
h
dz,
(34)
which, together with (25) and (26), yields T,, To, M~ and M o as follows:
To
,
.
(35)
(36)
Mo
Relations (3), (4), (8), (9) and (33) furnish the radial displacement, the deflection and the strain components : sa, ar sin(y~+#) (37)
2V3c F(1)
dw _
d~
so
(38) (39)
sT } = T
By these relations, a system of four simultaneous nonlinear ordinary differential equations with respect to s, a, % and ~f~ is obtained as the fundamental equations. The details of these equations are again omitted. On the calculation, the independent variable r is transformed into R ~ rlh for convenience. 26*
344
IngenieurArchiv
3. Numerical s and its Results. As an example of numerical calculation of the aforementioned fundamental relations, analysis is performed for the large deflection of a simply supported annular plate with free inner edge under a lateral load P (total value) uniformly distributed along it. For numerical calculation, geometry of the annular plate and material constants have to be specified. I n this case, annular plates of b/h = 40 and a/b = 0.25, 0.50 and 0.75 are selected. Values of as, G and u are determined as 30.1 kg/mm 2, 8160 kg/mm ~ and 0.28, respectively, which were obtained for a mild steel (0.13~ C) which will be used for the corresponding experimental examination. To integrate the fundamental equations mentioned above, the values of Z1, al, COo and 091 at inner edge R = A ( = a/h), (Z1)a, (~l)a, (co0)a and (col)a and P must be given. I n t h e present calculation, (Z1)a is selected as a parameter, and the calculation is performed for seven values of (Z1)a, i.e., Zla = 1.0, 0.8, 0.6, 0.4, 0.2, 0 and  0.2. Then, the values of (al)a, (co0)a, (col)~ and P corresponding to the given Z1, must be specified. A n y two values of (al),, (CO0), and (COl) can be determined by the boundary conditions T, = M, = 0 at inner edge, and the rest and the magnitude of P can be determined b y the conditions T, = M, = 0 at outer edge. Therefore, (a~), and P of these four values are chosen as the values to be determined with outer edge conditions. Assuming the value of (al)a corresponding to the given values of ZI,, values of (CO0)a and (CO~)a are obtained as roots of simultaneous transcendental equations T, = 0 and Mr = 0 given by (27) and (28). These equations are solved numerically b y the NewtonRaphson 5 method with two variables. With these boundary values, the fundamental differential equations are integrated numerically towards outer edge, where the Predictorcorrector '5 method and the RungeKuttaGill 5 method were used. Then, the values of Z1, al, coo and cox at each value of R can be obtained, and T~, To, Mr, Mo, u, w, e~ and e0 are calculated by the relations (27) to (32). On the way of calculation, Z2 becomes  1 at R = R, ( = r,/h, Fig. 1). At R = R b ( = %/h, Fig. 1) Z~ becomes 1, and accordingly s also becomes 1. Then R ~ R~ corresponds to the boundary of elastoplastie bending range and pure elastic one, and the fundamental equations are replaced by those for pure elastic range there. At R = B, corresponding to the outer edge, the boundary conditions T, = M~ = 0 are satisfied only when (a~)~ and P assumed at the inner edge are chosen properly. The values of (~)a and P, therefore, must be determined b y double trial method to meet these conditions.
,Z 1,o
o,8~
O,6 0,~
~ 
O2
I
0,25
a/~aso
Examples of the results thus obtained are shown in Figs. 2 to 6. Fig. 2 shows the plastic regions corresponding to various values of Zla in a meridian section. Though the upper elastoplastic interface at inner edge proceeds towards middle surface while Zla > O, it returns thereafter, and an unloading process occurs there. Therefore, strictly speaking, the present analysis cannot be applied to the case Zla < 0. However, as the region where an unloading occured is quite small for Zl~ =  0.2, the results corresponding to it can be accepted with sufficient accuracy, As shown 5 s. f. i. in A. Ralston and tl. S. Wilt', Mathematical Methods for Digital Computers, New" York, 1960, pp. 95: 110, 240.
XXXV.Band 1967 Y. Ohashi et al: Elastoplastic Bending of an Annular Plate at Large Deflection
345
Y~/~
r
g
l 84
\
.. \ ",q , /I /~ ..o,8 I /.a~ t a9,8
~o8
16
I '
5
T~/~
0
"% /~
/ / /'/az i
~',o,2    
~57
213~
kg ~n 43
~
8 76
4~
O4
0,6
0,7
0,8
O,8
I,o
3g ,
86
O,?
0,8
Q9
1,0
Me/M
32 ~ 
, M;/,V ~'~ 2 g, mm
kg/ ~m~
~a / o
185
i!
~ ]
I /
~   , l
28 Z4 2O
o,8
16
12 0,8 L _ 89,& /
9 ~5
O,6
O,7
a8
0 0,9
Zo
0,5
0,6
0,7
0,8
O,S
1,o
in Figs. 3 to 6, the radial components of membrane force and bending moment are quite small compared with the circumferential components. The ratios of the maximum values of T~ and Mr to those of T o and M0 are approximately 1/6 and 1/20, respectively. These trends show that the circumferential component of stress is essential for the plastic deformation of the present annular plate.
4. Experimental Examination and its Results. To examine the validity of application of the above mentioned procedure to the present problem of annular plate, and to determine the extent to which it can be applied with satisfactory accuracy, the experiment corresponding to the prcceeding analysis was performed with mild steel specimens. The material constants of it were already described above. Dimensions of the specimens and the apparatus used are shown in Fig. 7. The lateral load P was applied by the testing machine of Amsler type. The deflection was determined by measuring displacement of the upper surface (the surface which contacts with compressing block, Fig. 7) by using dial indicators. The components of strain were measured by means of
346
Ingenieurhrchiv
electric resistance strain gauges of 10 ram length and 2.8 m m width attached radially and circumferentially to both surfaces of the specimen. As the strain distribution on the surface in plastic state is highly inhomogeneous because of the formation of Lftders' hands, the experimental values are determined by measureing the strain components at two positions of equal radius situated symmetrically with respect to the center of the plate.
b/ock Speoirne n
~5o I 2oo I
~"
s 1 6 2 2b=400r 402r
sfrain gage
The propagation of plastic region in the specimen was also examined. An annular plate which has been deformed up to a certain load was cut at several sections perpendicular to radius and the strain figures are observed by etching these sections. The average depth of the tips of strain figures in each section was measured by means of magnifying projector. Figs. 8 to 15 show the results of the experiment together with the corresponding results of calculations. I n these figures, full lines represent the results of present calculation and dashed lines the results of calculation without membrane force 8. The small circles show the results of the experiment. As shown in Figs. 8, 9 and 10, the extents to which the theory neglecting the effect of membrane force can be applied within the accuracy of 10 percent are 70 percent (a/b ~ 0.25), 75 percent (a/b = 0.50) and 85 percent (a/b ~  0 . 7 5 ) , respectively, of load carrying capacity for annular plates of b/h = 40. Though the experimental results deviate gradually from the theoretical curves taking account of membrane force after the incipience of plastic deformation, in the mean
,W
" / 6  mm
14
a/b=o,25
/2
10 8 6
/ /
/ ,Ij
9 ~/b=o,2~l i
'Oz~O / zI
~SO~ rxperzmenTaqI
, J
I
/
~80J!
/_.r
2
.1oa kg
while, they again begin to approach the theoretical ones and run parallel with the extension of theoretical curves. The most significant case of such trend is observed in the case of a/b ~ 0.75. The maximum differences between the experimental values and the full lines are about 10 percent, 12 percent and 45 percent for a/b ~ 0.25, 0.50 and 0.75, respectively. I n the case of circular plate investigated previously, on the other hand, the theoretical and experimental results coincided almost completely for the whole range of loading.
XXXV.Band 1967 Y. Ohashi et al : Elasfoplastic Bending of art Annular Plate at Large Deflection
347
16 84 mm
12
I0
P mg~g 7 "
W
mm
Ti
7heorefica/wi~h mombrone force '~Theore/icaL withou/ membrane force ' ~ fxperimen/e/l~~
I
i 
14 12 10 8 
_ _ 2 _ _
o
6 [
4
I
~ i /
"
I il J
! ~r~
2
I 2
~~"~~~+r176
~ .
~ 3
'p
Figs. 11 to 14 show strain components of the annular plate of a/b ~ 0.50. While the full lines in these figures increase rapidly after the incipience of plastic deformation, the corresponding experimental values increase almost linearly for a while and at the load around load carrying capacity they begin to increase more intensively and to approach the full lines. I n other words, whenever the increase of theoretical curve is nearly linear, the experimental values coincide well with the corresponding theoretical values, but when the gradient of theoretical curve increase rapidly, a transient delay of the corresponding experimental values becomes significant. Comparing Figs. 11 and 12 with Figs. 13 and 14, absolute values of (er)h and (e,)_h are about 1/3 of those of (e0)h and (s0)t,. The differences between the experimental and theoretical values in (er)h and (r are much larger than those in (e0)h and (e0)h, which are almost the same as those in u,. Fig. 15 shows the propagation of plastic region observed by the method mentioned above (in the plate of a/b = 0.25 loaded up to P = 4680 kg, i.e., Zta ~  0.2) and the theoretical elastoplastic interfaces as well. Though the depth of strain figure excellently coincides with the corresponding theoretical elastoplastic interface in the region where [ZI ~ 0.4, the former is much shallower than the latter elsewhere.
348
Y . O h a s h i e t a l : E l a s t o p l a s t i c B e n d i n g of a n A n n u l a r P l a t e a t L a r g e D e f l e c t i o n
IngenieurArehiv
fer)/s .1o~
Z8 2,4
7II
BI
I~
lI
I ii
!l
~V
t [ /i
i I
/ [
J ]
2,0  
. [ n..~E
o~5
o
7;,2 02 0,~ 
i
~
I l ]}:
!/,/i'!/
9 ~
I
,~++.
.r
bY
c o
+ ~" I .
[
...47_/~~&.D~
I I Ifo;

Theoretico/.withmembruneforce
I I /II~l
,.
lZ
9 TIb=~551~i + 065 I
"~75~ E
o
~eo.~
~,l~0a,m.~br~
.
//
//
,, ! ii ~
_~,
"
' o,35J
II~ ~
~ +
i
1
I~ I~
I ~+."
.ioUkg 7 "
# f%)/s
a .1o3 I! I 
,,
a/~<,o
/
III illlt~
',!i!l~
l ~t~t r I
tjlll

It
IiIil~
"
"
+
+

+
x
o'.95j !
l
i
I
I liYl
/i / ol / I'!Jl I#1 l
/Vil I 'fill(
llt
V" '
.
+
'
Z
.....
<+"1 eb ~_
. I
II
I
_.
"
 
~l~~!~
.~.",,,t.,,+
./,,i/i+
, ~1
11
!14"
5
F i g . 13, R e l a t i o n b e t w e e n s t r a i n e 0 on t h e surface z = h a~td t o t a l load P
10% 7"
(aib = 0.50).
XXXV.Band 1967
349
5

..... 9 flU=O,,551
'1 I,
'
!/i'lilil ii
 i
~ + :7
,I
70 k~ 7
F i g . 14. R e l a t i o n b e t w e e n s t r a i ~ e 0 o n t h e s u r f a c e z =   h a n d t o t a l l o a d
P (a/b
= 0.50).
5. Discussion. As shown in the proceeding figures, some discrepancies between the theoretical and the corresponding experimental results are observed, especially in relatively early stage of plastic deformation. In general, the cause of the discrepancy between the theoretical and experimental results in the problems of elastoplastic bending of thin plates m a y be divided into the following factors: errors of the fundamental equation used due to large deflection, the difference between the boundary conditions postulated in the analysis and those realized in the experiment and the factor due to the property arising from the microstructures of material which is difficult to formulate from a macroscopic view point. In view of the fact that both results coincide fairly well in the purely elastic bending stage, cause of discrepancies in the elastoplastic bending stage cannot be attributed to the former two factors. The last factor, on the other hand, relates to the validity of the macroscopic formulation postulated in the analysis, i.e., assumptions of the homogeneous and isotropic elastoperfectlyplastic body following the Mises' yield condition and the Hencky's stressstrain relation. The mild steel used in the present experiment reveals a well defined upper yield point and has a range without strain hardening up to ~ = 20 10 3 under uniaxial tension. Since the maximum value of measured in the experiment was less than 8 X 10 8, it turned out t h a t the experiment was performed within a elasticperfectlyplastic range of the material. The Mises' yield condition and the Hencky's stressstrain relation postulated in the present analysis were also employed in the previous analysis of circular plates. The part of the Mises' yield curve relevant to the present analysis is involved in those of the previous problems in which excellent coincidences between the theoretical and experimental results were ascertained. Moreover, considering the above mentioned fact t h a t the deviations tend to diminish after relatively large deformation, the discrepancies cannot be referred to these causes. I n general, real material is far from homogeneous and isotropic on microscopic view point. After all, the interaction of these inhomogeneity with those of stress distribution m a y be accounted as the last cause. The upper yield point of the present material was about 7 percent higher than the corresponding lower yield point even under uniform stress state. This phenomenon may be interpreted as a kind of delay of yielding. This will be more significant under inhomogeneons stress state due to a restriction of the surrounding parts of lower stress level. To verify this assumption, it seems convenient to note the propagation of strain figures in the plate. I n view of Fig. ] 5, the depth of strain figures and the corresponding elastoplastic interface agree well with each other when the depth of the latter attains to more than about 0.6 h, but the former cannot attain to the latter as long as the depth of the latter is less than that. Moreover, the former does not appear when the depth of the latter is not so large. The above mentioned observation, t h a t is, the trend t h a t strain figure does not appear continuously following the postulated yield condition, is also tO be explained by the fact that Liiders' bands on the surface propagate intermittently with ~ncreasing load. Consequently, the influence of the delay during the incipient stage of plastic deformation will give a considerable discrepancy to the result of calculation obtained by assuming a continuous yielding.
350
IngenleurArehiv
As already described, the larger the inhomogeneity of stress distribution, the more significant the delay of this kind. Then, it may be easily understood that, while the present procedure furnished almost complete coincidence in the case of circular plate, the discrepansy is more significant in the annular plate in which the value of a/b is large, that is, when the gradient of stress distribution is large. Though the inhomogeneity of plastic deformation must be taken into account to improve the discrepancies observed in the present investigation, it may be difficult to formulate mathematically these microscopic phenomena. Moreover, since the discrepancies are not so large and are observed only transiently in reality, the procedures used in the present calculation may be sufficiently valid for practical investigation of the large deflection of annular plate.
/,0' Z
o
88
~
o,4 0,6 0,8 /,0
~8 ~
o ~
o,5o
o,75
h= 50,Omrn P=4680 kg Z;a = 0,2
1.o
Fig. 15. Comparison of theoretical elastoplastic interface and the corresponding experimental result.
6. Conclusion. The numerical results were obtained for simply supported annular plate with free inner edge subjected to concentrated load along the inner edge, and the corresponding experiment was also performed to examine the validity of the analytical procedure. The theoretical results were also compared with the numerical results of the theory without membrane force. In the course of experiment, some discrepancies were observed between the theoretical and the corresponding experimental results. I t turned out that the discrepancies can be ascribed, not to the yield condition or stressstrain relation used, but to the complexity of the transient phenomena which accompany to yielding. 1Nevertheless, since the discrepancies are not so large in usual and are restricted only to the incipient stage of yielding, it may be said that the analytical procedure employed here will give a satisfactory landmark for the investigation concerning the large deflection of annular plates. (Eingegangen am25. Mai 1966) Anschriften der Verfasser:
Dr. Y. Ohashi, Professor, Dr. S. Murakami, Assistant Professor, and A. Endo, Graduate Student, Department of Applied Physics, Faculty of Engineering, Y~agoyaUniversity, Chiknsaku, Yqagoya, Japan.