thermal deformations of a sphere by radiation from a point source
TRANSCRIPT
THERMAL DEFORMATIONS OF A SPHERE BYRADIATION FROM A POINT SOURCE
ICY HARINDFR SINGH * AND A. SINGH **
Received October 8, 1971
(Communicated by Dr. B. R. Seth, F.A.SC.)
ABSTRACT
In this paper the temperature and thermal stresses set up in an elasticsphere due to a point source at a finite distance from the centre of thesphere and outside it, have been studied. The final results are obtainedin terms of series involving Legendre polynomials. The nature of the stressand temperature distributions have been studied for two different distancesof the point source from the centre of the sphere Numerical calculationsdone on an IBM 1620 Computer have been represented in graphs.
NOTATION
_( .
where x-, y-, z- form cartesian co ordiwate system withorigin at the centre of the sphere.
D — the displacement vector.
T — the excess of temperature over that of zero stress andstrain.
A, µ — Lame's constants.= A/2 (A + µ) Poisson's ratio.
a — coefficient of linear expansion.
8 =2(1 + v)a.
a — radius of the sphere.
* Lecturer, Mathematics Department, G. N. University, Amritsar (India).
** Reader, Mathematics Department, Punjabi University, Patia'ia (India).
A2 247
248 HARINDER SINGH AND A. SINGH
d — the distance of the point source from the centre of thesphere.
= a; d.
1. INTRODUCTION
THIS paper deals with the steady-state thermal stresses which arise in anelastic sphere by radiation from a point source at a distance d (a < d < co)from the centre of the sphere. It is also assumed that the sphere is losingheat to the surroundings by radiation from all over its surface. The surfaceof the sphere is taken to be stress-free.
The particular integral used for solving the thermoelastic equilibriumequations is same as used by J. T. Holden' in his paper where the sourcehas been taken at infinity. The residual surface stresses are removed byusing the method introduced by E. Sternberg et. al. 2
The numerical calculations have been done with the help of IBM 1620Computer by taking the first thirty terms. It has been found that if d islarge as compared to the radius of the sphere the results approach those ofRef. I for the temperature. However, there is no such agreement in thecase of stresses because of some mistakes or misprints in Ref. I. Thefirst mistake is that the second series on right-hand-side for aPq [eqs. (4.3)Ref. 1] should start from 2, not from 3. The second is that in the equationfor v 1 [eq. (5.1 , Ref. 1] the expression n 2 + (1 + 4v) n + 1 + v shouldbe replaced by n 2 + (1 + 2 v) n + 1 + v. After these misprints havebeen rectified there is complete agreement in our results and thoseof Ref. 1 in the limiting case d -^ oo.
2. STATEMENT OF THE PROBLEM
With the origin at the centre of the sphere we take the z-axis in the
direction of the source and k as unit vector in that direction. To ensuresymmetry with respect to z-axis, spherical polar coordinates (r, 0, ¢) havebeen used for the description of the problem.
In the steady state the heat conduction equation and the thermoelasticequilibrium equations are
V 2 T =0 (2.1)
O(O.D)+(i — 2v) V 2 D— VT=0. (2.2)
Thermal Deformations of a Sphere by Radiation 249
The temperature boundary conditions are
—K jr =— Ecos^ll+hT 0 <0 <arccosdo
= hT arc cos d° G B < 7r (2.3)
at r = a. K denotes the thermal conductivity of the sphere; h denotesthe heat transfer coefficient of the surface; E denotes the energy absorbed
per unit area per unit time by an element of surface held normal to k, andis the angle between the incident ray and the outward normal to the
surface of the sphere.
We now have
00— cosO— d° f ( s_ s+ 2
cos — l — 2d° cos 8 4 d° 2 L 2s — 1 l̂0 2s } 3 d0 Js=o
00
>: PS (cos B) _2J lsPs (cos B) (say). (2.4)3=U
Further the elastic boundary conditions for a stress -free surface are
Urr = are = 0 (2.5)
at r = a.
While solving the problem it is convenient to use the non-dimensionalquantities
KT ,p = r/a; ®_ , k = ha/K; u - -D/a and c' = z/a. (2.6)
3. TEMPERATURE DISTRIBUTION
Using the non-dimensional quantities as introduced in (2.6), theboundary conditions (2.3) take the form at p = 1
—— =—cos0+k® 0 <6 <arccosd°^p
= k® arc cos d° < B < i. (3,,1)
250 HARINDER SINGH AND A SINGH
To combine -these two conditions into one we introduce a function definedas
f (cos 0) = cosh 0 < 0 < arc cos d0
=0 arc cos d„<0<7r.
Developing the function ,f (cos 0) as a series involving. Legendre poly-.nomials, we have
f (cos 0) -_ E bnP,, (cos 0)
where
bn = I (2n + 1) f f (cos 8) P (cos B) d (cos 0)
3 (2n +1) f cos q, Pn (cos 6) d (cos 0)d^
00 1
(2n + 1) ? is f PS (cos 0) Pn (cos 0) d (cos 0).$=0 du
Making use of the results for integrals involving Legendre polynomiais 3
we get
bn = In [1 — 2d0 (P1 2 + P22 + ... + Pn__1 2) — d0Pr ' 2
+ 2 (P1P2 + P2PA + ... + Pn-iPn))12
Co
+ 2n 2 1 ^1 ^s sPs_4Pn (s
nP nP (S (s+ -
n) dOPSPnL ).—Oxn
(3.2)n = 0, 1, 2, 3, 4, 5, .. .
where in this equation P S , Pn , etc., are functions of d0 .
The temperature boundary condition (3. 1), therefore, becomes00
—_— bnPn(cos0)+kO 0 <7r (3.3)Jp.«u
at p = 1.
Thermal Deformations of a Sphere by Radiation 251
The heat equation satisfied by © has the solution
e = E anPn Pn (cos 0)n-o
where an 's are constants.
Substituting this in the boundary condition (3.3) we get
= bn l (n + k) n = 0, 1, 2, 3, 4, .. .
When do = 0, the as are same as those in (1). When n is finite the
sum "n-1
+ `ith term in (3.2) is finite and therefore the behaviour of seriess=o
for bn is same as the behaviour of 2 • When n is large the Factors=n}1
In =in n 1 don i — 2nn + 2 -{- 3 don+i (3.4)
is sufficiently small, as 0 < do < 1, and tends to zero as n --* oo. So inthis case we can take
001 sPs_iPn — 1PA-, — (s — n) doPSPn
bn = 2 (2n + 1) Z lg — (s — n) (s -F- n =^- 1) --8=U
Thus it is sufficient to study the series
` sPs_1Pn- -- nPsPn--t — (s — a) doPsP^, (3.5sn+ )2(21) (s-n)(s-i- n-I-1)
a=u
for the investigation of the convergence of the fog b,1 a.id ti;c^: ^, e or
a n .
Using the value of is from (2.4) the general term of (3.5) can be broken
up into six components:
s2 dos-' Ps-,Pn — s (s + 2) dos r`
(2s-1)(s—n)(s { a+1)' ( +3)(s—n)(s +n + 1)
252 HARINDER SINGH AND A. SINGH
nsd s-1 PsPn. - 1 n (s + 2) dos+1PsPn_1,— (2s— 1)(s—n)(s+n+l)' (2s+3)(s=n) (s+n+1)
sdo' PsPn (s -i- 2\,1S+2 Pspn(2s-1)(-1)' s+n+l)'
Now consider the series
1 (2n + 1) s2 dos-' Ps-iPn
2 ^j (2s — 1) (s — n) (s + n + l )a><o
00
(2n + l) !!z s2 dos 'P5- i2 (3.6)2n jea (2s — 1) ^n — 1 ) \
S n 1 + 1
Comparing (3.6) with the series with the general term PS = s-413 we getfor large s and all n
s2 do s p-1
s-1
Ids (2 — 1) (n — 1' ( s n 1 + 1
S2 s4/3 dos-1 PS-1
<1,
(2s-1)`n -1J(s n +1 )
as do < 1 and J Ps (cos 0) < '2%(s sisin B) for s large. Since the series
his converges absolutely so does the seriesa—U
00
s2 doS-1 Ps-1
a=e (2s — 1) U — O n 1 + 1 ^
Similarly we can prove that the other five component series convergeabsolutely, and their sums are proportional to either (2n + 1)/(2n 2) P. or(2n + l)/(2n) Pn- i . Both these factors are finite for finite values of n andtend to zero as n —. oo. Hence the series (3.2) for b„ is absolutelyconvergent.
Thermal Deformations of a Sphere by Radiation 253
Now consider the series
00©_ E anp Pn (cos B) (3.7)
n-0
with an = bnl(n + k). Since b,,, are all finite we can find a positive numberM such that I bn I < M and therefore
E anP Pn (cos 0) 1 <M E I Pn Pry (cos 8)/(n + k) 1. (3.8)s=0 n=0
Again comparing the series (3.8) with the series n w-- get for large nn=o
n413 n I n" 3 2Pn cos B)n + ( < I n { 1c Pn Vyr;r sin d l <1.
Hence the series (3.7) is absolutely convergent.
4. STRESS DISTRIBUTION
Writing (2.2) in the non-dimensional form w:. get
V(V .u)+(1 —2v) 0u— OT=O (4.1)
where
V = a p.
co
T — p ka ® _ 1nPn Pn (cos B) (4.2)w=e
^n = $Eaan /k. (4.3)
The usual procedure to find the solution of (4. 1) is to put
u = V0 4 This method leads to intricate calculatio.is. A simplermethod developed in Ref. 5 will be used here.
254 HARINDER Sm GH AND A. SINGH
We seek a solution of the form
u=ck
where V = 0, whieb gives
(4.4)
Now T is given by (4.2). Putting
00_^nPn+iPn+i (cos 0)
no
we get ti = çtk to be the solution of (4.2). The stresses corresponding tothis solution are
00
app =f2n2n 1 fn-2Pn-2 Pn (COS'I22
00
1 PnB— 2n 1 ^nP" (cos( 0)
no
100
_I^g^ Ln + 3 pnpn Pn (cos 0)
n-e
00
2n
— 2n — L In-2Pn 2 Pn (cb5'8)• n-2
00
Q^¢ _ — nPn Pn (coS 0)n=o
00
ap® —= — —12 2n_2 Pn ( ) (cos F)2n
ww=
Thermal Deformations of a Sphere by Radiation 255
Co
2: Bp_ P(l) (cos 6)(n+1)(2 ^ + ,^)
Q#f =&.p 0
where
P(i) (cos B) _ (sin d) ! OS 8) Pn (cos 0); app = app/1i., etc.
To make the surface, stress-free, as required by the boundary condi-tions (2.5), we make use of the Boussinesq solutions
2u=77; 17 2 4=0. (4.5)
2u = 0 (z'O) — 4 (1 — v) Ok; '71 0 = 0. (4.6)
We consider the harmonic functions
0 = pn P. (cos B) and 0 = pn Pn (Cos 0)
and denote the resulting solutions (4.5) and (4.6) by [A n] and [Ca] respec-tively. For convenience we construct the solution
[Bn] _ — (2n + 1) [Cn] + (n — 3 + 4v) [A 1 ].
itTt is found that the surface is stress-free under the solution
-^ -,U = u + [S]
where
[S] — 2 un [An] + IE vn [Bn]R-2 n=1
— $Ea (an,an_2
un— k L 2 (n 2 -̂- n+2nv+v+1) -- (2n-1)(n-1)J' n> 2 .
2)12+1k '2(n2 ±n+2n ±v+1)(2n-I-3)(n+1)'n-0.
256 HARINDER SINGH AND A. SINGH
The stresses at any point of the sphere are given by the relations
E 00 00
Qpp — 2n^n 1 Pn-2 Pn (cos 8) — 2^ + 3 Pn (cos 0)
s
n=o
00
+ E n (n — 1) unp"i-2 Pn (cos 8)n=2
00
—E n (n2 — 3n — 2v) v np'^- ' Pn_1 (cos 8) (4.8)
n00
BBB E 2f + 3 Pn (COS 8) 2n$n-1 Pn-2 Pn (Cos 8)
N0 w=s
00
+ 2 upn-2 [P', (cos 8) — n (n — 1) Pn (cos 0)]n=2
00
+ vnPn-1 [n (n2 + 3n + 3 — 2v) Pn_1(cos 0)way
— (n+4 -4v)P'n(cos 8)] (4.9)
Q^ f = — 2+' fnPn Pn (cos 0) — unPn-2 P'n-^. (cos 0)n=o a=2
00
+ 2' vnPn-1 [(n +4- 4 v) P'n(cos8)
— n (2n + 1) (1 — 2 v) P_1 (cos B)] (4.10)
00
= 24-2 n-2 P {z> (cos 8)pI — 2n-1P nR=:
00
Snpn1 )(2n+3) Pn ^l^ (cos B)(n+
n=1
00
[(n — 1) unpn-2 Pn( l ) (cos 0)n=f
Thermal Deformations of a Sphere by Radiation 257
2 +2v) Vpfl-1 P(i) n-1 (Cos B)] (4.11)
og^,=acbp =0
where
P' (cos B) = dd O) pn (cos B).
5. NUMERICAL EXAMPLE
The value of h lies between 1.8 x 10-4 and 3.3 x 10-4 cal. per sec.cmz C°6 and conductivity K is found to lie between 0.001 and I .00 cal.
o•
qd 9e°
go•
FiG. 1-a
per cm.sec. cm2 for metals and stone.' If a is measured in centimetersthen the range of k is found to be
1.8 x 10-4 a <k <3.3 x 10-' a.
258 HARINDER SINGH AND A SINGH
b
0
O
90
1P
i9 0 *
FIG. 1 -b
,. 1
12 /i
e- .)
for., 2•,ft r4vair-.b.
FIG. 2-a FIG. 2-b
The temperature distribution given by (3.7) and the stresses given bythe relations (4.8) to (4.11) have been computed for k = 0.1, 0.25.do = 0.01 and 0.1 on the surface of the sphere and for p = 0•5.
•014
0 12
'008
•004
0
I ux0-
-004
-•008
Thermal Deformations of a Sphere by Radiation 259
The results have been represented in graphs. The isotherms have alsobeen drawn in cross-section through z-axis for d„ = 001 and d, ) = 0.1.The calculations were done with the help of a hand calculator and an IBM1620 Computer at Department of Mathematics, Punjab University,Chandigarh.
FIG. 3
6. DISCUSSION OF RESULTS
As the source comes nearer the surface of the sphere the non-dimensionaltemperature O at any point of the sphere decreases as is clear fromFigs. 2 a and 2 h, which is due to the fact that the area exposed to the radia-tion decreases as the source approaches the sphere. However, the actualtemperature T may not have this behaviour because as the source is removedto a large distance from the sphere, the amount of radiation falling on anelement of area will decrease in the inverse square proportion to its distance
260 HARINDER SINGH AND A. SINGH
from the sphere. Comparing Figs. 1 a and 1 b it becomes clear that thecurvature of isotherms is greater at d =0l than at d 0 = 0.01. FromFigs. 2 a and 2 b, it is clear that the temperature is maximum for B = 0 andminimum for B = i•. The variation of temperature is more pronouncedfor 0 < 0 < ir/2 than for ir/2 <0 < ,r, this variation being similar fordifferent values of d.
--*I*
-•as
-•014
Je<b^
- ago
d=g jo
5'c at
5'
-'010
-°l^, nla lx Z73 S"/b
n
a -- b
Fiu. 4
The curves for the stress components tend to be symmetrical about0 = 7T/2 for small values of d0 . For do = 0 they are symmetrical about0 = ir/2 as in Ref. 1 (shown in figures by dotted lines). However, this sym-metry is not observed when the source is near the surface of the sphere.The asymmetry is more pronounced in case of stress component kaosj(fEa)at p = 1 (Fig. 4). The asymmetry is due to the fact that when d 0 is large,
•012
'008
. 004
o
<(^Q°
Ir 004
-'008
Thermal Deformations of a Sphere by Radiation 261
the terms in the series (4.8) to (4. 11) which change sign at 0 = IT/2 havecoefficients of significant magnitude. These coefficients decrease numericallyas do decreases and they vanish all together as do --0.
The stress component kaaP/(iEa), at p = 0 5 is maximum for 0 = 0and r, and is minimum for 0 = 7r/2 (Fig. 3). The stress component ka.9 1($Ea) is maximum for 0= 2Tr/3, minimum for 0 = ir/3 and vanishes at0 = ir/2 for small do (Fig. 5).
FIG. 5
The stress component ko99/(f Ea) is maximum for 0 =r and is minimumfor 0 = Tr/2 on the surface of the sphere (Fig. 4). For d„ = 0 maximumis attained both at 0 = 0 and n.
262 14ARINDER S1NGH AND A. SING I
REFERENCES
1. Holden, J. T. .. Quart. J. Mech. and App!. Math., 1962, 15.
2. Sternberg, E., Eubanks, R. A. Proc. of 1st U.S. Nat. Cong. of App!. Mech., 1951.and Sadowsky, M. A.
3. Copson, E. T. .. Theory of Functions of a Complex Variable.
4. Goodier, J. N. .. Phil. Mag., 1937, 23.
5. Fox, N. .. Proc. of Lon. Math. Soc., 1961, 11.
6. Bosworth, R. C. L. .. Heat Transfer Phenomena, Ass. Gen. Pub!. Phy. Ltd
7. McAdams, W. H. .. Heat Transmission, McGraw-Hill.