International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) Computational Modeling of Heating and Cooling Effect of a Thin Annular Disc Sunita S. Patil*1 and J.S.V.R.Krishna Prasad*2 *1 SSBT’s College of Engineering & Technology, Bambhori, Jalgaon (M.S.), India *2 Abstract M.J. college, North Maharashtra University, Jalgaon (M.S.), India - The paper is concerned with the inverse unsteady-state problem of determining the 2. A THIN ANNULAR DISC IN THE PLANE STATE OF STRESS: Consider a thin annular isotropic disc of temperature (in heating and cooling process). Homogeneous boundary conditions of the third kind thickness h occupying the space D : a h. are maintained on curved surfaces of the disc. The finite Marchi-Zgrablich and Laplace r b, 0 z The differential equation governing the displacement function U(r, z, t) is transform techniques are used to find the solutions of the inverse transient themoelastic problems of a thin 2 U r2 annular disc. Keywords Thermoelasticity, Annular Disc, Thermal Stresses. problem of determining the temperature (in heating cooling process), displacement and functions of the disc occupying the space D: a 0 z )atT and at are Poisson's ratio and the linear coefficient of thermal expansion of the material of the The inverse unsteady state thermoelastic and (1 (2.1) where Introduction - 1 U r r disc respectively, and T(r, z, t) is the temperature of the disc satisfying the differential equation stress r 2 T r2 b, h with the stated boundary conditions. 2 1 T r r T z2 1 T , k t (2.2) The inverse problem is very important in view of its relevance to various industrial machines subjected to heating such as main shaft of the lathe and turbine and roll of a rolling mill. subject to the initial condition T(r,z,0) = 0 (2.3) the boundary conditions A related problem of determining the temperature, displacement and stress functions due to partially distributed heat supply at z = a thin annular disc [2] (0 < T r , z, t < h) in is reconsidered to study the T r , z, t temperature (in heating and cooling process). Discussion on some similar problems may be found in Ozisik [3] , Sneddon k2 r r T (r , z , t ) 0 (2.4) r a T (r , z, t ) 0 r b (2.5) [8] , Hetnarski and Eslami [9]. The corresponding correct expressions are derived in the present paper. k1 Also, the numerical z T (r , z, t ) 0 z 0 (2.6) and the interior condition results are obtained and presented graphically. ISSN: 2231-5381 http://www.ijettjournal.org Page 266 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) Taking Laplace transform of Eqns. (2.10), (2.12) and z (2.7) T (r , z , t ) f (r , t ) using the condition (2.11) gives z 2 where k is the diffusivity and k1, k2 are the z radiation constants on the two curved surfaces of the 2 T* q 2T * 0; q 2 s k 2 n (2.14) disc. In order to solve the differential equation (2.2) subject to the conditions (2.3) – (2.7) we define the finite Marchi-Zgrablich (FMZ) transform of T as b T( n , z, t ) rT (r , z, t ) S0 (k1 , k2 , n r )dr a (2.8) z T* 0; z z 0 T* f *( n , s) z (2.15) Solving the differential equation (2.14) and using the conditions (2.15) yields T *( n , z, s) f *( n , s) with the inversion formula cosh( qz ) . q sinh( q ) (2.16) T( T (r , z, t ) n n 1 , z , t ) S 0 (k1 , k 2 , Cn n r) In order to find the inverse Laplace transform of T* in (2.16) we need to find the inverse Laplace transform for G(s) given by (2.9) where S0 and Cn are given by Eqns. (19) – (21) and (25) in the Appendix and n are the positive roots of G ( s) the transcendental equation (22). Taking the FMZ transform of Eqns. (2.2)-(2.3), (2.6)-(2.7) and using cosh(qz ) ; q2 q sinh(q ) 2 n s k (2.17) the operational property (18) with p = 0 in the Appendix and conditions (2.4) –(2.5) we get 2 T z2 Inverse Laplace transform of G(s)= sum of residues 1 T k t 2 n T According to the method described in [5] we have est of G(s) at the poles (2.18) (2.10) T( n , z ,0 ) The poles and residues of estG(s) are given by 0 (i) q (2.11) T z T z 0; z 0 f( 2 mn n ,t) im 2 n or 2 m ; s 2 n k( m m 2 m ) , m 1,2,... z (2.19) (2.12) Now we introduce the Laplace transform with respect to the variable t and define T *( e stT ( n , z, s) (2.20) n , z , t ) dt. (ii) q 0 or s k 2 n 0 (2.13) ISSN: 2231-5381 (2.21) http://www.ijettjournal.org Page 267 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) where m and mn are given by (2.19) and n are the Now we can find the inverse Laplace transform of positive roots of the transcendental equation (22) in G(s) by using (2.18) – (2.21). Then, using (2.16) – the Appendix. * (2.17) the inverse Laplace transform of T ( n , z, s) is given by The mathematical formulation of the problem described in this section has been studied earlier by [2] Pidadi and Khobragade (2.23) where we have made use of the convolution theorem of Laplace transform. Thus the temperature T(r,z,t) satisfying the conditions (2.2) – (2.7) is given by (2.9), . The first term not involving the infinite series in (3.1) is missing from the corresponding result of [2] . In the second term also there is an error of a negative sign which may be corrected by replacing f(r, t) by –f(r, t). (2.22) – (2.24) and (2.19). COOLING PROCESS The temperature change T0(r, z, t) for the 3. THIN ANNULAR DISC SUBJECTED TO cooling process satisfies the equation PARTIALLY DISTRIBUTED HEAT SUPPLY 2 T0 r2 In this section an analysis is carried out for the inverse unsteady state thermoelastic problem of determining the temperature (in heating and cooling (3.2) process), displacement and stress functions of the disc the boundary conditions occupying the space D : a r b, 0 z 2 1 T0 r r T0 z2 1 T0 k t h with the stated boundary conditions. T0 HEATING PROCESS k1 T0 r 0; T0 k2 r a T0 r 0 r b (3.3) Consider a thin annular disc occupying the space D. For time t = 0 to t = t0 the disc is subjected to a partially distributed and axisymmetric heat supply f(r,t) from the interior point (r, ). After that, the heat supply is removed and the disc is cooled by the surrounding medium. The heating temperature T(r,z,t) of the disc at T0 z T0 z 0; z 0 0 z h (3.4) and the continuity condition T0 (r , z , t0 ) T (r , z , t0 ) (3.5) time t satisfies the equations (2.2) – (2.7) with u(r,t) where T(r,z,t) is the heating temperature of the disc at 0. The solution of these equations is given by (2.9) time t given by (2.9) and (3.1). Applying FMZ transform with respect to r together with in Eqns. (3.2), (3.4), (3.5) and using the operational property (18) with p = 0 in the Appendix together with the conditions (3.3) we get (3.1) 2 T0 z2 2 n 0 T 1 T0 k t (3.6) ISSN: 2231-5381 http://www.ijettjournal.org Page 268 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) h T0 z T0 z 0; z 0 ( 1) ( cos(p z ) cos( m z )dz 0 0 m 2 m sin(h p2 ) m ) . z h (3.14) (3.7) In view of (3.13) –(3.14), the finite cosine transform T0 ( n , z , t0 ) T ( n of (3.1) is given by , z , t0 ) (3.8) (3.15) With respect to the variable z we define the finite cosine transform as ˆ T0 ( n , p , t ) Thus h T0 ( n , z, t ) cos(p z)dz ; p 0 Tˆ0 n , p , t is given by (3.12) and (3.15). Inverting the finite cosine transform and noting the h remark after Eqn. (3.13) we find (3.9) Taking the finite cosine transform of (3.6), (3.8) and using (3.7) gives t k p2 2 n Tˆ0 Tˆ ( n , p , t0 can be determined by means of the conditions (3.6) and (3.8). Indeed, from (3.6) and (3.16) we have (3.10) F t n , p , t0 ). k 2 n F 0 (3.17) (3.11) Solving the differential equation (3.10) and using the n , p , t Tˆ n , p , t0 e k p 2 2 n t t0 . (3.12) In order to find Before applying (3.8) we note the Fourier series expansion condition (3.11) we get Tˆ0 (3.16) where the unknown function F( n,t) Tˆ0 t Tˆ0 t0 Tˆ n m ( 1) m cos( mx) 2 m2 1 1 2 2 cos( x)` , 2 sin( ) (3.18) , p , t0 we calculate which yields the finite cosine transform of the expression in (3.1). First, we note that ( 1) cos( p z ) ( m2 p2 ) 1 h cos( m z ) , sin( h ) m m 1 1 2 m2 h cos(p z )dz 0, (3.19) 0 where (3.13) which shows that the finite cosine transform of a term m and p are given by (2.19) and (3.9). Now using (3.8), (3.1), (3.16) and (3.19) we get 0 independent of z is zero. Therefore, while taking the T0 ( n , z , t0 ) T ( n , z , t0 ) inverse finite cosine transform, a term independent of z can be added. Also note that (3.20) ISSN: 2231-5381 x http://www.ijettjournal.org Page 269 h z h International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) Then F( n parameters are t0 = 0.05, z = 1, a = 0.5, b = 1.0, k , t ) satisfying (3.17) and (3.20) is given = 0.38, k1=k2= 0.25, = 0.7501, h = 1 and r = 0.5, by 0.7, 0.85, 0.95, 1.0. t 4. t0 SPECIAL CASE OF PARTIALLY special case DISTRIBUTED (3.21) HEAT SUPPLY The temperature T0(r, z, t) for the cooling process is Consider then given by (3.16), (3.21) and the inversion formula a wherein the prescribed function f(r,t) in (2.7) is given by S0 (k1 , k 2 , T0 (r , z , t ) n 1 n r ) T0 ( Cn n , z, t ) , f (r , t ) (1 e t )( h (3.22) where S0 and Cn are given by (19) – (21) and (25) in the Appendix. It may be pointed out that (3.16) p ) m that 1 and therefore Following the simplifications, we find that T(r, z, t) and h must be such p for any values of m and . For (4.2) m example, if h = 1 and m (4.1) given by (2.9) and (3.1) reduces to contains a factor ( ) (r 0.9) p = ¾ and then where which vanishes for every (4 m / 3) choice of k with m = 3k and n, Bmn and Dn are given by (h n )( 0.9) S0 (k1 , k2 ,0.9 n ) 2 = 4k. k (1 e t ) (e k mn t 1) / 2 (k mn 1) Bmn (t ) 2 k (1 e t ) (e k mn t 1) / (k n2 1) Dn (t ) These n, 2 mn 2 n ; and ; Bmn and Dn are given in special case of [1]. Similarly T0 in (3.16) reduces to (4.3) and F in (3.21) simplifies to (4.4) FIG.1: Variation of T T T (r , z, t ) (0 t t 0 ) together with T0 (r , z , t ) (t t0 ) where T(r, z, t) is (3.22), (4.3) – (4.4) , where as given by (5.2) and T0(r, z, t) is given by (4.22), (5.3), (5.4).For clarity in the figure, the scales on the two parts of the t-axis demarcated by t t0 are taken to be different. The values of other ISSN: 2231-5381 Thus the cooling temperature T0(r, z, t) is given by [1] n, Bmn and Dn are given .Using the series (3.19) it is easily verified that the expression for T and T0 given by (3.22), (4.3) – (4.4) and (4.2) satisfy the continuity condition (3.5). Taking t0 = 0.05 the variation of heating temperature T(r, z, t) (0 http://www.ijettjournal.org t t0) together with the Page 270 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) cooling temperature T0(r, z, t) (t t0) with t is The pair of Eqns.(4) – (5) has a nontrivial solution shown in Fig. 1. The values of other parameters are only if taken to be z = 1, a = 0.5, b = 1.0, k = 0.38, k 1=k2 = J p (k1 , a )Yp (k2 , b) J p (k2 , b)Yp (k1 , a ) 0 0.25, = 0.7501, h = 1 and the curves are drawn for r (8) = 0.5, 0.7, 0.85, 0.95, 1.0. Let be the nth positive root of the transcendental n APPENDIX equation (8). Using (4) – (5) we can write the solution FINITE MARCHI-ZGRABLICH TRANSFORM (3) in the following two forms Let the Bessel differential equation of order p be given by x2 y (x2 xy 2 p2 ) y A J p ( n x)Yp (k1 , n a) Yp ( n x) J p (k1 , n a) , Yp (k1 , n a) yn ( x ) (9) 0 and (1) and let the boundary conditions be given by y(a) k1 y (a) 0 ; y(b) k2 y (b) 0 (2) yn ( x ) A J p ( n x)Yp (k2 , nb) Yp ( n x) J p (k2 , nb) . Y p ( k 2 , n b) (10) Define a function S p as follows: The general solution of Eqn.(1) is S p ( k1 , k 2 , y ( x) A J p ( x) B Yp ( x) n x) An J p ( n x) Bn Y p ( n x) , (11) (3) where Jp( x) and Yp( x) are Bessel functions of the An Yp (k1 , n a) Yp (k2 , nb) ; Bn J p (k1 , n a ) J p (k2 , nb). (12) first and the second kind, respectively, of order p. Then the boundary conditions (2) yield In view of (9) – (10) the function Sp is a solution of Bessel’s differential equation (1) of order p and A J p (k1 , a ) B Yp (k1 , a ) 0 B Yp ( k 2 , b) 0 satisfies the boundary conditions (2). Because of this (4) A J p ( k 2 , b) fact, the sequence of functions S p ( k1 , k 2 , n x) n 1 is orthogonal in the interval (a, b). Finite Marchi(5) Zgrablich (FMZ) integral transform is defined as (see where, for i =1,2, we have [1], [2], [4], [7] ) b J p ( ki , x ) J p ( x) ki J p ( x), f( xf ( x) S p (k1 , k2 , n x)dx a (6) Y p ( ki , x ) n) Yp ( x ) ki Yp ( x). (13) and its inversion formula is given by (7) ISSN: 2231-5381 http://www.ijettjournal.org Page 271 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) f( f ( x) n ) S p (k1 , k2 , n x) S0 (k1 , k2 , Cn n 1 n r) An J 0 ( n r ) Bn Y0 ( n r ), (19) (14) where the norm Cn may be written An Y0 ( n a) k1 nY1 ( n a) Y0 ( nb) k2 nY1 ( nb), b x S p2 (k1 , k2 , Cn n x)dx (20) a (15) The value of the integral Cn is given by (see Sneddon Bn J 0 ( n a) k1 n J1 ( n a) J 0 ( nb) k2 (21) [8] , where [ Prob. 8 – 14]) Cn S p (k1 , k2 , n x) 2 1 p2 2 n x2 S p2 k1 , k2 , n n is the nth positive root of the transcendental equation b x2 2 J ( nb), n 1 x a J0 ( na k1 J ( n a) Y0 ( nb) k2 nY1 ( nb) n 1 (16) Since S p satisfies the boundary conditions (2) we can J 0 ( nb k2 J ( nb) Y0 ( na) k1 nY1 ( na) n 1 write (22) For a = 0.5, b = 1.0 and k1 = k2 = 0.25 the first 20 roots are listed in the Table. With p = 0 the FMZ (17) transform may be rewritten Thus the FMZ integral transform and its inverse are b given by (13) – (14) and (17). Also, if we carry out f( n ) rf (r ) S 0 (k1 , k 2 , integration by parts twice we can prove the operational property n r )dr a (23) together with the inversion formula b 2 f 1 f p2 b f x 2 f S p (k1, k2 , n x)dx S p (k1, k2 , nb) f k2 2 x x x x k2 x a xb f( f (r ) n n 1 - a S p ( k1 , k 2 , k1 n a) f k1 f x 2 n f( n ). Cn SPECIAL CASE p = 0 In our analysis we need the results only for b2 1 1 2 2 k2 a2 1 1 2 2 k1 the special case p = 0. For this special case we have ISSN: 2231-5381 n r) (24) where x a (18) ) S0 (k1 , k2 , Cn http://www.ijettjournal.org 2 n 2 n S 02 (k1 , k 2 , S 02 (k1 , k 2 , n n b) a) (25) Page 272 0 International Conference on Global Trends in Engineering, Technology and Management (ICGTETM-2016) Finally, the operational property is given by (18) with p = 0 where f ( n ) is now given by (23). REFERENCES [1] Khobragade N.W. and Durge M.H.,( 2006) An inverse 4.100035 21.114158 38.493181 55.917084 7.399590 24.582173 41.976075 59.403888 10.781796 28.055419 45.460120 62.891112 14.206770 31.532288 48.945081 66.378691 17.654071 35.011765 52.430780 69.866573 unsteady-state thermoelastic problem of a thin annular disc in Marchi-Zgrablich transform domain. Bull. Cal. Math. Soc., vol. 98, pp. 255-266. [2] Pidadi Sonali S. and Khobragade N.W.,( 2008.) Study of stress functions in a thin annular disc due to partially distributed heat supply. Bull. Cal. Math. Soc. Vol. 100, pp. 691 – 698. Ozisik [3] M.N,( 1968 ) Boundary Value Problems of Heat Conduction. International Text-Book Company , Stranton, Pennsylvania. [4] Pidadi Sonali S. and Khobragade N.W.,( 2008) Numerical study of inverse transient thermoelastic problem of a finite length hollow cylinder. Bull. Cal. Math. Soc., vol. 100, pp. 599 – 608. [5] Churchill R.V. and Brown J.W.,( 1990.) Complex Variables and Applications. 5th Ed. McGraw Hill, New York. [6] Nowacki W., Thermoelasticity, Addison-Wesley Publ. Co. Inc., 1962. [7] Marchi E. and Zgrablich G., (1964) Heat conduction in hollow cylinders with radiation. Proc. Edinb. Math. Soc., vol. 14, pp. 159 – 164. [8] Sneddon I.N., ( 1972.) The Use of Integral Transforms. McGraw Hill, New York. [9] Hetnarski R.B.and Eslami M.R.,( 2009) Thermal StressesAdvanced Theory and Applications , Springer ,New York. TABLE CAPTION TABLE: First positive 20 roots of equation (22) for a = 0.5, b = 1.0, k1= k2 = 0.25. ISSN: 2231-5381 http://www.ijettjournal.org Page 273