APROXIMATE ANALYTIC ANALYSIS OF ANNULAR FINS WITH UNIFORM THICKNESS BY WAY OF THE MEAN VALUE THEOREM FOR INTEGRAL TO AVOID MODIFIED BESSEL FUNCTIONS Antonio Acosta-Iborra* Departamento de Ingeniería Térmica y de Fluidos, Universidad Carlos III de Madrid, Avda. Universidad 30, Leganés 28911, Madrid, Spain Antonio Campo Department of Mechanical Engineering, The University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249, USA * Corresponding author Phone: (34) 916248465 Fax: (34) 916249430 E-mail: aacosta@ing.uc3m.es Abstract In the analysis of annular fins of uniform thickness, the main obstacle is without question, the variable coefficient 1 r multiplying the first order derivative, dT dr in the quasi onedimensional heat conduction equation. A good-natured manipulation of the problematic variable coefficient 1 r is the principal objective of the present paper on engineering education. Specifically, we seek to apply the mean value theorem for integration to the variable coefficient 1 r viewed as an auxiliary function, the proper fin domain extending from the inner radius r1 to the outer radius r2 . It is demonstrated in a convincing manner that approximate analytic temperature profiles of good quality are easy to obtain without resorting to the exact analytic temperature Page-1/26 profile embodying modified Bessel functions. Instructors and students in heat transfer courses may be the beneficiaries because of the easiness in computing the temperatures and heat transfer rates for realistic combinations of the two controlling parameters: the normalized radii ratio c and the thermo-geometric fin parameter. Keywords: annular fin with uniform thickness; mean value theorem for integrals; simple approximate temperature solution. Nomenclature ht k Bi transversal Biot number, c normalized radii ratio, EBi enlarged transversal Biot number, EBi E relative error for the fin efficiency Et relative error for the dimensionless tip temperature (1) h mean convection coefficient ……………………………………………… Iv modified Bessel function of first kind and order v k thermal conductivity ……………………………………………………….. Kv modified Bessel function of second kind and order v L length, r2 r1 ……………………………………………………………….. MR mean value of the auxiliary function f (R) = Q actual heat transfer ………………………………………………………….. W Qi ideal heat transfer ……………………………………………………………. W r radial coordinate ……………………………………………….…………...... m r1 inner radius ……………………………………………………..……………. m r1 r2 ht r2 r1 k t 2 W m-2 K-1 W m-1 K-1 m 1 in [ c , 1] R Page-2/26 r2 outer radius ……………………………………………………..……………. R normalized radial coordinate, S exposed surface………………………………………………………..…. t semithickness …………………..…………………………………………… m T temperature …………………………………………………………………... K Tb base temperature …………………………………………………..……..…... K Tf fluid temperature ………………………………………………..….………… K m r r2 m2 Greek letters h …………………....………………..……….. kt β2 thermogeometric parameter, dimensionless group, η fin efficiency or dimensionless heat transfer, θ normalized dimensionless temperature, 1,2 roots of the auxiliary equation (13) dimensionless thermo-geometric parameter, L fin effectiveness 1 c Q Qi T Tf Tb T f Subscripts b base i ideal f fluid t tip Page-3/26 h kt m-2 1. Introduction One passive method for augmenting heat transfer between hot solid bodies and surrounding cold fluids increases the surface area of the solid body in contact with the fluid by attaching thin strips of material, called extended surfaces or fins. The enlargement in surface area may be morph in the form of spines, straight fins or annular fins with various cross-sections. Obviously, the problem of determining the heat flow in a fin bundle requires prior knowledge of the temperature profile in a single fin. There are two fin shapes of paramount importance in engineering applications, one is the straight fin of uniform thickness and the other is the annular fin of uniform thickness. In the great majority of textbooks on heat transfer, the section devoted to fin heat transfer begins with a mathematical analysis of the straight fin of uniform thickness that leads to exact expressions in terms of exponentials or equivalent hyperbolic functions for calculating a) the temperature profile, b) the heat transfer rate and c) the fin efficiency. However, this is not the case with the annular fin of uniform thickness where the cross-sectional area and the surface area are functions of the radial coordinate. In view of the impending difficulty, most textbooks skip the mathematical analysis and present only the fin efficiency diagram to facilitates the calculation of the heat transfer rate. However, there are exceptions in the old textbooks by Boelter et al. [1] and Jakob [2] and the new textbooks by Mills [3] and Incropera and DeWitt [4], who do explain the mathematical analysis that eventually supplies exact analytic expressions for a) the temperature profile, b) the heat transfer rate and c) the fin efficiency. Moreover, it is worth adding that the efficiency diagram is included in [4], but not in [1-3]. From a historical perspective, several exact solutions to the problem of heat conduction in an annular fin of constant thickness have been developed by Harper and Brown [5], Murray [6], Carrier and Anderson [7] and Gardner [8]. These exact solutions are based upon the standard assumptions of quasi one-dimensional conduction in the radial direction. Under the quasi one-dimensional formulation, the temperature descend along an annular fin with uniform cross-section is governed by a differential equation of second order with one variable coefficient 1 r that multiplies the first order derivative dT dr . The homogeneous Page-4/26 version of the differential equation is named the modified Bessel equation of zero order wherein the variable coefficient 1 r is troublesome. A review of the heat conduction literature reveals no previous efforts aimed at solving this modified Bessel equation by means of approximate analytic procedures. The present study addresses an elementary analytic avenue for the treatment of annular fins of uniform thickness in an approximate manner. The central idea is to replace the cumbersome variable coefficient 1 r by an approximate constant coefficient. Invoking the mean value theorem for integration, one viable avenue is to substitute 1 r , viewed as an auxiliary function, in the proper fin domain [ r1 , r2 ] by its mean value. A beneficial consequence of this approach is that the transformed fin equation holds constant coefficients. Herein, the two controlling parameters are the normalized radii ratio and the thermo-geometric parameter. It is envisioned that the simple computational procedure to be delineated in the paper on engineering education may facilitate the quick determination of approximate analytic temperature profiles and heat transfer rates for annular fins of uniform thickness without the intervention of modified Bessel functions. 2. Modeling and Quantities of Engineering Interest An annular fin of uniform thickness dissipating heat by convection from a round tube or circular rod to a surrounding fluid is sketched in Fig. 1. The three fin dimensions are: uniform thickness 2t , inner radius r1 and outer radius r2 . In the modeling, the classical Murray-Gardner assumptions (Murray [6], Gardner [8]) are adopted: steadiness in heat flow; constant thermal conductivity k ; uniform heat transfer coefficient h ; unvarying fluid temperature Tf; prescribed fin base temperature Tb; preponderance of radial temperature gradients over those in the transversal direction; negligible heat transfer at the outermost fin section (i.e., adiabatic fin tip); and null heat sources or sinks. Correspondingly, the resulting quasi one-dimensional fin equation framed in cylindrical coordinates is Page-5/26 d dT h r r (T T f ) = 0 dr dr kt in r1 r r2 (1) or in expanded form 2 d T + 1 dT 2 (T T ) = 0 f dr 2 r dr in r1 r r2 (2) where 2 h tk is the thermo-geometric parameter [1]. As far as the classification is concerned, Eq. (2) is a non-homogeneous differential equation of second order with variable coefficients, wherein the variable coefficient 1 r is of intricate form. The proper boundary conditions implying prescribed temperature at the fin base r1 and zero heat loss at the fin tip r2 are T (r1 ) = Tb and dT (r2 ) =0 dr (3a, 3b) Fundamentally speaking, when a body surface is extended by a protruding fin of any shape, the external convective resistance decreases because the surface area is increased, but on the other hand the internal conductive resistance increases because heat is conducted through the fin before being convected to the fluid. Calculation of the heat transfer from an annular fin of uniform thickness can be carried out directly in two ways: (1) differentiating T(r) at the fin base r = r1: Q1 kAb dT (r1 ) dT (r1 ) 4 k t r1 dr dr (4a) Page-6/26 or (2) integrating T(r) over the fin surface S: r2 Q2 h (T T f ) dS 4 h (T T f ) r dr S (4b) r1 Clearly, Eq. (4a) is easier to implement. Upon introducing the normalized dimensionless temperature T T f Tb T f and the normalized dimensionless radial coordinate R r r2 , the parameter c r1 r2 emerges as the normalized radii ratio. Thereby, Eq. (1b) is transformed to 2 2 d 1 d + =0 d R 2 R dR (1 c)2 in c R 1 (5) along with the boundary conditions (c ) = 1 and d (1) =0 dR (6a,6b) Eq. (5) is classified as the modified Bessel equation of zero order (Polyanin and Zaitsev [9]). Despite that the transversal Biot number Bi ht k is the natural reference parameter in fin analysis, the enlarged Biot number ht r r EBi 2 1 k t 2 (or 2 for short) is another viable parameter. Herein, the fin slenderness ratio is defined as the length L r2 r1 divided by the semi-thickness t. Page-7/26 Conversely, the heat transfer Q from any fin can be estimated indirectly using the concept of dimensionless heat transfer or fin efficiency Q Qi as proposed by Gardner [8]. In here, Qi is an ideal heat transfer from an identical reference fin maintained at the base temperature Tb (equivalent to a material with infinite thermal conductivity k ). For the specific case of annular fins of uniform thickness, the ideal heat transfer is given by Qi 2 (r2 r1 ) h (Tb T f ) 2 2 As noted before the computation of may be carried out in two ways: (1) by differentiation of ( R ) at the fin base leading to 1 = 1 2c 1 c d (c) 2 1 c dR (7a) or (2) by integration of ( R ) over the fin surface, leading to 2 1 ( R ) R dR 2 c 1 c 2 = (7b) At the end, the magnitude of heat transfer Q is obtainable with Q Qi 2 (r2 r1 ) h (Tb T f ) 2 2 (7c) where η comes from Eq. (7a) or (7b). 3. Exact Calculation Procedure The exact analytic solution of Eqs. (5) and (6) gives way to the exact dimensionless temperature profile ( R ) found in [3]: Page-8/26 (R) = I1( )K 0 ( R ) I 0 ( R)K1 ( ) I1( )K 0 ( c) I 0 ( c)K1( ) (8) where I v (*) and Kv (*) are the modified Bessel functions of first and second kind, both of order v and stands for the dimensionless group /(1 c ) (see Nomenclature). The safe-touch temperature of hot bodies is an important issue for the safety of technical personnel in plant environments (Arthur and Anderson [10]). In this regard, the fin tips of annular fins of uniform thickness are prone to be touched accidentally. Because of this, the fin tip temperature T(r2 ) is considered by design engineers as a “parameter of relevance”. Therefore, the exact dimensionless tip temperature θ(1) follows from Eq. (8). That is (1) I1 ( ) K 0 ( ) I 0 ( ) K1 ( ) I1 ( ) K 0 ( c) I 0 ( c) K1 ( ) (9) Further, the two avenues in Eqs. (7a) and (7b) coalesce into the exact fin efficiency 1 2c I1( )K1( c) I1( c) K1( ) = 1 c I1( )K0 ( c) + I 0 ( c) K1( ) (10) Numerical evaluations of the sequence of equations (8)–(10) are elaborate even with contemporary symbolic algebra codes, like Mathematica, Maple and Matlab. 4. Approximate Calculation Procedure The idea behind the mean value theorem for integrals boils down to replacing an auxiliary function in a certain closed interval by an equivalent number. From Differential Calculus (Stewart [11]), the mean value theorem for integrals can be stated as follows. Let f(x) be a continuous function on [a, b] and the mean (or average) value of f (x) is: Page-9/26 f 1 b f ( x) dx b a a Why not extend this idea to replace a variable coefficient appearing in a differential equation by an equivalent number so that the variable coefficient becomes a number i.e., a constant coefficient? The derived benefit is self-evident, because differential equations with constant coefficients are easier to solve than differential equations with variable coefficients. The disturbing variable coefficient 1 r in Eq. (5) may be viewed as a continuous function f ( R) 1 R outlining a hyperbola segment on the closed R–interval [c, 1]. Upon applying the mean value theorem for integrals to f (R ) , the result is MR 1 1 dR ln c 1 c c R 1 c (11) Next, replacing the variable coefficient 1 r with the constant coefficient MR in Eq. (5), it is converted to 2 2 d + M d =0 R d R2 dR (1 c)2 in c R 1 (12) This is a differential equation of second order with constant coefficients and homogeneous. Hence, the general solution of Eq. (Error! Reference source not found.) is (Boyce and DiPrima [12]): ( R) C1 e R C2 e R 1 (7) 2 The two distinct roots of the auxiliary equation are 1 , 2 M R M R2 4 2 (1 c) 2 (8) 2 Page-10/26 The combination of Eqs. (6), (7), and (14) culminates in the particular solution of Eq. (12). In other words, the approximate dimensionless temperature profile 2e R 1 1e R 1 ( R) 2e c 1 1e ( c 1) 1 2 1 2 (9) At this juncture, it is reasonable to pause for a moment to contrast the complex structure of the exact temperature profile in Eq. (8) involving four modified Bessel functions against the simple exponential structure of the approximate temperature profile in Eq. (9) involving four exponential functions. Consequently, Eq. (15) being of ultra compact form, constitutes the centerpiece of the present work. Additionally, by virtue of Eq. (9), the approximate dimensionless tip temperature θ(1) is (1) 2 e 2 1 1e ( c 1) 1 ( c 1) (16) 2 In the fin efficiency diagram for annular fins of uniform cross-section in [4], the family of curves is parameterized by the radii ratio r2 varying from 1 (straight fin of uniform thickness) r1 up to a maximum of 5. In terms of the normalized radii ratio c, this ample span is synonymous with the reduced c-interval 0.2 ≤ c ≤ 1. In passing, it should be mentioned that the latter format was chosen in the fin efficiency diagram in the textbook by Chapman [13]. In this sense, realistic numbers for the emerging MR in terms of c values of practical significance are listed in Table 1. Returning to Eq. (15), the two approximate fin efficiencies can be generated through the tandem of Eqs. (7a) and (7b). That is, (1) by differentiation of ( R ) at the fin base: Page-11/26 1 c 1 c 1 2c e 1 e 2 1 c 2 2 e1 c 1 1e2 c 1 (17a) or (2) by integration of ( R ) over the fin surface: 2 2 1 c2 23 1 1 1 c1 e1 c 1 13 2 1 1 c2 e2 c 1 1 c 1 2 c 1 2 2 1 2 2 e 1e (17b) It should be expected that the differentiation approach in the short Eq. (17a) could produce numbers that are different than those related to the integral approach in the large Eq. (17b). The explanation for this disparity is that the approximate temperature profile in Eq. (9) does not satisfy exactly the governing fin equation (5). From physical grounds, the heat by conduction entering the fin at the base and the heat by convection dissipated along the surface of the fin could be unequal. This is the reason why Arpaci [14] recommended that whenever ( R ) is approximate, the integration approach η2 in Eq. (17b) must be preferred over the differentiation approach η1 in Eq. (17a). 5. Presentation of Results Inspection of the fin efficiency diagram found in [4] reveals that the smallest radii ratio is c 0.2 (corresponding to r2 5r1 ). This radii ratio was deliberately selected here as a critical test case in order to analyze the totality of the results. The exact dimensionless temperature profiles calculated with Eq. (4) are compared against the approximate temperature profiles in Eq. (9) deduced in this work in Fig. 2. Combined with c 0.2 , three temperature curves for a small = 0.5, an intermediate = 2 and a large = 10 are plotted in the figure. The comparison for the three values reveals satisfactory quality between the approximate and exact temperature profiles. Interestingly, the approximate temperature profiles do not degenerate for the large = 10 because Eq. (9) is physically consistent. In other words, the approximate temperature profile tends rapidly to zero whenever Page-12/26 . Using the approximate analytic temperature of Eq. (15) for both the smallest c = 0.2 and the largest 0.8, the fin efficiencies estimated via the differential approach 1 in Eq. (16.a) and the integral approach 2 in Eq. (11.b) are listed in Table 2 for the trio = 0.5, 1.5 and 3. The exact fin efficiencies computed from Eq. (10) range from 0.1720 for the pair c = 0.2 and = 3 to 0.9160 for the pair c = 0.8 and = 0.5. In Table 2, the relative error in the efficiency E is defined as: E approx. exact exact (18) First, for the smallest c = 0.2, the relative errors E for the differential-based 1 vary from -1.62e-1 when = 3 to -1.77e-1 when = 0.5. Second, for the largest c = 0.8, the relative errors E for the differential-based 1 vary from -2.88e-3 when = 3 to -4.08e-1 when = 0.5. Third, for the smallest c = 0.2, the relative errors E for the integral-based 2 vary from 3.5e-2 when = 3 to 4.37e-3 when = 0.5. Fourth, for the largest c = 0.8, the relative errors E for the integral-based 2 vary from -8.43e-4 when = 3 to 4.08e-5 when = 0.5. As may be seen, all relative errors E are insignificant. Moreover, the fin efficiency conveyed through the integral-based furnishes more accurate results than the alternate derivative-based . This statement is in harmony with the recommendations made in [Error! Reference source not found.]. As the numbers listed in Table 2 demonstrate, the differences between the efficiency results based on the integral and derivative approach diminish for large values of c . In fact, in the limiting case corresponding to c 1 , the approximate and the exact predictions coincide, both collapsing to: c 1 tanh( ) (13) This expression can be easily deduced from the approximate Eqs. (11) taking into account that the Page-13/26 roots confirm that 1,2 (1 c) whenever c 1 . It should be noted that Eq. (13) is the fin efficiency for a longitudinal fin of uniform thickness [3,4] and same , which is a logical similitude owing to the null curvature in the annular fin when c tends to unity and L is maintained constant. Fig. 4 depicts the tip temperature as a function of the dimensionless thermo-geometric parameter and the radii ratio c. For the smallest c = 0.2 and the largest 0.8, and the trio = 0.5, 1.5 and 3, the exact dimensionless tip temperatures range from (1) range from 0.879 for the combination c = 0.8 and = 0.5 to 0.0559 for the combination of c = 0.2 and = 3. The latter having a nearly zero value is representative of an infinite annular fin. As before, the relative errors Et for the dimensionless tip fin temperature calculated with Et (1)approx. (1)exact (1)exact (19) are listed in Table 3. First, for the smallest c = 0.2, the relative errors Et vary from -8.35e-3 when = 0.5 to -3.65e-1 when = 3. Second, for the largest c = 0.8, the relative errors Et vary from 1.50e-5 when = 0.5 to -4.08e-1 when = 0.5. All Et are very small. If c tends to unity while maintaining L constant, the approximate equation (10) for the tip temperature simplifies to c 1 (1) 1 cosh( ) (15) In general, the approximate results for both (1) and η deteriorate when the radii ratio c decreases because the differences between the constant mean values MR coefficient 1/R of the descriptive fin equation are higher. Page-14/26 and the variable Finally, in the event that the instructor or design engineer decides to maximize the heat transfer from a single annular fin of uniform thickness, pertinent information about the three optimal fin dimensions r1, r2 and t was first published in the textbook by Jakob [2] and years later in the articles by Brown [15], Ullmann and Kalman [16] and Arslanturk [17]. 6. Conclusions In this study on engineering education, concepts from courses on calculus, differential equations and heat transfer have been blended in a unique way. In calculating the performance of annular fins of uniform thickness, the use of the mean value theorem for simplifying the fin equation, i.e., the modified Bessel differential equation, gives way to approximate temperature solutions endowed with an unsurpassed combination of accuracy and easiness. Differences between the analytic temperature approximations developed in the present work and the exact analytic temperature profiles relying on Bessel functions are probably below the level of inaccuracy introduced by the Murray-Gardner assumptions on both exact and approximate temperatures. References 1. Boelter, L. M. K., Cherry, V. H., Johnson, H. A. and Martinelli, R. C., Heat Transfer Notes, pp. IIB18-IIB19, University of California Press, Berkeley and Los Angeles, CA, 1946. 2. Jakob, M., Heat Transfer, Vol. 1, pp. 232-234, John Wiley, New York, NY, 1949. 3. Mills, A. F., Heat and Mass Transfer, 2nd ed., pp. 90-93, Prentice-Hall, Upper Saddle River, NJ, 1999. 4. Incropera, F.P. and DeWitt, D.P., Introduction to Heat Transfer, 4th edition, pp. 139-140, John Wiley, New York, NY, 2002. Page-15/26 5. Harper, D. R. and Brown, W. B., Mathematical Equations for Heat Conduction in the Fins of Air-Cooled Engines, NACA Report No. 158, 1922. 6. Murray, W. M., Heat dissipation through an annular disk or fin of uniform thickness, Journal of Applied Mechanics, Vol. 5, Transactions ASME, Vol. 60, 1938, p. A-78. 7. Carrier, W. H. and Anderson, S. W., The resistance to heat flow through finned tubing, Heating, Piping, and Air Conditioning, Vol. 10, 1944, pp. 304-320. 8. Gardner, K. A., Efficiency of extended surfaces, Transactions ASME, Vol. 67, 1945, pp. 621631. 9. Polyanin, A. D. and Zaitsev, V. F., Handbook of Exact Solutions for Differential Equations, CRC Press, Boca Raton, FL, 1995. 10. Arthur, K. and Anderson, A., Too hot to handle?: An investigation into safe touch temperatures. In Proceedings of ASME International Mechanical Engineering Congress and Exposition (IMECE), pp. 11-17, Anaheim, CA, 2004. 11. Stewart, J., Single Variable Calculus, 3th edition, Brooks/Cole, Pacific Groove, CA, 2002. 12. Boyce, W. E. and DiPrima R. C., Elementary Differential Equations and Boundary Value Problems, 7th edition, John Wiley, New York, NY, 2001. 13. Chapman, A. J., Fundamentals of Heat Transfer, 5th edition, MacMillan, New York, NY, 1987. 14. Arpaci, V., Conduction Heat Transfer, Addison-Wesley, Reading, MA, 1966. 15. Brown, A., Optimum dimensions of uniform annular fins, Int. J. Heat Mass Transfer, Vol. 8, pp. 655-662, 1965. 16. Ullmann, A. and Kalman, H., Efficiency and optimized dimensions of annular fins of Page-16/26 different cross-section shapes, Int. J. Heat Mass Transfer, Vol. 32, pp. 1105-1110, 1989. 17. Arslanturk, C., Simple correlation equations for optimum design of annular fins with uniform thickness, Applied Thermal Engineering, Vol. 25, pp. 2463-2468, 2005. Page-17/26 List of Figures: Fig. 1. Sketch of an annular fin of uniform thickness Fig. 2. Comparison between the approximate and exact dimensionless temperature profiles for a fixed normalized radii ratio c 0.2 combined with three different fin parameters . Fig. 3. Comparison between the approximate and exact fin efficiencies as a function of the dimensionless fin parameter for different radii ratios c . Fig. 4. Comparison between the approximate and exact tip temperatures as a function of the dimensionless fin parameter for different radii ratios c . Page-18/26 List of Tables: Table 1. Functional mean MR in terms of the normalized radii ratio c Table 2. Comparison of the fin efficiencies Table 3. Comparison of the dimensionless fin tip temperatures θ (1) Page-19/26 FIGURE-1 r 2t r1 L r2 Page-20/26 FIGURE-2 Page-21/26 FIGURE-3 Page-22/26 FIGURE-4 Page-23/26 Table 1 Functional mean MR for typical radii ratios c c MR 0.2 2.012 0.4 1.527 0.6 1.277 0.8 1.116 1.0 1.000 Page-24/26 Table 2 Comparison of the fin efficiencies Procedure c Relative error E (%) Exact derivative 0.2 3 -1.62e-1 0.1720 integral 0.2 3 3.50e-2 0.1720 derivative 0.2 1.5 -1.83e-1 0.4020 integral 0.2 1.5 -1.25e-3 0.4020 derivative 0.2 0.5 -1.77e-1 0.8470 integral 0.2 0.5 -4.37e-3 0.8470 derivative 0.8 3 -2.88e-3 0.3068 integral 0.8 3 8.43e-4 0.3068 derivative 0.8 1.5 -3.69e-3 0.5760 integral 0.8 1.5 3.34e-4 0.5760 derivative 0.8 0.5 -4.08e-3 0.9160 integral 0.8 0.5 4.08e-5 0.9160 Page-25/26 Table 3 Comparison of the dimensionless tip temperatures (1) c Relative error Et Exact (%) 0.2 3 -3.65e-2 0.0559 0.2 1.5 -3.04e-2 0.2918 0.2 0.5 -8.35e-3 0.8159 0.8 3 -1.02e-4 0.0921 0.8 1.5 -7.18e-5 0.4061 0.8 0.5 -1.50e-5 0.8790 Page-26/26