Professional Documents
Culture Documents
{-
DP-MS-81-118
LWMERICAL
SOLUTION OF NATUW
CONVECTION
IN ECCENTRIC ANNULI
by Darrell [(.Peppert and Roger E. CoopertT I. du Pent de Nemours & Co. Savannah River Laboratory Aiken, South Carolina 29808
F ~.
. .
,..
Proposed for publication and presentation AIAA/ASM Fluid, Plasma, Thermophysics, and Heat Transfer Conference St. Louis, HO June 7-11, 1982
T Research Supervisor, Nuclear Engineering Division tt Research Engineer, Environmental Transport Division
This paper was prepared i.nconnection with work done under Contract with the U.S. Department of Energy. By No. DE-AC09-76SROOO01 acceptance of this paper, the publisher and/or recipient acknowledges the U.S. Governments right to retain a nonexclusive, royalty-free license ina dn to any copyright covering this paper, along with the right to reproduce and to authorize others to reproduce all or part of the copyrighted paper.
This document was prepared in conjunction with work accomplished under Contract No. DE-AC09-76SR00001 with the U.S. Department of Energy.
DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. This report has been reproduced directly from the best available copy. Available for sale to the public, in paper, from: U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161, phone: (800) 553-6847, fax: (703) 605-6900, email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/ordering.htm Available electronically at http://www.doe.gov/bridge Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062, phone: (865 ) 576-8401, fax: (865) 576-5728, email: reports@adonis.osti.gov
DP-MS-81-118
Numerical Solution of Natural Convection by D. W. Pepper and R. E. Cooper E. I. du Pent de Nemours & Co. Savannah River Laboratory 29808 Aiken, South Carolina
in Eccentric Annuli
eccentric annular space a!re solved with two highorder numerical algorithms. Tk~e equation set is transformed
equations:
coordinates
elements are used in the direction the pseudospectral Transient ld<Ra<106, technique
0.70<Pr<102, with the inner cylinder near the top, Results are similar to for low Ra
published
INTRODUCTION Natural many years. convection Numerous in concentric annuli has been studied for
articles can be found in the literature which and numerical results.1-4 Natural con-
sively, has begun to receive more attention over the last few
years .
annuli occurs
the electric and nuclear energy fields, as well as in solar energy and thermal storage systems. ground electric transmission For example, the cooling of undercables is significantly affected by Simi-
larly, the cooling of nuclear fuel and target tube assemblies following irradiation must be carefully monitored in the event of Likewise,
assembly burnout due to inner tubular eccentricities. natural convection cooling of casks containing exceed thermal ~nidelines if positioning
transformation algorithms.
state results are compared with existing theoretical mental data. Forced convection Trombetta5
and experi-
in eccentric annuli was investigated by methods and a least squares techSimilar studies
using approximation
nique for various radius ratios and eccentricity. for laminar flow through an eccentric
Both Newtonian and non-Newt:onian fluid flows were calculated bipolar coordinates and a finite difference technique.
An ana-
-2-
transformed
Analytical
for slightly eccentric annuli with Ra<10 4 for constan~ cylinder wall temperatures . Bofh numerical and experimental and thermal conduction et ala studies of natural convection
A finite element procedure using isoparametric Cylinder walls were held constant;
employed.
were obtained
for
1.2x104<Ra<2.2x104 . Free convection flow in eccentric annuli was modeled by coordinate transfor-
curvilinear
finite difference
procedure.
tricities were com- pared with data obtained by Kuehn and Goldstein. 11 Bipolar coordinate recently used to calculate transformations have also been distributions
through an eccentric annular duct by Feldman, et al. ,12~13 following the work of Synder and Goldstein.14 Recently, Prusa and Yau15 numerically annular free convection between <16,900 simulated eccentric for Gashof numbers
two cylinders
for steady laminar flow in polar coordi and transformed into a unit circle
-3-
(the boundary of the inner cylinder becomes the unit circle). The transformed
half of the circular domain using finite differences mesh spacings. Computation
hours on a CDC 175 for 10-4 residual convergence. favorably with experimental using perturbation data and analytical
solutions obtained
techniques
at low Gr numbers. 7
Problem Analysis The problem to be analyzed consists of free convective developed Figure 1. in the cross section of an eccentric The geometry and configuration flow
annulus, as shown in
bipolar coordinates,
=_ c sin 5 coshncos$
(lb)
where c=ri sinh~i=ro sinhno and ~i and denote inner -and outer annulus surfaces defined no as ni = cosh-l
[
(2a)
no = cosh-l
[y
(2b)
with y=ri/ro
and ~=e/(ro-ri).
-4-
The inner surface, as depicted in Figure 1, is assumed to be heated tclconstant temperature, Ti. The out?r surface is held a tempera-
constant
(3)
= PrD~2w+
PrRa (B~+~)
A aT a UT
%-
a$
a UT __BavT Ban
as
- avT - D72T an
(4)
streamfunction,
sinhrl), a,ld D = (coshrl - cos$)2/(c sin$ sinh~), Pr is the prandtl number, and Ra is the Ray Leigh number. are defined as: u = -~ (1 - cosh~ cos~) ~+ sinhq sin$ ~ 1 (1 - coshq COSS) ~ 1 associated with Equations (3-5) are (8) (7) The velocities, u and v,
c [
(6)
v = ~ [
sinhq sin$ ~-
Y(~ij $) = Y(~oj 5) = 0> (~, - ) = ~(~j ) D A a2y W(ri, - Tr) = W(n, Tr) a~2,
(9)
w(Tli, s) =W(no,$)
= -
(lo)
-5-
Numerical Solution Technique Application of bipolar coordinates results in :~.? ,l~?TT5ical as shown
into a rectangular
domain,~~~:g
and a pseudo-
The problem is strongly governed by the nonperiodic ary conditions, suited. for which the finite element technique
is well
The pseudospectral
Cubic spline techniques and other compact operators are also viable alternatives to solving the equation set. (3) and (4) are solved in a fractional sequence,
Equations
allowing the equation set to be solved with basic one-dimensional algorithms . The resulting equation system is solved sequentially The advantages of using these methods lie in
in each coordinate direction. methods over more conventional their high order accuracy, of implementation. Application example, Equation
finite difference
equivalent
computational
of the techniques
is straightforward.
For
(ha)
(llb)
-6-
For a one-dimensional
ment lengl:h, the basic recursi~n rzlation for the n direction is written as20 _l-2ui)
+
$i(i-l -
i+l
(12)
where
Ui
is
the velocity at node point i, $ is either vorticity tile time derivative, k is either
terms are expressed as new and old values using Boundary relations at ~i and no over one
time averaging.
i.e. , the boundary and first interior node. 20 transport equation in the
n direction, $ is solved in the $ direction by a one-dimensional pselldospectral technique. A (~,t) in onedimensional from the distribution ~(k,t) ~; The :ime dependent Fourier components spectral space are determined
$ exp(ik~)
-7-
(14) where nj ass~nes integer values as W/2<nj<X12. Fourier cornponel.ts are obtained, directly by the relations (15a) the derivatives After the are evaluated
32+ := ~
~ $2 i
-k2A(i,t)exp(it5)
(15b)
These transforms are evaluated using a Fast Fourier Transform algorithm.24 Equation nique. ?eriodicity of the boundary conditions, as i[~posed by techfor the
by the pseudospectral
spatial derivatives
and forward-in-time
Results Solutions to the transformed 102<Ra<10;. equation set were calculated for
Solutions were begun with w and T set to zero within domain for Ra<10. A 20x20 mesh was used with were made on an IBM to achieve 10-3
the computational
Computations
3081; an average of 500 iterations were required Convergence fOr VOrticitY ((Wnew-Wo~d)/Weld).
Studies were
conducted with the inner cylinder oriented near the bottom, top, and side of the outer cylinder. To assess model accuracy, a nearly concentric simulated, and values of local equivalent vs. azimuthal (A=.001) case was
-8-
Projahn, et al. ,9 and Kuehn a~d Goldstein 25 for Ra = 5 x 104, Pr=O.7, and y=2 .6. A perfectly concentric as A=O causes situation is not possible using to becor,e degener-
the equations
Figure 3a compares results from the present study with both and numerical data. 9,23 The model predicts Figure similar
experimental conductivity
parisons for two eccentric cases, A=.652 and A=-.623. for ~=.652
(near the top of the outer cylinder) and A=.623 The appearance
(bottom) were 4.8 x 104 and 5 x 104, respectively. of a premature maximum in the outer cylinder
This effect is due to flow inhibition between the two cylinder surfaces; convection heat transfer decreases and conduction increases as 3 + OO. Model predictions for Keg vs. o agree fairly
conductivity
cylinder offset laterally from the center of the outer cylinder for A=.65 and Ra=104. Results from the present model are not in close
agreement with results obtained by Projahn, et al. ,9 but appear to follow the same general trend and curve shape. Streamline and isotherm plots are shown in Figures 4 and 5 for Ra=102-106 with A=-.625 and A=.625, respectively. and temperature distributions are essentially The flow field Half-
symmtric.
plane results obtained by Projahn, et al.g are similar in appearance; differences in values are principally due to boundary
-9-
transformations
the order of numerical accuracy between finite diff~rence and Galerkin mthods. An increase in Rayleigh number creates an pattern for all the inner cylinder the overall heat
positioning
transfer and flow strength, and supports similar conclusions obtained in previous studies. 9,15,25 Solutions tended to be
independent of Prandtl number when ?r >10. Velocity vector plots are shown in Figures 6a through 6C for Ra = 104 and Ra = 105 with A = f0,325 and A = fO.625. without
cells
Vectors
appear witnin
previously
cylinder walls, Jelocity gradients become large and the local heat transfer increases. The effect of increasing the eccentricity 6a and 6b for Ra = 104. is
buoyancy <:auses the fluid to become more vertically boundary separates layer decelerates
thermal p~lume. Upon hitting the outer cylinder wall, the flow cools very rapidly and travels downward. dominates between cylinder; Conduction heat transfer
the two cylinders at the bottom of the inner becomes dominant as the flow travels around
convection
-1o-
In this configuration
large velocity gradients near the inner and outer cylinder walls. Conduction now becomes the dominant mechanism of heat transfer in As the fluid
returns along the outer cylinder wall, the flow cools and approaches a region of relatively stagnant fluid where convection in Figure 5. The effect of
convection becomes more inhibited in the region between the inner and outer cylinders for A = 0.625; this is understandable upper region is now inadequate since the of
fluid motion around the inner cylinder. The (effect of increasing Figures 6b and 6c. the Rayleigh number is shown in
velocity );radients become steeper near the cylinder walls. onset of i~small secondary vortex is just beginning
A = -0.625 immediately above the two large cells adjacent inner cylinder.
large cel;ls shift slightly upwards and the bottom stagnant region appears relatively unchanged.
-11-
It i.sapparent that the mesh spacing near the cylinder walls cannot accurately resolve the boundary layer development Although and local is
the =sh
somewhat coarse, the numerical methods do p~ovide a reasonable approximation to the bulk fluid motion and temperature pattern.
A plot of the overall Nusselt number, NU, vs. Rayleigh number is shown in Figure 7 for y=2.6 and A=.325 with Pr=O.7 and Pr=100. The average Nusselt number is based on the local values at Results are in close agreement with values The transformed is. (16) which is integrated to yield an overall Nusselt number at ri. The integral in Equation (16) is solved by Simpsons rule. A relation for
(convection)
similar relation is obtained for the value at r=ro. heat transfer is essentially (<103). one of conduction
The mode of
at low Ra numbers
equation8P 25 Nucd = cosh-l (r~+~~-e2)/2r r oi [ 1 overall Nusselt number is calculated by the relation Nu =
[
(17)
me
(18)
where Nut,, in the averaged overall Nusselt number for convection. Equa!:ion (18) is used to compromise conduction and convection in obtaining the effects of both number.
-12-
The exponential value of 15 was chosen by Kue;l;l ~iliGolds tein25 t:] fit experimental data, and was similarly used i> t;lis stud~: to
compare with their results. The effect of pos~tion of the inner cylinder significantly influences the local Nusselt numbers. A decrease in the inner
upper region of the outer cylinder; accordingly number increases. An increase in Prandtl number
ntiv5er.
annular space betveen two isothermal cylinders has been obtained for 102<Ra<106 and .7<Pr<102. The transient equation set of from rectangular to
bipolar coordinates and solved with two high order accurate onedimensional algorithms. A linear finite element recursion relation
was used [:0solve the transport equations normal to the cylinder surfaces; equations
a
pseud~spectral
technique was used to solve the periodic Both algorithms are simple
in the pseudospectral
The abilities of both methods to resolve steep gradients documented in the literature. The use of one dimensional and computational
-13-
required for Ehe problems addressed t!lan 100 k byte stora;: .{i~s t-his s tudy. Stzady-sta:z
in
CPU for Ra >105 on an IBM 3081 computer. results support data previously published in the heat transfer }Ieat transf2r
configurations.
rates are lowest when the inner cylinder toward the outer cylinder. tions, local conductivities cal half-plane; streamlines axis.
In horizontally
are nearly symmetric about each vertiand isotherms are displaced slightly
As the Prandtl number increases, heat transfer rates increase slightly; the recirculation center tends to shift slightly upwards ?ositions. A
for Pr>10 in concentric and negative displacement multicellular eccentricities within flow
regime
when Ra=105; flows appear to be nearly symmetric At Ra=106, a multicellular occur, although flow regime is the flow is
slightly
asymmetric.
and non-dimensional
variables20
aT
aT
aT = 72T _
E+u%+vay
T2Y =
-w
(2A) (3A)
-14-
*here w i;ivort icity, T is temperature, Pr is the Prandtl number, Xa is the Rayleigh number, Y is stream function, and u and the velocity components defined as (4A)
v are
readily by employing
Uch that
= J-l
(6A)
-sinhn sin$ c
[
cosh~ COS$-l
(7A)
of the chain
cosh~-cos$ 2 72 c )
(8A)
Temperature
involving the Nusselt number require temperature to the two cylinder surfaces, is transformed by the relation i.e. , ~ ar/ri,ro me
-15-
(9A) and ernploymeat of Equation aT _ cos$-coshn aT m z c ) ( A prclliminary transformation inner cylinder is horizontally tion. (6A); thus, Equation (9A) becomes (lOA) of the equation can be made if the (rotated) in the y direc can be
displaced
coordinates i.e.,
to bipolar coordinates,
(11A)
Acknowledgment The information contained in this article was developed during with the U.S.
-16-
REFEMNCES 1. Powe:, R. E., C. L. Corley, and S. L. Carrath, A )Jumericai Solution for Natural Convection Transfer, May, 210-220 (1971). . 2. Charrier-llojtabi, M. C., A. Mojtabi, Numerical Horizontal (1979). 3. Custer, J. R. and E. J. Shaughnessy, Liquid Metals (1977). 4. Schws~b, T. H. and K. J. de Witt, Numerical Free Convection between Two Vertical (1970). Eccentric Annuli, Investigation of in an Enclosure, Natural Convection in and J. P. Caltagirone, in in Cylindrical Annuli, J. Heat
101, 171-173
J. Heat Transfer,
99, 675-676
Coaxial Cylinders,
Trombetta, M. L., Laminar Forced Convection Int . J. Heat Mass Transfer, 14, 1161-1173
(1971). Fluids in an
6.
Eccentric Annulus, J. Engr. Industry, ASME, 498-506 (May i975). 7. Yao, L. S., Analysis of Heat Transfer Annuli, J. Heat Transfer, 8. in Slightly Eccentric
for
-17-
REFERENCES, 9.
Contd. of
Projahn, V., H. Reiger, and H. Beer, Numerical Analysis Laminar Natural Convection Between Concentric
and Eccentric
4, 131-146 (1981).
Thompson, J. F., F. C. Thames, and C. W. Mastin, Automatic Numerical Generation of Body-fitted Curvilinear Coordinate Two-
15, 229-310
Correlating
Equations
for
Between Horizontal
Circular (1976).
10, 1127-1134 .1
12.
and J. F. Osterle,
AnnL~lar Ducts, Int. J. Heat Mass Transfer, (19812). 13. Feldman, E. E., R. W. Hornbeck, Numerical Developing Transfer, . 14. Solution of Developing
An Analysis
of Fully
-18-
REFERENCES, 15.
presented
and Heat
Tran:~fer Conference, June 7-11, 1982, St. Louis, MO., ASME paper 82-HT-43. 16. Synd~r, W. T., An An;ilysis of Slug Flow Heat Transfer Eccentric Annulus, A.I.Ch.E.J. , 9, 503-506 (1963). 17. E1Saden, M. R. Heat Conduction in an Eccentrically Hollow, J. in an
Infinitely Long Cylinder with Internal Heat Generation, Heat Transfer, 18. 83, 510-512 (1961).
Redberger, P. J., and M. E. Charles, Axial Lamirlar F1OW in a Circular Pipe Coritaiininga Fixed Eccentric Chem. Engr., 463-467 (1962). in an Core, Can. J. of
19.
Eccclntric Annulus with Constant Heat Flux Boundary Condition. Proceedings of the Tenth Southeastern Conference on
latirig Flow by a Simple Finite Element Recursion Computers 21. and Fluidsfi 8, 213-223 (1980).
100-113 (1973).
-19-
REFERENCE!;, Continued 22. Myers, R. B., T. D. Taylor, and J. W. Murdock, Pseudospectral Vortex Flow in a Stratified,
Fluid, J. Comp. Physics, 43, 180-188 (1981). P. and R. Sweet, Efficient FORTRAN Subprograms Equations,
for the Solution of Elliptic Partial Differential NCAR Tech. Note, TN/IA-109, 24 130 pp. (1975).
297-301 (1965). 25 Kuehn, N. H. and R. J. Goldstein, Nat~~ral Convection Heat Transfer Horizontal (1978). 26 Kuehn, T. H. and R. J. Goldstein, Theoretical Horizontal (1976). An Experimental and Cylindrical An Experimental In Concentric Study of
J. Fluid Mech.,
-20-
Nomenclature
= c/(sin$
sinhn) sinhn)
= (1-t:oshn cos$)/(sin5 .
= (coshn-cos$)2/(c = eccentricity,
X*Xi
= acceleration of gravity = heat transfer coefficient = Jacobian = PrD or D = wave number ;<eq = local equivalent conductivity L Nu Pr = cavity gap = Nu~sel~ number = Prandtl number, v/a
q r Ra i =0 t T
= heat flux, 2~hi (Ti--To) = radial direction = Ra:yleigh number, g6(Ti0-To-)L3/va = radius of inner cylinder = radius of outer cylinder = time = internal temperature of inner cylinder wall
Ti = temperature
Nomenclature,
Continued
x Y a a d Y A $ n 9 A v y 6
rectangular
coordinate coordinate
direction,
rectangular
= angle of rotation of x,y axis = thermal diffusivity = thermal expansion coefficient = ri/ro = gria interval = bipolar coordinate, = bipolar coordinate, . azimuthal direction = e/(ro-ri) = kinematic viscosity = Streamfunction = time dependent variable normal to ~ normal to $ (=$)
170ndimensional Variables t = ciLfL2 T = (TO-To)/(To-Ti) u = uLla v = vL/a w = L2WIV x = XIL y = y/L Y f/La
Subscripts cd = cond~lction Cv = convection i o a = inner cylinder wall = outejt cylinder wall = rotated x,y axis
FIGURE 1.
FIGURE 2.
FIGURE 3.
FIGURE 4.
FIGURE 5.
for Ra = 102-106,
FIGURE 6.
with
A = *0.325
and A = tO.625
FIGURE 7.
Overall Nusselt
Number
vs. Rayleigh
To
FIGUW
1.
Eccentric
Annul:~r Cross-Section
FIGU~
2.
Transformed
Eccentric Annulus
I
R.(
I
1.=?
I klte,
G
\ \ \ \ \ \ \ \ .
25 0 Py=.: :,
\ \ k\
. e, 3 =. -.. *
1:0 I.O 1% 180
1 .,0
I
,0
I
60 80 0 ,00
A
.,625 -.623 -.623
I..,,
0.,.7 v o 1
0 _ ----
0
b. VerticalEccentricDisplacement ~~ I
5
R,! Y
P.=.,,,
u ml x
() /
, /0
.-----. 0
\, \
& i,
0
c.
FIGURE 3.
Local Equivalent
STREAMLINES
G
ISOTHERMS
... .. ,..
,,----.
Fla = Id
lRa = Id
:. ,/:
........ ., ..
Ra = 104
,.
Ra =
105
Ra = 106
,:\ ,.-.,
FIGURE 4.
Streamlines
and
Isotherms
for Ra =
102_106 ,
STREAMLINES
Ra =102
lRa=103
Ra = 104
.......,+-,
-/-
,,,
Ra = ld
Ra =
106
FIGURE 5.
for Ra = 102-106,
Ra = 104
[0>
.-~,,, -,. -.,< ,/ - ,--,,,
,,~1
...
, . . .
,,
.,,
,-.
-.
.
,.
,,
(,,
.,
,,,
,,
f.....~ ,
,,
,.,
,~:
!,
!.
!~$ .,
,,. . . ...,~
.
b!, ~, ./,
.,
,, ,, ,,
,, ,, ,,
-.
.
..,,,?
--, --/ , -.
,,
,
,.
,,
~,
.,,
,. ..
/./,
...
-.
(MAX ~=
(MAX
Ra = 104
,,, , ,, \oy
-,--.-,,!
,-,
. ----
... .,,
,. ,,, ~~,~
, ,,, ,,,
. . -. ,--,
,,
-J,
,--
,--
,,.
--
, .,
,,.
//,
/.,
,,, .
--. , .< .
+4
,,,
,.
,
!!,,, J ,,
., ,.. ,..
,$.
,,, ,
-,
,,
!!,.. ,,
,, --
..
-., -/,
..
--
--. ..-.
. . .
..
.-
,!.
-.
.-
(MAX k=
,.,
r.. ,-. .,. ,, . . .$,> .~,,, .,.
. . . . . ,
,,, -.. , ---. . ,
/,-,
Ra = 105
,.
-
t,, ,,
,,. -,,.i~
.. , t
,-, .. Eo:
(MAX VEL = 35.1) ~= 0.625
...
. .
!-.
,.,~
,,-.
,.
,,j
I
,-,
(.
~k
,(,
I,-*
,1
!,
,t,
,t
,dl,~
$..\.,,
,,
Il.
IL.
,,
. .
w
\\- (MAX
1,<
,,,
.11
-,,
(MAX
VEL = 247.8)
~= -0.625
c
for M
~=0.625
with
VEL = 229.1)
FI(;URE 6.
= 104-105
A = fO.325
I
PR 07 I
I
Ref
I I I I Ill
I I I
I]j
1, I,
1000
~
L+
Prmnt Results Pre=nt Results -
11
PR=loo-
L-~=07
1 , ~2
111!!1
,03
Ill
I
i!+ 104
Ill,
,05
Ra
FIGURE 7. Overall Nusselt Number vs. Rayleigh Number for A = -.325