Beruflich Dokumente
Kultur Dokumente
00
Printed in the U.S,A. ~' 1986 Pergamon Press Ltd.
(Received 26 October 1984; revision received 23 September 1985; accepted 17 October 1985)
Abstract--Improved equations are formulated for the transfer of heat in flat-plate solar collector ab-
sorbers, including a shape factor accounting for details of flow duct designs, and developing heat
transport in the ducts. Approximate, analytical solutions are obtained in terms of perturbation series,
which are applicable to practical collectors with lengths much greater than the distance between flow
ducts.
The primary result of the analysis is a universal-type design equation, or chart, which determines
the collector efficiency factor for any duct design, heat transfer development, or other absorber-plate
parameter. Collector performance is stated in terms of an effectiveness/number-of-transfer-unit rela-
tionship. Ease of use and utility of the design chart is illustrated by examples.
443
444 K. O. LUND
qs qL
I
where kp is the plate thermal conductivity. The con- hydraulic diameter, and the factor Dh/t is intro-
vective boundary condition at y = b is complicated duced to render S = 1. By comparison of the heat
by the geometry surrounding the flow duct, and transfer process of the actual duct geometry with
would also be limited to the particular duct design that of the idealized geometry (matching the heat
in Fig. 1. Hence, for generality and mathematical fluxes and temperatures at points A and B in Fig.
simplicity, the plate of thickness t is imagined to 2) the shape factor can be determined. The axial
extend to the duct centerline, as indicated in Fig. variation of h s is conveniently expressed in terms
2, and the boundary condition for y = b is written of the Nusselt number ratio R(X) = N u ( X ) / N u ' ;
in terms of an effective heat transfer coefficient, hs(X) = h'sR(X) = (ks/Dh)Nu'R(X), where the
hse(X): prime denotes the fully developed condition, and
ks is the fluid conductivity.
or, With these definitions, the equations can now be
qse = - k p - ~ = hse(Tp - TS) (4) stated in scaled, nondimensional form. The largest
temperature potential in this analysis is Tpm - Tsi.
where Ts(X) is the fluid temperature. Using this as reference, the nondimensional plate
In eqn (1), UL represents a linearization of com-
plex radiative and convective heat-transfer pro-
cesses, and depends on the average plate temper-
ature; it can be determined iteratively by standard
methods[l, 2]. The absorbed flux, qs, is obtained
from the incident irradiance at the particular time
and location, and from the effective transmittance/
absorptance product of the collector cover and
plate surface.
Useful heat is removed from the absorber plate
at y = b by the fluid with inlet enthalpy CsTsi,
where Cs = (rhCp)s is the half-duct fluid heat ca-
t ~.K~'~"~qA'~-~, DdDO
pacity rate, and Ts,. is the fluid inlet temperature.
Thus, the steady-state convective heat transport TI
equation for the idealized flow duct is
d~s
C s --~ = hsd(Tpb TS) (5)
-
(qs'-'qL)e
where Tpb(X) = Tp(X, b), and Ts(X ) = Tsi; at X _t_
= 0.
In order for this formulation to have practical
t ,~'~.B~ B~-.~'~''~," qfe
validity, it is necessary to relate hse(X) to hs(X), T I I.<_Do/2_..,.
the heat transfer coefficient for the actual duct de-
sign. To this end a shape factor, S, is introduced I
such that hse = S(Dh/t)hs, where Dh is the duct Fig. 2. Shape factor determination.
Parallel-flow flat-plate solar collector absorbers 445
and fluid temperatures are defined by seen from eqn (7) that % = 0(1); moreover, from
eqns (A.33) and (A.35) of the Appendix, % - N
O(x, y) - Tpm - - Tp (6) relationship is seen to be, to an excellent approx-
Tpm - Tf i ' imation,
FR = -0 = % / N . (17)
with
Equations (14)-(17) show that the performance
0O (9) of a collector depends directly upon F ' for a given
-- = 0atx = 0, 1,
Ox number of transfer units, N (i.e., a given size of
oO collector); in fact, collectors are frequently de-
-- = 0aty = 0, (10) signed for small N where, from eqns (15)-(17), F n
Oy
F ' as N --~ 0, thus emphasizing the importance
O0 ~2 (11) of accurate determination of F ' .
- - = - - SR[1 - (~b + O ) ] a t y = 1;
Oy K Because developing flow and heat-transfer pro-
cesses are described in terms of the Graetz number,
and Gz = C s / k s X = h N u ' / x , where X = K / N = Cs/
h} ZDh, the development function, R ( x ) , will also
d0 N depend on parameter k; hence, from eqns (A.12),
- - - -- SR[1 - (tb + 0)] = 0 , 0 < x < 1, (12) (A.14) and (16), F'/rl = P(1) depends only on the
dx K
parameters, h and [3, where 13 = ~lv,/S. In other
with words, F'/.q = f c n ( k , 13) constitutes a universal de-
sign function for the thermal performance of the
0 = 0 at x = 0, and ~b(x) = 0(x, 1). (13) class of collector absorber plates considered.
To illustrate application to a specific develop-
ment, the form reported by Kays[8] for a Langhaar-
Here, ~ = b / L is the plate width/length ratio (see
type developing laminar velocity profile and con-
Fig. 1), ix2 = ( b / t ) U L b / k p is a Biot-type modulus,
stant heat flux in a circular duct is considered:
K = U L b / h } Dh is a measure of the relative mag-
nitudes of heat loss and useful heat removal for the
plate, and N = U L b L / C s represents the number of (4/'rr)K1Gz
Nu = Nu' + (18)
transfer units for the collector. As in heat exchanger 1 + (4hr)K2Gz'
theory, N is here a measure of the size of the equip-
ment required to effect the heat transfer; this is seen or, equivalently,
by writing N = U L A c / C s c , where Ac and Cs¢ are
the area and heat capacity rate for the entire col- Clh x + C3k
R = 1 + - - - - - (19)
lector, respectively[6, 7]. X + Czk x + C2k'
Here, Tsm = Tp,, is the maximum possible fluid out- × 1 + C1 In 1 + kC3 + 13 " (20)
let temperature, and % = (Tso - Tsi)/(Ts,,, - Tsi)
is the collector thermal effectiveness (i.e., the ratio Equation (20) is the universal-type design function
of the actual rate of heat removal to the rate for sought; it is shown in Fig. 3. Although eqn (20) was
maximum possible temperature change)[6, 7]. It is obtained using the result of eqns (18) and (19) from
446 K.O. LUND
1.0
0.8
• 20
F"
0.6
I"00 ~ ~ ~ : ~ ~ : ~ Z ~ ® o
0.4
'o
0 9 0 ~
• 0 0.02 0.04 0.06 0.08 0.0
0.2 t I n I , I I I ,
0 0.2 0.4 0.6 0.8 1.0
xS'
Fig. 3. Design chart for determination of the collector efficiency factor, F'.
a circular duct with constant heat flux and flat inlet- is distributed throughout the duct and the surround-
velocity profile, it is likely to be applicable to other ing plate materials. The heat flux entering the flow
geometries and conditions with slightly different duct is
values of constants CI and C2, the main effect of
non-circular geometry being included in shape fac- q s = h s ( T a - Ts), (21)
tor S. In any event, results are obtained in this work
for a n y heat transfer development, R ( x ) , and duct where Td is the local surface temperature of the
shape, S, as shown in eqns (A.35), (A.14), and duct. In the idealized system, the heat flux at point
(A.12). It is noted that k ~ 0 in eqn (20) is equiv- B (i.e., the location equivalent to point A) is qB,
alent to the fully developed condition throughout and the flux entering the idealized duct is
the ducts (R = 1), which would be approximated
in turbulent flow. In laminar flow, the fully devel- q¢e = hse(Tpb - TS) = S ( D h / t ) h s ( T p b - TS).
oped condition is approximated only for xe ~ 1 (22)
where, from X f l D h -~ 0.05 Re Pr, the nondimen-
sional heat transfer entrance length is Xe ~ 0.3h. Consider now the remainder of the plate (to the
Although eqn (20) and Fig. 3 were obtained for a left of point A and point B, in Fig. 2). The temper-
constant plate thickness, they are applicable to ature and heat-flux fields in the plate of the equiv-
variable thickness plates provided the appropriate alent system will be precisely the same as those of
fin efficiency, 0, is used. the actual system, provided
Utilization ofeqn (20) or Fig. 3 requires the eval-
uation of the shape factor, S, which is introduced qB = qA (23)
in the problem formulation to render the i d e a l i z e d
plate/duct system equivalent to the a c t u a l design and
under consideration. One such design and its ideal-
ized equivalent are shown in Fig. 2, where the flow- TB = Ta, (24)
duct material is surrounded by the plate material.
At point A the heat flux from the plate is qa; this where TA and TB are the temperatures at points A
flux, along with the net irradiance flux, ( q s - qL), and B, respectively.
Parallel-flow flat-plate solar collector absorbers 447
1 "iT
qAt + ~ Do(qs - qL) = -~ Ddqa
(26)
~r Ddhc(Ta Td),
2 I I
2 4 6 I0
c
where qd is the heat flux at the outer duct diameter,
and hc is the unit contact conductance between the Fig. 4. Plate/duct shape factors.
plate and duct materials. Similarly, it also follows
that
4. DISCUSSION
Ddqa = Dhqf = Dhhs(Td - 7/':). (27)
Equations (15)-(17), (20) and Fig. 3 stand as the
Now, condition (24) implies that (qs - q L ) e primary result of this analysis. Their importance to
(qs - q L ) . Thus, taking :In = TA ~ Tpb, eliminating collector design can be realized by comparison with
Ta between eqns (26) and (27), and substituting eqn conventional methods for the determination of F',
(23) into eqn (25), the combination of eqns (25)-(27) which usually are valid for only one duct design,
yields and without consideration for flow development[l,
2]. For example, customary charts for illustration
"IT ¢ display F ' versus b, with kpt as a parameter, such
S- 2 1 + c' (28) that a separate chart is used for illustrating each
value of UL and hy[l, pp. 219-221]; by contrast, the
present results can be summarized as one universal
where c = D d h c / D h h s is a nondimensional contact-
design equation [eqn (20)] or chart (Fig. 3), includ-
conductance factor. For S to be as large as possible,
ing the effects of flow development, duct design and
c must not be a small number (i.e., hc should be of
contact conductance.
the same order as hy); furthermore, S ---> rr/2 as c
The key to obtaining a result of this generality
lies with the nondimensional formulation of the
For a circular duct below the plate, the shape
problem, with the introduction of the shape factor,
factor is also given by eqn (28), except that, in this
S, and with the analytical perturbation-type solu-
case, c = w h c / D h h f , where w is the width of the
tion of the equations (cf. the Appendix). Thus, the
contact area between the plate and the duct; for a
descriptive differential equations (8)-(13) contain
half-circular duct below the plate, the result for the
the shape factor and relevant nondimensional pa-
shape factor is
rameters; the approximate analytical solutions are
obtained for any heat transfer development, R(x).
('tr + 2) 2 2 + 4c
S =
- -
4~ ~r + 2 + 4c
.
K heat-loss/heat-removal ratio parameter, K = Ut.b/ radiation and external heat losses. Proc. A S M E Solar
h 'yDh Energy Conf., pp. 201-208 Las Vegas, Nevada (April
h duct flow development parameter, h = K/N = C r~ 1984).
h'sDhL 5. W. F. Phillips, The effects o f axial conduction on col-
txz Biot-type parameter, i~z = UcbZ/kpt lector heat removal factor. Solar Energy 23, 187
0(x, y) evaluated at y = 1 (1979).
q~ fluid nondimensional temperature, + = (T~ - T r Y 6. W. M. Kays and A. L. London, C o m p a c t Heat l£~--
(T,,. - T,) changers, McGraw-Hill, New York (1964).
7. D. K. Edwards, Solar Collector Design. Franklin In-
stitute Press, Phila., Pa. (1977).
REFERENCES 8. W. M. Kays, Numerical solutions for laminar-flow
heat transfer in circular tubes. Trans. ASME 77, 1265
I. J. A. Duffle and W. A. Beckman, Solar Engineering (1955).
o f Thermal Processes. Wiley & Sons, New York 9. K. O. Lund and W. B. Bush, Two-dimensional ther-
(1980). mal analysis o f parallel-flow fiat-plate solar collec-
2. J. Howell, R. B. Bannerot and G. Vliet, Solar Thermal tors. Proe. A S M E Solar Energy Conf., pp. 289-296,
S y s t e m s . McGraw-Hill, New York (1982). Las Vegas, Nevada (April 1984).
3. P. P. Rao et al., Two-dimensional analysis (~f a fiat- 10. J. D. Cole, Perturbation Methods in Applied Math-
plate solar collector. J. Energy 1, 324 (1977). ematics. Blaisdell, Waltham, Mass. (1968).
4. N. E. Wijeysundera, Foreed convection in absorber I 1. A. H. Nayfeh, Perturbation Methods. John Wiley and
ducts subjeeted to uniform and nonuntlfbrm wall ir- Sons, New York (1973).
APPENDIX
Central-region solutions where the Di's are constants of integration (to be deter-
For the central region (B ---, 0, x fixed: ~ --~ ~), mined). Hence, to these orders of approximation, the cen-
substitution of eqn (A.1) into eqns (8)-(13) yields, for tral-region solutions for the plate-temperature functions
450 K. O. LUND
can be written as from eqns (A.26) and (A.13) that DI = 0. T h u s , with re-
spect to the first-order approximations in 8, F~ (x, y), J~ (x),
Fi(x, y) = DiQ(x)e -'Ne(x) cosh ~y (A.15) Gdx), and g d O are zero, but f~(~, y) is not. To determine
cosh tx this function it is written as a Fourier series:
Edge-region solutions
fl(~, y) = ~ klne -°'"~ cos vny. (A.28)
For the inlet edge region (8 ~ 0, ~ fixed: x ~ 0, with n=l
O/Ox = (1/~) O/d~, etc.), the boundary-value problem takes
the form, for i = 0, 1, Now, with gl(O = 0, application of the convective bound-
ary condition, eqn (A.22) to eqn (A.28) yields the following
t92fi 02fi relations for determining the eigenvalues, v,, ton:
O~--T + --Oy2 - la,2fl = 0, (A.16)
a7 _ Ix2 _ b T _ v2 '
ai bi The parameter, ~, is evaluated by considering the second-
order t e r m s in the series (i = 2). In particular, for a con-
w h e r e v is the separation constant. T h u s , with to = (ix2 stant fluid heat transfer coefficient (R(x) = 1), the inte-
+ v2) n/2, the solutions of a7 , to2a~ = 0, and b" + v2b~ grated result is
= 0, subject to al---* 0 as ~ ~ ~, bS = 0 at y = 0, are
82 { ll+'q}
fi(f~, y) = kle ~ cos vy, (A.27) = ~- ( F ' N ) 3 e -F'N 1 IX2 .q2 , (A.37)
w h e r e the k f s are constants.
F o r i = 0, eqn (A.18) requires that ko = 0, such that w h e r e F ' = "q/(l + -q~S).
fo(~, Y) = 0 and Jo(0 = 0; furthermore, the solution of T h e above solution has general validity w h e n e v e r the
eqn (A.23) that satisfies eqn (A.25) is go(~) = 0. T h u s , to t e r m s of 0(~) are negligible (i.e. for ~ ,~ 1); since • d e p e n d s
lowest order, there are no edge-region correction t e r m s primarily on 82, as s e e n in e q n (A.37), this is true for m o s t
for either the plate or the fluid temperatures• Also, for go applications. U n d e r s o m e conditions where O, is as small
= 0, it follows from eqn (A.26) that Do = 1; thus, the as 8 (e.g., the case of infinite thermal conductivity), ~ =
integration c o n s t a n t in e q n s (A. 13) and (A. 15) is evaluated. (~(N) 3 is small only i f N is also small (F', "q ~ 1); if, in this
For i = 1, the solution of eqn (A.24), withjo, go = 0, case, N = (~(1), the effect o f axial conduction in the plate
subject to eqn (A.25), yields g~(~) = 0. In turn, it follows would have to be included.