Professional Documents
Culture Documents
Abstract-The heat transfer between a fluid flowing through a packed bed and the wall is interpreted
according to a one-dimensional model characterized by an overall heat transfer coefficient, h, and by
two-dimensional models with one parameter, the effective thermal conductivity, k, or two parameters,
the effective thermal conductivity A, and the wall heat transfer coefficient o, The experimental results
for h,, k,, X, and a, are correlated as a function of Reynolds number, packing and tube diameter.
The correlations with respect to the Reynolds number are all linear. Correlations between the para-
meters of the various models are also given.
more or less drastic conditions (Froment[5]). The thermocouples are connected to the com-
This conclusion led to the work reported here. pensating cables in the lower cross. They extend
The purpose of this work was to determine 18 cm down from this cross. It was found
precise parameter values for the three models necessary to hold the junction of the couple
discussed above and to correlate them as a exactly in the desired radial position. Therefore,
function of the flow rate and the tube to particle a very thin third cross was attached to the above
dia. ratio. mentioned lower cross at 1 cm above the thermo-
couple junction.
2. APPARATUS AND EXPERIMENTAL The radial location of the couples can be seen
TECHNIQUE from Fig. 1. The thermocouples were located
The experimental equipment is similar to that slightly above the packing, at a distance of one
used by previous investigators [ 1,161. The particle diameter. From this distance onwards
packed bed itself was contained in a copper the radial temperature profile was completely
cylinder surrounded by a concentric cylinder smooth.
also made of copper. Saturated steam con- The wall-temperature was taken equal to the
densed in the annulus at 100C. The construc- temperature of the steam condensing on the
tion permitted easy interchange of the inner outer surface of the copper wall.
cylinder in order to investigate the influence For every bed height and flow rate eight profiles
of packing to tube dia. were measured by turning the crosses over 45.
The fluid, air, was delivered by a two-stage This was done in order to minimize deviations
blower with a capacity of 160 m3/hr at zero from the radial position and effects of irregular-
pressure drop and 40 m3/hr with a packed bed of ities in the bed surface.
1 m height. Flow rates larger than 30 mYhr were Steady state was attained after about 1 hr.
measured with an orifice, British norm B.S.
726, having a pressure drop of 1 m water at the 3. EXPERIMENTAL RESULTS
maximum flow rate of 100 m3/hr. The experimental program and conditions are
Flow rates lower than 30 m3/hr were measured shown in Table 1.
by means of a rotameter. At high flow rates the Figure 1 shows a typical radial temperature
air was heated considerably by the blower. profile.
Therefore, a heat exchanger was installed before
the packed bed, so as to maintain the inlet
temperature of the air to the bed at 30C -+O*lC
in all cases.
The experiment consists of measuring the
radial temperature profile in the fluid just above
the packing. This is done for several bed heights.
For each bed height a wide range of fluid
velocities is used. The thermocouples were
chrome1 alumel Philips Thermocoax AB-AC 10
with diameter 1 mm. The emf was read on a Pye
potentiometer thermocouple testing set.
Twelve such thermocouples were mounted on
a cross which has springs guiding it along the I 1 I
inner cylinder wall. The center of the cross is 20 40 60 7
r, mm
attached to a vertical rod extending to the top
of the cylinder, where another cross helps keep- Fig. 1. Radial temperature profiles; V&-catalyst.
ing the rod in the center axis of the inner cylinder. dt = 0.1575 m; d,, = 04057 m; z = O-293 m; Re = 252.
568
Heat transfer in packed beds
F=L(T,-T,)
35
0 _
&
I I
100 200 300 400 5 )
Re
Table 2. Static contributions to the parameters of the different bed heights extrapolate to the same
different models static contribution, keo, except again that of
0.218 m. The slope tends to a limiting value
Lo k he0 %I which depends on dAd,. In the limit the follow-
d,, $2 d,, d,, d,, d,, d,, d,, ing correlation is obtained:
SO, cat 5.2 14.0 0.138 0.223 0.176 0.240 17.0 5190 0.0022
PA cat 6.6 13.3 0.173 0.219 0.241 0.224 16.2
NH, cat 10.0 20.5 0.270 0.320 0.360 0.350 31.0
70.0
85.0
ke = keo+ 1+ 120 (dJdr)Re (8)
d,, = 0.1575 m. where the k,Ofor the different catalysts are given
dt2 = 0499 m. in Table 2. The mean deviation between cal-
culated and experimental values is 3.75 per cent.
The mean deviation between calculated and From the analogy between heat and mass trans-
experimental values is 3.75 per cent. fer the dynamic contribution to k, may be written
572
Heat transfer in packed beds
I 1
200 300 400 0
Re
Re
Bunnell et al. correlated their data according to the higher c&/d, the longer the bed has to be, to
this model[15]. If their dP and dt values are reach the limit for A,.
substituted in (8) a value of 040153 is found for Figure 6 shows the relation h,=f(Re) is
the slope, in excellent agreement with their value linear, with the slope tending to a limit value as
of O-0015. the bed height increases.
The correlation for the limit value is as follows:
(c) Two dimensional model with two parameters
(1) Radial e$ectiue thermal conductivity. The 04025
trends are analogous to those observed for k,: = *O+ 1 +46(dP/dJ2 Re (9)
573
A. P. DE WASCH and G. F. FROMENT
where the h,O for the different catalysts are where the olwoare given in Table 2.
given in Table 2. The correlation for (Y, is far superior to
The mean deviation between calculated and those published until now. The mean deviation
experimental values is only 1.1 per cent. This between calculated and experimental values is
linear relationship was found already by previous only 6 per cent.
investigators using this model (Coberly and Figure 8 finally shows the agreement between
Marshall [l] ; Campbell and Huntington [ 17,181;
Calderbank and Pogorskyll61, Yagi and Kunii
[2,3]) and is predicted by theoretical models
(Yagi and Kunii [2,3]; Kunii and Smith [4]).
The slope of the A, vs. Re line is dependent on
ddd,, as can be seen from (9). When the slopes
obtained by previous investigators are com-
pared with those predicted by Eq. (9) for corres-
ponding d, and d,, excellent agreement is found
for those presented by Plautz and Johnstone [ 191
and Quinton and Storrow [20].
(2) Wall heat transfer coeficient. (Y, also varies
up to a certain extent with bed height. The ten-
dency is analogous to that obtained for h,, A,:
the higher dJd, the longer the bed has to be to
reach the limit value. 701 I I 1
20 40 60 7 7
Figure 7 shows the relation (Y, =f(Re) for
r, mm
the phthalic anhydride-catalyst. Again a linear
relation is found, in contrast with Coberly and Fig. 8. Calculated and experimental radial temperature
Marshall and Calderbank and Pogorsky, but in profiles. V,O,-catalyst.
d( = O-1575m; d, = 040!95 m; Z_= 1.345 m
agreement with Yagi and Kunii. In the limit the curves 1, 2, 3: calculated profiles. Two-dimensional models
following correlation has been obtained: with 2 parameters A, and %. Correlations (9 and 10).
Curve 1 :Re = 81
Curve 2: Re = 188
Curve 3 : Re = 358
%J = a,O+0*01152$Re (10) Points : corresponding experimental measurements.
*,
Fig. 7. a, as a function of Reynolds number. Same catalyst, d,, dt and Las in Figs. 4,5 and 6.
574
Heat transfer in packed beds
the experimental profiles and those calculated at which this limit is attained increases with
on the basis of the two-dimensional model with W,.
two parameters using the correlations (9) and All correlations of the parameters in function
(10). There is only a slight difference in the of the Reynolds number lead to straight lines
immediate vicinity of the wall. which intersect the heat transfer coefficient axis,
indicating there is a static and a dynamic con-
6. RELATIONS BETWEEN THE PARAMETERS tribution. The static contribution is independent
of bed height, except for very shallow beds, but
Some relations between the static contribu-
clearly dependent on the conductivity of the
tions of the different models were established:
particle. Correlations were also set up between
the parameters of the various models. The mean
h,O = (j.,$
t deviation on the correlations is very low: the
data show very little scatter. This is believed to
be due to the great precision with which the
with the dimensions used in this paper. The
thermocouples were located and to the optimiza-
form of this expression is analogous to that
tion procedure used for determining the para-
observed by Verschoor and Schuit [2 11. Further:
meters.
A
t!
0
-- k=0413
_
RbumC-Les auteurs interprttent le transfert de chaleur entre la paroi dun tube et un fluide coulant
B travers un lit garni daprks un mod&e B une dimension caract&isC par un coefficient du tmnsfert
global de temptrature, h,, et par des modeles B deux dimensions et B un paramttre, la conductivitt
thermique effective K,, ou ?Ideux paramktres, la conductivitC thermique effective Ae et le coefficient
G du transfer? de chaleur dans la paroi. Les r&ultats expirimentaux pour h,, k,, A, et CU,sont des
fonctions du nombre de Reynolds, et du diametre du tube et des grains. Les corrklations par rapport
au nombre de Reynolds sont toutes linaires. Let auteurs donnent aussi les corr&lations entre les
parametres des diffhrents modeles.
576