This article was downloaded by: [University of Texas Libraries]

On: 10 December 2014, At: 09:21

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office:
Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Heat Transfer Engineering

Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/uhte20

Evaluation of Immersion Coil Designs for

Natural Convection-Driven Batch Cooling or
Heating in Tanks
Donald D. Joye, Michael A. Smith
a

Department of Chemical Engineering, Villanova University, Villanova,

Pennsylvania, USA
Published online: 29 Oct 2010.

To cite this article: Donald D. Joye, Michael A. Smith (2000) Evaluation of Immersion Coil Designs for
Natural Convection-Driven Batch Cooling or Heating in Tanks, Heat Transfer Engineering, 21:5, 47-54, DOI:
10.1080/01457630050127946
To link to this article: http://dx.doi.org/10.1080/01457630050127946

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed in this publication are
the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising out of
the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions

Heat Transfer Engineering , 21:4754, 2000

Copyright C 2000 Taylor & Francis
01457632/00 $12.00 + .00

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

Evaluation of Immersion
Coil Designs for Natural
Convection-Driven Batch
Cooling or Heating
in Tanks
DONALD D. JOYE and MICHAEL A. SMITH
Department of Chemical Engineering, Villanova University, Villanova, Pennsylvania, USA

Various designs of immersion coils for batch cooling or heating of liquids in tanks were investigated
to determine which design was most ef cient. This equipment has direct utility in chemical batch
processing operations and, particularly, wort coolinga critical step in small-scale batch brewing
of beer. This operation does not use a stirrer or agitation for a variety of reasons, hence it is
dependent on natural-convection mechanisms for heat transfer, yet quick cooling is desired. Four
basic coil designs with several permutations were evaluated. The results fell into four groupsbest,
good, mediocre, and worst performers. We found that coil placement and shape were design
parameters that had a critical effect on the rate of cooling. Coil spacing also had a signi cant effect
on the cooling rate when it was less than about 2 diameters.

The present work falls into the category of heating

and cooling in tanks, which has a paragraph section
devoted to it in Perrys Chemical Engineers Handbook [1]. Heating and cooling in agitated vessels are
often done by coils or by jackets in the vessel. Early
work in this area was done by Stuhlbarg [2], in which
calculations for immersed coils were presented. Other
correlations for such cases can be found in textbooks,
e.g., McCabe et al. [3], and a fair amount of commercially designed equipment is available. In virtually all
of these cases the liquid being heated or cooled is agitated. This is to increase the heat transfer coef cient
and obtain economic design of such units.

The present problem, however, is constrained to operate under natural-convectio n mechanisms for the liquid in the vessel. Very little fundamental scienti c information has been published on this subject, as it has
only become interesting recently. In chemical processing, particularly pharmaceuticals production, reasons
for not agitating the liquid to be heated or cooled could
include shear sensitivity of the product, or the prevention of aeration (oxidation of the product ) during
cooling or heating. In melting and freezing of paraf n materials [2], for example, where a solid phase is
part of the heat transfer, agitation may not be possible.
In the wort cooling step in small-scale beer making,
natural-convectio n cooling is desired to avoid oxidation
in the wort and to maintain sterile conditions . Large,
industrial-scal e brewing operations, those described by
Strauss [4], for example, use external heat exchangers

Address correspondenc e to Prof. Donald D. Joye, Department of Chemical Engineering, Villanova University, 800 Lancaster Ave., Villanova, PA
19085-1681 , USA. E-mail: donald.joye@villanova.ed u

47

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

or other equipment common to the art. Plate-and-frame

exchangers are favored here because of their ease of
cleaning. Further process information is available from
Hough et al. [5].
Brewing begins with malted barley and other grains.
Enzymes in the malt convert starch to sugars during
mashing. The brewer collects the sweet extract called
wort and boils this with hops to extract avor. The
brewer wants only the right micro ora to turn this hopped wort into beer. The wrong microorganism s can spoil
the batch. Beer yeast, however, cannot be introduced
or pitched into the wort unless the 100C starting
temperature is reduced below 25C. In some cases this
can be as low as 10C. So the hot boil must be cooled.
A precipitate (hot break ) consisting of spent hops and
proteins forms in the kettle and needs to be separated
by decanting or ltration before the liquid is cooled.
Hot liquid is put into a cooling vessel, usually a tank
nowadays, for the wort cooling operation. If the liquid is aerated by agitation, oxidation of the brew can
occur, thereby giving rise to off- avors and a stale or
unstable product. Because agitation could introduce air
bubbles into the liquid, no agitation is used in cooling of the wort. Another reason for not having agitation is that a precipitate (trub or cold-break ) forms
during wort cooling that is often removed before the
wort is fermented into beer. This can be done by ltration or settling, but sometimes it is just carried over
with the wort into the fermentor. Opinions vary on
the usefulness of cold-break separation. In large-scale,
semicontinuou s operations, both slow and careful
agitation and ltration can be, and indeed are, used.
However, in small-scale batch operations favored by
artisan brewers, including many of the booming microbreweries, these operations are neither possible nor
desirable. Thus, our problem is constrained to naturalconvection cooling.
Conventional cooling coils in a variety of designs
are commercially available for this operation and are
presently in wide use. DeLange [6] describes some
ne points of the wort chilling operation, but these
allow agitation, which the author feels is asking for
trouble. McConnell and Schramm [7] present a performance comparison of commercially available cooling
coil units, but the differences they report are due mostly
to differences in heat transfer area (longer coils cool
faster ). Immersion coils are favored over jackets or external coils mostly for economic reasons, but putting the
coils directly in the liquid maximizes surface area and
minimizes cooling time. The rate of cooling the wort
is critical, not only from a process time standpoint, but
also to give the best chance to ensure sterile conditions .
The purpose of this work was to investigate the design
of the cooling coils to see how the operation could be
48

heat transfer engineering

Figure 1

Schematic representation of coil designs.

improved by the speci c design features of the coil.

Perhaps a differently designed coil could work better.
To that end we designed a series of experiments to cool
a hot tub of water with cooling coils of different design
(see Figure 1 ).
EXPERIMENTAL
Figure 2 shows the apparatus, which consisted of
a 37.86-liter (10-gal ) stainless steel tank, 0.345 m
(13.6 in. ) in diameter and 0.406 m (16 in. ) in height.
The tank was uninsulated , open at the top and lled with
22.7 liters (6 gal ) of water. The tank was supported on a
stand, and a propane heater was used to heat the water to
100 C before the start of the cooling cycle. As shown in
Figure 2, the tank contained a copper cooling coil, generally of 7.62-m (25-ft ) length, having an outside tube
diameter of 0.953 cm (0.375 in. ) and an inside tube diameter of 0.714 cm (0.281 in. ). These dimensions are typical for home-brewing equipment. Copper-constanta n

Figure 2 Experimental setup with coil in the tank and thermocouple placement.

vol. 21 no. 5 2000

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

thermocouples were used to measure temperature.

These were formed from typical two-wire thermocou1
ple stock. The bead size was about 0.16 cm ( 16
in. ), and
the bead was placed directly in the liquid whose temperature was to be measured. This small bead size has
a very fast response time, e.g., about 1 2 s to measure
correct temperature when suddenly placed into boiling
water from room temperature.
Cooling water was run through the coils for 1 h, and
temperature readings of the inlet and outlet cooling water temperatures and the wort temperatures at ve vertical locations (temperature variations were found to
be minimal in the radial direction, except very near the
coils ) were recorded every 2 min for the rst 15 min and
about every 5 min thereafter. The setup is illustrated in
Figure 2. About 20 s were required to read these temperatures, so the individua l temperature readings were
not taken at exactly the same time. When the temperatures in the tank are not changing rapidly this is not a
problem, but when they are, such as in the rst 5 10 min
of the experiment, some uncertainty is introduced. The
cooling-water ow rate was high enough to maintain a
good turbulent ow inside the tube and entered at the
bottom to maintain a countercurrent ow situation. We
also reversed the leads to try a cocurrent ow in one
trial. The cooling-water ow rate was kept constant to
eliminate that as a variable. Reynolds numbers (Re ) for
the cooling water in the tube side ranged from 10,000
to 25,000. The change in Re was due mainly to the
change in viscosity as the average bulk temperature of
the cooling water increased. Under these conditions , the
thermal resistance of the copper tubing is relatively very
small, and the resistance of the tube-side cooling water
is much less than the outside resistance due to natural
convection. However, both of the smaller effects were
accounted for in the calculations. The inlet water temperature was not able to be controlled precisely, and it
varied from run to run in the range 9 15C; in most runs
it was in the range from 10 to 12C.
A summary of the coil geometry is given in Table 1.
The heat transfer area varied somewhat with the vertical coil placement in the tank, because more or less
Table 1

Coil design summary

Coil
Commercial helix
Horizontal-coil helix
Conical coil
Asymmetrical
horiz coil
Vertical tube

Coil diameter
(m)

Heat transfer area

(m2 )

0.203
0.279
0.279 large end,
0.139 small end
0.254 width,
0.102 depth
Not applicable

0.228
0.231
0.231
0.249
0.221

heat transfer engineering

exit/entrance tube length would be submerged. This was

a small factor, but was taken into account in each instance. All coils were made from the same tube size.
THEORY
Central to all of the calculations is the determination
of a true average temperature for the wort. This was
done by volume-weightin g the thermocouple readings
taken at speci c vertical positions (depth ) in the wort
liquidfrom just under the surface (about 1 cm ) to just
above the bottom of the tank (about 1 cm ). Two variations of log-mean driving force D T were calculated.
The classic log-mean D T was calculated from the inlet and outlet cooling water temperatures and the wort
temperatures at the top and bottom of the immersion
coil. The second was a log-mean calculated from the
volume-average wort temperature and the inlet and outlet cooling water temperatures. The agreement in most
cases was close, within a few degrees celsius. In some
cases there was strati cation in the wort temperature
pro le resulting from a layer of hot water on the top
left virtually untouched by the cooling coils due to their
placement in the tank. In this case the weighting did not
take into account this layer for the calculation of lm
coef cient (the second D T log-mean was used ), but it
did for the tank cooling curve.
We calculated the experimental values of lm heat
transfer coef cient (h ) due to natural convection on the
outside of the tubes by rst calculating an overall coef cient (U ) in the usual manner set forth in any standard
heat transfer textsee, for example, Holman [8]:
Q = wC p (Tcw,out Tcw,inlet ) = Uo A o D Tlm

(1 )

where Q is the heat transfer rate, w is the coolingwater ow rate, C p is the heat capacity of water, Tcw,out
is the cooling-water outlet temperature, Tcw,inlet is the
cooling-water inlet temperature, Uo is the overall heat
transfer coef cient based on A o , the outside surface area
of the cooling coils, and D Tlm is the log-mean average
temperature driving force as de ned above.
The de nition of the overall heat transfer coef cient
is well known [3, 8]:
Uo =

1
(2 )
(1 / h o ) + (D r / (kDav / D o )) + (1 / (h i Di / Do ))

where h o is the outside lm coef cient due to natural

convection, D r is the wall thickness of the cooling coil
tube, Do is the outside diameter of the cooling coil tube,
Di is the inside diameter of the cooling coil tube, Dav
is the arithmetic average diameter of the cooling coil
tube, k is the thermal conductivit y of the cooling coil
vol. 21 no. 5 2000

49

tube, and h i is the inside lm coef cient due to turbulent

ow of cooling water. The inside lm coef cient can be
calculated from the well-known Sieder-Tate correlation
[3, 8] as follows:

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

h i = 0.027

()

k
(Re )0.8 (Pr )1 / 3
Di

(3 )

A correction for curvature of the coil is not necessary,

because the bend diameter is rather large compared to
tube diameter and the Reynolds numbers were all in the
fully developed turbulent region, where mixing due to
turbulence is much stronger than currents induced by
curvature of the tube. In Eq. (3 ), Reynolds number, Re,
and Prandtl number, Pr, have the usual de nitions. Cooling water properties were evaluated at the bulk average
temperature. The viscosity correction was not used here,
because we could not measure the wall temperature of
the tube, and it would have been a small effect anyway.
Additionally, this resistance is not the major one, and
neglecting the viscosity-correctio n term would not have
much effect on the nal outcome.
From the U calculated experimentally from Eq. (1 ),
and the h i calculated from Eq. (3 ), Eq. (2 ) can be used to
calculate an experimental value of h o . This can be put in
Nusselt number format and compared to expectations
from general natural-convectio n correlations. For example, for horizontal coils the correlation for Grashof
number in the laminar regime is [1, 3, 8]
Nu = 0.53 (Gr f Pr f )1 / 4

(4 )

and for vertical surfaces under the same conditions ,

Nu = 0.10 (Gr f Pr f )1 / 3

(5 )

where Grashof number has the usual de nition for both

vertical surfaces where the distance measure is L, the
vertical length, and for horizontal tubes, where the distance measure is D, the outside diameter of the tube.
In both cases above, the Grashof and Prandtl numbers
are evaluated at the lm temperature, which is an arithmetic average of bulk uid temperature and wall temperature. We did not measure the outside wall temperature of the tubes, and in this case we cannot ignore it.
Therefore the wall temperature was estimated by subtracting the D T due to the outside lm resistance as
follows:
Twall = Twort (h o / Uo ) D Tav
50

(6 )
heat transfer engineering

RESULTS AND DISCUSSION

Major Results
The major results from the experimentation yielded
lm heat transfer coef cient as a function of time of
cooling, or by cross-plot, the temperature of the wort
being cooled. We compared these values obtained experimentally with values of h calculated by correlation.
Different designs could be evaluated by comparing heat
transfer coef cients at the same time, and heat transfer
enhancement could be traced by how the heat transfer
coef cient compared with standard correlations.
Five basic designs were evaluated. They included a
commercially manufactured, conventional helical cooling coil with xed tube spacing (about 1 D ) by doublestrand wires woven through the coils in an axial direction at four locations around the circumference of
the coil, a similar horizontal cooling coil in a helical
shape with variable spacing of the coils, an asymmetrical horizontal coil in which one side of the helix was
pushed inward so that the cross-sectiona l shape looking down the axis of the coil was more like a C than
an O (width and depth of the cross section are given
in Table 1 ). This pushed the coils to one side in the
tank and was an attempt to create a natural-convectio n
cell in which the uid motion was downward on one
side of the tank and upward on the other. A design with
the tube coiled vertically in hairpin-type shape with the
straight long sections vertical was also tried, and two designs with conically coiled tubes, one with apex up and
the other with apex down, were also investigated . The
conical designs were an effort to eliminate the stacked
tube problem with horizontal, helical coils, in which
tubes below have less heat transfer effectiveness due to
interference of the tubes above, or vice versa depending on the direction of the natural-convectio n currents.
Figure 1 shows a schematic representation of the coil
designs. Figure 2 shows the experimental setup of the
coil in the tank. Figure 3 shows how four of the designs
performed with respect to rate of cooling of the charge
in the tank. The time scale does not start at zero for this
gure, in order to make the differentiation of curves at
later times as clear as possible. Table 2 shows how the
designs compare with respect to thermal resistance or
heat transfer coef cient of the natural-convectio n layer
outside the cooling water tubes calculated at two different temperatures.
From the table one can see that the best designs were
the asymmetrical coil with 1D spacing (actually this
was a bit larger ) and the horizontal coil with 2D spacing. They both gave essentially the same performance
(h of 761 versus 727 at 25C and 1,000 versus 994 at
vol. 21 no. 5 2000

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

Figure 3 Comparative performance of different coil designs measured by rate of cooling.

40C ). The horizontal coil with 1 D spacing was a bit

less ef cient, but still good; but with little or no spacing
between the coils, the performance became much less
good: 761 for 2D, 676 for 1D, and 213 for 0 spacing.
Part of this trend is shown in Figure 3 also. When it
rested on the bottom, the performance was very poor,
almost 5 times worse than the best design (227 versus
1,000 at 40C ).
The commercial helical coil cooler, which has about
a 1D spacing, performed well when the top coil was at
the surface of the wort, but the performance fell somewhat when it rested on the bottom (630 versus 505 at
25C ). Figure 4 shows the comparative data. Here the
time scale has been adjusted also, in order to show differentiation of curves more clearly. An improvement of
almost 20 min (cooling time cut by about one-third )
results when the coil is moved from resting on the bottom to a position where the top coil is at the surface
Table 2 Ef ciency of cooling coil designs based on lm heat
transfer coef cients
Designtop coil mounted
at liquid surface,
h o @ 25C, h o @ 40C,
except where noted
W/m2 K
W/m2 K
Best

Asymmetrical horizontal
coil, 1D spacing
Horizontal coil, 2D spacing
Good
Commercial helical coil
Conical coil, apex down
Horizontal coil, 1D spacing
Mediocre Commercial helical coil,
resting on bottom
Conical coil, apex up
Vertical coil, top at liquid
surface
Worst
Horizontal coil, no spacing
Horizontal coil, no spacing,
resting on bottom

761

1,000

727
630
693
676
505

994
909
892
880
812

545
505

812
750

312

528
227

heat transfer engineering

Figure 4 Cooling rate of the commercial helical coil design as a

function of vertical placement in the tank.

of the wort. This is very signi cant. Mounting the coils

on the bottom, as is currently practiced, is the worst
situation for obtaining ef cient heat transfer in a cooling operation. Mounting the chiller such that the upper
coils are at the surface of the wort is the single most
important modi cation to immersion chiller operation
that we found. This is a point that is not well appreciated by current practice, but stands to reason after some
thought (in fact, there is a recommendation in Perrys
[1] to that effect ), and the data in Table 2 and Figure 4
show this conclusively. We discuss why this might be
so a bit later. This nding is actually contrary to some
literature. In Stuhlbarg [2], for example, there is an explicit statement that raising the coils off the bottom of
the tank in a batch, natural-convectio n cooling situation
did not lead to better performance. In this case paraf n was undergoing freezing and melting cycles in a
tank with natural-convectio n cooling and heating. Perhaps raising the coils off the bottom, which was done
for cleaning purposes, did not place the top coil at the
surface. In any event, our ndings are clear, but every
situation may be somewhat different.
Our data also show a signi cant performance improvement with a controlled separation between the
coils of 1 2D. With the exception of the commercial
unit used here, most commercial coil designs do not
separate the coils and suffer from that disadvantage.
There was a layer of hot liquid at the surface (strati cation ) in all cases with the chiller coils resting on the
bottom of the tank, even at the end of the cooling cycle.
It is possible to improve this situation by stretching the
coils so that they reach from top to bottom. However,
this would require spacers to keep them there.
Figure 5 shows the temperature pro les as a function of time during the cooling cycle of one of the runs
with the commercial helical coil. It is dif cult to infer
the ow patterns from the temperature pro le alone, but
there appears to be a strati cation zone in the top layer
vol. 21 no. 5 2000

51

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

Figure 5 Time temperature position pro le of liquid being

cooled in the tank. Top thermocouple (T1) placement is 5 cm below
liquid level and 23.36 cm from the tank bottom; bottom thermocouple (T5) placement is 0.32 cm above tank bottom. Other thermocouple positions indicated are distance from the tank bottom.

(the top layer was signi cantly hotter than the rest of
the liquid in the tank after substantial time of cooling ).
In the case of the horizontal coils with no spacers, this
zone was even more marked. In Figure 5 the hot zone is
about 5 cm deep, and there appears to be some mixing
from below, as the top layer does show some significant cooling. In visual observation, Schlieren waves
[1, 3, 8] were observed hitting the bottom of the tank
and radiating out from there. Some waves moved substantially upward, mixing with the hot, upper liquid.
However, the wort seems to cool primarily from below,
with some mixing between the hot region on top of the
cold region. The cooled uid near the coils drops to
the bottom of the tank and spreads radially to form a
cool disk, which then expands upward as more cooled
liquid ows into it. There seems to be an intermediate
region between the two that is the result of some mixing, and there may be some small, local recirculation
currents. It would be quite interesting to do a ow simulation or dye experiments with these currents to verify
the transient temperature pro les of Figure 5, and this is
planned for future work. This mechanism explains why
there is strati cation when the top coil is not placed at
the liquid surface.
The conical design was an attempt to eliminate the
upper coils from interfering with heat transfer of the
lower coils. As is evident from Table 2, the apex-down
design worked better than the apex-up con guration.
This has partly to do with the above-mentioned mechanism. With the apex up, there is a toroidal or doughnutshaped zone of hot liquid at the top, between the upper
coil and the tank wall, that is not in contact with the
52

heat transfer engineering

cooling coil. There is more heat transfer area at the top

in the case with the apex down (the large-diameter coil is
at the top, and hence there is no strati cation zone ) and
consequently more rapid cooling at that point because
of the larger temperature difference between hot liquid
in the tank and the water coolant temperature inside the
coils. Thus, the overall cooling rate is faster.
The vertical coil design, which was an attempt to encourage stronger natural-convectio n currents, did not
perform particularly well. We believe the lower heat
transfer coef cients of the vertical coil as opposed to the
horizontal situation was the major cause, and apparently
this was not overcome by faster-moving currents of the
liquid being cooled. The performance of the vertical
tube design could probably be improved by adding longitudina l ns to the straight portions of the tubes. This
would not interfere with the natural-convectio n currents
and would provide more surface area to increase heat
transfer. We did not test this because of fabrication dif culties, but this basic design is commercially available
in other applications and may be useful for this one,
although the cost may be prohibitive .
Fit to Correlation s
The data for every run were compared to correlations used to predict lm heat transfer coef cient. This
is illustrated in Figure 6 for the commercial cooler, for
example. The agreement between measured and predicted values agreed very well over the whole range
of data in all but two casesthe vertical coil and the
horizontal coil without spacers.
For the vertical coil, performance exceeded predictions based on comparison of lm heat transfer coef cients. One possible explanation is that about one-third
of the total coil length is not vertical, but rather horizontal (in the loop sections ). These have higher heat

Figure 6 Comparison of experimentally calculated outside lm

coef cient of the cooling tube and that predicted by correlation for
horizontal coil design, 1D spacing.

vol. 21 no. 5 2000

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

Figure 7 Effect of tube spacing in helical coil designs for naturalconvection cooling.

transfer coef cients and would skew the data upward.

Additionally, the asymmetrical arrangement of the coils
in the tank could also have enhanced the heat transfer
by channeling natural-convectio n currents.
For horizontal coils with no spacing, the experimental results were well below the predicted values. This
is to be expected, because the coils interact and reduce
the heat transfer if they are too close to one another.
We reduced all the data taken with horizontally coiled
tubes with different coil spacing in an attempt to correlate the spacing function (see Figure 7 ). We found that
it was related to the Nusselt number by the 14 power of
the spacing to diameter ratio, and could be expressed as
follows:
Nu = 0.53

Gr f Pr f
Sp / D

0.25

for 0 < S p / D 2 (7 )

Reduction in heat transfer occurs until the single-tube

situation is reached at about S p / D 1.5.
Sources of Error
Two major sources of error were differences and uctuations in the cooling water temperature, and the time
it took to obtain the thermocouple readings. The cooling water temperature uctuated about 0.5C during
the run and did not always start at exactly the same
place. Inlet temperatures varied mostly between 10 and
12C, but one run was taken at 15C. We compensated
for uctuations by taking the correct time averages, but
comparisons with different coils had cooling water temperatures that were not exactly the same. The variation
was about 1C; this would affect the temperature driving force and the heat transfer results. Overall the error
in the heat transfer coef cient results is about 12%,
and this is indicated by error bars in Figure 5.
heat transfer engineering

Because the full recording of temperature from thermocouple readings took about 20 s, not all the temperatures were taken at exactly the same times. This is not a
problem when the time between readings is 5 10 min,
but clearly affects the results when much shorter times
are used. Therefore, the data at the very beginning of
each run are not as reliable as those later. Thus, the error in the rst 5 min is somewhat higher, about 20%
in h.
The tank was uninsulated and therefore lost heat from
natural convection to the air as well. This, however, is
so small as to be negligible, and we did not include it
in our calculations. Also, there is a point in the cooling
cycle when the wort temperature is below the ambient air temperature, and heat transfer to the tank from
this mechanism would be taking place. The coef cients
from water to air are in the ratio of about 100/1, and the
heat transfer surface areas are comparable, so that less
than 2% error is estimated from this source.
CONCLUSIONS
1. The most signi cant design parameter for enhanced heat transfer in batch cooling by immersed coils is
coil placement. The top coil should be at the surface
of the liquid being cooled. The opposite should also
be true. That is, for batch heating, the coils should be
placed at the bottom, or slightly above. These placements enhance natural-convectio n currents and hence
give better heat transfer.
2. Spacing in the coils of at least 1D is highly desirable, and 2D spacing is superior still. When coils are
too closely spaced, natural-convection currents are inhibited and heat transfer ef ciency falls. At about 1.5 D
spacing and higher, h is equivalent to the single-tube
value.
3. Asymmetrical coil con guration helps heat transfer ef ciency by channeling natural-convectio n currents
from one side of the tank to the other. This effect is
weaker than the rst two above.
4. Our results show that cooling in the tank is done
by complex natural-convectio n currents that give a mild
degree of mixing. Cooled liquid appears rst at the outside of the cooling coils; this liquid falls to the bottom,
pools, and works its way upward as its mass is increased
by further cooled water dropping from the cooling coils.
At the beginning of cooling only, some of the cooled
liquid rises in jets after hitting the bottom (when it has
enough velocity ) and does the mixing. This mechanism
explains why there is a hot strati cation layer at the top
if the top cooling coil is not near the surface. It also explains why a conical coil is better with the apex down
rather than up.
vol. 21 no. 5 2000

53

NOMENCLATURE
A
Cp
D
Dav
g
Gr

Downloaded by [University of Texas Libraries] at 09:21 10 December 2014

h
k
L
Nu
Pr
Q
r
D r
Re
T
D T
D Tlm
U
v
w
b
l

heat transfer area, m2 or ft2

heat capacity, kJ/kg K or Btu/lb F
diameter, m or ft
average diameter of tube wall [ = ( Di + Do )/ 2]
gravitationa l constant, 9.81 m/s2 or 32.17 ft/s2
Grashof number ( = Do3 q 2 gb D T / l 2 ), dimensionless; D replaced by L for vertical surfaces
lm heat transfer coef cient, W/m2 K or Btu/
hr ft2 F
thermal conductivit y, W/m K or Btu/hr ft F
length of vertical surface, m or ft
Nusselt number ( = hD / k ), dimensionles s
Prandtl number ( = l C p / k ), dimensionless
heat transfer rate, W or Btu/hr
radius (generally of tube ), m or ft
tube wall thickness, m or ft
tube Reynolds number ( = Di v q / l ), dimension less
temperature, K or C, or F
temperature difference, C or F
log-mean temperature difference, C or F
overall heat transfer coef cient, W/m2 K or
Btu/hr ft2 F
velocity of cooling water inside tube, m/s or
ft/s
mass ow rate of liquid, kg/s or lb/s
thermal expansivity, 1/K, or 1/R
viscosity, mPa s or lb/ft s
density, kg/m3 or lb/ft 3

Subscripts
av
cw
f
i

54

average, generally arithmetic

cooling water (temperature, generally )
properties evaluated at lm temperature ( Tw +
Tbulk ) / 2
inside (of tube, generally )

heat transfer engineering

lm
o
w

log-mean, standard de nition

outside (of tube, generally )
wall (of tube, generally )

REFERENCES
[1] Green, D. W., and Maloney, J. O. (eds.), Perrys Chemical En[2]
[3]
[4]
[5]
[6]
[7]
[8]

gineers Handbook, 6th ed., sec. 11-29 to -31, McGraw-Hill,

New York, 1984.
Stuhlbarg, D., How to Design Tank Heating Coils, Petrol. Re ner, vol. 38, p. 143, April 1959.
McCabe, W. L., Smith, J. C., and Harriott, P., Unit Operations of
Chemical Engineering, 4th ed., McGraw-Hill, New York, 1985.
Strauss, K. M., Wort Cooling, in H. M. Broderick (ed.), The
Practical Brewer, pp. 147 159, Master Brewers Association of
the Americas, Madison, WI, 1977.
Hough, J. S., Briggs, D. E., Stevens, F., and Young, T. W., Malting and Brewing Science, 2nd ed., vol. 2, Chapman & Hall,
New York, 1982.
DeLange, A. J., How to Gain Maximum Control over Wort
Chilling Operations, Brewing Techniques, pp. 42 53, July/
August 1996.
McConnell, D., and Schramm, K., CoilsImmersion Chiller
Road Test, Zymurgy, pp. 54 60, Fall 1996.
Holman, J. P., Heat Transfer, 8th ed., McGraw-Hill, New York,
1997.

Donald D. Joye is professor of chemical engineering at Villanova University, having had several
years of experience both in industry and teaching
at other universities. He holds degrees in chemical engineering from Princeton University (1967)
and Lehigh University (1969, Ph.D. 1972). His research interests in heat transfer are in curved rectangular channels and vertical, mixed-convectio n
ows. He is a registered patent agent and has additional research interests in
mass transfer, rheology, and the thermo- uids aspects of breathing mechanics in deep diving.
Michael A. Smith received his M.S. degree in
Chemical Engineering at Villanova University. He
is a 1980 graduate of Lafayette College and is
pursuing the Ph.D. in chemical engineering at
the University of Delaware. Before Villanova, he
spent 17 years at DuPont in a variety of process
engineering manufacturing and development assignments. He has been active in the American
Homebrewers Association for 14 years.

vol. 21 no. 5 2000