System Calculation

APPENDIX
EXAMPLE CALCULATIONS FOR

MASS TRANSFER EFFECTS
J. J. Perona
Consider a direct contact boiler in which isobutane is vaporized by contact with

hot brine. Estimate the concentration of isobutane dissolved in the effluent brine.
Let us choose boiler conditions similar to those of the 500 kw geothermal
pilot plant at East Mesa:
Brine flow rate 97,200 lb/hr
Inlet brine temperature 340 F
Boiler pressure 467 psia
Isobutane flow rate 99,400 lb/hr
The rate of dissolution of isobutane into the brine is given by a material bal-
ance over the boiler:
R = B(XI - Xo) (1)
where R = rate dissolved, lb/hr
253
264 DffiECT-CONTACT HEAT TRANSFER
B = brine flow rate, lb/hr

Xl = mass fraction of isobutane in brine phase leaving the boiler
Xo = mass fraction of isobutane in brine entering the boiler
Of course, Xo is zero. Observation of experimental direct contact boilers indicates

that good mixing takes place. Let us assume that the boiler is perfectly mixed;
that is, the temperature and compositions of the bulk brine and isobutane phases
are uniform throughout the boiler, and therefore are the same as the exit streams.
With the assumption of perfect mixing, the rate of transfer of isobutane to
the brine phase may be written in terms of the mass transfer coefficient as follows:
R = (k.. a) V (X" - Xl) (2)
where k"a = the volumetric mass transfer coefficient
for the brine phase, lbs/(hr)(cu. ft of boiler volume)
X" = equilibrium concentration of isobutane in brine, mass fraction
V = boiler volume, cu. ft.
The isobutane phase is essentially pure isobutane, and there is no resistance to

mass transfer in the isobutane phase.
Equations (1) and (2) may be combined by eliminating R, and solving for Xl
(k.. a) (V)(x")
Xl = (k.. a) (V) +B (3)
Equation (3) provides the answer for the example, but it requires a value for
a mass transfer coefficient in a direct contact boiler. No experimental correlations
are available for this; therefore, we will make an estimate from experimental heat
transfer coefficients. Thus, from the Colburn (or j-factor) analogy between heat
and mass transfer
k a =1-
.. 0,
[N-Ns f/3 U.
pr
(4)
where U. = volumetric heat transfer coefficient, Btu/hr ft3 F.

0
A well-accepted general correlation for predicting values of U. is not yet

available. Values for UlI for the isobutane-water boiler as East Mesa are reported
by Lawrence Berkeley Laboratory to be in the 30,000 to 40,000 range.
The saturation temperature is 250 F and the bulk brine temperature is
slightly higher, at about 260 F. Under these conditions the following property
values are found:
NPr = 1.4
Ns = 45
X" = 360 ppm
APPENDIX 1 255
Using a value for U. of 35,000 Btu/hr ft8 F, we obtain

k"a = 3560 Ib/hrft 3
and V = 15 ft 3
Finally, Xl = 130 ppm
APPENDIX
Am/MOLTEN SALT DffiECT-CONTACT

HEAT TRANSFER ANALYSIS
Mark S. Bohn
NOMENCLATURE
a interfacial surface area per unit volume, m-I (ft-I )
A" finned-tube heat exchanger surface area, m2 (ft~
AO levelized annual cost, S/yr
o specific heat, J/kg C (Btu/Ibm F)
0 0
0, parameter in HI. correlation

0, plant capacity factor
0" specific heat at constant pressure, J/kg C (Btu/Ibm F); 2.394 X 10-4
0
OJ capital cost, $
Originally prepared under Task No. 4250-00, WPA No. 431-83, for the U.S. Department of
Energy Contract No. DE AC02 83CH10093, SERIfTR-252-2015, UC Category: 59c, DE 84000080,
November, 1983.
257
258 DIRECT-CONTACT HEAT TRANSFER
D mass diffusivity, m2/h (ft2/h); 10.76

d, packing size, em (in.)
d, column diameter, m (ft)
11 viscosity function
12 density function
13 surface tension function
G gas-flow rate per unit bed area, kg/h m2 (lbm/h ft~; 0.2044
H column height or finned-tube exchanger height, m (ft)
Htl height of a transfer unit, m (ftl
h heat transfer coefficient, W /m C (Btu/h ft2 F); 0.1761
0 0
k thermal conductivity, W /m C (Btu/h ft F); 0.5777

0 0
kg gas side mass-transfer coefficient, kg mol/h m2 atm (lb mol/h ft2 atm);
0.2044
L liquid flow rate per unit bed area, kg/h m2 (lbm/h ft~; 0.2044
L length of finned-tube heat exchanger in flow direction, m (ft)
m mass flow rate, kg/h (Ibm/h)
m, n exponents in Htl correlation
OM operating cost, $/yr
P total pressure, atm
PBM logarithmic mean partial pressure of component B, atm
Pr Prandtl number
Q heat transfer or heat duty, W (Btu/h); 3.412
R universal gas constant, mS atm/ o C kg mol (ftS atm/ R lb mol); 8.918
0
Be Schmidt number
T absolute temperature, 0C (0 R)
Ua overall volumetric heat transfer coefficient, W /ms C (Btu/h ftS F);
0 0
0.05368
V flow velocity, m/h (ft/h)
V, volume of packing bed, m8 (ftS)
W finned-tube heat exchanger width, m (ft)
Greek
c heat transfer area per volume, mol (ft-l)
.1p heat exchanger pressure drop, N/m2 (psi); 1.451 x 10-4
.1 Tm log mean temperature difference, 0C (0 F)
p absolute viscosity, N h/m2 (lbm/ft h); 8.690 x 106
v kinematic viscosity, m /s (ft2/s); 10.76
p density, kg/ms (Ibm/ftS); 0.0623
(J' surface tension, dyne/em
tP, 'If; parameters in Htl correlation
APPENDIX 2 269
Subscripts
a aIr
9 gas
i inlet
0 outlet
8 salt
l liquid
NOTE ON SYSTEM OF UNITS

Since a majority of the chemical engineering literature, especially product litera-
ture, continues to use the English system of units, we have not attempted to con-
vert such information into the SI system of units. Any data generated in this
study, however, is presented in the SI system. To facilitate conversion between
the two systems, the factor for converting SI units to English units follows the
more important quantities in the preceding list. For example, Ua = WOO W fm 3
o C = 53.68 Btufh ft 30F for a volumetric heat transfer coefficient.
1 INTRODUCTION
Direct-contact heat transfer is the transfer of heat across the phase boundary of
two immiscible fluids, either two liquids or a liquid and a gas. Conventional heat
exchange technology involves heat transfer across a solid boundary such as the
wall of a steel tube in a shell-and-tube exchanger or across a plate in plate heat
exchangers. Where the two fluids do not react and can be separated after the heat
exchange has been effected, direct-contact heat exchange (DCHX) has several
advantages over conventional heat exchangers: (1) without the intervening wall, a
lower thermal resistance is present, and there is not heat exchange surface to foul;
(2) intimate mixing of the two fluid streams can produce very high rates of heat
transfer; and (3) the heat exchanger design can be simpler, require fewer materials
of construction, and allow more flexibility in choice of materials.
Figure 1.1 depicts a conventional finned-tube heat exchanger, and Fig. 1.2
shows a DCHX with a packed column. In the finned-tube heat exchanger one
fluid is pumped through the tubes, and the other fluid is pumped over the outside
of the tubes. The entire tube bundle is enclosed in the shell, which contains the
latter fluid. Sufficient heat transfer area is provided so, given the temperature
differential between the fluids and the heat transfer coefficients, the required heat
duty can be met. IT one or both of the fluids is a gas, it is common to provide fins
on the gas side to increase the heat transfer surface because of the poor heat
transfer characteristics of gases. The DCHX is a column substantially filled with a
packing material. The packing material consists of rings or saddles (Fig. 1.3) that
are generally 2-3 in. in size for large columns and are dumped in the column in a
random arrangement. As shown in Fig. 1.2, one fluid enters the top of the vessel
and flows countercurrent up through the vessel. It is also possible to have a
crossflow configuration.
260 nffiECT-CONTACT HEAT TRANSFER
Air Flow
Figure 1.1 Conventional finned-tube heat exchanger.

When the DCHX uses a gas and a liquid, the liquid flows downward by
gravity and the gas flows upward in the countercurrent configuration. By prop-
erly distributing the liquid at the top of the packing, the liquid forms many small
rivulets that flow over the packing. These rivulets give a large surface area
between the two phases and increase the time during which the liquid stream is
exposed to the gas, greatly increasing the rate of heat transfer per unit volume of
heat exchanger. We can eliminate the packing by simply spraying the liquid
downward and having the gas flow upward in an empty column (i.e., a spray
column). Although we did not study the spray column, results of the economic
analysis (Section 4) indicate that it could offer some advantages over the packed
bed at very high temperatures and, therefore, it warrants further investigation.
Applications in which DCHX is especially attractive include those in which
it is necessary to transfer heat between a gas and a liquid because large heat
transfer rates can be achieved without the added expense of finned tubes. In solar
thermal technology, two examples include high-temperature process air and the
Brayton cycle (shown in Fig. 1.4). In both examples, solar energy provides a heat
source at a central receiver in which molten salt cools the receiver and transfers
the solar energy to a storage device. Molten salt is the logical heat transfer fluid
at high temperatures because it exhibits very low vapor pressure, has high sensible
heat storage, has excellent heat transfer characteristics, and is relatively benign in
relation to the receiver containment materials (at temperatures below 600 C for
state-of-the-art nitrate salts).
APPENDIX 2 2M
Air
Wire Mesh Mist

Eliminator
1~~:::;:;:kJS~~~~-salt Inlet
Column Packing
Packing
Support Plate
Figure 1.2 Direct-contact heat exchanger.

There are two configurations for storing the energy in the molten salt to
accommodate the diurnal variation in energy supply or variations in the load. In
the first configuration, the molten salt is stored in an insulated vessel providing
storage by the sensible heat in the salt. In the second, the salt is used to heat air,
which in turn heats a rock bed to provide sensible heat storage. The first method
will be necessary for very high temperature applications in which the only
material that can tolerate the temperature cycling in the storage is the salt itself.
The second storage method is attractive for lower temperature applications
600 C) because it provides economical, long-term storage. In either storage
0
concept it is necessary to transfer heat from the molten salt to air.

Using packed columns is very common in the chemical process industry for
mass-transfer operations. One example is the removal of carbon dioxide from a
gas stream by contacting the gas stream with monoethanolamine (an organic
liquid) or with a hot carbonate solution. The gas is blown up through the bottom
of the column, and the monoethanolamine or carbonate solution enters through
the top of the column and is distributed over the packing. As a result of this com-
mon application and other similar applications, numerous data and design correla-
tions are available for mass-transfer applications, but very little of this information
is available for heat transfer applications.
262 DIRECT CONTACT HEAT TRANSFER
Metal
Pall Ring
Metal
Rasch ig
Ring
Ceramic
Intalox
Saddle
Figure 1.3 Three types of packing for a direct-contact heat exchanger.

APPENDIX 2 263
Molten Salt
Receiver
/
//
///1/
/ / / /
/ / / /
/ / I /
// / / /
Salt Salt Air to Process or

b.,..,'7n..F-" Brayton Cycle
Salt
Dlrect-
Con tact
Mol ten Salt Heat
Receiver Exchanger
t:t..,..,......JL-...:~_--4=----::::I--_ Air to Process
Hea t
Exchanger
Figure 1.4 Applications of direct-contact heat exchangers.

Because of this lack of heat transfer data or design correlations, we cannot
accurately assess the economic potential of direct-contact heat exchange. Such an
assessment requires one to determine the rate of heat transfer per unit volume in
the DCHX. This determines the required size and cost of the column to deliver
the required amount of heat to the air. It also helps one in determining the costs
associated with operating the equipment, primarily the cost of blowing the air
through the column.
It is possible to use mass-transfer data by invoking the mass-transfer/heat

transfer analogy (Fair, 1972). However, there are several reasons to suspect this
approach. Some mechanisms of heat transfer have no analogy to mass transfer.
Fair's mass-transfer data and correlations are generally for experiments on
water/carbon dioxide systems or water/sodium hydroxide systems. The wetting
of the packing by the molten salt probably differs from that of water, and this
affects how much interfacial surface area is created by the flow down the packing.
Heat may be transferred by conduction in the packing, thereby transferring heat
from the dry parts of the packing to the air. This fin effect has no analogy in
mass transfer. At high temperatures, radiation heat transfer may be significant,
and this mechanism also has no mass-transfer analogy. One may conclude that
the calculations of heat transfer based on the mass-transfer analogy as given in
Fair (1972) may underestimate the heat transfer coefficients.
The objective of the present work is threefold: (1) to experimentally deter-
mine the heat transfer coefficients in direct-contact heat exchange between molten
salt and air, (2) to calculate these heat transfer coefficients based on the mass-
transfer analogy and compare them with the experimental data, and (3) to analyze
the economics of this system by using the experimental data and comparing
DCHX with conventional finned-tube heat exchangers. In general, we want to
determine if, and in what applications, DCHX is a cost-effective technology.
In the following sections, we describe calculations of the heat transfer
coefficient based on the mass-transfer analogy; describe the experimental
apparatus, methods, and results; compare the results with the calculated values;
and, finally, describe an economic analysis that compares the cost effectiveness of
DCHX and finned-tube heat exchangers in several applications.
2 THE HEAT-TRANSFER/MASS-TRANSFER ANALOGY

The dimensionless heat transfer coefficient (Stanton number) may be related to
the dimensionless mass-transfer coefficient (Sherwood number) (Kreith, 1976) by
_h_ _ kgRTPBM
(2.1)
GO, lIP
which is based on the Reynolds analogy and holds only if the Prandtl number (Pr)
and the Schmidt number (Se) both equal unity as shown here:
pO,
Pr = -k- = 1 = -;D = Se .
Il
(2.2)
In the case where Eq. 2.2 does not hold, heat transfer can often be related to mass
transfer by
_h_ Pr2/3 = kgRTPBM Sc2/3
(2.3)
GO, VP
from which we can calculate the heat transfer coefficient
APPENDIX 2 2116
A= [;J -.--_G_O.::..,---:- (2.4)
[kgR:BM 1
For a packed column, transfer coefficients are commonly presented as volumetric
coefficients by multiplying the surface coefficients by the interfacial surface area
per unit volume a
ha = [ Sc JfJ GO, . (2.5)
Pr [ kgaRV:PBM 1
The denominator in parentheses in Eq. 2.5 has dimensions of length and is called
the height of a transfer unit Hd Correlations of mass transfer are often expressed
in terms of Hd Fair (1972) gives such a correlation expression for Hd for packed
columns
,pSc;f2 dr
Hg,d = (Lld2/s)m ' (2.6)
and
(2.7)
for the gas side and liquid side, respectively. Note that Eqs. 2.6 and 2.7 are
dimensional. Dimensions of Hg,d are in ft, L is in lbm/h ft2, and de is in ft. We
can then calculate the gas side and liquid side heat transfer coefficients from
h a -[ - -
SCg 12fJ -O,G
- (2.8)
g - Prg Hg,d
hla = [sc l 1fJ

2
OlL . (2.9)
Prl Hl,d
The overall volumetric heat transfer coefficient is then calculated from
Ua = [_1_ + -1-fl
hga hla
(2.10)
The value of the parameters m, n, W, t/J, and 0, in Eqs. 2.8 and 2.9 may be found
in Table 2.1. The parameters It. 12, and Is are functions of the liquid viscosity,
density, and surface tension, respectively, and are defined as
II = pt l8 (2.11)
12 = pr1.26 (2.12)
Is = (ITt/12.8)-{)8 . (2.13)
Dimensions are: Pl ("'Cp), Pl (~/cm8), ITl (--dyne/em).
266 DmECT-CONTACT HEAT TRANSFER
Table 2.1 Parameters for packed-column heat and mass transfer

ij.aschig Rings Berl Saddles
I-m. ~m. I-in. 2-in.
m 0.6 0.6 0.5 0.5

n 1.24 1.24 1.11 1.11
I/t
40% flood b 110 210 60 95
60% flood 105 210 60 95
c
80% flood 80 -
~
L =2450 0.045 0.059 0.032
L -4900 0.048 0.065 0.040
L =24,500 0.048 0.090 0.068
L - 49,000 0.082 0.110 0.090
0,
<50% flood 1.00 1.00 1.00 1.00
60% flood 0.90 0.90 0.90 0.90
80% flood 0.60 0.60 0.60 0.60
~air 1972.
bFlooding is defined 118 a column pressure drop of 1.5 in.
(water column) per foot of bed height.
CData not available or extrapolated.
Experimental conditions can then be used to calculate the heat transfer

coefficient based on the heat transfer/mass-transfer analogy. The air and salt inlet
and outlet temperatures determine all property values. The air and salt flow rate
and column diameter determine G and L for Eqs. 2.8 and 2.9, respectively. For
the experimental apparatus it will be necessary to use Table 2.1 data for the I-in.
Raschig ring packing (because there is not data for O.5-in. rings) even though the
experiment used O.5-in. Raschig rings. Extrapolation from the I-in. to O.5-in. rings
from the I-in. and 2-in. ring data would be nothing more than guesswork. How-
ever, note that I/t nearly halved in changing from I-in. to O.5-in. rings. This, in
turn, increases the gas side heat transfer coefficient. Thus, using I/t for the I-in.
rings should give conservative values of heat transfer coefficients. We also need to
estimate the percentage of flooding to determine ~ and 0,. Constant values of tf;
= 110, tP = 0.048, and 0, = 1 were used. We determined the property values as
discussed in section 7. Figure 2.1 gives results for air and salt flows and tempera-
tures typical of the experimental apparatus.
The film coefficient on the gas side is much smaller than that on the liquid
side, and, therefore, the overall coefficient Ua very nearly equals h(la. From Eqs.
2.6 and 2.8 we see that
Ua '" mIl m,O.6 . (2.14)
It will be useful to compare the exponents in Eq. 2.14 with the experimental
APPENDIX 2 267
4000r---------------~--------~
3000
-
<.:>
..,0
E
...... 2000
~
til
:::l
1000
O~ ______ ~ ______ ~ ________ ~
30 40 50
rna (kg/h)
Figure 2.1 Overall volumetric heat transfer coefficients based on mass-transfer
data.
results. In addition to comparing the relative magnitudes of the calculated and
experimental results, comparing the trends is important to determine whether the
transfer phenomena are actually equivalent.
The range of experimental parameters (air flow and salt flow) is restricted
(see Section 3.4). Within this range we should expect heat transfer coefficients Ua
from about 1000 to 3250 W 1m 3 C.
3 EXPERIMENTAL MEASUREMENTS OF VOLUMETRIC HEAT

TRANSFER COEFFICIENTS
3.1 Purpose
As previously mentioned, there are uncertainties associated with calculating the
heat transfer coefficients from mass-transfer data. In addition, there are mechan-
isms of heat transfer for which no mass-transfer data exist. Therefore, we have
developed an experimental program to determine actual volumetric heat transfer
coefficients in direct-contact heat exchange between air and molten salt. Using
these data to determine the economic value of DCHX should give us more
confidence in the results than if only the calculated heat transfer values were used.
Upper Storage
Tank -
Salt
Valve
Air Flow Meter
r" ~~=::[==J::==== "-Compressor

From
Preheater
Bu bbler-~+-+-lL~:"
Figure 3.1 Flow diagram of DCHX test loop.

3.2 Description of the Apparatus
A flow diagram of the experimental apparatus is given in Fig. 3.1, a detailed
diagram of the packed column is shown in Fig. 3.2, and a photograph of the
apparatus is shown in Fig. 3.3. The test loop is a batch operation with regulated
air pressure on the upper tank providing regulated salt flow through the salt valve.
The upper tank is filled with molten salt and pressurized to approximately 50 kPa.
In this way, the salt flow is affected minimally by loss of salt head in the upper
tank. The salt flows through the salt valve into the top of the column and into a
salt distributor (a can with three holes in its bottom) that distributes the salt uni-
formly over the top of the bed. Salt flows from the salt inlet pipe directly into the
distributor where it flows out the three holes. The distributor is slightly smaller
than the column inside diameter, allowing air to flow in the resulting annulus.
The packing bed is supported by a gas-injection support plate that allows
the salt to flow downward while providing a uniform air distribution at the bot-
tom of the packing. After the air passes up through the bed, it flows around the
annular gap between the salt distributor and the column inside diameter. The air
then flows through a wire-mesh mist eliminator that removes any small salt dro-
plets present before the air flows out of the column. Salt flowing out of the bot-
tom of the bed is collected at the bottom of the column and flows to the lower salt
tank.
APPENDIX 2 260
Salt Inlet
Thermocouple
Salt Inlet
----;~II
Pipe Air Outlet
Thermocouple
Air Outlet ~-- Salt Distributor

Thermocouple
6-in. Schedule 40
Pipe
-~iIIWl
T Ali Materials
304 Stainless Steel
3 ft . 7 ft.
1122'i1!i'lM8-t--- Packed Bed
Support -~EI!Z~
Plate Air Inlet
Thermocouple
Salt Outlet
Thermocouple
Figure 3.2 Details of the DCHX packed volume.

The column size used in this experiment is typical of pilot-scale studies. We
determined the column design (size, distributors, bed height, packing size) with the
assistance of Norton Chemical Company, Rolling Meadows, TIL The entire test
loop, with the exception of the salt valve, is constructed of 304 stainless steel. The
salt valve is 316 stainless steel.
We tested two types of commercially available packing: stainless Raschig
rings, and stainless Pall rings (see Fig. 1.2). Data on the Raschig rings were useful
because of the large amount of mass-transfer data available for them. Supplemen-
tal data on the Pall rings were taken because the Norton Chemical Company
determined that Pall rings would be the most effective packing for heat transfer
duty. For proper liquid distribution, a general guideline is to use a packing size
about 1/10 of the column diameter. The Raschig rings we tested were 0.5 in., and
the Pall rings were 0.6 in. (both the height and diameter of the packing).
Air supplied at the bottom of the column is preheated by a 9-kW electric
preheater powered by a silicon-controlled rectifier (SCR) power supply. A
proportional-integral process controller supplies the control signal to the SCR
power supply based on the desired air temperature and the measured air tempera-
ture at the preheater outlet. A two-cylinder, 10-hp compressor supplies air to the
preheater.
Figure 3.3 Photograph of the DCHX test apparatus.

Two heat transfer salts were tested in the apparatus. Both were supplied by
the Park Chemical Company, Detroit, Mich. The first salt tested is called Parth-
erm 430, which has a melting point of 222 C (430 F). The nominal molar com-
0 0
position of the salt is 43% potassium nitrate and 57% sodium nitrate. Although
the melting point is listed as 222 C by the manufacturer, most of the salt melts in
0
the 240 -305 range, according to differential scanning calorimetry tests. Since
0
0
this high melting point caused considerable problems in operation (see Section 3.5),
a salt with a lower melting point was used for most of the tests. This salt was
Partherm 290 with a molar composition of 40% sodium nitrite, 52% potassium
nitrate, and 8% sodium nitrate. The melting point of Partherm 290 is listed as
143 C (290 F) by the manufacturer. One would assume that considerable melt-
0
0
ing does not occur until approximately 200 C for this salt.
0
3.3 Instrumentation
Primary instrumentation is shown in Figs. 3.1 and 3.2. Air flow rate is measured
by an inline mass flow transducer manufactured by Datametrics, Inc.
A bubbler system determines the salt flow rate by continuously monitoring
the level of salt in the lower salt tank. This system consists of a tube that passes
APPENDIX 2 271
through the top of the tank to within a few centimeters of the tank bottom and a
similar tube short enough so its end is always above the salt surface. By measur-
ing the pressure required to force a bubble of air out the bottom of the long tube,
one can determine the salt depth because it determines the pressure head at the
end of the long tube. H the lower tank is pressurized, this pressure is sensed by
the short tube and subtracted from the pressure at the long tube. The major
advantage of this type of system is that molten salt does not come into contact
with any parts of the flow measuring system. The output of the bubbler is pro-
portional to the salt depth, and differentiating this output gives the salt flow rate.
The lower tank was calibrated by filling it with water in 5-1 increments and
recording the bubbler voltage output. In this way we could directly measure the
liquid flow rate because mass flow rate determined by this type of measurement is
independent of liquid density. When attempting to measure the salt depth, one
must be sure that the salt is completely molten and free of entrained air bubbles.
All thermocouples were Chromel-Alumel (type K). A probe inserted into
the vertical portion of the pipe from the upper tank measured the salt inlet tem-
perature (see Fig. 3.2). The probe should be an accurate measure of salt inlet
temperature since it is totally immersed in salt just before it flows into the salt dis-
tributor. This temperature was typically within 2 C of the upper-tank salt tem-
perature.
A probe inserted in the pipe leading out of the bottom of the column meas-
ured the salt outlet temperature. We inserted the probe just to where the cone at
the bottom of the column begins to expand. The probe was exposed to rivulets
dripping from the packing support plate and is the best compromise for measuring
salt outlet temperature. Constraints on this measurement include the trace heat-
ing on the column wall, which could affect the temperature of salt flowing along
the wall, and air entering the column at a lower temperature than the salt leaving
the column, which could reduce the outlet salt temperature reading if the probe
were inserted further into the column. We observed the responses of this probe to
sudden changes in the salt flow, air flow, and air inlet temperature and determined
that the probe gives a good indication of salt outlet temperature.
A probe inserted into the horizontal portion of the air inlet pipe measured
the air inlet temperature, sensing the temperature of the air about 20 cm from the
column. A probe inserted through the column just below the mist eliminator
measured the air outlet temperature. Secondary measurements included upper-
tank salt and surface temperatures, lower-tank surface temperature, column sur-
face temperature, bed temperature, and the pressure differential between the
preheater outlet pipe and the column air outlet pipe.
Operation of the primary thermocouples was checked by placing the probes
in condensing steam at local atmospheric pressure (622 mm Hg) corresponding to a
saturation temperature of 94.3 C. The five primary probes read 94.4 , 94.7 ,
95.1 , 94.2 and 95.1 for the salt in, salt out, air in, air out (column), and air
0
,
out (outlet pipe) temperatures, respectively.

Readings from the outlet air probe indicated a close approach (",10 C) to
the salt inlet temperature. This implies that either the heat exchanger is very
effective or that some of the salt flowing in the top of the bed is carried up beyond
the distributor and contacts the probe causing it to read too high. Any salt
trapped in the mist eliminator could also drip onto this probe. To resolve this we
inserted a second air outlet probe through the air outlet pipe to just inside the
mist eliminator (see Fig. 3.2). This probe read about 7 C lower than the probe
near the mist eliminator. Pulling the probe out about 2 cm so it was not inside
the mist eliminator increased the difference to about 14 C. With the probe pulled
out to about 20 cm from the mist eliminator the discrepancy was about 50 C.
Therefore, a large temperature gradient exists in the air outlet port area, and it is
difficult to measure the air outlet temperature. For other reasons (see Section 3.6)
we used the temperature measured by the probe inserted in the column just below
the mist eliminator as the actual air outlet temperature.
Data were recorded by a Hewlett-Packard Model 85 computer that gave
printed, displayed, and plotted information. A Leeds and Northrup strip chart
also recorded surface temperatures, air flow rate, and bubbler output.
3.4 Column Sizing

Flooding constraints determine the column design in terms of allowable liquid and
gas flow. Flooding occurs when a large quantity of the liquid is entrained with the
gas and carried upward with it. This situation is caused by increasing the liquid
flow rate at a fixed gas flow rate. Decreasing the gas flow allows more liquid flow
before flooding. Flooding produces excessive pressure drop and must be avoided in
commercial applications. The generalized pressure drop correlation (Norton
Chemical, 1977), seen in Fig. 3.4, gives this relationship in general terms. This
correlation gives lines of constant pressure drop across the column bed as a func-
tion of flow rates and properties of the gas and liquid. When this correlation is
made specific to the experimental design with air and molten salt and for O.S-in.
Raschig rings, the pressure drops shown in Fig. 3.5 results. Also shown in Fig. 3.5
are the limits of salt flow and air flow for the apparatus and the points where
actual data were taken.
Table 3.1 gives nominal design values for the experimental apparatus. We
chose the maximum operating temperature of 350 C because common nitrate
salts do not cause excessive corrosion with the stainless steel alloys at this tem-
perature. We felt that from what is known about materials compatibility, ade-
quate operating time could be expected from the apparatus by limiting operation
to 350C.
It is clearly necessary to maintain all surfaces with which the salt comes in
contact at a temperature above the freezing point of the salt, including the exte-
rior surface of the column. In practice, the column surface would probably be
heated only for start-up; and when operating conditions were reached, the heat
tracing would be turned off. Then, the insulation applied to the outside surface of
the heat exchanger would control the heat losses. In this experimental apparatus
APPENDIX 2 273
4.00
~ 0.40
u.. ~
'"
(!J ... 0.20 0.05 (4)
(OS
""
c:i ~ 0.10
0.06
0.04
0.02
0.01 ~--L_-"'-"'---L--'-_...L.-...L--L_..I..---J
0.01 0.02 0.06 0.20 0.60 2.00 4.00
~
G
jPG
PL
Figure 3.4 Generalized pressure drop correlation.

we left the heat tracing on during heat transfer experiments to control the ten-
dency of cold spots to form frozen salt, which blocks the column. In addition, the
heat tracing was set to maintain the column surface near the operating tempera-
ture to minimize start-up transients and to act as a guard heater, minimizing
losses from the salt to ambient.
Heat tracing required to initially bring the loop up to operating temperature
and to maintain this temperature was a high-temperature tracing supplied by Nel-
son Electric (type A-846K-016-07). The tracing has a stainless steel shell (0.25-in.
outside diameter) with nichrome wire inside that is protected from the shell with a
refractory insulation. This heat tracing was secured to all exterior surfaces of the
test loop (including both tanks, piping, and the column) with baling wire supplied
with the heat tracing. Approximately 50 m of the heat tracing was required to
provide adequate heating. We then insulated the test loop with a Johns-Manville
Cerawool blanket to a thickness of approximately 15 cm.
3.5 Operational Problems

Most operational difficulties were caused by localized cold spots in the transfer
piping. It was difficult to apply the heat tracing, because it was rather stiff and
could not be easily formed to fit all contours uniformly especially near valve
bodies, on the transfer pipes, and on the bottom conical portion of the column.
6-in. diameter column

0.5-in. Raschig rings
1 in. 1.5 in.
\
200 0.25 in.lft 0.5 in.lft
\
\
3o 100
u::
,
'" " "
Domain of Experiments
o~ ____________ ____________________ ____
~ ~ ~
10 20 30
Air Flow (scfm)
Figure 3.5 Pressure drops and limits of salt and air flow.
We alleviated some of these problems by replacing the original salt with one hav-
ing a lower melting point (143' C versus 221' C for the first salt).
Only one materials-related failure occurred. The heat tracing overheated on
the connecting tube about 30 cm from the upper tank, corroding the tube, and the
entire contents of the upper tank leaked out. On examination we found the tub-
ing had a dark brown discoloration about 3 cm long on either side of the hole. We
subsequently replaced all the tubing with some that had a larger diameter and a
heavier wall (0.7S-in. schedule 40 pipe with a O.Il3-in. wall versus O.S-in., 0.03S-
in.-wall tUbing). Using a larger diameter tubing allowed better application of the
Table 3.1 Nominal column operating conditions.
Salt inlet temperature 350C

Air inlet tempera.ture 200C
Salt flow ra.te m. = 80 kg/h
L = 4390 kg/h m 2
Air flow ra.te mil = 40kg/h
G == 2190 kg/h m2
Hea.t duty Q .. 2kW
APPENDIX 2 275
Table 3.2 Analysis of upper tank residue

Salt after As-received
Residue ",,3600 h Salt
% Fe 0.70
% Mg 1.50 0.005
% Ca 0.50 0.003 0.003
% Ti 0.03
% Si 1.0
%AIO.2
ppm Mn 100
B 20
Ba 30
Cr 70
Cu 15
Mo 15
Ni 150
Pb 20
Sr 30
V 20
~one detected.
heat tracing because it was easier to form the tracing to the contours of the larger
tubing. In addition, the salt tends to freeze more easily in small-diameter tubing.
Most operational difficulties were eliminated after installation of the larger-
diameter tubing.
A brownish residue collected in the bottom of the upper salt tank. Analysis
of this residue is given in Table 3.2 along with an analysis of the as-received salt.
It appears that at some point in the loop, the salt is reacting with the containment
materials; that even at 350 C, long life could be a problem; or that temperatures
in parts of the test loop are substantially above 350 C. Operational problems
resulted, as this viscous residue tended to plug the salt valve, making it difficult to
maintain a contant salt flow. Foreign particulate matter also became trapped in
the valve orifice. These particles were very hard and brittle (similar to small peb-
bles) and were either related to the viscous residue or in the salt as delivered.
The apparatus shown in Fig. 3.1 is somewhat simplified from the original
design, which had two features that caused operational problems and were ulti-
mately abandoned. The salt valve originally was electrically actuated and could
be operated remotely from the control room. This provided a way to control the
salt flow as the salt head in the upper tank was reduced. Unfortunately, the valve
did not operate as smoothly as required, and, if salt froze in the valve, it was
difficult to diagnose the lack of salt flow because of the remote location of the
actuator. Finally, the weight of the entire valve/actuator assembly caused one of
the tubing welds to break. We replaced the valve with a manual bellows valve,
and the pneumatic system described previously provided good salt flow control.
The second feature we abandoned was a pneumatic system for transferring

the salt in the lower salt tank to the upper salt tank. The system involved a pipe
from the lower tank to the upper tank and associated valves for isolating the
lower tank. Applying air pressure to the lower tank forced the salt into the upper
tank. Problems with this system were primarily related to salt freezing in the
return line; plus, the extra valves provided more locations where we could not
apply the heat tracing. We removed the additional valves when we replaced the
small-diameter tubing, and we manually transferred the salt to the upper tank
thereafter.
3.6 Heat Transfer Measurements and Procedures

Using the inlet and outlet salt and air temperatures and the salt and air flow rates,
we can determine the rate of heat transfer from
Q. = m.C.(T.. - T.o) , (3.1)
Qa = ma C" .. ( Tao - Tai) . (3.2)
Equation 3.1 gives the rate of heat transfer from the salt, and Eq. 3.2 gives the
rate of heat transfer to the air. We determined the specific heat for the air and
the salt using the method described in section 7. A comparison of Q. and Qa
gives a quantitative measure of the quality of the heat transfer data since in the
absence of heat losses and measurement errors we would have Q. = Qa' There-
fore, we will refer to the absolute value of the quantity 100(1 - Q./Qa)% as the
heat balance for the experiment.
We can then calculate the volumetric heat transfer coefficient from
Ua = _--.:Q:l<....-._ (3.3)
VpLlTm '
where Vp' is the volume of the packing bed [15 cm (inside diameter) x 0.914 m =
0.0167 m8] and Ll T m is the log-mean temperature difference, defined as
(T.. - Tao) - (T.o - Tai)

Ll Tm =
1
~---,-...!..--~---,------!... (3.4)
[ T.. - Tao
In T.o - Tai
The value of Q in Eq. 3.3 can be either Q. or Qa, and the error in Ua is therefore
equal to the heat balance for the experiment.
From Eq. 3.4 it is clear that for a close approach (Til. = T.. ), which is typi-
cal of these experiments, large errors in Ll T m and therefore in Ua can result.
Table 3.3 demonstrates this from the baseline of actual measured data for one run;
the value of Tao was perturbed to show the effect on QII and Ua. This shows that
for values of Tao lower than the measured value (typical of the probe in the air
outlet line), the heat balance is poor. For values of Tllo larger than the measured
value at the bottom of the mist eliminator, the heat balance is good, but Ua
APPENDIX 2 277
Table 3.3 Measured data showing effect on Q. and Ua
Tai Tao T,; T.o m. ma Q. Qa Ua

(" 0) (kg/h) (W) (W) W/m 3 '0
baseline 193.7 334.8 341.9 309.9 167.8 50.9 2305 2065 3536
294.0 1468 1791
300.0 1556 1895
340.0 2141 4955
341.8 2167 8389
increases very rapidly as Tao approaches T,;. For a 5.2' 0 increase in Tao, Ua
increases by 40%.
Although the thermocouple should not generate errors greater than 1 0 0
(see Section 3.3), placing the air outlet probe where it is influenced by salt drain-
ing from the mist eliminator, heat tracing on the column walls, etc., could cause
large errors. The solution is to totally separate the two phases to eliminate the
influence of salt in the measured air temperature and at the same time to place
the probe close enough to the top of the bed to get a true air outlet temperature.
Based on the two air outlet temperature probes (one inserted through the column
wall just below the mist eliminator and the other inserted through the knockout
pot just inside the mist eliminator), it appears that one probe may read slightly
high because of entrained salt and that the other may read lower by a few
degrees. The best temperature measurement to use, therefore, is the one just
below the mist eliminator.
Experimentally, it is possible to vary the salt flow rate, the air flow rate, and
the salt and air inlet temperatures. The last two variables are of secondary
importance (as long as the air inlet temperature is above the salt freezing point),
so we did not vary them in any systematic way. The salt flow rate was varied
from 50 to 200 kg/h, and the air flow was varied from 30 to 50 kg/h (see Fig. 3.5).
We could attain higher salt flow rates, but this would result in run times too short
to establish steady conditions-a crucial requirement for good data; i.e., small
values of the heat balance. Figure 3.5 shows that air flow rates much larger than
50 kg/h ("'-'25 scfm) produces column flooding.
The system was not temperature-cycled but was left at operating tempera-
ture for about six months continuously with the exception of downtime for repairs,
as described previously. To minimize the time required to reach steady state and
to minimize losses, we set the heat tracing so the bed temperature was fairly close
to the upper tank salt temperature. Pressure was applied to the upper tank from
the regulated air supply, and the salt valve was opened. To achieve a constant
salt flow rate, as indicated by the bubbler output trace, generally required 20
minutes. (As long as nothing lodged in the valve, this flow rate was steady until
the upper tank was empty. ) We then set the air flow to the desired value (it was
4000~------------------------~Sa~l~t~F~IO-W------'
kg/h _9.z,..
." ........ 168
I" .,,'"
3000 e; "\1~~'"
~s ."
.,,~~
() ...... ~
'"'
~ ...... -Iii 170
~ 2000 168 110/ '57
------
S 170::: 60 \(g/'"
__ _---
83 rns _-
-- Salt Inlet Temperature: 350C
1000
O.S-in. Raschig rings; EHn. column
Datae
Calculated - - -
O~----~------------~------------~--~
15 20 25
m. (scfm)
30 40 50
rna (kg/h)
Figure 3.6 Overall volumetric heat transfer coefficients based on experimental
data.
helpful to heat the preheater to about 200 C before turning on the air), and,
0
when steady state was achieved, we could adjust the air flow to a new setting.
Examination of the data indicated that the best heat balances were achieved
when the salt flow was the most uniform and when no adjustments of the salt
valve or tank pressure were necessary. A typical run of two hours provided data
on one salt flow rate and five air flow rates.
3.7 Results and Discussion

Experimental data for Raschig rings are presented in Fig. 3.6 in the form of
volumetric heat transfer coefficient versus air flow rate with salt flow as a parame-
ter. The data are also shown in Table 3.4 with the heat transfer coefficient calcu-
lated from the mass-transfer analogy, Eqs. 2.6 through 2.9.
The heat transfer coefficients do not appear to depend on salt flow rate, as
all the data for rna 0::::: 40 kg/h and for 57 ~ m, ~ 170 kg/h vary by only a few
percentage points. The variation with air flow is relatively strong-a best fit pro-
duces
TT 1.28
va = 21.1 rna (3.5)
APPENDIX 2 279
Table 3.4 Heat-transfer data and comparison with mass-transfer calculations
Salt Inlet Measured Calculated

Temperature Air Flow Salt Flow Ua Heat Balance Ua
("C) (kg/h) (kg/h) (W /m 3 C) (%) (W/m3 C)
342 30.8 168 1820 3 2171
348 30.9 83 1771 2 1429
341 40.7 171 2252 6 2896

353 40.5 57 2203 3 1478
349 40.3 170 2164 4 2854
348 40.2 110 2228 1 2189
348 50.5 96 3520 5 2535

342 50.9 168 3351 5 3574
where m. is in kg/h and Ua is in W /ms C. This is clearly at variance with Eq.

2.14.
As shown in the last column of Table 3.4, heat transfer coefficients calcu-
lated from mass-transfer data underestimate measured heat transfer coefficients
except at large salt flows. Because the experimental data do not correlate with
Eq. 2.14, it is doubtful that using the mass-transfer/heat transfer analogy will
work. Apparently, the heat transfer mechanism does differ significantly from the
mass-transfer mechanism, as discussed in Section 1.
Data for the Pall rings are compared with the Raschig ring data correlation,
Eq. 3.5 in Fig. 3.7. Data were for a salt flow rate of only ......,130 kg/h; we did not
test the dependence of heat transfer on salt flow rate. The three data points fall
fairly close to the Raschig ring curve, although the point for the highest air flow is
somewhat below (......,20%) the curve.
Overall heat transfer coefficients calculated from mass-transfer data are
shown in Fig. 3.6 for two salt flow rates, 170 and 60 kg/h. These results further
demonstrate the lack of sensitivity to salt flow rate for the heat transfer data com-
pared to the mass-transfer data. There are several possible explanations as to why
the heat transfer data do not depend on salt flow rate while the mass-transfer data
do, as explained in Section 1.0. These explanations are the (1) different wetting
characteristics of the packing by the salt, (2) heat conduction through the metal
wall of the packing, and (3) radiation heat transfer.
IT the salt totally wets the packing, any increase in salt flow beyond some
minimum will not provide more heat transfer surface area per volume of packing,
causing the volumetric heat transfer coefficient to be insensitive to salt flow. Since
Ua !:::: hga, the volumetric coefficient Ua may only be affected by changing the
gas-side film coefficient hg or the surface area per unit volume a. This is con-
sistent with the similarity between the Pall ring data and the Raschig ring data
0.5-in. Raschig Rings

3000 21.1 m~ 28

-
U
E 2000
o
0.6-in.
Pall Rings
O~ __________ ~ __________ ~-J
15 20 25
rna (scfm)
Figure 3.7 Comparison of measured heat transfer coefficients for Pall rings and
Raschig rings.
(Fig. 3.7). From Fig. 1.3 one can see that the two packing types should provide
similar surface areas per unit volume since the only difference is that the Pall rings
have the spokes punched in from the periphery of the ring.
Peters and Timmerhaus (1980) give surface areas per unit volume for several
types and sizes of packings. For the O.S-in. metal Raschig ring (with O.06-in.
wall), approximately 118 ft 2 of surface are provided per ft3 of packing volume.
For the Pall ring, the value is 104 ft2jft3 The two types of packing provide simi-
lar heat transfer areas and, therefore, if fully wetted by the salt, should exhibit
about the same heat transfer performance.
On the other hand, if conduction heat transfer in the packing is important,
then it would not be important how much of the surface of the packing is wet by
the salt because heat could then be transferred from the dry areas of the packing
to the air.
A better understanding of the heat transfer mechanism is required. By per-
forming tests at various temperatures with liquids having various wetting proper-
ties and with pac kings of various thermal conductivities and surface areas, it
should be possible to separate the different heat transfer mechanisms.
Overall system pressure drop is plotted in Fig. 3.8. Recall that this is a
measure of the differential pressure from the column air inlet pipe to the column
air outlet pipe. Therefore, it includes not only pressure drop across the bed (Fig.
3.5), but expansion and contraction losses at the column inlet and column outlet
and loss across the air distributor, salt distributor, and the mist eliminator. We
took additional pressure drop data with zero salt flow to determine the contribu-
tion of all these column components. These data allow only a qualitative assess-
ment of the bed pressure drop because it is only about 30% or 40% of the
APPENDIX 2 281
3r-------------------------------_
50 kg/h =
.
rna
.J
2-
,.... e
~40
51
e
-30
.40 } Pall Rings
e31
I I I
o~------~------~------~----~
o 50 100 150 200
ms (kg/h)
Figure 3.8 Overall system pressure drop.
measured system pressure drop. We could not find a satisfactory method for
measuring bed pressure drop because of the difficulties associated with isolating
the high-temperature salt from a pressure-sensing port or isolation diaphragm.
The data in Fig. 3.8 clearly show the benefits associated with using Pall
rings. At a given air flow the overall system pressure drop for the Pall rings is
about half that of the Raschig rings. From Fig. 3.4 we see that constant bed pres-
sure drop, G ,...., YF, in the region of the map where the Llp lines are level. Since
the packing factor F for Pall rings is about 0.17 that of Raschig rings, we could
operate with approximately V6 = 2.4 times as much air flow with Pall rings com-
pared to Raschig rings in the experimental column. The ratio for large (2-in.)
packing is close to 2, so we could operate with about 41 % more air in a large
column with Pall rings than with Raschig rings. From Eq. 3.5 the Pall rings could
provide about 1.411.28 = 1.55 more heat transfer per unit volume than Raschig
rings in a large column. Extrapolation of Eq. 3.5 to such large air flows should be
tested experimentally to determine if flooding is approached. This would most
likely cause the volumetric heat transfer to fall below that predicted by Eq. 3.5.
4 ECONOMIC ANALYSIS
4.1 Purpose
With experimental and calculated values for the heat transfer coefficient, we can
now determine the economic value of DCHX. Rather than basing economic calcu-
lations solely on the mass-transfer data, using actual heat transfer data should give
us more confidence in the results. As expained in Section 4.2, the experimental
data cannot be applied directly to a commercial-size DCHX, and, therefore, even

these results require some caution in interpretation.
4.2 Method of Analysis

The economics of DCHX and finned-tube exchangers are compared by considering
all capital and operating costs associated with the heat exchanger. Using a metho-
dology described by Bohn (1983), one can calculate the annuallevelized cost. This
is the constant annual cost (in fixed dollars) that, if paid over the lifetime of the
heat exchanger, would have a present value equal to the present value of the
actual costs incurred over the lifetime of the heat exchanger. In computing one
single cost this method can easily consider: escalation rates, depreciation, discount
rates, lifetime, and tax rates, among other parameters. Bohn (1983) describes the
method and also gives the values for these parameters used in the analysis.
We will assume that only one capital cost in incurred and that the only
operating cost is that associated with the power required to pump the air through
either heat exchanger. Maintenance costs will be taken as a constant annual cost
equal to 3% of the capital cost. With these assumptions the annual cost may be
computed from
AO = 0.2299 X OJ + 1.886 X OM , (4.1)
where OJ is the capital cost expressed in 1981 dollars and OM is the annual cost
of pumping the air, also expressed in 1981 dollars. We will generally give results
in the form of AO/Q, which is the annual cost per unit heat transferred in $/GJ.
Note that this is quite close to $/106 Btu transferred.
Considering the pumping cost first, for the cost of electricity given in Bohn
(1983) ($12.89/GJ = $O.0464/kWh), we see that
(4.2)
which assumes an isentropic efficiency of 0.70 for the compressor, an electrical

motor efficiency of 0.96, and a plant capacity factor 0, of 0.8. We can calculate
the pressure drop ..1p through the heat exchanger once we know the air flow rate
and the characteristics of the heat exchanger (friction factor versus Reynolds
number plot for the finned-tube exchanger and the generalized pressure drop
correlation curve for the packed column).
To determine capital cost, we must know the size of the heat exchanger and
the materials of construction. The first is determined by the required heat duty,
overall heat exchange coefficients, and log-mean temperature differences. Materi-
als of construction are determined by operating temperature and working fluids
(see Section 4.3). For consistency, we used the data given by Peters and Tim-
merhaus (1980) for all the capital costs.
For the finned-tube heat exchanger, the size is best expressed as heat
exchange area including fins. Peters and Timmerhaus (1980, p. 669) give a graph
of the cost of carbon steel, finned-tube, and floating-head heat exchangers
APPENDIX 2 283
operating at 10 atm. This curve has been generalized to a correlation curve

Of = 2051 A~6622 ($) (4.3)
that includes a factor of 2.3 for installation cost, a factor of 1.8 for the use of
stainless steel in the entire heat exchanger (see Peters and Timmerhaus, 1980, p.
677), and a factor of 1.12 to escalate the 1979 cost to 1981 dollars. For atmos-
pheric pressure operation the capital cost is reduced by a factor of 0.92 (Peters
and Timmerhaus, 1980, p 673). For materials other than stainless steel the cost is
adjusted according to the table given in Peters and Timmerhaus (1980, p. 677),
which lists relative cost factors for entire heat exchangers of several different
materials of construction. For carbon steel the factor is 0.56; and for fucoloy, it is
1.67. The factor between carbon steel and stainless steel (0.56) is consistent with
cost data in Dubberly et al. (1981), which gives 0.60. It is also consistent with a
rule-of-thumb (0.5) used by Mercury Fin Tube Products to scale the cost of a
stainless steel, finned-tube heat exchanger to a carbon steel unit.
Peters and Timmerhaus (1980, p. 772) also give installed costs for packed
towers (excluding cost of packing) as a function of the height and diameter of the
column. For a stainless steel column from 1 m to 5 m in diameter, the correlation
curve is
Of = 10,762 HdfJ.9 ($), (4.4a)
and for carbon steel it is
Of = 2620 Hdl'34 ($), (4.4b)
which includes a factor of 1.03 to account for the cost of installing insulation and
the same 1.12 factor to escalate costs to 1981 dollars. For operation at pressures
other than atmospheric the cost is escalated by the same factor used to rate shell-
and-tube heat exchangers (peters and Timmerhaus, 1980, p. 673) applied to the
fraction of column cost attributable to the shell, ",55%. Note that we did not
increase the cost of the finned-tube heat exchanger to include insulation costs
because we assumed that such a unit would ordinarily be insulated, and the insula-
tion would be part of the 2.3 factor for installation costs. Since packed columns
are not ordinarily insulated, we included the factor of 1.03 for the packed-column
capital cost calculation. For the high-temperature DCHX a more complex column
design required costing of individual components related to the insulation (see Sec-
tion 4.3).
For the finned-tube heat exchanger we used the characteristics of the heat
exchange core, denoted CF-8.8-1.0J by Kays and London (1964), which consists of
I-in. tubes (outside diameter) on 1.96-in. spacing using spiral-wound fins with 8.8
fins/in. and a O.OI2-in. fin thickness. Data on the core includes the Colburn i
factor (the heat transfer coefficient on the air side) and the friction factor as a
function of the air-side Reynolds number. The salt-side Reynolds number was
chosen as a constant (10,000) because the overall heat transfer coefficient is not a
strong function of the salt-side Reynolds number as long as the salt flow is
turbulent. Also, salt-side pumping work is negligible, so we need to consider only

the salt-side pressure drop from the standpoint of tube stress at elevated tempera-
tures.
The layout of the core is shown in Fig. 4.1. Salt flows through the tubes,
and air flows across the finned tube banks (crossflow arrangement). Given the
flow rates (determined from the heat duty and terminal temperatures), we can
determine the heat exchanger effectiveness and the required number of transfer
units (NTU) from equations for crossflow exchangers. Beginning with the lowest
air-side Reynolds number for which the heat exchanger core data are given, we
can determine the value of W (H is arbitrarily set equal to W), calculate the air-
side heat transfer coefficient and the fin efficiencies, and then calculate the
required total heat transfer surface required, which determines the core dimension
in the air flow direction L, which also determines the core pressure drop. With
the surface area and pressure drop we can calculate the cost AG. This procedure
is repeated for increasing air-side Reynolds numbers until a minimum heat
transfer area (900 m 2) for which cost data exist in Peters and Timmerhaus (1980).
This corresponds to a shell diameter of 2.77 m (9.09 ft) for 4.87-m (l6-ft) long,
2.54-cm (I-in.) (outside diameter) tubes. Multiple heat exchangers are specified
when heat duties require more than this maximum heat transfer area.
Optimization of the DCHX is somewhat different because of flooding con-
straints. Outlet air temperature cannot be specified a priori. Beginning with a
column diameter of 1 m (the smallest diameter for which cost data are available),
we calculated the volume of packing. (We used the shortest practical column
height, equal to the diameter, because this always minimized annual cost. Column
diameters larger than the column height produce problems with uniform salt and
air distribution in the column.) For an assumed value of the overall heat transfer
coefficient, we determined the log-mean temperature difference, which gives us the
air outlet temperature. This determines the air flow rate, and from the general-
ized pressure drop correlation we can then determine the pressure drop. We
rejected diameters that produce operating conditions off the generalized pressure
drop map (Fig. 3.4). We used a maximum column diameter of 5 m since this is
the largest for which Peters and Timmerhaus (1980) give cost data. Knowing the
column size and pressure drop, we could then calculate the annual cost.
We repeated the procedure for increasing column diameters giving annual
cost as a function of approach temperature (air outlet temperature). The overall
volumetric heat transfer is taken as a parameter; for most calculations we used the
values Ua = 2000, 3000, 4000 W 1m3 C. We can then use the experimental
0
values of Ua or the calculated values of Ua along with the results of this economic
calculation (which also goves sensitivity to Ua) to appraise the economic viability
of DCHX relative to finned-tube heat exchangers.
APPENDIX 2 286
I+---H--_~
0.060
0.050
+=11-
0.040
---------
0.030 S= 1.73
tttltl
0.020
IHtltl
+
1.024 i~iAA1":m
'" I---~~~=
~ 0.010 0.1137in. 0.012 in.
"- 0.008
0
(!)
...... 0.006
:S
1.0 2.0 3.04.0 6.0 10.0
Re x 10.3
Tube outside diameter = 1.024 in.
Fin pitch = S.S/in.
Fin thickness = 0.012 in.
Fin area/total area = 0.825
Flow passage hydraulic diameter. 4rh = 0.01927 ft
Free-flow/frontal area. a = 0.439
Heat transfer area/total volume. a = 91.2 ft,
Note: Minimum free-flow area IS In spaces transverse to flow
Figure 4.1 Layout of the finned-tube heat exchanger core.

281l DffiECT-CONTACT HEAT TRANSFER
4.3 Materials
From exposure tests of up to 4500 hours at temperatures up to about 600 C, Tor-
0
torelli and DeVan (1982) demonstrate that stainless steel alloy 316 oxidizes at the
rate of 5 mil/yr or less, typically 2 mil/yr. A 9O-mil tube wall thickness, there-
fore, should be adequate for a 30-year life. Fin material will also have to be stain-
less steel because of manufacturing constraints and materials compatibility con-
siderations.
For the packed column at 600 C or lower, we specify stainless steel Pall
0
rings for the packing and a stainless steel column with external insulation. Inter-
nal insulation with a liner to contain the salt and support the packing would allow
us to use a carbon steel column, greatly reducing the cost. This insulation method
has been tested by Martin Marietta (1979) but is not commercially available at
this time; therefore, it involves some technical risk. An alternative design that
may prove economical at intermediate temperatures is an externally insulated,
carbon-steel column with an Inconelliner. This configuration was not examined in
this study.
At temperatures above 600 C, common nitrate heat transfer salts decom-
0
pose. SERI, Oak Ridge National Laboratory, and other laboratories are presently
researching candidate heat transfer salts and compatible containment materials at
temperatures above 600 C. Based on state-of-the-art knowledge it appears that
0
Incoloy 800 alloy is a reasonable choice for finned-tube heat-exchanger materials

for temperatures ~800 C.0
The DCHX for temperatures ~800 C consists of an internally insulated,

0
carbon steel column similar in design to the storage tank described in Martin
Marietta (1979). A high-purity (99% alumina) packing is required to resist attack
by the salt. To seal the insulating firebrick from the salt, an Inconelliner is neces-
sary. A layer of fiberglass insulation will cover the outside of the column, and
aluminum lagging will weatherproof the assembly.
High-purity alumina packing is not commonly available. We determined
the cost of this packing from the cost ratio (3:1) between 99% and 57% alumina
catalyst support from Norton Chemical Company. To account for unknown
manufacturing difficulties that could arise when fabricating saddles from the 99%
alumina, a cost factor of 4 (suggested by Norton Chemical Company) was applied
to the 1983 second quarter price ($19.80/ ft 3, 100 ft 3 order) quoted by Norton
Chemical Company, giving an equivalent 1981 cost of $70.49/ft 3 The packing is
the major cost item for the DCHX, which suggests that a spray column could be a
viable, high-temperature alternative.
The Inconelliner cost was taken from Martin Marietta (1979) who priced an
Incoloy liner with a wafHed design to accommodate thermal cycling. Since the
DCHX does not operate in a cycling mode, the liner can be a simpler design that
allows for thermal expansion. Allowing a 2 cost factor between Incoloy and
Inconel and a 0.5 factor for a simpler liner design, the cost of the installed liner is
$283/m 2 internal column area.
APPENDIX 2 287
The firebricks (Krilite 30) need to be thick enough to allow the carbon-steel
column to operate below 316' C for an internal temperature of 760' C and to
keep thermal losses to less than 1% of the transferred heat. The cost of the
firebrick from Martin Marietta (1979) is $809/m 3 .
The cost of the carbon-steel shell is calculated in Eq. 4.4b. For the fiberglass
outer insulation (sufficient thickness to keep losses to 1% of the transferred heat
when the ambient temperature is 20' C and the carbon-steel column is 316' C)
Martin Marietta (1979) used $265/m 3 of insulation and $25/m 2 of aluminum lag-
ging.
Temperatures above 800' C will most likely require ceramic finned-tube
heat exchangers. Although the cost of these ceramic heat exchanger tubes is not
prohibitive (most likely less than the cost of higher alloy tubes), fabricating the
tubes into a heat exchanger is a relatively unknown art. Therefore, cost projec-
tions are difficult. Construction of the packed column, however, would not change
drastically for temperatures above 800' C and could be costed with the same level
of confidence as the 760' application.
Table 4.1 summarizes the assumed construction of both types of heat
exchangers used for the calculation as a function of salt inlet temperature.
Assumed air inlet temperature is also given for the three salt temperatures that
were run. Table 4.2 summarizes the component costs for the 760' C DCHX with
internal insulation. These are installed costs for large storage vessels. Costs for
smaller units, such as aim x 1 m DCHX, would be greater for field erection, but
if the units could be fabricated in a shop, the costs given in Table 4.2 would prob-
ably be conservative.
Table 4.1 Materials of Construction
Temperatures
at Salt Inlet Direct-Contact Finned-Tube
Heat Exchanger Heat Exchanger
Salt In Air In
350C 2OOC Stainless steel Pall rings Carbon steel tubes and fins
in an externally insulated,
carbon steel column
550C 25OC Stainless steel Pall rings Stainless steel tubes and
in an externally insulated, fins
stainless steel column
750C 55OC 99% alumina saddles in an Incoloy 800 tubes and fins
internally insulated, carbon
steel column with Inconelliner
Table 4.2 Installed Materials Cost, 760 C DCHX
Component Material Cost 1981$
Packing 99% alumina saddles $2489/m3
Liner Inconel $283/m 2
Firebrick Krilite 30 $809/m3

Column Carbon steel (Eq.4.4b)
Insulation, outer Glass fiber $265/m3
Lagging Aluminum $25/m2
4.4 Effect of Packing Size and Type

We need to relate the experimentally measured, heat transfer coefficients (Section
3.0) to those expected at full-scale. The experimentally measured, heat transfer
coefficients were for a pilot-scale experiment with O.5-in. stainless Raschig rings,
O.6-in. stainless Pall rings, and a 350 C salt inlet temperature. The economic cal-
culations assumed 2-in. stainless pall rings or 2-in. ceramic Intalox saddles and salt
inlet temperatures of 360 C, 560 C, or 760 C.
The pressure drop characteristics of all these packings are well understood.
Since the viscosity of molten salt (1.5-3.9 cp) does not differ greatly from that of
water (1 cp), using the generalized pressure drop correlating Fig. 3.4 with
appropriate property values for salt should be adequate for pressure drop calcula-
tions for any packing.
Heat-transfer performance, however, is not well characterized. Even if one
wishes to use mass-transfer data, results will be restricted to I-in. or 2-in. Raschig
rings or BerI saddles. As mentioned in Section 2.0, we had to extrapolate down to
O.5-in. Raschig rings to calculate expected heat transfer coefficients for the experi-
ment from mass-transfer data. For the 2-in. Pall rings or 2-in. Intalox saddles
used in the economic calculations, no information is available either in the form of
mass-transfer or heat transfer data. Therefore, we must comment on the applica-
bility of O.5-in. Raschig ring heat transfer data (Section 3.0) to the 2-in. Pall ring
or 2-in. Intalox saddle data used in the economic calculation and O.6-in. Pall ring
data.
For a given packing type, increasing the size reduces the pressure drop at
fixed G and L because the column flow area is less restricted. For fixed G and L,
however, the mass-transfer coefficient is reduced for a larger packing, presumably
because the larger packing cannot provide as much interfacial surface area
between the gas and the liquid. However, the higher flow capacity of the larger
packing offsets this, and the maximum mass-transfer coefficient (which occurs near
column loading) is only weakly dependent on packing size. This is illustrated in
APPENDIX 2 280
3.0
-c: 2.0 Cl
c:
-
.9!
u .lie:
U
;;:
-...
m
Q)
0
1.0 Il..
...
(J .......
-...
Q)
III
c:
m 0.4
0.6
-Q)
m
3:
0.4 c:
I-;-
III
III
m Co
~ 0.2 0.2 <l
0.1~ __~__,-~____~____~__~__~~
1,000 2,000 4,000 10,000 20,00040,00060,000
Liquid Rate (lb/ft2 h)
Figure 4.2 Mass-transfer coefficient for various sizes of Rashig rings.
Fig. 4.2 where mass-transfer and pressure-drop data from Norton Chemical (1977)
on metal Raschig rings 0.6 in., 1 in., 1.5 in., and 2 in. in size were superimposed
on one plot. The maximum mass-transfer coefficient for the 2-in. rings is about
75% of the maximum for the O.6-in. rings. The four points are the mass-transfer
coefficients for DCHX operation at a pressure drop of O.5-in. of water per foot.
With this comparison, the largest rings again have a mass-transfer coefficient
about 75% of that of the O.6-in. rings (mass-transfer data for O.5-in. Raschig rings
was not available). It appears that heat transfer coefficients for 2-in. Raschig
rings should be about 75% of that measured in the present column at a given
pressure drop.
Comparing the measured heat transfer coefficients with calculated values
based on mass transfer (see Fig. 3.6), we found that the mechanisms of heat
transfer and mass transfer appear to differ and that the mass-transfer/heat
transfer analogy does not apply. We have some experimental evidence (Fig. 3.7)
suggesting that heat transfer does not depend on packing type, at least when
changing from Raschig rings to Pall rings. Fig. 4.3 depicts the effect of packing
type on mass transfer. However, since the analogy with heat transfer is suspect,
we must assume that the 2-in. Pall ring (as used in the economic calculations)
transfers the same amount of heat as the O.5-in. Raschig ring at a given air mass
velocity G. Pressure-drop performance for the 2-in. Pall rings is accounted for in
the calculation procedure, so the reduced pressure drop caused by the Pall rings is
taken into account.
2QO DIRECT-CONTACT HEAT TRANSFER
4.0
OJ
c:
..- 2.0 2.0 .:.!
c: U
-- ,...
CI)
C"Cl
Q..
.S! ..-
CI)
0 1.0 1.0 CI)
...
()
0.8 i
3:
enc: 0.6
CI)
0.6
0.5 c::
...
C"Cl
I- 0.4
0.4 I
I
a.
(f) 0.3
(f) <l
C"Cl
~ 0.2
1,000 2 3 4 5 6 10,000 2 3 4 60,000

Liquid Rate (lb/fF h)
Figure 4.3 Mass-transfer coefficient for various types of packing.
Based on the experimental data presented in Section 3.0, an appropriate
range of heat transfer coefficient Ua for the O.5-in. Raschig rings and therefore the
2-in. Pall rings is 1800-3500 W 1m 3 C. From the previous discussion it is reason-
able to perform the economic calculations based on a range or Ua from 2000-4000
W 1m 3 C. We can then assess the sensitivity of the economics on Ua until a full
range of data at full scale is made available.
4.5 Results
Figures 4.4 through 4.9 present the results of the economic analysis. Each figure
shows the cost of transferring 1 GJ of energy as a function of air-outlet tempera-
ture. For each temperature range two graphs give results for 1 atm and 5 atm
operating pressure. The 1 atm case represents process-heat applications, and the 5
atm case represents a Brayton-cycle application. The Brayton cycle probably will
not apply to the two lower temperatures, but higher pressure operation is gen-
erally more economical because of increases in air density, and, therefore, the
low-temperature, high-pressure case may have application in process heat.
As the air-outlet temperature approaches the salt-inlet temperature, the
finned-tube heat exchanger needs a larger surface area and the DCHX requires
more packing. This is because the log-mean temperature difference (Eq. 3.4) is
reduced, and this increases the packing volume for a fixed, heat transfer coefficient
APPENDIX 2 201
Heat Duty: 1 MW1h

Finned-Tube
1.50 Salt In: 560e
Air In: 250 0 e
Operating Pressure:
"0
Q)
....
~ 1.00 Ua
<II
c: (W/m 30 e)
....(\I
I-
-, ~20/,3000
_~_--~-
C)
.......
~
__ 4000
0.50
o
52~O------530------5~4-0------55LO------5~60------J
Outlet Air Temperature (0C)
Figure 4.4 Cost comparison for 360 C, 1 atm, IMWth .
1.0
Heat Duty: 5 MW tr Finned-Tube
"0 0.8 Salt In: 360C

....
Q)
.... Air In: 200e Ua
2<II 0.6 Operating Press: 5 atm W/m3 e
c: 2000 3000
....
(\I
~4000
I-
-, 0.4
C)
.......
~
0.2
0
320 330 340
Outlet Air Temperature (0 C)
Figure 4.5 Cost comparison for 360 0, 5 atm, 5MWth
and heat duty (Eq. 3.3). We can compensate for this by increasing the air flow,
which increases the heat transfer coefficient, but this drives up operating costs.
Generally, DOHX provides closer temperature approaches than the finned-tube
Heat Duty: 1 MW'h Finned-Tube

1.50 Salt In: 560C
Air In: 250C
Operating Pressure:
'0
...
Q)
~ 1.00 Ua
(/)
(W/m 30 C)
...t-c:
I
-, ~20/.3000
_~_-~--
~
(!)
......
Eoo9-
__ /4000
0.50
O~ ____ ~ ______________ ~ ______ ~ ______

520 530 540 550 560
Outlet Air Temperature (OC)
Figure 4.6 Cost comparison for 560 C, 1 atm, IMWth .
0
heat exchanger before the cosu! increase rapidly. [The curves for the DCHX for a
given Ua increase for low outlet air temperature because the volumetric heat
transfer coefficient has been artificially fixed, and the only way to reduce outlet
temperature is to increase air flow, which drives up the cost. For
Ua = 2000W 1m 3 C this effect is generally not seen for the temperature
0
approaches presented (less than 30 C).] 0
Calculations at the higher operating pressure (5 atm) were for 5

MWth or 2 MWth heat duty, while those for 1 atm pressure were for 1 MWth .
These generally resulted in maximum-sized packed columns (5 X 5m) at the close
approaches. Larger columns can be built, but since the cost data were restricted
to 5-m-diameter columns, we restricted the calculations to this diameter for con-
sistency. Also, the finned-tube heat exchanger tended to reach maximum size (900
m 2 before the DCHX. This is seen in the curves for the finned-tube heat
exchanger, which show the change of slope (Fig. 4.5, for example). These slope
changes occur because multiple heat exchangers are used to meet the heat duty.
Therefore, one major conclusion we can make is that DCHX provides substan-
tially more heat transfer capacity for a given size of equipment.
The cost advantage of DCHX relative to finned-tube heat exchangers is not
a strong function of approach temperature (difference between salt inlet and air
outlet temperatures) except for small temperature approaches where the cost of
the finned-tube heat exchanger increases more rapidly. Table 4.3 gives the cost
APPENDIX 2 2Q3
1. 50 r--------------------_
Heat Duty: 5 MWth
Salt In: 560C
Air In: 250C
'0 1.00 Operating Pressure: 5 atm
......
m
2VI
c(1) Finned-Tube
...
~
-,
CJ
.......
0.50
~ua W/m 30 C
~
~2000
~4000
O~ ________________ ~ _____________ L________________ ~ ____________________~~
520 530 540 ~50 560

Figure 4.7 Cost comparison for 560 C, 5 atm, 5MWth
ratios from all six graphs at a 10 C approach. We used the DCHX curve for
Ua = 3000 W 1m 3 C in each case. 0
There appears to be no great decrease in this ratio in going from 360 C to 0
560 C. This is because the cost of construction materials increased substantially

0
for both types of heat exchangers. Using an internally insulated, carbon steel
column with stainless steel Pall rings for 560 C operation significantly improves
0
the cost ratio. (Since this represents a larger technical risk than the externslly
insulated, stainless steel column, it is not a fair comparison to make with commer-
cially available, stainless steel, finned-tube heat exchangers.)
The large reduction in relative cost in going from 560 C to 760 C occurs 0
because the DCHX does not need high alloy steels, while the finned-tube
exchanger does require such materials. Even though the high-purity alumina
packing is more costly than stainless steel packing, using a carbon steel column
provides a very large cost advantage over the Incoloy finned-tube heat exchanger.
5 CONCLUSIONS AND FUTURE RESEARCH

We measured volumetric heat transfer coefficients in the range of 1800-3500
W 1m 3 C in a 6-in.-diameter column with a 3-ft bed of O.5-in. metal Raschig
0
rings and O.B-in. metal Pall rings. The heat transfer coefficient depends on air
flow rate but not on salt flow rate. Heat-transfer coefficients based on mass-
4.00r-------------~-~~-__,
Finned Tube
Heat Duty: 1 MWth
Salt In: 760C
Air In: 550C
3.00 Operating Pressure: 1 atm
'U
-
CD
~
~
CD
en
c
to
~
2.00
...,
I-
C)
......
~
2000
1.00
/3000
__~==-
___~ 4000
o~ ______ ______ ________ __________
~ ~ ~ ~
720 730 740 750 760

Outlet Air Temperature (OC)
Figure 4.8 Cost comparison for 760 C, 1 atm, IMW'h.
transfer data show dependence on both air flow and salt flow. Thus, the mechan-
isms controlling heat transfer appear to differ from those controlling mass transfer.
The measured heat transfer coefficients are large enough so one can say with
confidence that direct-contact heat exchangers are more cost-effective than con-
ventional finned-tube heat exchangers. At low- to mid-temperatures (360-
560 C), the cost (capital and operating) ratio should be about one-half, while at
high temperatures (600 -800 C), where high alloy steels are required in the
finned-tube heat exchanger, the cost ratio is about one-fifth. The cost advantage
occurs because of the high rates of heat transfer and the ability to use materials
other than high alloy steels to contain the salt in the DCHX.
Future research should be directed toward experimentally determining the
effect of packing size and type, since in lieu of such data one must project this
effect based on mass--transfer data. Since we showed that the heat- and mass-
transfer mechanisms are different, this analogy is not a satisfactory approach. At
high temperatures, radiation heat transfer will become important. Thus, high-
temperature testing, perhaps with internally insulated columns, will be necessary.
To ultimately produce heat transfer correlations valid over a wide range of
operating conditions that will aid designers, it is necessary to understand the
mechanisms of heat transfer. A study on separating the effects of radiation, fin-
effect, packing wetting, etc., will be helpful.
APPENDIX 2 2DS
2.00,---------------------,
Heat Duty: 2 MW'h Finned-Tube
Salt In: 760C
Air In: 550C
1.50 Operating Pressure: 5 atm
"......
Q)
2en
c: 1.00
...
~
f-
....,
c:l
---..
~
2000 3000 4000

0.50
=======---::::.------::::--:::;::/
o~____~~------~~----~~----~~~

Figure 4.9 Cost comparison for 760 C, 5 atm, 2MWth .
Any high-temperature (>600 C) experiments on heat transfer must be

delayed until materials research has identified compatible heat transfer salts and
containment materials.
Table 4.3 Cost ratios of transferring heat via DCHX

relative to finned-tube heat exchanger.
Temperature Operating Pressure

CC) (1 atm) (5 atm)
360 0.44 0.46
560 0.46 0.57
760 0.18 0.26

2g6 DffiECT-CONTACT HEAT TRANSFER
6 PROPERTY VALUES
Values
We took the constant values for specific heat density and thermal conductivity of
the salt from data provided by Park Chemical Company (1983). Constant values
of the diffusion coefficient and viscosity and surface tension as a function of tem-
perature were taken from NBS (1981). The values or functions are
G = 1553 J/kg K
p = 1820 kg /m 3
k = 0.573 W/m K
D = 2.91 X 10-9 m2/s
Jl = 90.811 - 0.3517T + (4.665 X 1O-4)T2 - (2.086 xlO-7)T3 (cp)
u = 155.678 - 0.0627T - (2.315 X 1O-7 )MT2

+ (5.9877 X 1O-7)M2T (dyne/cm)
M= mol % KN0 3
The temperature used in the equations for viscosity and surface tension is the
average of the salt inlet and outlet temperatures.
We evaluated air properties at the average of the air inlet and outlet tem-
peratures by the following equations derived from tabular data in Kreith (1976).
p = 350.8 P /T (atm, K, kg/m 3)
Jl = (3.5158 X 10-6) + (4.8240 X lO~)T - (9.2908 X 1O-12)T2 (kg/ms)
k = (2.719 X 10-.'3) + (7.8017 X lO--6)T - (1.1598 X 10~)T2 (W /m K)

Gp = 997.9 + 0.143T + (1.10 X 1O-4)T2 - (6.776 X 10~)T3 (J/kg K) .
We derived the diffusion coefficient following the procedure in Sherwood, Pigford,
and Wilke (1975).
D _ (3.555 X 10--6) Tl.6
- 0.00285 [Lf -
78.6
0.06063 [L] +
78.6
1.0739
In the above equations for air properties, we used the average of the inlet and
outlet temperatures. All temperatures are in kelvin.
REFERENCES
Bohn, M. S., 1983 (Nov.), Air/Molten Salt Direct Contact Heat-TraTl8fer Experiment and Economic
Analysis, SERI!Tr-252-2015, Golden, CO: Solar Energy Research Institute.
APPENDIX 2 297
Dubberly, L. J., J. E. Gormely, J. A. Kochmann, W. R. Lang, and A. W. McKenzie, H181(Dec.), Cost

and Performance of Thermal Storage Concepts in Solar Thermal SysteTYl8, SERI/TR-XP-O-
9001-1-B, Golden, CO: Solar Energy Research Institute.
Fair, J. R., 1972(June), 'Designing Direct Contact Coolers/Condensers", Chemical Engineering, pp.
91-100.
Kays, W. M. and A. 1. London, 11164, Compact Heat Ezchangers, New York: McGraw Hill.
Kreith, F., 1976, Principle8 of Heat Transfer, New York: Harper and Row.
Martin Marietta Aerospace, 1979(Dec.), Internally Insulated Thermal Storage System Development
Program, MCR-79-1369, Denver, CO: Martin Marietta Aerospace.
National Bureau of Standards, 1981, Physical Properties Data Compilations Relevant to Energy
Storage, Vol. II: Molten Salts Data on Single and Multi-Component Salt SysteTYl8, NSRDS-NBS
61, Washington, DC: NBS.
Norton Chemical Company, 1977, "Design Information for Packed Towers", Bulletin DC-ll, Akron,
OH: Norton Chemical Company.
Park Chemical Company, 1983, Technical Bulletin J-9, Detroit, MI: Park Chemical Co.
Peters, M. and K. Timmerhaus, 1980, Plant Design and Economics for Chemical Engineers, New
York: McGraw Hill.
Sherwood, T. K., R. L. Pigford, and C. R. Wilke, 1975, Mass Transfer, New York: McGraw Hill.
Tortorelli, P. F. and J. H. DeVan, 1982(Dec.), Thermal Convection Loop Study of the Corrosion of
Fe-Ni-Cr Alloys by Molten NaN0 3-KN0 3, ORNL/TM-8298, Oak Ridge, TN: Oak Ridge
National Laboratory.
APPENDIX
DESIGN OF DffiECT-CONTACT
PREHEATER/BO~ERSFORSOLAR
POND POWER PLANTS
John D. Wright
1 INTRODUCTION
Salt-gradient solar ponds may provide the simplest and least expensive method of
converting solar to thermal energy. The pond combines the functions of both col-
lection and storage. Because the salt gradient suppresses thermally induced con-
vection, temperatures of 750 _100o C may be achieved in the storage layer. Ther-
mal energy from ponds may be used to generate electricity, because the low collec-
tion cost offsets the inherently poor efficiency of the low-temperature power cycle.
One promising method of power production is the organic Rankine cycle. Because
of the low efficiency of the conversion cycle, the shell-and-tube heat exchangers for
heat addition and rejection are the major capital expense. Replacing these heat
exchangers with direct-contact heat exchangers can result in major cost reduc-
tions.
Originally prepared under Task No. 9123.00, WPA eo. 21-99g..99, for the U.S. Department of
Energy Contract No. EG-77-C-OI04042. SERI/TR-252-1401, VUC Categories, 62c, 63e, May, 1982.
This report critically reviews the methods available for sizing direct-contact
heat exchangers used to couple an organic (pentane) Rankine cycle to a solar
pond. Conceptual heat exchanger designs are developed, and areas requiring
further research are identified. Section 2.0 describes the overall operation of the
pond, power cycle, and heat exchanger. Section 3.0 describes methods for deter-
mining the cross-sectional areas of the liquid/liquid and vaporization zones, while
Sec. 4.0 discusses heat transfer in each zone. Section 5.0 describes complete heat
exchanger designs, and Sec. 6.0 describes the research required for confident sizing
of such devices. For a more detailed discussion of the overall system design,
choice of working fluid, and system economics, see the earlier report, An Organic
Rankine Cycle Coupled to a Solar Pond by Direct-Contact Het ExchangeSe/ec-
tion of a Working Fluid (Wright, 1981).
2 SYSTEM DESCRIPTION
2.1 Salt-Gradient Solar Ponds

In a salt-gradient solar pond, salt is dissolved in high concentrations at the bottom,
decreasing to low concentrations near the surface. Solar radiation enters the pond,
and most of the energy which is not absorbed on its way down is absorbed on the
dark bottom, warming the storage layer. The salt concentration gradient estab-
lishes a density gradient. The warmer bottom waters typically exhibit a specific
gravity of 1.2, while the cooler, nearly salt-free surface waters have a specific grav-
ity of approximately 1.0. Pure water becomes less dense when warmed. In the
absence of the salt gradient, warm water from the bottom would continually rise
to the surface and lose its heat. However, the density gradient prevents thermally
induced convection. In the absence of convection, heat loss to the surface is by the
much slower process of conduction, and high temperatures (600 _100o C) may be
achieved at the bottom of the pond.
An actual pond has three layers (Fig. 2.1). The virtually salt-free top layer
is vertically mixed by wind and evaporation and should be kept as thin as possible.
The next layer, approximately one meter thick, contains the salt concentration
gradient. It is essentially salt-free at the top and saturated at the bottom. The
lowest layer is saturated with salt and provides thermal storage. Since the salt
concentration is similar throughout the storage region, convection can occur
within the layer, and the temperature throughout the region is constant.
\========::::::=========::::'1- Surface Convecting Layer

Nonconvecting Layer
(salt concentration increases
with depth)
Storage Layer
(constant salt concentration)
Figure 2.1 Salt-gradient solar pond.

APPENDIX 3 301
Surface
Convective
Solar Pond Layer
Nonconvective
Insulating Layer
Convective
Storage Layer
Mechanical Draft
Wet Cooling Tower
Figure 2.2 Organic Rankine cycle coupled to a solar pond.

Temperatures of up to 1000 0 have been achieved in the storage layer of solar
ponds. However, as in all collectors, the heat losses are proportional to the operat-
ing temperature, and, therefore, the collection efficiency decreases with increasing
temperature (Jayadev, 1980).
2.2 Organic Rankine Cycles

The thermal energy contained in the storage layer may be converted to electricity
using a generator driven by an organic Rankine cycle engine. In an organic Rank-
ine cycle, the organic working fluid vaporizes as it absorbs heat from the hot pond
fluid. The organic vapors are passed through a turbine and condensed. Five to
ten percent of the energy absorbed is converted to electricity in the turbine, and
the remainder is rejected to the condenser. The liquid working fluid is then
pumped back to boiler pressure and the cycle repeated (Fig. 2.2). Organic fluids
are used instead of steam because of their much higher vapor densities at the low
temperatures prevailing in the cycle.
The efficiency of a Rankine cycle engine increases with increasing inlet tem-
perature. The theoretical Oarnot-cycle efficiency for a power cycle operating
between 1000 and 300 0 is approximately 19%. When the limitations of real
working fluids, equipment, internal power consumption, and operating conditions
are considered, a maximum cycle efficiency of 10% is reasonable.
2.3 Cost Considerations

The relatively low cycle efficiency of an organic Rankine cycle engine coupled to a
solar pond (conventional fossil fuel plants have thermal efficiencies of 30%-40%)
leads directly to three observations. When a power cycle of 10% thermal
efficiency is coupled to a pond with a collection efficiency on the order of 12%, the
overall efficiency of converting sunlight to electricity is 1.2%. This suggests that
the pond surface areas required will be very large, and, therefore, the pond cost
per unit surface area must be very low. The capital costs of solar ponds are not
well established and are strongly dependent on location.
The most optimistic cost projection is $5/m2 by Ormat Turbines (Israel) for use
of preexisting bodies of water and a local salt supply, Jayadev (1980) estimates
$15/m2 plus salt expense for artificaial ponds lined with Hypalon. Capital costs on
the order of $5/m2 will be necessary for ponds to be practical for power genera-
tion.
The major capital cost in a fossil fuel power plant is the turbine. In the less
efficient low-temperature cycles, much larger amounts of heat must be transferred
to produce a similar quantity of electrical energy. Consequently, the heat
exchangers which supply heat to and reject heat from the power plant become the
major cost items. Furthermore, because the size and capital cost of both the pond
and generating plant are inversely proportional to the net cycle efficiency,
efficiency will be of paramount importance in plant design.
2.4 Direct-Contact Heat Exchangers

Direct-contact boilers and condensers have the potential to significantly reduce the
capital cost of the power conversion cycle. For example, a shell-and-tube boiler in
a 5-MWe plant would contribute between $2 and $2.5 million out of the total $7
million direct capital cost of the plant (Wright, 1981). A direct-contact boiler, to
perform the same function, should cost between $100,000 and $400,000. The sav-
ings could be considerably greater if it were necessary to overdesign the shell-and-
tube heat exchanger to compensate for scaling caused by the high concentration
of salt in the pond storage layer, or if spare shell-and-tube exchangers were needed
for periods of time when the main exchangers were out of service for cleaning.
Because extremely large surface areas are achieved in direct-contact equipment,
proper design of the boiler can result in small temperature differentials between
the water and organic fluids and in increased plant efficiency.
One potential design for a direct-contact heat exchanger (DCHX), combin-
ing the functions of preheater and boiler, is shown in Fig. 2.3. Liquid pentane
drops are injected into the column at the bottom, while hot brine enters at the
top. The pentane drops rise through the brine continuous phase and absorb heat.
When the vapor pressure of the pentane reaches the column pressure, the drops
begin to vaporize. By the time the pentane reaches the top of the active volume,
it has completely vaporized. The pentane vapor disengages from the brine and is
piped out to the turbine. The column is baffled or filled with packing to minimize
large-scale mixing of the aqueous phase.
This report is concerned with the design and sizing of the direct-contact
heat exchanger. It is necessary to predict the height required for heat transfer and
the cross-sectional area required to allow the two fluids to pass through. The
height of the preheater and boiler are calculated separately and added together to
give the total height. The volumetric heat-transfer coefficient in the preheater,
which is set by the interfacial area and the mechanism of heat transfer from the
continuous phase to the drops, is calculated from the flow rates and the physical
APPENDIX 3 303
Wire Mesh Mist Eliminator
rr==r==::;::=t:=_Brine I n let
Wier-Trough Bnne
Distributor
Bed Limiter-~=========~
Column Packmg
Active Heat-Exchange
Volume
Ladder-Type Pentane
Packing Distributor
Support Plate
Wire Mesh Entrainment Separator
1m .........:~_ _ BrineOutiet
L--I
1ft
Figure 2.3 Direct-contact preheater/boiler.

properties. The coefficient, with units of (W /oCm 2)(m2/m 3), is the product of
the heat-transfer coefficient at the surface of a single drop and the surface area per
unit volume. From a knowledge of the preheater duty Qp, the log mean tempera-
ture difference Ll T (driving force), and the volumetric heat-transfer coefficient Uv ,
the volume V required in the preheater can be calculated:
Qp = Uv VLlT.
The cross-sectional area of the liquid/liquid preheater is determined by the flow
rates and physical properties of the two phases. For a given ratio between the
flow rates in the continuous and dispersed phases, there is a minimum cross-
sectional area through which the two phases will flow stably. Dividing the volume
by the cross-sectional area, we determine the required height.
The height and cross-sectional area of the boiling zone are determined by
similar methods. Models to predict the volume and cross-sectional area of the
preheater exist, but they do not agree well. Models for the boiling zone are for the
most part inadequate.
Throughout this report, the various correlations are applied at the heat
exchanger entrance and exit. The flow rates and physical properties of the fluids
in the 50-MWt heat exchanger are shown in Table 2.1. Figure 2.4 shows the
Table 2.1. Flow rates and physical properties in a 50-MWt Preheater/Boiler
Continuous Phase Dispersed Phase

Property Brine (NaCI-H2O) Pentane (C5H I2 ) State
Mass /low rate 2120 k/s 115

Volumetric /low rate 1.84 m /s 0.193 Liquid
15.8 Vapor
Temperature 77C 72 Top

70C 27 Bottom
Density 1150 kg/m 3 596 Liquid

7.25 Vapor
Heat of vaporization 324.7 kJ/kg
Specific Heat 3.55 kJ/kg 2.4 Liquid
Viscosity 1.01 ep 0.185 Liquid

0.0075 Vapor
Surface tension 73 dyne/cm Water-Air
Interfacial tension 51 dyne/em Water-

Liquid
Pentane
temperature versus percent energy exchange achieved in an optimized counter-

flow solar pond heat exchanger (Wright, 1981). The important points are the
small temperature drop in the brine, the relatively small enthalpy change occur-
ring in the preheat section, and the low driving force for heat exchange available
in the boiling section.
3 CALCULATION OF COLUMN CROSS-SECTIONAL AREA

A single drop or bubble of pentane will rise through the brine continuous phase
with a terminal velocity set by a balance between buoyancy and drag. IT many
drops are rising simultaneously through the brine, they interfere with each other,
and their upward velocity is decreased. As the flow of the dispersed phase
increases, the drops crowd closer together and are further slowed. When a critical
flow rate is passed, the downward velocity of the continuous phase is greater than
the upward velocity of the drops, and drops are swept out the bottom with the
brine. This critical velocity is the flooding velocity and represents the maximum
throughput which can be achieved in a given system. The flooding velocity of a
phase is a function of the physical properties of the two phases, the drop size, and
the flow rate of the other phase.
APPENDIX 3 306
360
Brine
r-----------------~/
..
60 80
Percent Energy Exchange
Figure 2.4 Temperature vs. percent energy exchange in boiler.

Column cross-sectional area is determined from prediction of the flooding
velocity. From a knowledge of physical properties and the ratio of the continuous
and dispersed phase flow rates, the maximum possible velocity of each phase
through a given cross-sectional area is calculated. From this, the cross-sectional
area of the column is determined.
Interest in direct-contact heat exchange for desalination has generated a
large number of publications on the hydrodynamics of liquid/liquid spray columns,
and a much smaller body of literature on gases dispersed in liquids and on systems
in which the dispersed phase is vaporizing. The DOE Geothermal Program
(Jacobs and Boehm, 1980) has tested several preheater/boilers, but the primary
emphasis has been on heat exchange and on obtaining operational experience. A
small body of literature exists on liquid/liquid flow in packed columns, while an
extensive literature describes gas/liquid flows in packed columns.
3.1 Liquid/Liquid Spray Columns

Spray columns are simply empty towers in which the heavier continuous phase
(brine) flows downward, while drops of the lighter dispersed phase (pentane) rise
upward. Spray columns with length/diameter (L/D) ratios of less than 10 are usu-
ally subject to severe backmixing and rarely provide the equivalent of more than
one or two theoretical stages. In industrial practice, units with straight sides and
low L/D ratios are used, but most experimental data have been gathered in
laboratory-scale columns of the Elgin design with high LjD ratios (Jacobs and
Boehm, 1980).
The earliest attempts at defining the flooding velocity in spray columns as a

function of the physical properties of the fluids and the drop diameter were by
Minard and Johnson (1952) and by Sakiadis and Johnson (1954). In each case, the
Bernoulli equation was written for each phase, frictional losses were described in
terms of flow around submerged objects, and the equations were solved simultane-
ously to determine the form of the flooding relationship. The constants in the
equations were then determined from experimental data. The Minard-Johnson
correlation was fitted to a relatively small set of liquid/liquid data, while the
correlation of Sakiadis and Johnson was fitted to a much larger set of
liquid/liquid, solid/liquid, and gas/liquid data. Because it covers a much larger
range of properties and represents a later effort by the same research group, the
correlation of Sakiadis and Johnson is preferred. However, it should be remem-
bered that all the data were taken on laboratory-scale columns. Also, the ability
of the correlation of Sakiadis and Johnson to describe flows of gas through liquid is
doubtful, since the flows described used much larger bubbles and much smaller
columns than will be found in heat exchange systems. The correlations of Minard
and Johnson and of Sakiadis and Johnson are presented in section 9.
Letan (1976) recommends the semiempirical correlation of Richardson and
Zaki (1954) to describe the hindered upward velocity of a swarm of drops. Letan
defines flooding as the point at which, for a given flow rate of the dispersed phase,
the velocity of the continuous phase cannot be increased without gross entrain-
ment of the dispersed phase and at which the holdup (volume fraction of dispersed
phase) is at a maximum. An algebraic method is presented for determining the
flow rates and holdup at flooding and at the chosen operating condition.
A better picture of the relationship between the flow rates and holdup is
achieved by combining the rise velocity equations of Richardson and Zaki (1954)
with the graphical presentation of Mertes and Rhodes (1955).
A schematic operating diagram for a system with 4-mm-diameter drops is
presented in Fig. 3.1. The drop number is the ratio of the superficial velocity of
the dispersed phase to the terminal velocity of a single drop. The liquid number is
the ratio of the superficial velocity of the continuous phase to the terminal velo-
city. It is clear that for a fixed continuous-phase superficial velocity there is a
maximum dispersed phase flow rate. The locus of all these points is the flooding
curve. Also, for any given ratio R of dispersed to continuous phase flow rates that
is set by the system's energy balance, there exists a single flooding point where the
line of constant R crosses the flooding curve. Finally, for a given operating point
on the chart, it is possible to predict the effect of changing either or both of the
flow rates.
Like the correlations of Minard and Johnson and of Sakiadis and Johnson,
use of the correlation of Richardson and Zaki by either the method of Letan or of
Mertes and Rhodes is validated only in liquid/liquid systems in smaller-diameter
columns.
In designing a spray column, the superficial velocities of the two phases are
unknown, but the ratio of the volumetric flow rates (and therefore the ratio of the
APPENDIX 3 307
0.2r--"",""---~----~--_--.-_---.
Do = 4 mm
U dt = 24 cm/s
0.1
0.01 0.02 0.03 0.04 0.05

Drop Number
Figure 3.1 Schematic operating diagram for counter-current flow illustrating

relationships among dispersed phase flow rate, continuous phase flow rate, and
holdup.
superficial velocities) is set by the energy balance on the heat exchanger. Knowing
the ratio of the superficial velocities and the physical properties of the two phases,
we can solve for the superficial velocity of the continuous phase at flooding as a
function of drop size using any of the previously mentioned methods. These
superficial velocities represent the maximum possible throughput. Knowing the
superficial velocity and volumetric flow rate of the continuous phase, we divide to
obtain the minimum permissible cross-sectional area of the heat exchanger.
The maximum allowable superficial velocities in the liquid/liquid section as
determined by the three correlations are plotted as a function of drop size in Fig.
3.2, while the corresponding minimum cross-sectional areas for a 50-MWt heat
exchanger are plotted in Fig. 3.3. Drop size in the liquid/liquid section is deter-
mined by the dispersed-phase distributor design and may be set by the designer.
A review of methods for drop size prediction is given by Horvath (1978).
It is clear from the preceding figures that the superficial velocities at flooding
predicted by Richardson and Zaki are up to 50% higher than those predicted by
Sakiadis and Johnson, and that the predictions of Minard and Johnson are essen-
tially independent of drop size. However, little weight should be given to the
0.12 Richardson and

Q)
VI
III Zaki
~
Cl.. 0.10
VI
:::I
0
:::I
c: Sakiadis and
0.08
C Johnson
0
()
~--~""'~---Minard & Johnson
a- ......E
Q)
~
VI
0.06
~--'
<.J 0.04
0
Q)
>
~ 0.02
...
~
Q)
a.
:::I 0.00
en
0 2 3 4 5
Drop Diameter (mm)
Figure 3.2 Spray column continuous-phase flooding velocity VB. drop diameter.
40
ca
...
(1)
30 Sakiadis and Johnson
<ii 1 "2" Berl
c
o~
Saddle Minard and Johnson
.~ E 20
Richardson and
(1)~
U) 3" Pall Rings Zaki
:n
If)
::: 10
U
o----~----~--~----~--~
o 2 3 4 5
Do (mm)
Figure 3.3 Minimum cross-sectional area VB. drop diameter.

APPENDIX 3 30Q
prediction of Minard and Johnson because their work is superseded by that of

Sakiadis and Johnson, and because the dependence on drop size is at odds with
theory and the observations of all other researchers. It is obvious that consider-
able uncertainty exists in the prediction of the flooding velocity. This is not criti-
cal if the preheater and boiler are contained in the same vessel, since the cross-
sectional area required by the vapor flow is larger than that required by the liquid.
However, the uncertainty will be important if separate preheaters and boilers
must be used.
Each of the authors recommends a safety factor for design. Letan recom-
mends the operational holdup be set at 90% of the flooding holdup. Since the
slope of holdup versus superficial velocity at flooding is very steep, this amounts to
an increase in cross-sectional area of only 5%. Sakiadis and Johnson recommend
increasing the cross-sectional area by 25%, and Minard and Johnson recommend a
50% safety margin.
3.2 Backmixing and Packings

It is preferable to operate the heat exchanger as a counter-current device in order
to obtain the maximum possible driving force for heat exchange. Backmixing also
tends to flatten the continuous-phase temperature profile of Fig. 2.4 and make the
system behave as a mixed tank instead of a plug flow device. Backmixing tends
to keep the continuous phase at the brine outlet temperature. This can drastically
reduce the average driving force in the vaporization section, but has little effect on
the driving force in the preheat section because of the large difference between the
inlet pentane and outlet water temperatures. Backmixing has three basic causes.
Each rising drop of pentane carries with it a wake of continuous phase. Because
there is little convective interchange of heat or mass between the bulk phase and
the wake, cold brine is moved from the bottom of the column to the top of the
preheat section, decreasing the continuous-phase temperature and driving force
higher up in the column. Secondly, the dispersed-phase drops tend to channel
upward together in the center of the column and avoid the edges. The continuous
phase is dragged upward in the center and compensates by flowing downward fas-
ter at the sides. This also mixes the continuous phase and lowers the driving
force. Finally, the boiling section is violently agitated by the vaporization and
very high velocity of the pentane vapor. This turbulence can set up violent eddies
and large-scale mixing within both the vaporization and preheat sections.
Conventional tower packings can be used to suppress backmixing. A pack-
ing with a large void fraction (open packing) is desirable, so that backmixing will
be reduced, but the flow will be impeded as little as possible. An example of an
open packing is the three-inch metal Pall ring manufactured by the Norton Co.
(Fig. 3.4). A second possibility is a grid packing (Fig. 3.5) a design that is com-
monly used in cooling towers. Pall ring pac kings force the drops to rise in tortu-
ous paths, causing them to shed their wakes repeatedly. The packing increases
the pressure drop, preventing the dispersed phase from channeling up the center,
and the solid walls block the formation of large-scale turbulence. However, by
Figure 3.4 Metal pall ring tower packing (Norton Chemical CO.).
impeding the progress of the two phases, the rings make it necessary to increase
the size of the column. Grid-type pac kings offer less resistance to flow and would
allow the column to work essentially as a spray column, while preventing the for-
mation of large-scale circulation patterns. By offsetting the grids as the packing
grids are stacked on top of each other, the flow can be made to split and recom-
bine repeatedly. The column would behave like many small stirred tanks in
series, which is essentially indistinguishable from plug flow.
3.3 Liquid/Liquid Packed Columns

While the correlations for spray columns filled should apply to a large column
filled with an open grid packing, a column packing, such as a Pall ring, presents
enough resistance to flow that it must be described by different correlations. The
---.
2 - 4"
~-+--+--7f---T--""'J-----J-}
2 - 4"
/"------./
Figure 3.5 Grid packing.

APPENDIX 3 311
larger pac kings offer less resistance to flow and are less expensive per unit volume
and, therefore, are preferred in order to minimize both the size and cost of the
tower. Packings should be made of a material, such as ceramic or oxidized metal,
that is not preferentially wet by the dispersed phase. At> a rule of thumb, the
packing pieces should be less than one-eighth the diameter of the column, in order
to achieve good distribution and prevent the dispersed phase from tending to
migrate to the walls (Treybal, 1973). Since most of the data on liquid/liquid
flooding in packed columns were obtained in laboratory-scale columns, most corre-
lations deal with pac kings smaller than one inch, as opposed to the 3-in. Pall rings
which would be most suitable for this application. The only correlation which is
valid for both the flow rates and fluid properties in the preheater and for reason-
ably large packings was derived by Hofting and Lockhart (1954) (section 9). The
largest packing covered by this correlation is a 1-1/2-in. Berl saddle. The cross-
sectional area required in the liquid/liquid section for a 1-1/2 in. Berl saddle is
shown on Fig. 3.3 and is essentially in the middle of the range for spray towers. If
the correlation is applied to 3-in. Pall rings, the predicted cross-sectional area is as
low as the lower limiting values predicted for spray towers. It is not probable that
the correlation accurately predicts the flooding velocity of the 3-in. Pall ring pack-
ing, since the correlation predicts continually diminishing cross-sectional areas as
the packing size is increased, instead of reaching a limiting value equal to that of a
spray column as would be expected. All that can be conclusively stated is that the
cross-sectional area required for a column packed with 3-in. Pall rings would be
somewhat less than that for 1-1/2-in. Berl saddles.
3.4 Vapor/Liquid Spray and Packed Columns

When the vapor pressure of the liquid drops reaches the pressure in the vessel,
they begin to vaporize. There is a zone where three phases are present: liquid
water, liquid pentane, and pentane vapor. This zone is violently agitated and
characterized by small drops. There have been no attempts to date to character-
ize these flows in either packed or spray columns.
In a properly designed column, the entire pentane flow will be vaporized
before it reaches the top liquid surface. It is at the point where vaporization is
complete that the dispersed-phase volumetric flow rate will be the greatest,
because the pentane undergoes an 8~fold expansion on vaporization. If this large
flow can move through the column, the lower volumetric flow rate of liquid and
vapor pentane will be able to pass through the middle of the column.
The only correlation that has been proposed to predict flooding rates of gas
bubbling through a liquid-continuous phase is that of Sakiadis and Johnson (1954).
Figure 3.6 shows the calculated cross-sectional areas at flooding as a function of
bubble diameter for a 5~MW heat exchanger. The area is a strong function of
bubble diameter, especially at 10w diameters. No systematic or quantitative stu-
dies of bubble diameter in vaporizing systems have been conducted, although lim-
ited visual study of such systems suggests they are in the range of 2 to 5 mm.
Bubble diameter is determined by a balance between coalescence and turbulent
160
140
120
-'"E 100 ~13.-

as ~
~-
....
Q)
\j\
<C Q).,.
80
iii ()I
c: v'0-1
0
I)~
+=
u 01)
Q)
en 60
en 1%" Berl Saddles
en
0
....
U 40
3" Pall Rings
20
o~--~--~----~--~----~--~
a 1 2 3 4 5 6
Bubble Diameter (mm)

Figure 3.6 Heat-exchange cross-sectional area for packed and spray columns,
based on vapor flow_
forces that tend to break up the droplets. Liquid drop size in turbulent systems
(Heinz, 1955) and air bubble size in agitated tanks (Calderbank, 1967) have been
derived for significantly different systems, and both require knowledge of the
mechanical energy input per unit mass, a parameter unknown in the situation
being considered. H bubble diameters are of the order of 1 mm, the vessel cross-
sectional area required by the vapor section will be much larger than that required
by the preheater, while if diameters are larger, the required cross-sectional areas in
the two zones will be more similar. It should be remembered that the only
gas/liquid data which were used in the fitting of the correlation of Sakiadis and
Johnson came from bubbles of 1/4 to 1-1/2 in. in diameter, using a 2-in. diameter
column. While this correlation must be used for preliminary evaluation because it
is the only one available, its predictions should be used with caution.
APPENDIX 3 313
Extensive work has been done on gas/liquid flows in packed towers, because
packed towers are extensively used in the chemical industry for mass-transfer dev-
ices. Sherwood (1938) developed the first correlation of packed-tower flooding
velocities using an air/water system. Lobo (1945) modified the methood to use
experimentally determined packing factors to correlate the flow instead of meas-
ured surface-to-volume ratios. Leva (1954) included information on preflooding
conditions by including curves of constant pressure drop in the correlation, and by
extending the correlation to systems other than air/water. Eckert (1966) refined
the method for experimentally determining packing factors for dumped packings.
The plot relating flow ratio and pressure drop to the allowable column
diameter is shown in Fig. 3.7. The abscissa is determined from the energy balance
and the physical properties of the two phases. When the designer chooses a pres-
sure drop and a column packing material, the gas superficial velocity and column
diameter are fixed. In order to minimize both the volume and cost per unit
10.0
6.0
4.0
Parameter of curves IS pressure drop
In Inches of water/ft. Figures shown
In parentheSIS are mm of water/m
2.0 of packed height.
1.0 025 (21)
06
l,i 0..0
0.10 (8)
'"
u..
0.4
b 0.."
U
<i'
0.2
0.05 (4)
0.1
Property Symbol Metric Units

0.06 G kg:m" 5
Gas Rate
liqUid Rate L kg/m< 5 F = 16 3" Pall Ring
0.04 Gas Density PG kg/mY = 65 1-1/2" 8erl Saddles
Liquid DenSity P, kg/m J
LiqUid ViSCOSity v Centlstokes

Conversion Factor C 10.764
0.02 Packing Factor F
0,01 L-_....L_ _L-......L..._..l-_-.l._ _....I---L_--.J..._ _' - - _ - ' - _ ' - - - "

0.01 0.02 0.04 0.06 0.1 0.2 0.4 0.6 1.0 2.0
Figure 3.7 Gas-liquid packed tower correlations. (Source: Norton Chemical

Process Products Bulletin).
volume of the packing, a large packing is preferable. The calculated column

diameters for 3-in. Pall rings and 1-1/2-in. Berl saddles are shown in Fig. 3.6. It is
interesting to note that the predicted packed column diameters are less than the
spray column diameters. A possible explanation is that at these flow rates in
packed columns, the gas is the continuous phase and water is the dispersed phase.
The water channels through the vapor, and the pressure drop is greatly reduced.
In the model used to describe the spray column, many discrete bubbles of vapor
with a very large total surface area are rising through the water. Therefore, there
is greater resistance to flow and the required diameter is larger. This also means
that at some point in the boiling section of the packed column, the dispersed and
continuous phases must reverse. The vapor must be the dispersed phase where
boiling is just beginning and the vapor fraction is small, while at the top the vapor
may establish itself as the continuous phase. Whether this phase reversal, which is
a classic definition of flooding, will lead to unstable operation of the column must
be resolved by experiment.
4 HEAT TRANSFER
The considerable literature dealing with heat transfer to single drops, summarized
by Sideman (1966), applies mainly to systems where the drops are widely
dispersed and do not interact. Since the heat-transfer rate is proportional to the
surface area, it is desirable to operate at high holdups where many drops are
crowded together. Research on heat transfer in multidrop systems has been
1.50 A = 200
A
3000 UdS
A =
Uc 5

2500 0
2000
L>
~ LI
I
as ::::::::
0
:::l
1500 Legend
.B.
.r:
I-..l.-J
1000 DSS 0 '
0 0.97
"
":i isobutane ~ 1.51
1.93
.2.50
500
Garwin. Smith 1 0 6
Benzene 0.8
0
0 0.05 0.10 0.15 0.20
Holdup, (j)
Figure 4.1 Volumetric heat-transfer coefficient vs. holdup with flow ratio as
a parameter.
APPENDIX 3 315
focused on desalination in Israel, with the work of Letan, Kehat, and Sideman,
and on geothermal applications in the United States.
4.1 Liquid/Liquid Systems

Data from a large number of spray-tower studies (Suratt, 1977; Garwin and
Smith, 1953; and Plass, 1979) is presented graphically and in equation form by
Plase and is shown in Fig. 4.1. It is useful to note that the volumetric heat-
transfer coefficient increases with holdup for a constant R and decreases with
increasing R for a constant holdup. The data can also be fit with the empirical
equation:
Uv = 45,000 (tP - 0.05)e-o75R
+ 600 Btu/h ft 30 F tP > 0.05
Uv = 12,000 tP Btu/h ft 30 F <P < 0.05
where Uv is the volumetric heat-transfer coefficient, tP is the dispersed phase
holdup, and R is the ratio of the dispersed to continuous phase volumetric flow
rates.
The height of the liquid/liquid section may also be described by the method
of Letan and Kehat (1968). This model applies to drop Reynolds numbers greater
than 30 and laminar flow of the continuous phase in the column. The model
involves determination of the temperature distribution throughout the column
from energy balances on the fluids in the three zones of the column: the wake-
growth zone, the wake-shedding zone, and the mixing zone. In the wake-growth
zone, a wake builds up behind the newly formed drop. In this zone, the drop is
surrounded by fluid at the bulk continuous phase temperature, and the heat-
transfer rate is high. In the wake-shedding zone (the middle region), heat transfer
is poor, as the drop is in contact with an attached packet of continuous phase
which is essentially in thermal equilibrium with the drop. Heat transfer here is
limited by the frequency with which the drops shed their wakes. During the
coalescence or transition to boiling zone at the top, the drops again shed their
wakes, and the resulting agitation increases the rate of transfer. Nonlinear ordi-
nary differential equations are developed for each zone and are solved simultane-
ously to yield the temperature distribution, height, and outlet temperature. The
equations are described in section 9. The correlation has been successfully com-
pared with experimental data by Letan and Kehat and by Plass. Plass also found
good agreement between the correlation of Letan and Kehat and his volumetric
heat-transfer coefficient method. Urbanek (1979), however, in experiments on a
brine/isobutane preheater/boiler found that the Letan-Kehat model would only fit
the observed temperature profile with different constants. Figures 9.3 and 9.4
show that the constants are not yet known with reasonable accuracy. However, as
Urbanek merely adjusted the parameters in the model to fit his own data, it is not
known whether his parameters actually represent the physical processes that
occurred in his experiment.
No data are available for liquid/liquid heat transfer in packed towers, nor
can information be reasonably developed from the mass-heat-transfer analogies, as
mass transfer generally affects the hydrodynamics of the drops. Mass transfer
operations in liquid/liquid packed columns has not been successfully correlated,
and commercial equipment is generally designed by scaling up from pilot-plant
data. However, it can be said that the heat-transfer performance of packed beds
should be superior to that of spray columns because the convoluted path of the
drops through the packing ensures frequent wake shedding, and because the pack-
ing reduces the unmodeled reductions in performance due to backmixing. There-
fore, a conservative method of estimating heat transfer in liquid/liquid packed
columns is to estimate the holdup by the method of Pratt (Treybal, 1973) which is
shown in section 9, and then to estimate the heat-transfer coefficient from the
graph by Plass.
4.2 Boiling Systems

Little is known about direct-contact vaporization. Some experimental observa-
tions are available, but there is no theory or method of correlation to describe the
results in the region of interest. Sideman (1964) analyzed the vaporization of sin-
gle drops of pentane rising through water. In his study, the drops grew into single
large bubbles. However, in heat exchangers where many drops are vaporizing
simultaneously, the resulting turbulence breaks the large bubbles into many small
bubbles. In a later study, Sideman and Gat (1966) measured volumetric heat-
transfer coefficients and column heights required to vaporize pentane in a
laboratory-scale spray column. The data are presented in Fig. 4.2.
Measured volumetric heat-transfer coefficients are in the range 8,000-20,000
kJ/m3 hOC (5,000-12,500 Btu/ft3 h~). The coefficients decrease rapidly with
increasing driving force and are also a (unction of the ratio of the mass How rates.
As the ratio of the dispersed to continuous phase How rate is increased, the heat-
exchange coefficients rise steeply, pass through a maximum, and then slowly
decrease.
A considerable amount of testing of direct-contact preheater/boilers has
been carried out for the DOE geothermal program at the University of Utah and
at the East Mesa, Utah, test site (Suratt and Hart, 1977; Sims, 1976; Blair, 1976;
[me r
Deeds, 1976). Data were successfully correlated by the equation
lO
UA
St = = 2.0 Ja Pr -
(mOp)",P,l mIl
where
Ja = hIg
f /0.p v(Tc,ave - T sat')
Pr = (0 PP/k)d,v'
U = heat-transfer coefficients,
A = heat-transfer area,
APPENDIX 3 317
m= flow rate, and

C p = heat capacity.
A graph of the correlation is presented in Fig. 4.3. The correlation is useful
mainly for evaluating the relative sizes of heat exchangers working with different
fluids, and for suggesting the effect of different conditions on the heat-transfer
coefficient. Unfortunately, because no methods are available to predict the surface
area available in the heat exchanger, the equation cannot be used to predict the
heat-transfer coefficient independently. A second limitation on the correlation is
that it is derived from data where the driving force for heat transfer is an order of
magnitude greater than in solar pond systems. This is of concern because in
direct-contact boiling the heat-transfer coefficient is influenced by the magnitude
of the driving force. Finally, a typical solar pond heat exchanger would have a
JaPr product of approximately 30, which is well to the right of the bulk of the
data on the plot.
No heat transfer data are available at the flow ratio and continuous phase
superficial velocities which are typical of the pond heat exchanger. However,
20,000 kJ/m3hoC (12,500 Btu/ft3h~) will be used as the estimate of the heat
transfer coefficient in designing boilers (Jacobs and Boehm, 1980). It is encourag-
ing to note that in no case did Sideman require a depth greater than 0.2 m (6 in.)
to completely vaporize the pentane. Therefore, it is probable that the heat-
exchanger height will be determined primarily by the preheat section.
5 HEAT-EXCHANGE SYSTEM DESIGN

Because of the wide variation in the predictions of column cros&-sectional area,
three different heat-exchanger configurations will be discussed. The preferred sys-
tem appears to be the packed column, as this yields both a small cross-sectional
area and reduced backmixing. A second choice would be a spray column with
grid packing and large vapor bubbles. The least desirable design would be a spray
column with small vapor bubbles and a cros&-sectional area required for vapor
flow that was much larger than that required in the liquid/liquid section.
5.1 Packed Towers

The most desirable combination is the column packed with Pall rings. The
required vapor-section, cross-sectional area is 30 m2, if a pressure drop of I-in.
water/ft of column packing is assumed. This corresponds to a superficial velocity
which is 80% of that at flooding. The calculated cross-sectional area required by
the liquid is in the range of 15-26 m2. Therefore, if the columns are straight sided
and sized to meet the vapor flow rate, the column will be big enough to handle the
liquid phase throughput. A single column with a 30-m2 cross-sectional area would
have a diameter of 6.2 m (21 ft). This would be too large to ship by truck and
would have to be site fabricated (an expensive process). Therefore, it is preferable
to build three smaller columns with a 3.6-m (1l.7-ft) diameter which will be run in
parallel to handle the heat-exchange load.
20
.;
c::
Q)
-
.(3
;;::
,---,15
Q)
0
...
(.J
-...
(.J
Q) "ij0
UJ (Jr.
c::
as
~
E
- ......
I-j 1-.-.110
as
Q)
:r:
I
0
-
(J x
..::
Q) :5 5
E
:::I --- Gc = Constant
"0 - Gd = Constant
>
5 10 15 20
Gd
Mass Flow Rate Ratio. - x 100 [%]
Gc
(a)
.;
c::
-...
.~
(J 20r-~---------------------------'
;;::
Q)
0
(.J
-...
r-'7"'I
Q) (.J
UJ (ij 0 15
c:: (J r.
as
-
~
'"E
I-j
'---l
as
Q) ~
:r: ....
0
-
(J 10
..:: x
Q)
E :i
:::I
"0
>
5~~--~~--~~--~~--~~~~
o 1 234 5 6 7 8 9 10
Approach Temperature. ~T [0C]
(b)
Figure 4.2 Volumetric heat transfer coefficient plotted against mass flow ratio
at constant driving force and against driving force.
(a) coefficient vs. mass flow ratio at constant driving force
(b) coefficient vs. driving force
APPENDIX 3 3Ig
15
10
9
..
~ 8
C;; 7
II:
~ 6 o
0
u: 5 edt I
en
en
CO
:E 4
~
Q)
.a
E 3

~
z
c:::: Isobulane spray column IDSS engineers)
2 2 n-Hexane spray column (Sen Holl Co.)
c::::
CO A R-113 surface boiler (Unlv of Ulah)
U5
~} Jacobs. Plass el al. IR-113)
100 2 3 4 5 6 7 8 910 20 30 40
Jakob x Prandtl Number
Figure 4.3 Nondimensional vaporization heat transfer vs. Jakob x Prandtl

numbers.
To determine the liquid/liquid preheater height, the Pratt's method is used
to estimate the holdup in the packing at the design conditions (,p = 0.07), and the
volumetric heat-transfer coefficient is conservatively estimated from the graph of
Plass (U. = 28,000 W/m30C [1500 Btu/ft~~]). The preheater duty is 11 MWt ,
and the log mean driving force for counter-current flow is 15.50 C; therefore, the
required volume is 25.3 m3 , and the required height is 0.84 m (2.8 ft). As there
are large uncertainties in the calculations, a 50% safety factor gives us a l.2-m (4-
ft) tall preheat section.
The boiler section is sized bI using a volumetric heat-transfer coefficient of
20,000 W /m30C (12,500 Btu/ft~'1t') and an average driving force of 50 C. The
volume required is 30 m3 and the height is 1.3 m (4.26 ft). Again, using a 50%
safety factor we get an actual boiler height of 2 m (6.4 ft). We then have a heat-
exchange section comprised of three units, each with a diameter of 3.6 m (12 ft),
active height of 3.2 m (10.5 ft), and a total height of 5.6 m (18.5 ft). One of the
modules is illustrated in Fig. 2.3.
It is important to note some of the important assumptions inherent in this
design. Large uncertainties exist in the calculation of the cross-sectional area of
the liquid/liquid section, but as the sizing is set by the vapor flow, this is not a
major drawback. Likewise, the calculations of the heat-transfer coefficients in the
preheater are quite uncertain, but as the assumptions are conservative, this is
Table 5-1 Equipment Cost Estimates for 50-MWt Heat Exchangers (mid-1980 dollars)
Preheater Boiler
Packing Packing
Vessel and Vessel and Total
Internals Internals
Packed column 3-in.

Pall rings 50,300 69,000 119,300
Spray column
5-mm bubbles
combined prebeater /boiler 99,100 116,700 215,800
5-mm bubbles
separate prebeater /boiler 28,900 15,700 84,500 52,400 181,500
l.5-mm bubbles
combined prebeater /boiler 183,300 206,600 389,900
l.5-mm bubbles
separate preheater /boiler 28,900 15,700 169,000 33,900 247,500
Shell-and-tube HEX 2,000,000
again not critical. The vapor section is sized using the correlation of Eckert for
vapor liquid columns and is probably reasonably accurate. The uncertainty in the
heat transfer calculations in the vaporization section is important because no data
are available in the operating region under consideration. Data are not available
at the superficial velocities under consideration, and no data are available for heat
transfer in packed towers. While the cost of such packed-tower direct-contact
heat exchangers is low enough compared to shell-and-tube exchangers that almost
any necessary degree of over-design can be afforded (Table 5.1), it is necessary to
understand the operation to design an exchanger which will yield the expected
performance and ensure that the exchanger will perform stably under all operating
conditions.
The most important uncertainties are not those inherent in the calculations
but those in the processes that were not considered. For example, some degree of
backmixing will exist in the continuous phase, even with Pall ring packing. This
will decrease the available driving force and increase the required volume. The
major reason that the column diameter at the top is smaller in the packed tower
than in the spray tower is that the vapor phase is continuous in the top of the
packed tower. However, at the beginning of the boiler section the liquid is the
continuous phase. The point where the phases reverse is a Hooding point,
APPENDIX 3 321
characterized by large pressure drops. It is not known whether the transition will
cause large-scale flow instabilities in the column, such as slugging, or whether it
will have no effect at all on the overall column performance. In portions of the
boiling section where the vapor is the continuous phase, it is uncertain how the
liquid pentane drops will rise through the column. It is possible that the local
downward velocity of the water will be greater than the upward velocity of the
drops. This could cause accumulation of a layer of liquid pentane at the bottom
of the boiling section, which could in turn lead to liquid/liquid flooding or forma-
tion of large pockets of pentane vapor. Alternatively, the life span of a liquid drop
in the boiling section may be so short that these concerns are unwarranted. In
any case, research on the mechanism of fluid flow in the boiling section is needed
before such a column could be designed with confidence.
5.2 Spray Towers

The system configuration in a spray-tower system will be strongly dependent on
the equilibrium bubble size in the boiling section. Because there is no method
available for predicting bubble size, we will create two designs, one for a system
with 5-mm bubbles and one for a system with l.5-mm bubbles.
For a spray column with a grid packing open enough not to affect the flood-
ing calculations, the cross-sectional area of the vaporization section is 45 m3 at
flooding for 5-mm bubbles. Using a 25% safety factor, the total cross-sectional
area is 56 m2. If we assume a drop size of 3.5 mm in the liquid/liquid section, the
preheater area at flooding is in the range of 20-30 m2. Therefore, the area is set
by the vapor throughput.
From a knowledge of the boiler duty, heat transfer coefficient, and driving
force, we calculate the boiling depth as 0.67 m. With a 50% safety factor, the
boiler height is 1 m (3.3 ft). Using the correlation of Richardson and Zaki to esti-
mate the holdup in the preheater, we find it to be only 2.2%. This is because we
have spread the small liquid dispersed phase flow over a large area and have
reached the point where the drops rise without interference.
This holdup corresponds to a heat transfer coefficient of only 4,200 W /m300
(265 Btu/ft3h<T). The volume of the preheater is now 143 m3, yielding a height
of 2.6 m. Using a 50% safety factor, we have an actual preheater height of 3.9 m.
The overall active height of the heat exchanger is now 4.9 m (16 ft). In order to
shop-fabricate these units, we have 5 units with a 3.8-m diameter (12.5 ft) and a
total height of 7.3 m (24 ft). It is useful to note that the increase in the total area
required for the vapor flow did not increase boiler volume, but increased the
volume of the preheat section dramatically. The major uncertainties in this design
are in the degree of backmixing, the calculation of the vaporization heat transfer
coefficient, the flooding rate at 5-mm bubble diameter, and of course, what the
actual bubble diameter will be. No data are available on vaporization heat
transfer coefficients at these operating conditions. The predictions of the
gas/liquid flooding velocity in large diameter columns by the method of Sakiadis
and Johnson are suspect, even if the average bubble size is known.
The low holdup in the preheater lowered the volumetric heat transfer
coefficient and led to a relatively large and expensive unit. The size and cost may
be reduced by building a system with separate preheaters. Using the same
methods described earlier, we find that the preheating load can be carried out in
two parallel vessels, each with an active height of 3.3 m (11 ft), a total height of
5.3 m (17 ft), and a diameter of 4.1 m (13.5 ft). By regulating the pressure in the
preheater, the pentane is prevented from flashing. The penetane then leaves the
preheater and is injected into the brine just withdrawn from the pond in several
relatively shallow tanks, where it is vaporized. The boiler section consists of five
vessels in parallel, each with an active height of 1.8 m (6 ft), a total height of 4.3
m (14 ft), and a diameter of 3.6 m (12 ft). While this system has a somewhat
lower capital cost (Table 5.1), it will be more complex due to the piping connect-
ing the boiler and preheater. In this case it is necessary to be able to accurately
predict the heat transfer coefficients and holdups in the liquid/liquid section of the
exchanger as well as in the boiler. Because the preheater and boiler are separate,
the preheater is not necessarily oversized as it is in the combined system.
The situation with a 1.5-mm diameter bubble is an exaggeration of that at 5
mm. The cross-sectional area required by the vapor phase at flooding is now 100
m2, compared with 20-30 m2 for the preheater. H the preheater and boiler are
combined in a single vessel, the result is 10 vessels, each with an active height of
4.6 m (15 ft), a total height of 7 m (23 ft), and a diameter of 3.6 m (12 ft). H
separate preheaters and boilers are used, we again have two preheaters with a
total height of 5.3 m and a diameter of 4.1 m. The boiling is carried out in 10
vessels in parallel, each with an active height of 1 m (3.3 ft), a total height of 4.3
m (14 ft), and a diameter of 3.6 m (12 ft).
5.3 Cost Estimate

Table 5.1 presents the direct (fabricated) cost estimates for the five heat exchanger
designs described in Secs. 5.1 and 5.2, and for a conventional shell-and-tube heat
exchanger designed for the same duty. The packed-column exchanger is least
expensive because the phase reversal allows the vapor to flow through a smaller
cross section, and because the smaller cross section leads to a denser packing of
drops and a higher volumetric heat-transfer coefficient in the preheat section.
Because of the greater crOSfrsectional area required for vapor flow when the bub-
bles are small, the cost of the spray-column systems is larger than that of the
packed tower. The difference may be lessened if separate preheaters and boilers
are used, eliminating the effect of the enlarged vapor section crOSfrsectional area
on the preheater.
The columns were costed by the methodology described by Pikulik and Diaz
(1977). The vessels were assumed to be plain carbon steel with 7/16-in. walls. (It
is possible that these vessels could be built from fiber-reinforced plastic at a large
reduction in cost.) The conventional heat exchanger was sized using an overall
heat transfer coefficient of 450 W /m20c (80 Btu/ft~~) for the preheater and 790
W /m 20C (140 Btu/ft2h~) in the boiler (Bell, 1973). The heat exchangers were
APPENDIX 3 323
calculated to cost 215/m e0(20/ft2), a representative figure in 1980 dollars suggested

by Stearns-Roger Services.
6 RESEARCH REQUmEMENTS
Several issues need to be addressed before direct-contact preheater /boilers can be
designed with confidence. Because the packed tower appears most promising,
research should be done on flow patterns in the vaporization section. It should be
verified that the gas phase does indeed become the continuous phase in the top of
the packing. The flow behavior at the point of phase inversion should be studied
to be sure that the local flooding will not disrupt the operation of the rest of the
column. Also it should be determined whether the correlation of Eckert ade-
quately describes the liquid/vapor flow. It should be determined whether the
vaporizing pentane has a tendency to accumulate in large pockets and slug
through the column, as well as whether liquid pentane tends to accumulate in
pockets. Next, it should be determined whether salt water brines have a tendency
to create a stable foam at the liquid/vapor surface, as such a foam would eventu-
ally be carried over into the turbine.
In addition to the fluid flow questions, work is required on vaporization heat
transfer coefficients. The only heat transfer data available describe spray-tower
operation. However, the heat transfer coefficient is a function of flow ratio and
driving force, and no data are available at the conditions under which the column
operates. Also, no theory yet exists which gives an explanation for the type of
variations observed. Such an understanding would allow more confident scaleup
of bench- and pilot-scale experience to full-scale units. Also, experiments need to
be conducted to determine what effect the presence of packing has on the spray-
column heat transfer results.
The most important questions in spray-column systems are: what is the
equilibrium size of the bubbles formed in the vaporization section, what is the
maximum allowable vapor superficial velocity, and does the correlation of Sakiadis
and Johnson adequately describe flooding in the system? These questions are criti-
cal because the size, configuration, and cost of the heat exchanger are primarily
determined by the allowable vapor superficial velocity.
7 CONCLUSIONS
The use of a direct-contact preheater/boiler instead of a shell-and-tube unit can
significantly reduce the heat-exchange costs of a solar pond power plant. The size
and cost of direct-contact heat exchangers is related to both the heat transfer and
hydrodynamics of the system. The height of an exchanger is determined by the
driving force available for heat exchange, the interfacial area, and the enthalpy
change of the dispersed phase. The cross-sectional area is determined by the
volumetric flow rates of the two phases. Methods exist for predicting the height
and cross-sectional areas of direct-contact heat exchangers, but large uncertainties
exist. The liquid/liquid preheater section of the exchanger is reasonably well
understood, but it is the boiling section which usually determines the diameter of a
combined preheater /boiler column. Flow characteristics of packed columns with
vaporization are not understood. No heat-transfer data are available on the
vaporization of drops of a dispersed organic phase at the driving forces and flow
rates which will be encountered in this system. The prediction of flooding veloci-
ties by the method of Sakiadis and Johnson requires a knowledge of the bubble
diameter, a parameter that cannot be predicted. In addition, validity of the corre-
lation for gas/liquid flows in large diameter columns is suspect. Research is
required on the flow patterns in packed columns, flooding characteristics and bub-
ble size in spray columns, and heat transfer coefficients in the boiling sections of
both packed and spray columns. Packed- and spray-column heat exchangers were
designed and costed which covered the full range of possible designs. In all cases,
the exchangers would offer significant cost savings over shell-and-tube heat
exchangers.
8 REFERENCES FOR SECTIONS 1 THROUGH 7

Bell, K. J. 1973. "Thermal Design of Heat Exchangers, Condensers, and Reboilers." Chemical
Engineers Handbook, Section 10, 5th Edition. Edited by R. H. Perry and C. H. Chilton. New
York: McGraw Hill.
Blair, C. K.; Boehm, R. F.; Jacobs, H. R. 1976. Heat Transfer Characteristics of a Three-Phose
Volume Boiling Direct-Contact Heat Ezchanger. DOE Report 100/1523-1. University of
Utah.
Calderbank. 1967. Mizing. New York: Academic Press, Vol. 2.
Deeds, R. S.; Jacobs, H. R.; Boehm, R. F. 1976. Heat Transfer Characteristics of a Surface T1Ipe
Direct Contact Boiler. DOE Report DIO /1523-2. University of Utah.
Eckert, J. S.; Foote, E. H.; Walter, L. F. 1966 (Jan.). "What Affects Packing Performance." Chemical
Engineering Progress. Vol. 62 (No.1): p. 59.
Garwin, L.; Smith, B. D. 1953. "Liquid-Liquid Spray-Tower Operations in Heat Transfer." Chemical
Engineering ProgresB. Vol. 49 (No. 11): pp. 591-601.
Heinz, J. O. 1955 (Sept.). "Fundamentals of the Hydrodynamic Mechanism of Splitting in Dispersion
Process." American InBtitute of Chemical Engineers Journal. Vol. 2 (No.3): pp. 289-295.
Hoffing, E. H.; Lockhart, F. J. 1954. "A Correlation of Flooding Velocities in Packed Columns."
Chemical Engineering ProgreBB. Vol. 47: p. 423.
Horvath, M.; Steiner, S.; Harland, S. 1978 (Feb.). Canadian Journal of Chemical Engineering. Vol.
56: pp. 9-19.
Jacobs, H. R.; Plass, S. B.; Hansen, A. C.; Gregory, R. 1977. "Operational Limitations of Direct-
Contact Boilers for Geothermal Applications." ASME Paper No. 77-HT-5. 1976
ASME/AIChE National Heat Transfer Conference.
Jacobs, H. R.; Boehm, R. F. 1980 (Dec.). "Direct-Contact Binary Cycles." Sourcebook on the Produc-
tion of Electricity from Geothermal Energy. Edited by J. Kestin. U.S. Department of Energy;
pp.413-470. Available from U.S. Government Printing Office, Washington, DC 20402.
Jayadev, T. S.; Edesess, M. 1980 (Apr.). Solar Ponds. SERI/TR-731-587. Golden, CO: Solar
Energy Research Institute. Available from: NTIS, Springfield, VA 22161.
Letan, R. 1976. "Design of a Particulate Direct-Contact Heat Exchanger: Uniform Countercurrent
Flow." ASME Paper 76-HT-27. ASME/AiChE Heat Transfer Conference.
Letan, R.; Kehat, E. 1968. "The Mechanism of Heat Transfer in a Spray Column Heat Exchanger."
American Institute of Chemical Engineers Journal. Vol. 14 (No.3): pp. 398-405.
Leva, M. 1954. Chemical Engineering ProgreBB Symposium Series. Vol. 50 (No. 10): p. 57.
Lobo, W. E.; Friend, L.; Hashmall, F.; Zenz, F. A. 1945. "Limiting Capacity of Dumped Tower Pack-
ings." TransactionB of the AlChE. Vol. 41: p. 693.
APPENDIX 3 325
Mertes, T. S.; Rhodes, H. B. 1955 (Sept.). "Liquid-Pa.rticle Behavior: Pa.rt I." Chemical Engineering
Progress. Vol. 51 (No.9): pp. 429-432.
Mina.rd, G. W.; Johnson, A. I. 1952 (Feb.). "Limiting Flow and Holdup in a Spray Extraction
Column." Chemical Engineering Progress. Vol. 48 (No.2): pp. 62-74.
Norton Chemical Process Products. "Design Information for Packed Towers." Norton Bulletin DC-H.
Akron, OH 44309.
Pikulik, A.; Diaz, H. E. 1977 (Oct. 10). "Cost Estimating for Major Process Equipment." Chemical
Engineering. pp. 106-122.
Plass, S. G.; Jacobs, H. R.; Boehm, R. F. 1979. "Operational Cha.racteristics. of a Spray Column
Direct-Contact Preheater." American Institute of Chemical Engineer8, Symposium
Series-Heat Transfer. San Diego. Vol. 75 (No. 189): pp. 227-234.
Richa.rdson, J. F.; Zaki, W. N. 1954. "Sedimentation and Fluidization, Pa.rt I." Transactions of the
Institute of Chemical Engineers. Vol. 32: pp.35-53.
Sakiadis, B. C.; Johnson, A. I. 1954 (June). "Generalized Correlation of Flooding Rates." Industrial
and Engineering Chemistry. Vol. 46 (No.6): pp. 1229-1238.
Sherwood, T. K.; Shipley, G. H.; Ho\1oway, F. A. L. 1938. Ind. Eng. Chern. Vol. 30 (No.7): p. 765.
Sideman, S.; Taitel, Y. 1964. "Direct-Contact Heat Transfer with Change of Phase." International
Journal of Heat and Mas8 Transfer. Vol. 7: pp. 1273-1289. London: Pergamon Press.
Sideman, S. 1966. "Direct-Contact Heat Transfer in Immiscible Liquids." Adllances in Chemical
Engineering. Vol. 6: p. 207. New York: Academic Press.
Sideman, S.; Gat, Y. 1966. "Direct-Contact Heat Transfer with Change of Phase: Spray-Column Stu-
dies of a Three-Phase Heat Exchanger." American Institute of Chemical Engineers Journal.
Vol. 12 (No.3): pp. 296-303.
Sims, A. F. 1976. "Geothermal Direct-Contact Heat Exchange." Final Report, ERDA Contract No.
E(04-3) 1116. Pasadena, CA: The Ben Holt Co.
Suratt, W. B.; Ha.rt, G. K. 1977. "Study and Testing of Direct-Contact Heat Exchangers for Geother-
mal Brines." DOE Report ORO-4893-1. Ft. Launderdale, FL: DES Engineers, Inc.
Treybal, R. E. 1973. "Liquid-Liquid Systems." Chemical EngineerB Handbook" Section 21, 5th Edi-
tion. Edited by R. H. Perry and C. H. Chilton. New York: McGraw Hill.
Urbanek, M. W. 1979. "Development of Direct Contact Heat Exchangers for Geothermal Brines--
Final Report, Oct. 4, 1977-June 30, 1978." U.S. Government Report No. LBL-8558; available
from National Technical Information Service.
Wright, J. D. 1981 (June). An Organic Rankine Cycle Coupled to a Solar Pond by Direct-Contact
Heat Ezchange-Selection of a Working Fluid. SERIjTR-631-1l22. Golden, CO: Sola.r
Energy Resea.rch Institute. Available from NTIS, Springfield, VA 22161.
9 FLOODING AND HOLDUP CORRELATIONS

Minard and Johnson {1952} determined the form of the i r correlation by perform-
ing a force balance on the two phases and determined the constants and exponents
from experimental data. The data were taken in a column 32 in. tall with a diam-
eter of 8-518 in. The continuous and disper!ed phases were both liquids with den-
sity differences between 0.1 and 0.6 glcm. Continuous phase viscosity varied
r
from 0.9 to 36 cp, and drop diameters ranged from 0.55 to 1.27 cm. The form of
the correlation is
u;t, = [-1.8 D;'''' .;"'" [:~ 1ul"

where
U = superficial velocity ( ft 3 / ft 2It ),
D" = drop diameter (in.),
'7t = continuous-phase velocity (cp),
c = continuous phase, and
d = dispersed phase,
Sakiadis and Johnson (1954) developed the form of their correlation by wri~
ing a force balance on the two phases. The constants and exponents were derived
by fitting the equations to a large amount of liquid/liquid, solid/liquid, and
gas/liquid How data, gathered from experiments carried by a variety of investiga-
tors. The ability of the correlation to accurately predict the behavior of gas/liquid
systems is questionable, as the gas bubble diameter was on the same order of mag-
nitude as the column diameter. The final form of the correlation is
1 +1.8 [I::r I~ rl-O.56SD~~ I;:;; pl~r

where
D" = particle diameter (in.),
gt = acceleration due to gravity (4.17E8 ftfh2),

U = superficial velocity (Ct 3 /ft2It),
p = density (lb/ ft 3 ),
/J = viscosity (cp),
c = continuous phase, and
d = dispersed phase.
Letan (1974) surveyed the methods of predicting settling velocities in sys-
tems where holdup was important and determined that the method of Richardson
and Zaki (1954) correlated a wide range of data found in the literature. In 1976
Letan described a design methodology for spray columns that used the correlation
of Richardson and Zaki.
Richardson and Zaki correlated the hindered settling or rise velocity V. of a
particle with its terminal velocity VT at zero holdup (J = 0) and with the unhin-
dered Reynolds number Reo.
V, = VT (1 - J)m
where
m = 3.65 for Reo < 0.2
APPENDIX 3 327
m = 4.35 Re;<>03 - 1 for 0.2 < Reo < 1

m = 4.45 Re;<>'! - 1 for 1 < Reo < 500
m = 1.39 for 500 < Reo .
Defining the flooding velocity as
~
avo Iv. -- 0 0 ,
where V. and Vd are the continuous and dispersed phase superficial velocities, the
following quadratic equation relates holdup at flooding to the flow ratio and drop
properties:
(m + 1)(1 - R)q,] + (m + 2) Rq" - R = 0 ,
where R = V,i/v., and q" is the holdup at the flooding conditions. It is then sug-
gested that the column be operated so that the holdup is at least 10% below flood-
ing:
q, = 0.9q" .
The superficial velocity of the continuous phase is then
V T q,(1 _ q,)m+!
V = --=-'---'--
R(1 - q,) + q, .
Mertes and Rhodes (1955) developed theoretical upper and lower bounds on
the hindered settling (slip) velocity V, of a swarm of particles
V
low estimate V; = 1 - 1.209 q,2/3
V, 1- P
high estimate
VT - (1 - q,) + 1.209q,2/3
where V T is the terminal velocity of a single particle, and q, is the holdup.
An equation was also derived that related holdup, particle velocity number
(V,/VT ), continuous-phase throughout number (L = U./VT ), and dispersed-phase
throughput number (D = VD/VT ):
VD
D = VT = q, [ V,
VT + 1
1- q, u. 1
VT
By substituting in the high or low particle number predictions derived by Mertes

and Rhodes on the semiempirical correlations of Richardson and Zaki, the
behavior of the system is described.
A convenient method of presenting the results is to plot holdup versus
dispersed-phase throughput number, with the liquid-phase throughout number as a
parameter (Fig. 9.1). Allowing the dispersed-phase throughput number to take on
negative values, the plot shows the cocurrent downflow, cocurrent upflow
0.6~----------------~---,--~~~------------------~
0.5
Cocurrent
Upflow
(Lift)
Q.
:a" 0.3
X
0.2
0.1
O.O~ ____ ~ ____ ~ ____ ~~ ____ ~ ____ ~ ____ ~ ______ ~ ____ ~
-0.15 -0.10 -0.50 0.00 0.05 0.10 0.15 0.20 0.25
Drop Number
Figure 9.1 The method of Mertes and Rhodes (1955).

(pneumatic lift), and counter-current flow. The flooding line, which connects the
points where
21..,
aD L
= 00
divides the counter-current region into two parts-the n-phase (dense packing)
and p-phase (dispersed packing). In the dense packing zone, the holdup decreases
with increased dispersed-phase throughput, while in the dispersed zone, holdup
increases with increased dispersed-phase throughput. Although operation in the
dense region would be desirable, since high holdups result in increased interfacial
area and heat transfer, such operation is possible only in columns where the
dispersed-phase flow rate can be controlled by the coalescence rate at the outlet.
This is not possible in vaporizing systems. Additionally, densely packed systems
often exhibit phase reversal where the dispersed phase coalesces while rising
APPENDIX 3 329
through the column. The graphical method may be used to size columns. The
ratio UD/Uc is set by the energy balance-a line of constant UD/Uc can be con-
structed on the graph. The intersection with the flooding line is the flooding point
for the system. The margin of safety can then be specified, i.e., operate at 90% of
flooding throughput. It is useful to note that because of the steepness of the line
of constant L/D near the flooding point, small changes in throughput result in
large reductions in holdup, and therefore in the volumetric heat transfer
coefficient. Also, for any given drop size and ratio of flow rates, the holdup can be
calculated. Therefore, this plot can be used to evaluate the effect of variations in
operating conditions.
Hoffing and Lockhart (1954) studied packed-tower flooding in liquid/liquid
systems using a 6-in.-diameter column. Flooding velocities were correlated with
the physical properties of the two phases and with an experimentally determined
area/void fraction parameter (a/F1.2) similar to the packing factor defined by
Eckert (1966). The correlation is presented in figure form (Fig. 9.2). Given the
flow ratio R = UD/Uc , the abscissa is read to give f(R) and the following equa-
tion solved to yield the continuous phase superficial velocity:
U -
c-
[
3.33E - 5 p2 22
feR)
p~.l Jl2 08 J.t~.1 [tT:-IJ r [F~.2 r

,tjp .
05 ~ m r 25 RO.25
where
J.t = viscosity in centipoise,
p = density in g/cm s,
tT = actual interfacial tension, and
tT 1D-4 = surface tension of water in air.
The correlation was also tested against data from earlier studies. The largest
packing (smallest a/F1.2) described by the correlation is the 1-1/2-in. Berl saddle.
This packing is much less open than a modern high-capacity packing such as the
Pall ring.
Letan and Kehat (1968) developed a model of heat transfer in a spray tower.
Ai; drops enter the column, wakes begin to grow behind the drops. Heat transfer
in this region is rapid as the wakes approach the temperature of the drop. After a
short distance, the drops rise steadily through the column, periodically shedding
their wakes. As the wakes are essentially in thermal equilibrium with the drop,
and the majority of the heat transfer is from drop to wake, heat transfer in this
region, which occupies most of the column, is relatively slow and limited by the
rate at which wakes are shed. When the drops reach the top, they completely
shed their wakes and coalesce. Differential equations can be written to describe
the heat transfer rate in each zone. The equations can then be solved simultane-
ously to give the temperature difference between the two phases at the top of the
~
... II: ~
"!-
0 0.2
...-.. 0.1
alit
'-""
0 0.05
"l
0
...-..
"I 0
~I ~ 0.02
bib
'-""
8 I 0.01
0"
0.005
~{ 0Q.
.~~
0" <I
oQ. "!
~ 0 ..
oJ ::J 0.002
Packing (a/F12) 0.67
'l
::J
I
.. 0.001 1 1/2" Berl Saddles 12.6 (experimental)
0
3" Pall Rings 9.6 (calculated)
X 0.0005
M
M
M
0.0002
0.01 0.02 0.05 0.1 0.2 0.5 2 5 10 20 50 100 200 500 1000 2000 5000
R = U.
II..
Figure 9.2 Liquid/liquid flooding correlation of Hoffing and Lockhart.

APPENDIX 3 331
column.
Ttl,oat - Tc,in
Td,in - Tc,oat
where
p=1+--,
MR
r
G = volumetric flow rate,
L = column height (cm),
m = volume of wake elements shed per volume of drop and unit length
of column (cm-I ),
M = ratio of wake-to-drop volume,
r = (pGp)d/(pCp)c,
R = ratio of volumetric flow rates Gd/G c , and
T = temperature (OC).
The variation of M with holdup is shown in Fig. 9.3. The number of elements of
wake shed per drop volume and unit length of column (m) are shown as a function
of holdup in Fig. 9.4. The equations may also be solved to give the column height,
but only in the special case of Rr = 1.
Pratt et al. (Treybal, 1973) studied holdup in liquid/liquid packed towers.
For commercial packings larger than 1/2 in., and at low values of the dispersed-
phase superficial velocity (UD ), the holdup (q,) varies linearly with UD up to
q, = 0.1. With a further rise in UD , the holdup increases sharply. At a higher
value of UD there is an upper transition point, drops of the dispersed phase begin
to coalesce, and UD can be increased without an increase in holdup. At a still
higher superficial velocity, the system floods.
Drops of the dispersed phase reach a constant characteristic diameter as
they travel up through the packing. For any fluid system, there is a critical pack-
ing size above which the equilibrium drop size will be minimum and independent
of packing size. The critical packing size dp (usually greater than 1/2 in.) is given
by
...
0)
0.;:-- 0.06
E
-0'
0)
o Cooling drops at an average temperature of 35C
.r:.
t,)
A Cooling drops at an average temperature of 65C
m- 0.05
en:: Heating drops at an average temperature of 35C
'E~
0) 0)
E..J 0.04 A A
A i
0)-
jjj
O)~
'c
0.03

0

..111:-0
tV c: 0
3: tV
-0)
o E 0.02
0) ~
E-
~ 0
0
0
0
-> 0.01 A
00.
>0
II 0
E o 0.10 0.20 0.30 0.40 0.50 0.60 0.70
Hz = 0 = Holdup at Bottom of Wake Shedding Zone
Figure 9.3. Holdup at bottom of wake shedding zone, as a function
of wake elements shed.
0)
E 1.10
~ R = 1.0
"0
>
0..
...o
Cl
0.90 A
0

Q)
E 0.70

A
~
"0
>
0) 0.50 A
..111:
tV
3: 0 Cooling drops at an average temperature of 35C
II 0.30 A Cooling drops at an average temperature of 65C
~ Heating drops at an average temperature of 35"C
0.10 ~---_-'-_-'-_~_--L_---J~_.L....I
0.10 0.20 0.30 0.40 0.50 0.60 0.7.0
Hb = Average Holdup in Intermediate Zone
Figure 9.4. Average holdup in the intermediate zone, as a function of Wake
and Drop Volume.
APPENDIX 3 333
[~J'5,
,1pg
where
d = ft,
(J' = interfacial tension in lb r / ft,
gc = gravitational constant (4.18 X 108 Ibm ft/lb r h2),

,1p = difference in density (lb m/ft 3), and
g = acceleration of gravity (4.18 X lOS ft/h2).
For packings larger than dpack,c, the equilibrium drop diameter is
d = 0.92
p
[~J.5
,1pg
[VK
V
f.
D
if> 1'
where f. is the packing void fraction and VK is a characteristic drop velocity (ft/h)
which is obtained from Fig. 9.5. In Fig. 9.5, V T is the drop terminal velocity
(ft/h) and T is the tower diameter in feet. Because the drop diameter enters into
the correlation through the terminal velocity in the correlation, several iterations
are required to pick a drop diameter and characteristic velocity.
The following equation may then be utilized to determine the holdup:
UD Uc
T + I _ if> = f. VK(I - if .
1.0
::>
:::::.
>-
N
~
Q)
> 0.2
0.1 10
--z[
topg CTg c
0.38d pacK - 0.92 (';-p- )
05 1
PoV, u 9
Figure 9.5 Characteristic drop velocity for packed towers (Pratt, 1955).
The interfacial area (a) is then determined by
a= H .
dp
REFERENCES FOR SECTION 9

Eckert, J. S.; Foote, E. H.; Walter, L. F. 1966 (Jan.). "What Affects Packing Performance." Chemical
Engineering Progre88. Vol. 62 (No.1): p.59.
Hoffing, E. H.; Lockhart, F. J. 1954 (Feb.). "A Correlation of Flooding Velocities in Packed
Columns." Chemical Engineering Progre88. Vol. 47 (No.2): pp. 94-103.
Letan, R.; Kehat, E. 1968 (May). "The Mechanism of Heat Transfer in a Spray-Column heat
exchanger." AIChE J. Vol. 14 (No.3): pp.398-405.
Letan, R. 1974. "On Vertical Dispersed Two-Phase Flow." Chemical Engineering Science, Vol. 29:
pp.621-624.
Letan, R. 1976. "Design of a Particulate Direct-Contact Heat Exchanger: Uniform, Counter-Current
Flow." ASME Paper 76-HT-27. 1976 ASME/AIChE Heat Transfer Conference.
Mertes, T. S.; Rhodes, H. B. 1955 (Sept.). "Liquid-Particle Behavior: Part 1." Chemical Engineering
Progre88. Vol. 51 (No.9): pp.429-432.
Mertes, T. S.; Rhodes, H. B. 1955 (Sept.). "Liquid-Particle Behavior: Part 2." Chemical Engineering
Progre8s. Vol. 51 (No. 11): pp.517-522.
Minard, G. W.; Johnson, A. I. 1952 (Feb.). "Limiting Flow and Holdup in a Spray Extraction
Column." Chemical Engineering Progre8s. Vol. 48 (No.2): pp. 62-74.
Richardson, J. F.; Zaki, W. N. 1954. "Sedimentation and Fluidization: Part 1." Transaction8 of the
In8titute 0/ Chemical Engineer8. Vol. 32: pp.35-53.
Sakiadis, B. C.; Johnson, A. I. 1954 (June). "Generalized Correlation of Flooding Rates." Industrial
and Engineering Chemi8tr1l. Vol. 46 (No.6): pp. 1229-1238.
Treybal, R. E. 1973. "Liquid/Liquid Systems." Chemical Engineer8 Handbook. Section 21, 5th Edi-
tion. Edited by R. H. Perry and C. H. Chilton. New York: McGraw-Hill.
APPENDIX
DESIGN OF A DffiECT CONTACT

LIQUID-LIQUID HEAT EXCHANGER
R. Letan
1 SPECIFICATIONS FOR EXAMPLE

The theoretical background for the following example is presented in detail in
Chapter 6 ''Liquid-Liquid Processes". The example deals with sea-water which has
to be heated by kerosene in a direct contact heat exchanger. The operation is to
be carried out in a countercurrent manner to achieve close approach of tempera-
tures at both inlet-outlet ends. The kerosene is to be dispersed as droplets into the
sea-water. For the operation and design of the heat exchanger the following
parameters have to be referred to:
Operating Parameter8: flow rates of both liquids, inlet and outlet temperatures of
both liquids, holdup of the dispersed liquid, diameter of droplets.
Geometnc Parameter8: diameter of the column proper, length of the column
proper, the number and size of nozzles in the disperser.
The example herein considered relates to sizing. Therefore, the operating
conditions must be either specified or prescribed.
335
Operating Oondition8:
Sea-Water: specified flow rate Qe = 1.66 X 10-3 m3/s
specified inlet temperature Tei = 20 C
prescribed outlet temperature Teo = 70 C
Kerosene specified inlet temperature Teli = 75 C
disperse packing of
droplets of diameter d = 3.5 X 10-3 m
Physical Properties of Liquids:
Sea-Water: density Pe = 1000 kg/m3
specific heat cpe = 4200 J/kg'. C
viscosity Ite = 5 X 10-4 kg/m's
Kerosene: density Pel = 800 kg/m3
specific heat cpel = 2100 J/kg'. C
2 OPERATING PARAMETERS
2.1 Flow Rate Ratio

At steady state, with physical properties constant
(Q'P'cp)eI . (Teli - T do ) = (Q.p.cp)e . (Teo - T ei )
The flow rate ratio is defined as:
R = QeI = Vel
Qe Vel
The ratio of heat capacities is:
(P'cp)eI
r=
(P'cp)e
Therefore
R = 1. (Teo - Tei)
r (Teli - Telo )
For the same approach of temperatures at both ends of the column

Teli - Teo = Telo - Tci
and therefore,
Tdi - Tdo = Teo - Tei
That leads to
Rr =1
APPENDIX 4 337
and in the presently specified case:

r = 0.4
R = l/r = 2.5
R =2.5
2.2 Flow Characteristics or a Single Droplet

The size of the dispersed droplets is an independent variable which may be optim-
ized. In the case presented herein the size is specified as,
d = 3.5 X 10-3 m
The terminal velocity of the droplet is obtained from the drag coefficient - Rey-
nolds number relation
OD . Re 2 = .! . d3 Pc(Pc - p,,).g
o 3 2
Pc
The drag coefficient correlation for rigid spheres applies reasonably well also to
spherical droplets. For larger (d >4 X 10-3 m) and distorted droplets more
appropriate correlations have to be applied.
Thus, using the standard 0D(Re o ) curve of a rigid sphere provides for the
present case:
Reo = 1000 and OD = 0.44
Then the terminal velocity is obtained:
Reo Pc
UT = - - - -
d Pc
for d = 3.5 X 10-3 m, Pc = 1 X lOS kg/m3 , Pc = 5 X 10-4 kg/ms
UT = 0.143 m/s
2.3 Flow Characteristics or the System

Flow characteristics of a particulate system are defined by the relationship
between slip velocity and holdup. The relationship is unique for a system of
specified properties. Semi-empirical expressions like the one by Richardson and
Zaki (''Liquid-Liquid Processes", Eq. (4)) will provide a satisfactory approximation:
V
Us = _c_ = UT (1 - c/J)(m-l)
l-c/J
where for Reo > 500, m = 2.4.
Therefore in our case the slip velocity - holdup relationship is

Us = 0.143 (1 _ </1.4
and is also related to the superficial velocities in the countercurrent flow as fol-
lows:
m-1 _ ~ Vc
UT (1 - J) - J + (1 - J)
However, Vd = R Vc and therefore:
0.143 . (1 - </1.4 = Ve [2~5 + 1~J 1

2.4 Flooding and Operational Holdup
Flooding represents the state of disruption of the stable flow conditions. The
holdup at flooding in a disperse packing of droplets corresponds to the maximum
operable holdup and the combination of flow rates. On the flooding curve:
aVe
a:
IVd =0
Differentiation of the superficial velocity-holdup equation yields a relationship
between J, and Yd' Substituting RVe for Vd and combining again with the
holdup - superficial velocity relationship results in:
J/~0.35
for holdup at flooding. The operational holdup has to be at least 10% lower:
J < 0.315
2.5 Superficial Velocities

Superficial velocities of both liquids are calculated at the operational holdup.
Thus:
v = U. P'(I_p)(m+1)
e T R(I-J) + J
For UT = 0.143 mis, </> = 0.315, R = 2.5 and m = 1.4:
Ve = 8.95 X 10--3 mls
and
Vd = R- Ve = 22.4 X 10--3 mls
APPENDIX 4 3311
2.6 Temperatures
The inlet temperatures are specified
Tei = 200 and Tdi = 75 0
The outlet temperature of the sea water is prescribed, and the outlet temperature
of the kerosene is calculated at R = 2.5:
Teo = 70 0 and Tdo = 25 0
3 GEOMETRIC PARAMETERS
3.1 Diameter of the Column

The diameter of the column proper is obtained by using the flow rate and
superficial velocity of the dispersed or continuous liquid:
D=2 [~J,j,
1rVe
For Qe = 1.66 X 10--8 m 3/s, Ve !:::O 9 X 10--8 mls and
D = 0.485 m
D !:::OO.5 m
3.2 Length of the Column

To calculate the length of the column proper for the specific case of Rr = 1, the
appropriate equations ("Liquid-Liquid Processes", Eq. (34) - (39)) take the following
form:
L = [ [ Tdo - Teo
Tdi - Teo
1 exp (MR) - 11-m. S- -
. r -
ll'1
1 [1 + 1
-
S
1
where S and ll'1 are expressed by Eqs. (37) and (36) respectively in Chapter 6,
"Liquid-Liquid Processes".
At 1> = 0.315, M = 0.8, and m = 3.0 m-1
For these conditions, S = -1.11, ll'1 = -8.75 m -1. Substituting for R = 2.5, r = 0.4
and the respective temperatures provides the column proper length:
L = 8.2 m
Disperser
Nozzles of diameter
dN = 1.5 X 10-.'1 m
are required for the specified size of droplets d = 3.5 X 10-.'3 m. For uniformly
sized droplets the kerosene velocity through the nozzles is:
VN = 0.5 ml8
Therefore, the number of nozzles in the plate must be:
N = Q.I [~ . d% . VN 1
For Q. = 1.66 X 10-.'3 m 3/s,dN = 1.5 X 10-.'3 m and V N = 0.5 mls we obtain the
number of nozzles as:
N = 1900
The surface of the nozzles is then
SN = .!. . d% . N!::::: 3.4 X 10-.'3 m 2

4
If the disperser is located in an enlarged conical bottom, it may be designed of the
same diameter as the column proper, D. Thus,
DN =0.5 m
Then the relative drilled surface is:
4 VOLUMETRIC HEAT TRANSFER COEFFICIENT

The performance of a direct contact heat exchanger is usually assessed by means
of a volumetric heat transfer coefficient defined as:
where .dTm is the logarithmic mean temperature difference. For the present
design:
Uy = 4.45 X lOt W 1m3 K
APPENDIX 4 341
5 SUMMARY
Operating Oonditions:
Flowrates: sea-water Qe = 1.66 X 10~ m3 Is
kerosene Qd = 4.15 X 1O~ m3Is
flow rate ratio R = 2.5
Temperatures: sea-water, inlet Tei = 20 C, outlet Teo = 70 C

0 0
kerosene, inlet Tdi = 75 C, outlet Tdo = 25 C

0 0
Design:
Diameter of column proper D =0.5 m
Length of column proper L = 8.2 m
Disperser: nozzle diameter dN = 1.5 X 1O~ m
number of nozzles N= 1900
NOMENCLATURE
OD drag coefficient
cp specific heat capacity
D diameter of column proper
d diameter of droplet
dN diameter of nozzle
g gravitational acceleration
L length of column proper
M ratio of wake to droplet volumes
m wake elements shed per unit length of column and per unit volume
of droplet
N number of nozzles
Q volumetric flow rate
R flow rate ratio, QdlQe
r heat capacity ratio (p. cp)dl(p cp)c
S surface
Reo Reynolds number of a single droplet
T temperature
Us slip velocity
UT terminal velocity
Uv volumetric heat transfer coefficient
V superficial velocity
JI dynamic viscosity
p density
> holdup
SUBSCRIPTS
c continuous
d dispersed
f flooding
i inlet
0 outlet
APPENDIX
THERMAL AND HYDRAULIC DESIGN

OF DffiECT-CONTACT SPRAY COLUMNS
FOR USE IN EXTRACTING HEAT FROM
GEOTHERMAL BRINES
Harold R. Jacobs
ABSTRACT
This Appendix outlines the current methods being used in the thermal and
hydraulic design of spray column type, direct contact heat exchangers. It provides
appropriate referenced equations for both preliminary design and detailed perfor-
mance analysis. The design methods are primarily empirical and are applicable
for use in the design of such units for geothermal application and for application
with solar ponds. Methods of design, for both preheater and boiler sections of the
primary heat exchangers, for direct contact binary power plants are included.
Based on a report submitted to the U.S. Department of Energy, Contract No. DE-AS07-76ID
01523 with the Department of Mechanical and Industrial Engineering, The University of Utah, Salt
Lake City, Utah, April, 1985, Revised June, 1985.
343
1 INTRODUCTION
The spray column has been widely studied in the chemical industry for many
years due to its inherent simplicity as a counter-current device for heat or mass
transfer. Developments were enhanced in the 1960's due to increased interest in
desalination systems (Saline Water Conversion Engineering Data Book, 1971).
More recently, in the 1970's, Jacobs and Boehm (1980) suggested their use for
extracting heat from moderate temperature geothermal brines. They and a
number of other investigators have carried out a wide range of studies under U.S.
Department of Energy funding for nearly 10 years. This work culminated in the
construction of the 500 kW Geothermal Direct Contact Binary Cycle Power
Plant at East Mesa, Californfa, by Barber-Nichols Engineering under U.S.D.O.E.
funding, (Olander, et al., 1983). The 500 kW direct contactor was designed by
the present author as a combined working fl~id preheater-boiler. The working
fluid was isobutane with the continuous fluid being the immiscible geothermal
brine.
Based upon the relative success of this technique, a number of other applica-
tions have been spawned. Most closely related is the use of a modified spray type
direct contactor for the extraction of heat from a salt-stratified solar pond. The
low temperature design conditions for a solar pond dictate the use of pentane as
the working fluid if it is desired to utilize the working fluid vapor to generate
electricity.
Although both geothermal and solar pond applications have the same ulti-
mate purpose, to generate electricity from a moderate to low temperature source
and to obtain the energy exchange at small approach temperature differences,
many source-related characteristics cause significant differences in their design.
For the geothermal application, it has generally been conceded that the most
economical design is to utilize as much heat as possible from each unit mass of
geothermal brine. This leads to near equal mass flow rates of the working fluid
and brine. For the case of solar ponds, with much lower peak temperatures and
concerns about returning too cold a fluid back to the pond, the mass flow rate of
brine far exceeds that of the working fluid.
For a combined boiler-preheater, it is clear that for high-pressure, high-
temperature vapor generation (such as for geothermal applications) the heat duty
of the preheater can greatly exceed that of the boiler. For solar pond applications,
where the vapor is generated at temperatures as low as 67 C, the boiler duty can
0
be two to three times that of the preheater. Thus, design philosophy can be con-
siderably different. Nevertheless, in this Appendix, an attempt is made to provide
information for general purpose design of spray column type direct contactors. As
nearly as possible, the information provided herein is current and provides the best
available information.
APPENDIX 5 345
2 SPRAY COLUMN DESCRIPTION

A spray column is one of the oldest known devices for contacting two immiscible
fluids in order to transport either heat or some chemical substance from one fluid
to the other. It is basically an empty vertical column with injection devices for
each fluid and outlets for each fluid. In most common applications, each fluid is in
a liquid phase; however, for use with binary power cycles, a single column can
include a liquid-liquid preheating zone and a boiling or evaporation zone.
When only liquid-liquid heat exchange is desired, the column must have a
disengagement zone at both the top and bottom of the column (see Fig. 5.1). A
properly designed column needs only to control the flow rates of the two liquid
streams to insure a pseudo-steady operation.
A column in which both preheating and boiling takes place requires that
there be a disengagement section at the bottom of the column, a level control dev-
ice to insure that the column is not completely flooded with liquids, and a vapor
reservoir at the top of the column for the generated vapors. The vapor reservoir
must be sufficiently large to insure that a liquid phase does not exit as a mist with
the vapor mixture. Thus, mist eliminators may also be required (see Fig. 5.2).
3 DESIGN OF DISPERSED PHASE FLUID INJECTORS

In order to design a direct contact heat exchanger of the spray column type, it is
necessary to design a distributor which can produce regular uniform-sized drops of
one of the two fluids. Normally this is the lighter fluid. Thus, for geothermal or
solar applications, this would be the hydrocarbon working fluid. This is achieved
by designing a distributor which uses a perforated plate of a material not wetted
by the fluid to be dispersed to form the drops. Typically, punched holes which
leave a slight nozzle at the surface exposed to the continuous phase are used. For
geothermal applications, a mild steel plate pickled in sulfuric acid provides clean
jets and well-formed drops.
Actual design of the holes is not critical as long as the flow rate through
them is equal to the jetting velocity, but less than the critical jetting velocity.
Exceeding or equaling the jetting velocity is important due to the fact that lower
velocities can lead to situations where not all of the nozzles are flowing and due to
the fact that drops formed when VI is exceeded are very regular in size. Regular
size drops are important in order to predict column performance. Steiner and
Hartland (1983) recommend that the jetting velocity be calculated from
Vi = ~ [1.07 -
Ptl d"
0.75 ,tlpd;g
4u
1 (1)
where q is the interfacial tension and dn is the nozzle diameter.
*Interfacial tension is not surface tension. It can be predicted by Antonoff's rule which states that
for two saturated liquid layers in equilibrium, the interracial tension is equal to the difference
between the two individual surface tensions of the two mutually saturated phases under a common
It is also necessary not to exceed the critical jetting velocity. Above this
velocity the length of jet decreases dramatically followed quickly by automization
of the dispersed phase. This requires a large pressure drop across the nozzle and is
generally undesirable for spray columns. The critical velocity and corresponding
critical drop diameter can be calculated according to Skelland and Johnson (1974)
as follows
dic =
d" for K < 0.785 (2)
0.485](2 +1
or
d.
dic = for K ~ 0.785 (3)
1.51K + 0.12
where
K = YO'/.tJ.pg
(4)
Vic = 2.69
d~
J;
d,.
(5)
dic (0.514Pd + 0.472Pc)
At this critical velocity, Treybal (1963) recommends the following equations for
the critical drop size
2.07d.
dDC = 0.485 E: 1 for Eo < 0.615 (6)
and
2.07d.
dDC = for Eo ~ 0.615 (7)
1.51 EJ/2 + 0.12
where Eo is the EOtvOs number, defined as
Eo = .tJ.pgd; (8)
0'
For conditions between the jetting velocity and the critical jetting velocity,
the following correlation is recommended (Steiner and Hartland, 1983) to deter-
l
mine the drop size.
dD = dic r2.06 V.
Vic -1.47 In V.

Vic

(9)
vapor or gas, 0' = 0' d. - 0'c.' If the above values are not known saturated-phase surface tensions
can be used. Accuracy within 15% is claimed for organic-water and organic-organic systems for the
latter estimate. Saline Water Conversion Engineering Data Book, 2nd Edition (1971) gives values
for organic fluids and water and brine under air. These properties should be used.
APPENDIX S 347
I CONTINUOUS
PHASE INLET
DISPERSED PHASE
OUTLET
INTERFACE BETWEEN
DISPERSED PHASE AND
CONTINUOUS PHASE
o
0 0 0
00 0 0
000 0
00 00
o
0 0 0
0 0 0 0
00 00
00 0 0
00
00
00
00
00
00
00 0
00 0
o 0 0
0 000
000 0
0 000
t DISPERSED PHASE
INLET
CONTINUOUS
PHASE OUTLET
Figure 5.1 Direct contact spray tower for liquid-liquid heat exchange.
348 OffiECT-CONTACT HEAT TRANSFER
Steiner and Hartland (1983) recommend maintaining a minimum Weber

number~defined with nozzle velocity and the density of the dispersed phase, i.e.,
We = "d"Pd) greater than two to prevent seeping along the surface and secure
q
drop formation on all openings. Experience in the laboratory indicates that nozzle
or perforation spacing should not be closer than 1.5 dD to insure that jet or drop
coalescence does not occur.
4 BEHAVIOR OF DROPS
Drops formed from jets or nozzles may behave differently according to their den-
sity, interfacial tension, volume, and whether heat or mass transfer takes place
between them and the surrounding continuous phase. For a drop rising due to
gravity in an immiscible liquid, there are five dimensionless groups that govern the
motion of the drop:
PcdD VD
Reynolds number Re = (10)
Pc
Llpgd~
EOtvOs number Eo = q
(11)
gp~Llp
M - group M= (12)
p~a3
Viscosity ratio Kl = Pd/Pc (13)
and Density ratio '1 = Pd/Pc (14)
For any particular liquid-liquid combination M, K, '1 are constant in an

isothermal system. Thus, Grace (1983) correlated drop behavior by plotting Re
versus Eo for constant values of M for a large number of liquid pairs. Kl and '1
play a small role in the results. Figure 5.3 categorizes drops into three regimes:
the Spherical regime, the Ellipsoidal regime and the Spherical Cap regime. An
approximate curve is shown which separates the former two regimes. (Experi-
ments conducted at the University of Utah at high values of Re for Eo of near one
indicate that the spherical regime exists longer than shown.)
The spherical regime contains that region where drops are spherical, or
nearly so. For spherical drops, little or no internal circulation takes place.
Somewhat larger drops obtain, on a mean time basis, a shape like that of an
oblate ellipsoid of revolution. The instantaneous shape may depart radically and
undergo wobbling which, of course, would cause significant internal circulation.
When Eo ~ 40 for all M :::; lQ2 droplets have a leading surface which looks
spherical, but the rear may be Bat or concave. These drops may move randomly
and their behavior is hard to correlate. Thus, they should be avoided in the
design of a spray column.
APPENDIX 5 34U
I CONTINUOUS PHASE
~ INLET
MIST
ELIMINATOR
FLOAT LEVEL CONTROL
VAPOR -LIQUID
AEROSOL
DISENGAGEMENT
ZONE
BOILING
ZONE
o 0
0 0 0
00 0
00
o 0
00
o
00 0
o 0
o
0 0 0
o 0
PREHEATER
0 0 0
o ZONE
o 0
0 0 0
o
o 0
000
00 0
0 00
0 00
000
000
t DISPERSED PHASE
I
INLET
CONTINUOUS
PHASE OUTLET
Figure 5.2 Direct contact spray tower for preheating and boiling dispersed
phase.
CD
a:
a:
ILl
-
III
:!:
::::>
z
UJ
0
...J
0
Z
>
ILl
a: 10'
16'~U-~~&L-U-L~llll~-LJW~~~~~~~~~~WU
10- 2 10-' 10 10' 10 2 10 3
EOTVOS NUMBER, Eo
Figure 5.3 Drop characterization map.
For spherical drops, the terminal velocity in a quiescent fluid can be calcu-
lated by a simple balance of the gravity force by the drag yielding
f Vf = .! dD
3
[Pd - Pc
Pc
~
~
g (15)
If it is assumed that the drops behave like rigid smooth spheres, then f varies as
follows
APPENDIX 5 351
Re c< 0.1 f = 24/Rec

2 < Rec < 500 f= 18.5/Rec (16)
500 < Re c < 2 X 10& Rf = 0.44
2 X 10& < Rec Rf = 0.2
More recently Rivkind and Ryskin (1976) have proposed for the drag coefficient:
f -- K 1+ 1 [K[~
Re
+ _4_]+
Re;/3
c
~l Re~78
(17)
which accounts for the relative motion of the interface due to the differences in
fluid viscosities. It should be noted, however, that the presence of contaminants
such as found in geothermal brines or salt pond brines tend to make the interface
more immobile. However, data is missing on the influence of surfactants and
impurities.
For ellipsoidal drops Grace (1985) recommends that the terminal drop velo-
city be calculated from
VT = ~ M-o 149 (J - 0.857) (18)

PcdD
where
J = 0.9411J.767 for 2 < H < 59.3 (19)
H > 59.3
[r
J = 3.42H'l.441 for (20)
with
14
H = .! Eo M-o149 ~ (21)
3 0.0009
In the above equations Pc is in kgJmsec.
5 VELOCITY OF DROPS IN SWARMS

Drops in a spray column, depending upon the holdup, may move in dense swarms.
Ai> the drops get closer together they interact changing not only their own veloci-
r
ties but also that of the continuous phase. Steiner and Hartland (1983) recom-
mend
[ kLl:.dO' (1 - ,),..."
""r [1
(22)
[1 + + R!I: ,) ]
to predict the superficial velocity of the continuous phase. This compares with
Gt VT (1 - )m+l
(23)
A - (1 - )R +
which has been used by Letan, et al. (1968) and by Jacobs and Boehm (1980). In
Equation 22, recommended values for k and n are 2.725 and 1.834 respectively.
The value of m in Equation 23 is a function of the drop Reynolds number based
on terminal velocities, as follows:
Ret< 0.2 m = 3.65
0.2 < Ret < 1.0 m = 4.35 Re;o03 - 1
1 < Ret < 500 m = 44.5 R et-o 1 - 1 (24)
Ret ~ 500 m = 1.39
In the above equations, is the holdup.
In low holdup situations, the relative velocity between the drops and the
continuous phase is equal to the terminal rise velocity of a single drop.
Gt Gd
Vr = V DT = A(1 _ ) + A (25)
However, at higher holdup the close proximity of adjacent drops influences the
terminal velocity of a typical drop within the swarm, thus, v;.;I: VDT Kumar
(1980) recommends for this situation that
Vr = [2.725 ilpdDg [ 1 -
Pc 1 + 1/2
r834J/2 (26)
It was this expression together with Equation 25 from which Equation 22 was
developed. Comparison of Equation 26 with experimental data from other investi-
gators indicate mixed results especially at low holdup. Nevertheless, it is recom-
mended by Steiner and Hartland (1983). Deviation from Vr = V T , however, is
small unless dense packing is achieved. Thus, Equation 23 is preferred unless high
holdups are encountered.
8 HOLDUP OF THE DISPERSED PHASE

The holdup is the fraction of the total volume occupied by the dispersed phase. In
an isothermal spray cQlumn, it is constant along the column length. However,
when heat transfer occurs, the holdup can vary along the column length. Whereas
for many applications the variation may be small, in a geothermal application
using isobutane as the working fluid, changes in holdup must be taken into
account in the preheater much less the boiler (if a boiler-preheater combination is
used). This is due to the fact that density changes of 25% can readily occur in the
preheater.
APPENDIX 5 353
H the continuous phase flow rate is held constant and the dispersed phase
flow rate is gradually increased, a condition known as flooding can eventually
occur. (Similarly, flooding will eventually occur if the dispersed phase flow rate is
held constant and the continuous phase flow is increased.) At this point, it is
impossible to continue passing more oC the dispersed fluid through the column and
a Craction would then be washed out oC the column with the continuous phase or
the drops might all coalesce, changing either the drop size or which fluid is
dispersed. As this cannot be readily predicted, flooding is to be avoided. We,
thus, must be able to predict the flooding point.
There is not much data on flooding in the literature (Steiner and Hartland,
1983). Based on the relationship oC Richardson and Zaki (1954), Letan (1976)
recommends that
(m + 1)(1 - R)~ + (m + 2)R I - R = 0 (27)
be used to calculate the holdup at flooding where m is given in Equation 24 and
R = Gd/Ge , the ratio oC volumetric flow rates. For stable flow operation
(28)
is advised.
It should be noted, that it is possible to operate at a holdup greater than the
flooding point. This type oC operation occurs in what is called the dense packing
regime. Operating below the so-called flooding point is called the dispersed pack-
ing regime. Theoretically, it is possible to operate a spray column in each regime
Cor a given pair oC flow rates. Although considerable work has been directed
toward dense packing, in practice, it is difficult to achieve. It does have advan-
tages over the common operation with loose packing. The interfacial area can be
three to five times higher and back mixing is considerably reduced. The dense
dispersion can be controlled so that its interCace is 30-50 mm above the dispersed
phase distributor. However, iC the jets from the distributor plate enter into the
dense packing, it can lead to coalescence oC the drops. Once coalescence starts, it
continues through the column height and operation cannot be maintained. This
coalescence is the main danger in operating in the dense packing regime.
Not much is known about either the hydrodynamics or the heat transCer in
the dense packing regime. Thus, although it is possible to operate in this manner,
prudent designers have stuck to disperse packing operating. Further, research to
counteract coalescence by the adding oC surCactants to the drops could result in
practical dense phase operation oC a spray column. The influence oC the surCac-
tants on heat and mass transCer would also have to be studied.
7 AXIAL MIXING
Spray columns are designed with the intent to Cacilitate countercurrent contact
between two immiscible fluids. The degree to which this is achieved depends upon
the design oC the injectors Cor the two fluids. The design Cor the dispersed phase is
reasonably straightCorward and is essentially a perCorated plate covering a
manifold. The limiting factors on nozzle jet velocity have already been discussed.
The injection of the continuous phase is more difficult and flaws in its design are
probably the leading cause of axial circulation in a liquid-liquid spray column.
However, it should be noted that little research on internal back mixing or axial
mixing has been done.
In the analyses of spray columns, models used are normally one-dimensional
and transient or steady state. Back mixing can be introduced by introducing sin-
gle dispersion coefficients, Ee, and Ed which may be correlated by comparing
column operating results against different operating parameters. Principally, it is
assumed that an additional flux exists in the opposite direction to the main flow of
each of the phases. The magnitude of the back flow is assumed to be proportional
to the negative gradient of the parameter in question (temperature, concentration,
velocity, etc.). For example, for mass transfer, the conservation equations take the
form (Steiner and Hartland, 1983)
8x ax a2 x ke a _
- = Ve - + Ee - - - - (x - x ) (29)
8t az 8r 1-
for the continuous phase, and
2.JL = -Vd ElL + Ed ~ + -kc a (x - x - ) (30)
8t az az2
for the dispersed phase. The values for Ec and Ed will be dependent on column
geometry, injector design, etc., as well as the fluids being used. These points are
discussed by Steiner and Hartland (1983). They carried out experiments in a
spray column without any dispersed phase present. They noted strong circulation
in the continuous phase only. Other back mixing can be caused by the fact that
the drops carry wakes that are periodically shed as they rise. AB pointed out by
Letan and Kehat (1968), in a sufficiently tall column, this effect can even out along
the length of the column. In their heat transfer work this resulted in long
columns operating with a nearly constant value of the volumetric heat transfer
along the column length for columns with length to diameter ratios, L/De, greater
than eight.
In small diameter columns, bulk circulation can occur due to the fact that
drops concentrate in the middle of the column. In larger diameter columns, this
can lead to more difficulties as it has been proposed that the axial dispersion is
proportional to the diameter raised to a power (E ,....., DA). However, it is believed
that this is again caused by the difficulty in designing an appropriate continuous
phase inlet. Such was the problem, it is believed, with the 500 kW spray column
direct contactor at East Mesa. However, changes in other opera~ing conditions
were also made at the same time as the injector was changed. Thus, no one
knows for sure whether the injector modification alone led to improved operation.
However, testing with continuous phase nozzle design prototypes at the University
of Utah in its six inch diameter unit, seemed to indicate significant reduction in
back mixing. Thus, it is believed that appropriate design can significantly reduce
back mixing.
APPENDIX 5 366
No general rule is available to design the continuous phase injector. The

following rules of thumb are proposed:
1) The continuous phase injector should not release a strong jet of fluid. A
local strong jet produces recirculation regions about it. IT the fluid is
injected downward, strong axial recirculation cells may develop. Further, a
strong jet can lead to local flooding and potential breakup of the drops.
2) Lateral release of the continuous phase is desirable; however, again, strong
flows can lead to drop breakup. While this may be tolerated in a boiler-
preheater combination where the nozzle is in the boiler region it cannot be
tolerated in the preheater where countercurrent flow is strongly desired.
3) Multiple-spaced inlets can insure low speeds and small velocities, thereby,
minimizing any jet lengths and thus recirculation zones. No other guides
are available, although examples of inlets are provided in the literature.
8 HEAT TRANSFER CORRELATIONS
8.1 Liquid-Liquid Heat Transfer

Direct contact heat transfer between the drops of the dispersed phase and the con-
tinuous phase is complex. It depends not only on the thermal properties of each
phase, but also on the dynamics of the drops themselves. As was noted earlier,
the drops can be spherical, ellipsoidal or cap-shaped depending upon their Rey-
nolds number, their E<itvos number and their M-group. To date, most experi-
ments for spray columns were carried out with drops whose diameters were less
than 7.5 mm with most heat transfer studies being carried out for drops with 2.0
mm::; dD < 4.0 mm. In this range for use with geothermal brines or water as the
continuous phase, the drops are nominally spherical with the possibility of being in
the fluctuating ellipsoidal regime. For hexane, pentane or isobutane, experience at
the University of Utah indicates that the drops are spherical. In the presence of
impurities, such as occur in geothermal brines, there is further justification to
presume this behavior. This is due to the fact that impurities tend to immobilize
the interface. Thus, it is believed that little or no circulation will take place in
drops of light hydrocarbons in water, or brine for diameters less than 4 mm.
Further, in swarms, the terminal velocities of drops are decreased which makes it
less probable that they are in the fluctuating ellipsoidal regime.
When the drops have low thermal conductivity, as is the case with hydro-
carbons, it is likely that the governing resistance to heat transfer is internal to the
drops. However, until the recent work of Jacobs and Golafshani (1985, 1985), it
was generally assumed that external resistances were governing. Letan (1968,
1976) and coworkers postulated models based purely on the hydrodynamics of the
drops. Noting that at moderate Re e , Re e > 500, the drops periodically shed their
wakes, Letan and Kehat (1968) postulated that the reduced heat transfer as com-
pared to single drops was related to the shedding frequency and wake size. They
argued that for long columns most of the heat transfer occurred in the wake shed-
ding regime. Allowing for some empirical constants in the model, they were able
356 OffiECT-CONTACT HEAT TRANSFER
to fit their own and some other data. Further, for long columns they could justify
the use of a constant volumetric heat transfer coefficient. As their experiments
were conducted at small temperature differences between the incoming fluids, they
did not worry about temperature dependent fluid properties.
Following the lead of Letan and earlier investigators, Plass, Jacobs and
Boehm (1979) ran a series of experiments to determine a volumetric heat transfer
coefficient, Uv They correlated both their own data and that of other investiga-
tors for organic fluids dispersed in water or geothermal brine. They claim an
accuracy of 20% for the following correlation
U. = 1.2 X 104 Btu (Jor < 0.05) (31)

hrft3 of
U. = [4.5 X 104 ( - 0.05)e4J67 GD/Go + 600] (32)
Btu
----"'~- (Jor > 0.05)
hrft3 'F
where
U. = _---eQ"--_ (33)
Vol LMTD
The above equation was used in estimating the preheater length requirement
for the 500 kW East Mesa combined boiler/preheater spray column.
For drop; originally of 3.0--3.5 mm in diameter, Jacobs and Golafshani
(1985) showed that the heat transfer is reasonably well-represented by Equations
31-32 when actual local holdup values are used. However in deriving Equations
31-32, Plass, et al. (1979) used the correlation of Johnson, et al. (1957) for holdup.
The use of the latter correlation gave "sometimes agreement" with the data from
East Mesa (Letan, 1976). The degree of accuracy in predicting preheater length
was approximately 20% depending upon how the holdup was calculated. The
calculated, detailed temperature profiles did not compare as well with the experi-
ments. Jacobs and Golafshani (1985) also investigated a model using the assump-
tion of no drop internal resistance to heat transfer and one where the heat transfer
was governed by diffusion within the drop. This latter model showed better agree-
ment, especially when it accounted for drop growth.
For drops less than 4.0 mm in diameter, it is recommended that final
preheater spray column sizing be done using the conduction drop growth model of
Jacobs and Golafshani (1985) and Jacobs (March 1985). Preliminary sizing can be
carried out using Equation 33.
For drops greater than 4 mm in diameter, it is highly probably that the
drops will be in the fluctuating ellipsoidal regime. For this regime, both internal
and external resistances to heat flow would have to be considered. For single
drops, Sideman (1966) gives
NUd = 50 + 8.5 X 10-.'3 Reo PrO. 7 (34)
APPENDIX 5 357
for the external surface coefficient for oscillating drops for 150 < Re < 700. A
maximum deviation of 12% was reported. No reliable expression is ivailable for
Re> 700.
The internal resistance to heat transfer can be calculated by Equation 28 of
Golsfshani and Jacobs (1985). It gives
Re",Pr",
Nu", = 0.00375 / (35)
1 + J.td J.t c
This equation is reported to be accurate to 20%.
It is not clear from the studies conducted to date whether increases in the
surface heat transfer due to internal circulation will offset reduced surface area by
going to larger drops. Before settling on a drop size for a given applications, how-
ever, such a study is warranted.
8.2 Direct Contact Boiling Heat Transfer

The sizing of the boiler for the 500 kW unit at East Mesa was also done using an
estimated value for the volumetric heat\ransfer coefficient. Based on all available
data for light hydrocarbons and freons boiling in water, it was observed that
U = 48 000 Btu (36)

v , hrft3 of
where is estimated at the value just below the boiler in the preheater section.
Although this yielded a reasonable estimate and appears to agree well with the
500 kW facilities operation, it has no basis in the physics of the boiling
phenomeilOn. However, no correlation yet proposed does. Further, Walter (1981),
in a recent Ph.D. dissertation at the University of Tennessee concludes, "There
appears to be no way to calculate a heat transfer coefficient". For the lack of any-
thing better, Equation 36 is recommended.
9 DESIGN APPROACH FOR GEOTHERMAL APPLICATION

Based on laboratory and field experience with the 500 kW unit at East Mesa, it is
clear that we can safely design a DCHX for approach o~ pinch temperatures of
2.5 C (4.5 F). The pinch temperature is a necessary parameter in carrying out
0
0
the thermodynamic analyses to select the optimum direct contact binary cycle for
a given geothermal resource. It will set the flow rates of the two fluids, brine and
working fluid once the working fluid is selected. The complete power system can
be chosen utilizing the computer program DffiGEO described by Jacobs and
Boehm (1980) and Riemer, et al. (1976).
On the basis of the system thermodynamic analyses, the mass flow rates of
brine and working fluid would be available for a given geothermal resource. The
direct contactor pressure, working fluid boiling point, and the inlet and outlet tem-
peratures for both the brine and working fluid would be established. One could
now, utilizing this information, proceed to design the direct contactor.
The first thing that must be done is to decide on the drop size for the
dispersed phase. Typically the nozzle diameter, d,., will be from one half to two
thirds as large as dD At:, the equation for the drop size, Equation 9, depends upon
the critical jetting velocity as well as the critical jet diameter, which in turn
depend on d,., we must select d,. first. For light hydrocarbons in brine values of
d,. less than 1.58 mm (1/16 inch) should result in drops from 3.0 to 3.2 mm ("'-'1/8
inch) in diameter. Such drops should remain nearly spherical with little internal
circulation. A larger nozzle diameter will result in fluctuating drops whose
behavior at high holdup could lead to an unstable column. This, of course, would
need to be examined for the actual fluids selected in light of Fig. 5.3.
After selecting d,., the jetting velocity, Vi' can be calculated from Equation
1. This is the minimum velocity for the injection nozzles. Next the critical jetting
velocity and jet diameter are determined from Equations 2-5. The critical jetting
velocity, Vic' cannot be exceeded. At the critical jetting velocity, the drops
formed will be given by either Equation 6 or 7 depending upon the Eo number.
One can operate the injector at any velocity between the jetting velocity and the
critical jetting velocity at long as the Weber number is greater than two. This is
necessary to prevent seeping.
Knowing the desired drop diameter, Equation 9 can be used to calculate the
nozzle velocity, V,.. At:, long as the criteria mentioned above is satisfied, we will
have selected the nozzle size. We can now proceed to determine the required
number of perforations, or nozzles, in the dispersed phase distributor.
At:, the total mass flow rate of the disperse phase is known, as well as its
md
temperature and pressure, we can calculate the volumetric flow rate, Gd = -.
Pd
The number of nozzles required is
Gd
n =---- (37)
The number of nozzles, of course, must be a whole number. We round off the cal-
culation to the nearest one, making sure we do not cause a problem by exceeding
the value of Vic. Using the whole number, we calculate a new value of V,. from
Equation 37. We then recalculate dD from Equation 9. This is then the drop
diameter.
Depending upon whether or not the drops are spherical or wobbling ellip-
soidal, which can be checked roughly from Fig. 5.3, we are ready to calculate the
drop terminal velocity using either Equation 15 or 18, as is appropriate.
Next the flooding holdup, (. /' should be calculated from Equation 27. The
design value of holdup for the column should be selected as 0.9 (. / or less as
pointed out in Equation 28. Correcting for the amount of mass of geothermal
brine that is vaporized in the boiler with the working fluid, the mass entering the
preheater is determined prior to the actual calculation of (. /. Knowing the exit
conditions of the brine, the volumetric flow rate Gc is determined. We now know
APPENDIX Ii 3liD
both Cd and C e Their ratio C,,.fCe = R. Using Equation 24 and calculating Re e

for the drop we obtain m. I is the small positive root of Equation 27.
Having established the holdup for the bottom of the preheater, we can now
calculate the superficial velocity of the continuous phase at the bottom of the
column above the dispersed phase injector. Either Equation 22 or 23 can be used.
Equation 23 is sufficient unless the holdup is very high, i.e., > 0.35. The
superficial velocity is defined as the mean velocity of the continuous phase across
the entire column. Thus,
Dc = [! (CjA) ff2 (38)
gives the needed diameter of the column just above the dispersed phase injector.
If the column was isothermal, selecting the diameter just calculated would
insure its holdup being constant along its entire length. With heat transfer, the
holdup may vary. This is due to the fact that the density of the brine and
selected working fluid can vary considerably with temperature if the column pres-
sure is sufficiently high. If the working fluid density varies, the drop diameters
will vary and thus their terminal velocity, etc. No where along the column length
should the holdup exceed 0.9/. Thus, we should next calculate the conditions at
the top of the preheater in a manner similar to what was done at the bottom.
The column diameter should be whichever is larger.
Having determined the column diameter, we should next check the overall
size of the dispersed phase distribution plate. The nozzle holes should not be
placed on a center to center distance less than 1.5 dD With this type of layout,
the overall distributor plate injector should not be larger in diameter than the
column. This will insure that the drops can rise vertically.
The columns shown in Figs. 5.1 and 5.2 both have a conical section in which
the dispersed phase injector is located. Jacobs and Boehm (1980) and Treybal
(1963) indicate that the cone half angle should be about 15 0
The distributor

plate should be so located that the annular ring of open space around it should
equal the cross-sectional area of the column. This will insure that there is no
undercarry of the dispersed phase and that there is good separation of the phases.
Weare now ready to proceed with the calculation of the length of the
preheater part of the column. If the thermodynamic properties of the working
fluid liquid and brine do not vary significantly with temperature, it is possible to
calculate the length of the preheater from
mihbp - hi,.)d (39)

U.(:
Lp .H. = -"":"":'-=--"'-"--
D'6)LMTD
where hbp is the enthalpy of the working fluid liquid at the boiling point, hi,. is its
enthalpy at the inlet, U. is the volumetric heat transfer coefficient and LMTD is
the log mean temperature difference across the preheater assuming counterflow.
Unfortunately for fluids like isobutane, the specific heat varies considerably with
temperature as was noted in the design of the 500 kWe unit at East Mesa. Thus,
it is necessary to evaluate the heat transfer in a number of steps along the
preheater length. This was done by hand using Equations 31-33 for the 500 kWe
unit (Olander, et al., 1983). For design studies it is recommended that the steady
state computer program described in Jacobs (March, 1985) be used to determine
the preheater length. The computer program is easily modified to include a
variety of methods for estimating the heat transfer rate. For spherical drops, it is
recommended that the variable diameter drop conduction model described by
Jacobs and Golafshani (1985) be used to determine the preheater length. For
fluctuating ellipsoidal drops, Equations 34 and 35 can be used.
The length of the boiler section can be calculated using the value of U.
given in Equation 36. Due to the nature of the equation as discussed, it is not
warranted to divide the boiler into segments. Thus the length of the boiler should
be calculated as
L b= m(hezit - hb.pJd
!!... D't; U. LMTD
4
where LMTD is based on counterflow temperatures across the boiler.
Based on results of the 500 kWe design, the length of the spray column
should start at the top of the conical section housing the dispersed phase distribu-
tor.
It is recommended that the continuous phase injector or injectors be located
in the middle of the boiler section to insure maximum heat transfer rates.
A disengagement section at the top of the column needs to be included as
shown in Fig. 5:2. This should be sufficiently large to insure no liquid carryover
with the exiting vapor.
A sensitivity analysis should be made to determine possible input variations
and the time constant for the column such as was done by Golafshani and Jacobs
(1985) for the 500 kWe unit. The controls for the column should be designed
based on this information and the resulting information indicating possible
preheater and boiler length excursions.
Following the above procedures it is possible to design a highly reliable
spray column for geothermal power applications. The techniques posed appear to
be conservative based on experience with the 500 kWe East Mesa unit. Further
refinements will require additional laboratory studies as mentioned in the discus-
sions in this manual.
10 APPLICATIONS OF THE METHOD TO ISOBUTANE-WATER

SYSTEMS
Isobutane has been shown (Jacobs and Boehm, 1980) to be an excellent choice as a
working fluid for geothermal brines when the brine temperature is above 300 F 0
("'149 C). This section presents some calculations using the methods discussed in
0
this Appendix to show the influence of various parameters on spray column design.
APPENDIX 5 361
800
700 I SOBUTANE - H20

u
II)
TH20 = 150 of (65.5 Oe)
~ 600
E
E 500 Vjc (MAXIMUM VELOCITY TO
MAINTAIN DROP SIZE)
>-
~ 400
U W. = 2 (LOWER LIMIT TO
0 300 PREVENT SEEPING)
-'
w
> 200 VJ (M I N I MUM VELOCITY)
100
0
0 5.0 6.0 7.0
NOZZLE DIAMETER, mm
Figure 5.4 Typical limiting velocities for changes in nozzle diameter.
10.1 Establishing Drop Size
In designing a spray column, it is necessary to choose the drop size in order to
establish the column hydrodynamics. Figure 5.4 was developed for the isobutane-
H 2) system utilizing Equations 1-5. For a small nozzle diameter, it is clear that a
wlae range of nozzle velocities is possible between the working limits of Vi and
Vic' However, as the nozzle diameters increase, the range of permissible velocities
decrease. Thus, it would appear safer to design using nozzles of < 3.0 mm in
diameter if one wished to provide for significant variations in velocity. However,
since the area of the nozzles is proportional to their diameter squared, near
equivalent volume flow changes may occur for much smaller changes in velocity
for the larger nozzles.
H one considers the actual drop sizes as a function of the nozzle velocities, it
is clear from Fig. 5.5 that nearly the same variation in drop diameters is possible
over all the nozzle sizes shown. However, the approximate nozzle diameter limit
for non-circulating drops is less than 2.5 mm (7/64 inch) and at velocities near the
jetting velocity. As it is most easy to hydrodynamically design spray columns for
situations where the drops behave as rigid spheres, typical nozzle diameters of
1.5-2.0 mm are normally chosen. Plugging or partial plugging of some holes can
lead to velocities exceeding the critical jetting velocity. Reductions in working
fluid flow rate can cause drop diameters to exceed the limit for rigid sphere
behavior. Thus, fluctuations in spray column operation can occur even without
recirculation. Nonetheless, the advantages of spray column direct contact heat
exchange make them attractive to pursue. It should .also be noted that the limits
for rigid drop behavior and critical jetting velocity are experimentally established
and are, in general, conservative.
16 I I I
14 l- -
E
E 12 :- ISOBUTANE- H20 -
IX: - THi>= 150 OF (65.5 Oe)

-
.-
lIJ 10 I-
lIJ
~
8 I- -
<[
6 - ~:::~':OR RIGID DROPS
-
0
a.. 4 I- -
0
IX:
0 2 - VJc -
0 I I I I I I
0 2 3 4 5 6 7
NOZZLE DIAMETER, mm
Figure 5.5 Variation in drop sizes produced by nozzles within operating velocity
limits.
Figure 5.6 shows the terminal velocities calculated using Equation 15 for
rigid drops and Equation 18 for fluctuating drops. For all drops formed between
the limits on Weber number and Vic for the nozzle dimensions considered, isobu-
tane drops in H 20 fall within either the spherical or fluctuating ellipsoidal regime
shown in Fig. 5.3 as log M is approximately -12.16 and .7 ~ Eo ~ 2.6. It should
further be noted that the terminal velocity is higher for drops with internal circu-
lation by up to 50%. Thus, it should be possible to increase the throughput by
increasing drop size. However, as the path of larger drops is not vertical, coales-
cence is more likely to occur. Therefore, at this time, it does not appear prudent
to significantly increase drop size without further experimental data on drop
hydrodynamics, especially for drop swarms.
10.2 Establishing Flooding Limits

Data on flooding in spray columns has primarily been determined in small diame-
ter columns where change of phase of the volatile fluid does not occur. This has
been noted in the preceding sections, and is the case for Equation 27. This corre-
lation was developed for the spherical drop regime. Equation 23 presents a corre-
lation for the continuous phase superficial velocity also for the rigid sphere regime.
Utilizing these two correlations, the nondimensional, continuous phase, superficial
velocity as a function of holdup, , for a range of the volumetric flow ratio, R, is
shown in Fig. 5.7. Note the flooding limit, " and 0.9, and 0.8, are also shown.
The range of R is consistent with the use of DCHX systems with moderate tem-
perature geothermal brines and solar ponds. For this range, the flooding condition
APPENDIX 5 363
800
1129
ISO BUTANE - HzO
THO = 150 of (65.5 C)
2
1
0
Q)
600 RIGID SPHERES
TERMINAL VELOCITY FOR
'"
"-
REGIME
WOBBLING ELLIPSOIDAL
E
E
>- 400 TERMINAL VELOCITY FOR

t-
u RIGID DROPS
0
.J
lLJ
> 200 WOBBLING ELLIPSOIDAL
REGIME
DROP DIAMETER, mm
Figure 5.6 Terminal velocities for rigid spheres and fluctuating drops according
to Figure 5.3.
corresponds to a nearly maximum value of continuous phase superficial velocity.
The peak value of f / varies from nearly 0.34 at an R of 2.25 to 0.25 for R = 0.5.
In each case, the superficial velocity only undergoes a small change in order to
reduce to 0.8 / or 0.9f /.
The advantage for heat transfer of operating close to the flooding point is
shown in Fig. 5.8. For the case of small values of R, as is the case for solar ponds,
the correlation of Plass, et al. (1979), indicates volumetric heat transfer coefficients
of nearly 8,000 Btu/hrft3 , F (148.8 kW /m~() when operated near flooding. Of
course, it should be noted that in the liquid preheating regime, such a low value of
R leads to low utilization of the heating capacity of the continuous phase brine.
In typical geothermal applications, it is desired to utilize a considerable portion of
the thermal capacity of the hot liquid brine. With an organic working fluid such
as isobutane, a typical value for R would be around two. The correlation of
Reference 15 indicates a decrease in Uv with increasing R, but an increase with f.
Since the rate of increase of Uv with increasing f is less for higher values of R,
operating further away from the flooding point does not, as significantly, affect U.
as it would at lower R, more typical of solar pond applications.
Figure 5.8 can be used for preliminary design applications provided we keep
in mind the data from which it was determined. The limitations are: (a) the
dispersed phase is an organic fluid; (b) the continuous phase is brine or water; (c)
0.6 r-----.------.------.------r----~
0.5
0.4
-
N
::::
~
~ ~0.3
o ~
c> -a
Q.
<J
_0.2
0.1
o ~----~~----~------~------~------~
o 0.2 0.4 0.6 0.8 1.0
HOLDUP, E
Figure 5.7 Superficial continuous phase velocity as a function of holdup for
spherical drops.
the drops are less than 5.0 mm in diameter; and (d) the correlation is good to only
within 20%.
11 EXAMPLE OF DCHX DESIGN

Consider the design of a DCHX preheater-boiler for brine entering at a rate of
85,840 lb Ihr. Isobutane at a flow rate of 91,434 lb Ihr enters the DCHX at
86' F. Tlfese mass flow rates correspond to 327.9 gp: of IC4 and 192 gpm of
brine. The boiling point of the IC4 is 244.5' F (corresponding to a vapor pressure
of 400 psi), and the temperature of the brine entering the preheater is 251' F.
The brine exits the column at 129.5 F.
6 The change in enthalpy of the isobutane across the preheater is 10.332 X
10 Btu/hr. In order to size the column, we must first choose the size drops we
APPENDIX 5 365
104
R=
8Xl0 3
0.5
U.
-
0
'". 6x 103
~
.....
~ 4x 103
CO
l,.
::l
2x 10 3
0
0 0.1 0.2 0.3
HOLDUP, E
Figure 5.8 Volumetric heat transfer coefficient as a function of holdup from

Plass et al., 1979.
desire. Let us choose 3.5 mm diameter at entrance. This size can be achieved for
nozzle diameters less than 1.95 mm (see Fig. 5.5). The maximum velocity, Vic'
will be 540 mm/sec (1.772 ft/sec) for this size nozzle. Let us choose a smaller noz-
zle diameter. One-sixteenth inch diameter holes can be drilled (1.587 mm) to form
nozzles. For this size nozzle, the critical jetting velocity is 630 mm/sec and the
critical drop diameter dDC = 2.5 mm. Using Equations 2 and 9, we can solve for
die and Vn, the nozzle velocity. The equations yield die = 1.46 mm and Vn = 430
mm/sec (1.47 ft/sec). This velocity is in the midpoint range between the Vic and
We = 2.0 limits. The number of holes required for the distributor plate can be
calculated from Equation 37. T~e nearest w~ole number yields 23,328 nozzles.
The area of the nozzles is 0.497 ft or 0.0462 m .
Next we must calculate the terminal velocity of the drops . .Ai3 they are in
the spherical regime (see Fig. 5.6), the terminal velocity can be calculated from
Equation 15. The terminal velocity at the bottom of the column is 280 mm/sec or
0.919 ft/sec. Before determining the superficial velocity of either phase, we must
next calculate the holdup at flooding.
Using the terminal velocity, the Reynolds number of a drop is calculated.
This is necessary to establish the Re e regime. .Ai3 the Re e is greater than 500, Ret
= 2212, we obtain m = 1.39. From Equation 27, we next find the value of the
holdup at flooding, f,/, Given the ratio of flow rates R = 1.708 and m = 1.39, (/
is 0.326. If we choose to design for 0.9( /' then ( = 0.2934. From Equation 23, we
can calculate the superficial velocity at the bottom of the column. The chosen
conditions yield GclA = 0.0784 ft/sec or 23.88 mm/sec.
Since we know the volumetric flow rate of the brine Ge , the diameter of the
column proper at its base can be calculated from Equation 38, to be 2.80 ft or
0.855 m. Although this would be an appropriate diameter for the DCHX column
for conditions near the entry of the working fluid, large changes in the density of
isobutane can occur at pressures near 400 psi as it is heated to its saturation tem-
perature. Thus, the flooding conditions will have to be checked at several loca-
tions along the column length in the preheater section. At the top of the
preheater section, the 3.5 mm drops will have grown to a diameter of nearly 4.0
mm. The terminal velocity will increase and generally the deviation in holdup
will be to one further from the flooding point.
In order to establish the approximate length of the preheater, Equations 32
and 39 can be used. From Equation 32, U~ is approximately 4,650 Btu/hrft3 F or
86.5 kW/m 3K. The LMTD for the preheater is 21.8' F. Thus, from Equation
39, the approximate length of the preheater is 16.55 ft or 5.04 m.
In the boiler section, 5.85 X 106 Btu/hr will be transferred to the isobutane.
Equation 36 yields a U~ = 14.083 Btu/hrft3 F. The LMTD across the boiler is
31.5' F. Thus, the boiler volume is 13.19 ft3 . The boiler length is 2.14 ft or 0.65
m.
We now have a preliminary sizing of the column proper. The diameter is
2.8 ft (0.855 m) and the combined length of the preheater and boiler is 18.7 ft
(5.70 m). Using these preliminary dimensions, the steady state computer program
described by Jacobs (March, 1985) can be used to refine the size for the preheater.
Such a calculation yields an overall length of 19 ft (5.8 m).
For a vapor-mist disengagement section, a height of two diameters is sug-
gested (5.6 ft or 1.71 m). Two rows of chevron mist eliminators located mid-way
in the disengagement section would eliminate any droplet carry-over. The chev-
rons should be two inches high and inclined at 60' from the vertical.
The brine injector would be located near the midpoint of the boiler section.
An injector might be designed to distribute the brine horizontally from a single
tube. It should yield a horizontal velocity no greater than GjA at the wall of the
column. This would require the injector distribution over a height of 7.4 inches or
18.9 cm. A design such as used for the 500 kW East Mesa DCHX unit described
by Olander, et aI. (1983), would be satisfactory. e
The isobutane distribution pl~te, as noted, would require 23,328 1/16 inch
diameter holes spaced over 6.16 ft . This is equivalent to one hole every 0.038
square inches, or 0.22 inches or 5.59 mm between centers. The distributor should
be located in the conical frustrum section below the column proper where the
diameter is 3.96 ft. Ideally, the frustrum 1/2 angle would be 15'. This will allow
for the brine to pass by the IC 4 distributor with no increase in the continuous
phase superficial velocity. The orine exit would be located on the pressure head
below the column in a manner similar to that shown on the schematic of Fig. 5.l.
This design example used flow conditions identical to those for the East
Mesa DCHX observed on November 12, 1980 (Olander, et aI., 1983). The East
Mesa DCHX had a diameter of 3.67 ft. Under these flow conditions the holdup, f,
APPENDIX 5 367
for the East Mesa DCHX was only 0.227, which was only 70% of the flooding
value. ~he resulting U. calculated from Equation 32 would ~ave been 2,718
Btu/hrft F. The experiment indicated a U. of 2,390 Btu/hrft F. The com-
parison is within the 20% claimed by Plass, et al. (1979).
12 NOMENCLATURE
A Internal cross-sectional area of the column
dD Drop diameter
dDC Critical drop size according to Treybal
die Drop diameter at critical jetting velocity
d" Nozzle diameter
i1p I(Pc - Pd}1
Dc Column diameter
Ee,Ed Dispersion coefficients, empirically determined constants used in one
dimensional conservation equations to account for backmixing
EO'tvOs number, defined by Equation 8 and 11
Void fraction, local global fraction of volume occupied by the dispersed
phase
Void fraction at the flooding point
Drag coefficient defined by Equation 16 for rigid spheres and Equation
17 for surface mobile spheres
g Gravitational constant
Gc Volumetric flow rate of continuous phase
Gc/A Superficial velocity of the continuous phase
GD Volumetric flow rate of dispersed phase
hbp Enthalpy of liquid at the boiling point
hezit Enthalpy at exit from the column
hi" Inlet value of enthalpy
H Defined by Equation 21
J Defined by Equation 19 and 20
k,n Constants in Equation 22
K Defined by Equation 4
Kl Defined by Equation 13
L Length of column
LMTD Log mean temperature difference for countercurrent flow
Lp.H. Preheater length
m Constant in Equation 23 defined by Equation 24
Pc Dynamic viscosity of continuous phase
M-gro'Up Defined by Equation 12
n The number of nozzles required in the dispersion plate in Equation 37
NUd Nusselt Number defined on the basis of drop diameter
Pr Prandtl number
Q/Vol Total local heat transfer between phases per unit volume
31)8 DffiECT-CONTACT HEAT TRANSFER
Pc Continuous phase density

Pd Dispersed phase density
R GD/GC
Re Reynolds number, defined by Equation 10
Interfacial tension
Volumetric heat transfer coefficient defined by Equation 33
Maximum velocity in nozzles to insure near uniform drop formation
Jetting velocity, minimum velocity in nozzle to insure all nozzles are
flowing
Actual nozzle velocity corresponding to d", actual nozzle diameter and
flow rate of dispersed phase
Terminal drop velocity for surface mobile spheres calculated from
Equation 18
We
u
REFERENCES
Golafshani, M., H. R. Jacobs. Stability of a Direct Contact Spray Column Heat Exchanger,
ASME/AIChE National Heat Transfer Conference, Denver, CO (1985).
Grace, J. R. "Hydrodynamics of Liquid Drops in Immiscible Liquids," Chapter 38, Handbook of Fluids
in Motion, N. P. Cheremisinoff and R. Gupta, Editors, Ann Arbor Science, The Butterfield
Group, Ann Arbor, MI, pp. 1003-1025 (1983).
Jacobs, H. R. Stability AnalY8is of Direct Contact Heat Exchanger8 SubJect to System Perturbation8,"
Final Report-Task 2, U.S. Dept. of Energy Contract DE-AS07-76IDO 1523, Modification A014
(March 1985).
Jacobs, H. R., R. F. Boehm. "Direct Contact Binary Cycles," Section 4.2.6 Sourcebook on the Produc-
tion of Electricity from Geothermal Brines, J. Kestin, Editor, Published by U.S. Dept. of
Energy, Washington, D.C., DOE/RA/4051-1, pp. 413-471 (March 1980).
Jacobs, H. R., M. Golafshani. "A Heuristic Evaluation of the Governing Mode of Heat Tran8fer in a
Liquid-Liquid Spray Column," ASME/AIChE National Heat Transfer Conference, Denver, CO
(1985).
Johnson, A. I., G. W. Mirand, C. J. Huang, J. H. Hansuld, V. M. McNamara. "Spray Extraction
Tower Studies," AIChE Journal 3:101-110 (1957).
Kumar, A. D., K. Vohra, S. Hartland. Canadian J. Chemical Engineering, 53:158 (1980).
Letan, R. Design of a Particle Direct Contact Heat Exchanger: Uniform Countercurrent Flow,
ASME Paper 76-HT-27, ASME/AIChE Heat Transfer Conference (1976).
Letan, R., E. Kehat. "The Mechanism of Heat Transfer in a Spray Column Heat Exchanger," AIChE
Journal, 14(3):398-405 (1968).
Olander, R, S. Oshmyanshu, K. Nichols, D. Werner. Final Phase Testing and Evaluation of the 500
k We Direct Contact Pilot Plant at East Mesa, U.S. Dept. of Energy Report DOE/SF /11700-TI,
Arvada, CO (December 1983).
Plass, S. B., H. R. Jacobs, R F. Boehm. "Operational Characteristics of a Spray Column Type Direct
Contact Preheater," AIChE Symposium Series, 189:227-234 (1979).
Richardson, J. F., W. N. Zaki. "Sedimentation and Fluidization, Part I," Transactions of the Inst. of
Chemical Engineers, 32:35-53 (1954).
Riemer, D. H., H. R. Jacobs, R F. Boehm. Analysis of Direct Contact Binary Cycles for Geothermal
Power Generation (Program DIRGEOj, University of Utah, U.S. Dept. of Energy Report
IDO/1549-5 (1976).
Riemer, D. H., H. R. Jacobs, R F. Boehm, D. S. Cook. A Computer Program for Determining the
Thermodynamic Propertie8 of Light Hydrocarbon8, University of Utah, U.S. Dept. of Energy
APPENDIX 5 369
Report IDO /1549-3 (1976).

Riemer, D. H., H. R. Jacobs, R. F. Boehm. A Computer Program for Determining the Thermo-
dynamic PropertieB of Freon RefrigerantB, University of Utah, U.S. Dept. of Energy Report
IDO /1549-4 (1976).
Riemer, D. H., H. R. Jacobs, R. F. Boehm. A Computer Program for Determining the Thermo-
dynamic PropertieB of Water, University of Utah, U.S. Dept of Energy Report IDO/I549-2
(1976).
Rivkind, V. Y., G. M. Ryskin. "Flow Structure in Motion of a Spherical Drop in a Fluid Medium at
Intermediate Reynolds Number," Fluid DynamicB, 1:5-12 (1976).
Saline Water Conversion Engineering Data Boole, 2nd Edition, published by the M. W. Kellogg Com-
pany, Piscataway, NJ, for the U.S. Office of Saline Water, Contract No. 14-30-2639 (November
1971).
Sideman, S. "Direct Contact Heat Transfer Between Immiscible Liquids," AdvanceB in Chemical
Engineering, Vol. 6, Academic Press, New York, NY (1966).
Skelland, A. H. P., K. R. Johnson. Canadian I. Chemical Engineering, 52:732 (1974).
Steiner, L., S. Hartland. "Hydrodynamics of Liquid-Liquid Spray Columns," Chapter 40 Handbook of
Fluids in Motion, N. P. Cheremisinoff and R. Gupta, Editors, Ann Arbor Science, The
Butterfield Group, Ann Arbor, MI, pp. 1049-1092 (1983).
Treybal, R. E. Liquid Eztraction, 2nd Edition, McGraw-Hili, New York, NY (1963).
Walter, D. B. An EzperimentallnveBtigation 0/ Direct Contact Three-PhaBe Boiling Heat Transfer,
Ph.D. Dissertation, The University of Tennessee, Knoxville, TN (1981).
APPENDIX
THERMAL DESIGN OF
WATER-COOLING TOWERS
John C. Campbell
1 INTRODUCTION
Water-cooling towers are used for the removal of a major portion of the heat gen-
erated in industrial plants. The basic principle of operation is the direct contact
of the heat-laden water with atmospheric air, resulting in cooling the water by
evaporating a portion of it. Since the latent heat of vaporization of water is in the
order of 4,000 times the specific heat of the cooling air, the sensible heat
transferred by the latter is generally insignificant, and the process may be con-
sidered to be the removal of heat by means of mass transfer.
The basic heat transfer equations are fairly simple, since the only primary
fluids involved are air and water. Since the transfer is primarily by evaporation of
water, the temperature difference between the two streams cannot be used to
determine the driving force; instead enthalpy difference is generally used. The
problem of accurate evaluation of the magnitude of the effective surface through
which the heat and mass transfer occurs adds a complication not usually present
in indirect contact heat transfer. This is due to the fact that in many types of
cooling tower packing the fill faces provides only part of the transfer surface, the
371
rest being the surface of the water droplets. The number, size, and shape of these
droplets vary widely with such variables as fill arrangement, type of fill, operating
conditions, and air velocity. The difficulty of accurate evaluation of total transfer
surface is circumvented by using the transfer coefficient and corresponding area as
a single term. The familiar heat transfer equation
Q = (u)(A)(.1T) (1)
is thus modified to
Q = (Kg)(a)(v)(.1h) (2)
Substitution (L)(0,,)(T1 - T2) for Q, and simplifying by assuming that 0" = 1.0,
the integral form of Equation (2) is the famous Merkel [1] equation:
Tl
KaV = J dT (3)
L To
2
hw - h.
The term KaV/L has been given several names; popular ones are "tower
characteristic" and "number of diffusion units." Calculated using Equation (3), it is
representative of the required heat and mass transfer surface for given thermal
conditions.
r
A companion equation is convenient to express the capability of a specific fill
design and arrangement:
K~V = 0+ M [ ~ (4)
The constants 0, M, and n are normally developed from experimental data.

Equations (3) and (4) are the basic tools generally used for cooling tower
thermal design. The former, in single integral form, is applicable to counter-
current flow, and can be evaluated numerically by the Tchebycheff method [2], or
by other similar approximation methods. For crossflow, double integration is
required, and numerical evaluation is best achieved by aid of a computer. For
convenience, elaborate sets of curves have been prepared for both counterflow and
crossflow, relating required characteristic to L/G ratio for a wide range of thermal
conditions [3,4]. These curves, often termed demand curves, are particularly use-
ful to those who have only occasional need for design calculations and do not have
computer assistance.
For given thermal requirements, a solution is readily achieved by the follow-
ing procedure:
(a) Select a suitable type and arrangement of fill, and establish the allowable
L/G ratio. Direct solution can be made by determining the intersection of
the applicable demand curve and the fill character'8tic curve.
(b) From heat balance and psychrometric data establish properties and
volumetric flow rates of inlet and exit air streams.
APPENDIX 6 373
(c) Choose a cooling tower size and shape appropriate for the fill arrangement
and air and water flow rates.
(d) Compute air velocities and pressure losses through the cooling tower.
(e) Design equipment for providing the necessary flow rates and distributions of
the air and water streams.
Many combinations of
type of cooling tower
fill type
active volume
packed height
air flow rate
will yield workable cooling towers for a given heat removal rate. The skilled
designer will pare this multitudinous number down to the one selection of greatest
value to the user, considering first cost, operating power, ground area, flexibility,
maintenance costs, and other fertinent factors.
The following examples are presented to clarify the foregoing brief descrip-
tion.
2 MECHANICAL DRAFT COOLING TOWER

Problem: Design a cooling tower suitable for the following conditions:
Water circulation rate, gpm 40,000

Inlet water temperature, F 0 104
Outlet water temperature, F 0 86
Wet-bulb temperature, of 77
Dry-bulb temperature, F 0 95
Elevation sea level
Cost of power for fans, $/HP 2,000
Cost of power for water pumps,
$/foot of pumping head 22,000
Maximum ground area, LX W, feet 200 X 60
(a) The design conditions indicate that a counterflow cooling tower requiring
moderately low fan and pump power may be optimum; therefore this illus-
tration of one of many possible selections will follow these guidelines. A film
type fill requiring relatively low packed height and low resistance to air flow
will be used; and is identified as No. J-5. The characteristic curve for this
fill is shown in Figure 1. This curve may be superimposed on the
counterflow demand curves applicable to the design wet-bulb temperature of
*The units used in these examples conform to industrial practice.

5.0
4.0
3.0
2.0
KaV
-L-
1.0
0.5
0.2
Type J
Fi 11
0.1 1. 5 2.0 3.0 7.0 1.0 2.5

L
G
Figure 6.1 Characteristic curves for type J film fill.
77 F and cooling range of 18 F, as shown in Figure 2, establishing the
allowable LjG ratio of 1.41.
(b) The properties and volumetric flow rates of the inlet and exit air streams are
determined from equations and psychrometric data as follows:
APPENDIX 6 315
10
5
4
~----
3
KaV 1.5
-L-
1.0
0.7 24
28
5.0
4.0
3.0
Fi 11
2.0 J-5
0.1 0.2 0.3 0.40.5 1.0 1.5 2.0 3.0 4.05.0

L
G
Figure 6.2 Characteristic curve J-5 and "demand" curve 104/86/77.
L = (gpm) (lb/gal) = (40,000) (8.280) = 331,200 16/min
G = (L)/(LIG) = 331,200/1.41 = 234,900 lb/min dry air
Q = (L) (T1-T2 ) (Op) = (331,200) (104-86) (1.00) = 5,961,600 Btu/min
..1h = QIG = 5,961,600/234,900 = 25.38 Btu/lb dry air
hI @ inlet conditions = 40.57 Btu/lb dry air
h2 @ exit conditions = 65.95 Btu/lb dry air
Volumetric flow rate, AOFM = (G) (V)
Inlet air: AOFMl = (234,900) (14.563) = 3,420,850

Exit air: AOFM2 = (234,900) (14.891) = 3,498,000
(c) The economic advantage of construction standardization normally dictates

the bay sizes to be used by the thermal designer; for wood framed
counterflow cooling towers, column spacings of 6' or 8' are popular. For this
example, assume the designer will use 8' X 8' spacing. For large heat remo-
val rates, fairly large cell sizes will usually be cost effective. Based on these
guidelines, and on the stated ground area restriction, try a cell width of 6
bays, or 48'. Since the value of fan power is fairly high, assume moderately
low air velocity to ensure low resistance to air flow, say in the order of 300
to 400 feet per minute through the fill inlet plane. A rough estimate of a
trial tower length can now be made:
. AOFMl
estimated length, ft = (est. Ul
)(W f)
,t
= 3,420,850 = 203.6 ft
(.....,350)(48)
This slightly exceeds the 200' plot length limitation, and leads to the selection of
four 48' long cells. Thus this selection will be four in-line cells, each 48'W X 48'L,
with a total active plan area of 9,216 square feet.
(d) The design air velocities through the fill, based on nominal plan area, will
be:
Fill inlet: Ul = 3,420,850/9,216 = 371.19 ft/min
Fill exit: U2 = 3,498,000/9,216 = 379.56 ft/min
For convenience, pertinent air properties and flow rates are tabulated below:
INLET EXIT AVERAGE
WBT, OF 77.0 96.6 86.8
DBT, OF 95.0 96.6 95.8
V, ft 3 jlb dry air 14.563 14.891 14.727
p, lb mix./ft 3 mix. 0.06976 0.06974 0.06975
0, pip, 0.9314 0.9311 0.93125
ACFM, total 3,420,850 3,498,000 3,459,400

ACFMper cell 855,200 874,500 864,850
U through fill, ftjmin 371.19 379.56 375.37
The major pressure losses to cooling tower air flow are:

(1) tower inlet
(2) fill
(3) distribution system
(4) mist eliminators
(5) plenum and structure
APPENDIX 6 377
(6) water spray

(7) contractions, expansions, and changes in direction
(8) net velocity pressure at exit
(9) fan entrance
The summation of these is the total pressure differential at design conditions.
(1) tower inlet

Assume this example tower is equipped with air inlet louvers of a design charac-
terized by the manufacturer's empirical equation
t1P = (5.0)(U)1.82(O)
(5)
107
Also assume a design louver face velocity of 800 ft/min.
t1P = (5.0)(800)1.82(0.9314) = 0.0895 II H 2 0
107
(2) fill
The pressure drop through the fill is a function of:
air velocity: U a = 375.37 ft/min.
water loading: gpm/ft 2 = 40,000/9,216 = 4.34
relative air density: 0a = 0.093125
Using the manufacturer's data, Fig 3, and correcting for deviation from standard
air density,
t1P = (0.1490)(0.93125) = 0.1388 II H 20

This air pressure loss is primarily due to the energy of the water spraying counter-
currently to the air stream. The water distribution piping presents some addi-
tional restriction. Manufacturer's data, Fig 4, is applicable to this example. At U2
= 379.56 ft/min and a water loading of 4.34 gpm/ft2, the corrected pressure drop
IS
t1P = (0.0210)(0.9311) = 0.0196 II H 2 0

The pressure drop across the type of mist eliminators used in the example cooling
tower is given by the manufacturer's equation
(6)
0.6
0.5
0.4
Pressure
0.3
drop,
in. H 2 0
0.2
~ Water
loading,
gpm/fF
0.1
100 200 300 400 500 600

Air velocity,
U a , ft/min
Figure 6.3 Pressure drop data for film fill No J-5.
where u is the net air velocity. For this example the mist eliminator frames and
supports block 7.0 percent of the nominal face area.
net face area = (0.93) (9,216) = 8,570 ft 2
U net = 3,498,000/8,570 = 408.13 ft/min.
LlP = (0.39)(408.13)1.82(0.93125) = 0.0205 II H 20
106
APPENDIX 6 37D
0.08
0.07
0.06
Pressure
drop,
0.05 in. H 2 0
0.04
0.03
0.02
o
\ Water loading,
gpm/fF
0.01
Air velocity,
U2 , ft/min
100 200 300 400 500 600
Figure 6.4 Distribution system pressure drop.

The applicable equation Cor the example tower is
LlP = (1.70) [~r((J2)

4005
(7)
= (1.70) [ 3:~~6 f(0.9311) = 0.0142 II H 20

380 nffiECT-CONTACT HEAT TRANSFER
(6) water spray

The air pressure loss due to the water spray, exclusive of the distribution system
spray accounted for in item (3), is a function of the average air travel below the
fill, the water loading, and the average air velocity through this "rain" area:
1.&5
fjP = (0.0031) [ ulm. ] (Jf.)(~)0.40(/Jd (8)
2 X lOS 4 It Z
= (0.0031)[ 800
2 X lOS
]1.&5[~](4.34r40
4
(0.9314)

For the example tower, these flow abnormalities result in a total pressure drop
equal to 1.50 velocity heads, referred to the average face velocity:
fjP = (1.50) [~r(/Ja)

4005
(9)
= (1.50) [ 3~~~7 f(0.93125) = 0.0123" H 0 2

Assume that 30 ' diameter fans will be used for air movement, and that the velo-
city recovery stacks selected will recover, or "regain", 75% of the velocity pressure
difference between the fan and the stack exit plane.
net A, = (.!. )[(D,)2 - (DII)~

4
= (.!. )[(30.0)2 - (6.5)Z] = 673.7 ft Z

4
u, = ACFMz/A, = 874,500/673.7 = 1298.1 ft/min
VP , [~r(/J) = [1298.1
= 4005 24005
f(0.93ll) = 0.0978" H 0
Z
A. = ('!')(D.)Z = (.!.)(33.0)Z = 855.3 ft 2

4 4
u. = ACFM2/A. = 874,500/855.3 = 1022.4 ft/min
VP

= [~r(/J)
4005
= [1022.4
24005
f(0.93ll) = 0.0607" H 0
2
APPENDIX 6 381
VP recovered = (0.75)(VP, - vp.)

= (0.75)(0.0978 - 0.0607) = 0.0278" H 20
net VP = VP, - VPro<.
= 0.0978 - 0.0278 = 0.0700 " H 20
(9) fan entrance

An ideally designed inlet bell will virtually eliminate fan entrance loss, while a
very poor design will result in a loss approaching the velocity pressure at the fan.
The fan inlet bells on the example cooling tower are designed for a loss equal to 14
percent of VP,:
LlP = (0.14)(0.0978) = 0.0137" H 20
Summary of pressure losses
(1) tower inlet 0.0895" H20

(2) fill 0.1388" H20
(3) distribution system 0.0196" H 20
(4) mist eliminators 0.0205" H20
(5) plenum and structure 0.0142" H 20
(6) water spray 0.0114" H 20
(7) contractions, expansions, etc. 0.0123" H 20
(8) net velocity pressure at exit 0.0700" H 20
(9) fan entrance 0.0137" H 20
Total 0.3900" H 20
(e) The height of the example cooling tower is 30 feet, measured from basin
curb to fan deck. This dimension is determined by:
(1) sufficient space for air inlet louvers, fill, fill support structure, spray
chamber, mist eliminators, and plenum chamber, and
(2) acceptably uniform distribution of air and water streams, and
(3) manufacturer's standards.
The velocity recovery fan stack extends 20 feet above the fan deck, for an overall
height of 50 feet.
The fans and speed reduction gears chosen for the tower have overall
efficiencies of 78% and 96%, respectively. The required driver-output power is
calculated from the equation
(ACFM2)(TP)
HP/cell = (6356)(fJ,)(17 g )
-,(.8: . :. 74..:J.,5~OO~)u.::(0:..:::.3~900~) = 71.7

(6356)(0.78)(0.96)
Normally the cooling tower would be equipped with either 75 or 100 horsepower
motors for driving the fans, the larger size enabling operation of about 12 percent
above the specified heat removal capability.
3 NATURAL DRAFT COOLING TOWER

Problem: Design a natural draft cooling tower for the following conditions:
Water circulation rate, gpm 400,000

Inlet water temperature, F 110
Outlet water temperature, F 80
Wetrbulb temperature, F 60
Dry-bulb temperature, F 72
Elevation sea level
Cost of power for water pumps,
$/foot of pumping head 230,000
Maximum ground area, L X W, feet 700 X 700
(a) This example will describe the thermal design of a counterflow natural draft
tower. For convenience, the fill selected will be the same as used in the
Example 1 mechanical draft type. The applicable characteristic and
"demand" curves, shown in Fig 5, intersect at LIG = 1.40. In this type of
tower, the relatively large volume of the rain area below the fill contributes
significantly to the heat and mass transfer surface. Previous experience with
similar designs indicates that, at the specified thermal conditions, this added
surface will increase the allowable LIG by 9.0 percent. Therefore a net
effective ratio of 1.40 X 1.09 = 1.526 will be used.
(b) The properties and volumetric flow rates of the inlet and exit air streams are
determined as in Example 1:
L = (400,000)(8.269) = 3,307,600 lb/min
G = 3,307,600/1.526 = 2,167,500 lb/min
Q = (3,307,600)(100 - 80)(1.00) = 99,228,000 Btu/min
.1h = 99,228,000/2,167,500 = 45.78 Btu/lb dry air
hi @ inlet conditions = 26.46 Btu/lb dry air
h 2 @ exit conditions = 72.24 Btu/lb dry air
ACFMI = (2,167,500)(13.578) = 29,430,300
ACFM2 = (2,167,500)(15.098) = 32,724,900
(c) Air flows through the natural draft cooling tower due to the existence of a
pressure differential .1P. between inlet and outlet. .1p. is the product of air
density difference ilp and effective stack height He:
(11)
APPENDIX 6 383
10
5
4
3
1.5
KaV 1.0
T
0.7
4.0
2.0 b--';~---------
Fi 11
J-5
L
Ii
Figure 6.5 Characteristic curve for fill J-5 and 110/80/60 demand curve.
At equilibrium, AP. will equal the total air flow resistance of the system E-1PR'
The air density difference is a relatively small number, so the stack will be
correspondingly high; natural draft towers are much taller than mechanical draft
types of similar capacity.
A convenient thermal design procedure consists of assuming a cooling tower
size and shape, computing EAPR , and then checking the required net effective
stack height using Equation {11}. If the ratio of height to diameter HID is not
suitable, different selections are calculated to determine the one with suitable
shape and greatest value to the owner.
In this example, assume acceptable HID ratios are between 1.20 and 1.50.
For a first try, assume that the air velocity at the fill inlet plane will be in the
order of 360 feet per minute.

est. AFI = AGFMI / ",360
= 29,430,300 / ",360 = ",81,800 ft Z
,,' DF, ~ [(AF' l[! If ~ -322 It
The assumed shape of the trial design is shown in Fig. 6.6.

(d) Nominal diameters, areas, and face velocities are shown in the following
tabulation:
Dia., Area, u,
ft ft Z ACFM ftlmin
Base 340 90,792
Fill inlet 322 81,433 29,430,300 361.40
Fill exit 319 79,923 32,724,900 409.46
Eliminator exit 317 78,924 32,724,900 414.64
Throat 223 39,057 32,724,900 837.87
Tower exit 260 53,093 32,724,900 616.37
Pertinent air properties and flow rates at tower inlet and exit are:
INLET OUTLET AVERAGE
WBT, OF 60 100.28 80.14
DBT, OF 72 100.28 86.14
w 0.00837 0.043588 0.025979
V 13.578 15.098 14.338
P 0.074265 0.069124 0.071695
() 0.991289 0.922667 0.956978
ACFM 29,430,300 32,724,900 31,077,600
The major resistances to air flow are similar in scope to those tabulated for the
Example 1 cooling tower.
(1) tower inlet

Assume an inlet height of 34 feet, with air blockage due to support structure of
10%.
Gross inlet area = (34)(71')[(340 + 322)/2] = 35,355 ft Z
Net inlet area = (0.90)(35,355) = 31,820 ft Z
Inlet air velocity = 29,430,300/31,820 = 924.90 ft/min
Using a pressure drop equal to 1.08 velocity heads:
,1p = (1.08) [924.90 f(0.991289) = 0.057101/ HzO

4005
APPENDIX 6 385
-------
I
c
fi 11
Figure 6.6 Natural draft cooling tower.

(2) fill
The average plan area of the fill is (81,433 + 79,923)/2 = 80,678 ft2. The
corresponding water loading is 400,000/80,678 = 4.958 gpmjft2. The average air
velocity is 385.4 ft/min. From Fig 3,
tJP = (0.1602)(0.956978) = 0.15331 " H 20

At the level of the distribution system the inside diameter of the veil is 318 feet.
A = ( : )(318)2 = 79,423 ft 2
u = 32,724,900/79,423 = 412.04 ftjmin
water loading = 400,000/79,423 = 5.036 gpmjft2
From Fig 4, tJP = (0.0235)(0.922667) = 0.02168" H 20

Assume the mist eliminator design is the same as in Example 1.
net face area = (0.93)(78,924) = 73,399 ft 2
Unet = 32,724,900/73,399 = 445.85 ftlmin
t1P = (0.39)(445.85)1.82(0.922667) = 0.02386" H 20
106

The resistance of the plenum and structure, including the friction of the inner wall
of the veil, is equal to 0.25 velocity head, based on the throat velocity:
t1P = (0.25) [837.87 f(0.92267) = 0.01010" H 20

4005
(6) water spray

The air pressure loss due to the resistance of the falling water droplets below the
fill is given by Equation 8, except that the average air travel is a function of DF,
rather than W:
t1P = (0.0031) [ uinlet

2X103
r
ss DF
(_1 )(~)0.40(O )
4 ft2 1
= (0.0031) [ 924.90
2 X 1~
r
ss
[322][ 400,000
4 81,433
r 40
(O.991289)

For the example tower, these losses total 1.30 velocity heads, based on the average
velocity through the fill:
t1P = (1.30)( 385.4 )2(0.956978) = 0.01152" H 20

4005

Velocity pressure recovery for the particular hyperbolic shape used in this example
is 73 percent of the difference between throat and exit:
throat VP = ( 837.87 )2(0.922667) = 0.04038" H 20

4005
APPENDIX 6 387
exit VP = ( 616.37 )2(0.922667) = 0.02185" H 20

4005
recovered VP = (0.73)(0.04038 - 0.02185) = 0.01353" H 20
net VP @ exit = 0.04038 - 0.01353 = 0.02685" H 20
Summary or pressure losses

(1) tower inlet 0.05710" H20
(2) fill 0.15331" H2 0
(3) distribution system 0.02168" H20
(4) mist eliminators 0.02386" H 20
(5) plenum and structure 0.01010" H 20
(6) water spray 0.11226 " H20
(7) contractions, expansions, etc. 0.01152" H20
(8) net velocity pressure at exit 0.02685 " H20
TOTAL = sum~PR 0.41668 " H 20
(e) The required effective stack height may now be computed, using Equation
11.
PI = 0.074265 Ib/ft 3
P2 = 0.069124Ib/ft 3
I1p = 0.OO5141Ib/ft 3
EI1PR
He = (0.1924)(l1p)(g/g.)
0.41668
(0.1924)(0.005141)(1.00) = 421.3 It
The required overall height is the sum of the effective height and the height
of the air inlet is approximately:
H = 421.3 + 34.0 = 455.3 It
The assumed size and shape depicted in Fig 6 will yield cooling capacity slightly in
excess of that specified, consistent with good design practice. However an addi-
tional trial or two would readily determine a size and shape to satisfy the equality
jjp. = EI1PR . The H/D ratio of the example tower is within the allowable lim-
its previously stated.
4 SUMMARY
The thermal calculations presented in Examples 1 and 2 cover, in each case, only
one of many possible selections. Present day practice enables the design engineer,
with computer assistance, to investigate all practicable possibilities and choose the
particular cooling tower that will be the best buy for the potential owner.
5 NOMENCLATURE
SYMBOL DESCRIPTION UNITS
a area of transfer surface per unit sq ft/cu ft

of tower volume
A area sq ft
A, area of stack at plane of fan sq ft
All area of fan hub seal disc sq Ct
A, area of fan stack at exit sq ft
AOFM actual cubic feet per minute AOFM
AF flow area at fill inlet sq ft
a a constant dimensionless
0" specific heat at constant pressure Btu/lb per F
D diameter ft
D, diameter of stack at plane of fan ft
DII diameter of fan hub seal disc ft
D. diameter of fan stack at exit ft
DBT dry-bulb temperature OF
DF diameter of fill inlet plane ft
g acceleration due to gravity ft/sec 2
ge conversion factor 32.2 Ibm' ft (Ib c'sec 2)
gpm gallons per minute gpm
h enthalpy Btu/lb dry air
h,. enthalpy of air - water vapor
mixture at wet-bulb temperature Btu/lb dry air
htO enthalpy of air - water vapor
mixture at bulk water temperature Btu/lb dry air
H height ft
H. net effective stack height ft
H/D height/diameter ratio dimensionless
HP horsepower lIP
K overall enthalpy transfer coefficient lb per hr per sq ft
per lb water per lb dry air
Kg enthalpy transfer coefficient of gas film lb per hr per sq ft
per lb water per lb dry air
KaV/L tower characteristic dimensionless
L length ft
L water flow rate lb/hr
L/G water/air mass flow ratio lb water/lb dry air
M a constant dimensionless
n a constant dimensionless
Q heat load Btu/hr or Btu/min
T water temperature OF
TP total pressure referred to atmospheric inches of water
APPENDIX 6 389
U velocity ft/min
u, velocity at plane of fan ft/min
Ulo .. velocity through louvers ft/min
U overall heat transfer rate Btu per lb per sq ft per 'F
V effective cooling tower volume cu ft
V specific volume of air cu ft mixture per lb dry air
VP velocity pressure inches of water
VP, velocity pressure at fan inches of water
vp. velocity pressure at stack exit inches of water
w humidity lb water vapor per lb dry air
W width ft
WBT wet-bulb temperature 'F
..1 difference ..1
..1h enthalpy difference Btu per lb dry air
..1p density difference lb mixture per cu ft mixture
..1P pressure differential inches of water
..1PR air flow resistance inches of water
..1Ps pressure differential between
inlet and outlet inches of water
..1T temperature difference 'F
77 efficiency dimensionless
77, fan efficiency dimensionless
77 g gear efficiency dimensionless
0 density ratio pip, dimensionless
p air density lb mixture per cu ft mixture
P. standard air density lb per cu ft
E summation
SUBSCRIPTS
1 inlet
2 exit
a average
REFERENCES
(1) Verdun8tung8kUhlung by F. Merkel, V.D.!. Forschungsarbsiten, No. 275, Berlin, 1925.
(2) Acceptance Te8t Code for Water-Cooling Tower8, CTI Code ATC-105, Houston, Texas, June
1982 issue, Appendix m-D, p. 18.
(3) Cooling Tower In8titute Performance Curve8, Cooling Tower Institute, Houston, Texas, 1967.
(4) Kelly '8 Handbook of Cr088ftow Cooling Tower Performance, Neil W. Kelly & Associates, Kan-
sas City, MO, 1976.
AUTHOR INDEX
Abbasi, M. H., 195 Bauer, R. I., 125

Ackermann, G., 207, 221 Bauer, R., 53, 58, 116
Acrivos, A., 150, 165 Bauerie, G. L., 117
Adams, R. L., 149, 163 Beggs, G. C., 236
Aerov,150 Bell, K. J., 324
Ahlert, R. C., 117 Berg, B. V., 195
Akromenkov, A. A., 164 Bernhardt, S. H., 68, 81
Alpert, R. L., 56, 58 Beyaert, B. 0., 116
Althof, I. A., 222 Bharathan, D., 203, 209, 216, 221-222, 230,
Althof, I., 230-231, 235 235, 238-239, 240, 242
Amsden, A. A., 53, 59 Bikerman, J. J., 206, 221
Anderson, T. B., 131, 163 Blair, C. K., 316, 324
Archer, D. H., 193-194 Blanding, F. H., 117
Ayer, P. H., 194 Boehm, R. F., 1,22,24,99, 117-118, 120,
Ayyaswamy, P. S., 226-227,235,240, 125, 235-236, 245, 305, 324-325, 344,
242 3521356-357, 359, 360, 368
Azad, F. H., 152, 163 Bogart, I. A., 231, 235
Bohn, M. S., 282, 296
Bolles, W. L., 33, 39
Borishanskii, V. H., 234
Backovchin, D. M., 193-194
Borodulya, V. A., 24, 151, 163, 195
Baeyens, I., 52, 58
Botterill, J. S. M., 152, 154, 163
Bakhtiozin, R. A., 154, 164
Brauer, H., 120
Barbolin, V. S., 165
Bravo, J. L., 33, 39
Barile, R. G., 33, 39
Brazelton, W. T., 154, 164, 166
Baron, T., 77, 81 Brewster, M. Q., 152, 163, 167, 194, 199,201
Baskakov, A. P., 195
Brink, J. C., 240, 242
Batchelor, G. K., 149, 163
391
392 AUTHOR INDEX
Brinkman, H. C., 87, 116 Depew, C. A., 154, 164

Brown, D. H., 194 DeVan, I. H., 286, 297
Brown, G., 235 Diaz, H. E., 322, 325
Bruce, W. D., 75, 81 Dow, W. M., 154, 164
Butt, M. H. D., 163 Drew, D. A., 47, 58
Drew, T. B., 221-222
Dubberly, L. I., 283, 297
Cain, G. L., 163 DuckIer, A. E., 51, 58
Cairns, E. I., 117 Dukowicz, I. K., 56, 58
Calderbank, P. H., 77, 81, 312, 324 Dwivedi, P. N., 78, 81
Campbell, Iohn C., 371
Carye, I. S., 36, 39
Casile, E., 166 Eckert, E. R. G., 51, 58, 74, 81
Cassanova, R. A., 194 Eckert,I. S., 313, 324, 329, 334
Cavers, S. D., 177 Edesess, M., 324
Chan, C. K., 24, 152, 163 Einstein, V. G., 150, 154, 164
Chandran, R., 151, 163 Elgin, I. C., 77, 81, 86, 88, %, 116-117
Chandrasekhar, S., 186, 194 Epstein, N., 18,24
Chang, H.-C., 163 Ettahadieh, B., 164
Chang, T. M., 52, 59 Evans, G., 188, 191, 195
Chao, B. T., 163-164,232,236 Evans, I. W., 195
Charpentier, I. C., 76, 81 Ewanchyna, I. E., 117
Chen, I., 151, 163
Chen, I. C., 53, 59, 186, 194 Faeth, G., 236
Chen, K. L., 186, 194 Fair, I. R., 7-8, 24-25, 28, 30, 33-34, 39,
Chen, L. H., 128, 130, 146, 166 220, 222, 224, 234, 264, 266, 297
Chen, M. M., 52, 127, 142, 150-151, 163-165 Fairbanks, D. E, 52, 59
Cheremisinoff, N. P., 368 Fairbanks, F., 150, 165
Chilton, C. H., 324, 334 Falcone, P. K., 176, 186-187, 194
Chilton, T. H., 33, 39 Fannir, H., 224, 234
Chow, L. C., 58 Farber, L., 154, 164
Chu, C. M., 194 Filippovsky, N. F., 195
Chung, I. N., 58, 240, 242 Finkelstein, E., 116
Chung, Y. C., 150, 163 FinkIestein, Y., 235
Churchill, S. W., 194 Fisher, E. M., 234
Claxton, K. T., 81 Fisk, W. I., 194
Clift, R., 120, 125 Flamant, G., 24, 169, 170, 182, 191, 193
Coban, M., 118 Flamm, H. I., 166
Colburn, A. P., 33, 39,221-222,241-242 Florschultz, L. W., 236
Collingham, R. E., 150, 164 Foote, E. H., 324, 334
Cook, D. S., 225-226, 234, 240, 242, 368 Ford, I. D., 224, 226, 234, 240, 242
Crowe, C., 195 Forsythe, G. E., 207, 222
Crowe, C. T., 41, 45, 54, 56, 58, 131 Forusaki, S., 81
Culbreth, W., 120, 125 Friend, L., 324
Dalzell, W. H., 194 Gabor, I. D., 24, 151-152, 164-165

Daniel, K. I., 194 Galershtein, D., 24
Dankwerts, P. V., 73, 75, 81, 117 Galershtein, D. M., 151, 165
Dan Tran, K., 165 Ganic, E., 24, 153, 165
Danziger, W. I., 154, 164 Garner, F. H., 77,81
Davidson, I. F., 134, 164-165 Gartside, G., 166
Davies, I. T., 206, 222 Garwin, L., 315, 324
DeBellis, D., 22, 24, 234 Gat, Y., 18, 24, 316, 325
Deeds, R. S., 316, 324 Gauthier, D., 193
Dengler, I. L., 39 Geankoplis, C. I., 117
deNie, L. H., 77, 81 Gelperin, N. I., 150, 154, 157, 164
Dent, D. C., 166 Ghiaasiaan, S. M., 239, 243
AUTHOR INDEX 393
Gidaspow, D., 52-54, 59, 131, 164 Huang, C.-C., 28, 33, 39
Gier, T. E., 117 Huang, J. H., 368
Gilliland, E. R., 116 Humphrey, J. L., 220, 222
Glass, D. H., 146 Hunt, A. J., 194
Golafshani, M., 233, 236, 355-357, 360, 368
Goldobin, J. M., 195
Goldstick, R. J., 224, 234 Ikeda, Y., 81
Gomezplata, A., 132, 165 Incropera, E P., 194
Goodwin, P., 118 Isachenko, V. P., 235
Goosens, W. R., 52, 58 Isenberg, J. D., 232, 233, 236
Gorbis, Z. R., 154, 164 Ishida, M., 165
Gormerly, J. E., 297 Ishii, M., 47, 59
Grace, J. R., 74, 81, 120, 125,233,236,351, Iwashko, M. A., 154, 164
368
Green, H. J., 242
Gregory, R., 324 Jackson, R., 131, 163-164
Greif, R., 187, 194-195 Jacobs, H. R., 22, 24, 99, 117, 120, 125, 223-
Greskovich, E. J., 97, 117 226, 228-229, 231-232, 235-236, 238,
Grewal, N., 24 240, 242, 324-325, 343-344, 352, 355-
Grewal, N. S., 151, 165, 195 357,359,360,366,368
Gupta, R., 368 Jakob, M., 154, 164
Jayadev, T. S., 302
Jiji, L. M., 59
Johnson, A. I., 306-307, 309, 311-312, 321,
Hamaker, H. C., 194 325-326, 334, 356, 368
Handley, M. E, 134, 164 Johnson, D. H., 242
Handlos, A. E., 77, 81 Johnson, G., 120, 125
Hansen, A. C., 324 Johnson, K. M., 233, 236
Hanson, C., 125 Johnson, K. R., 346, 369
Hansuld, J. H., 368
Harland, S., 324
Harlow, E H., 53, 59 Kalman, H., 236
Harnet, J., 24 Kays, W. M., 283, 297
Harrison, D., 127, 146, 164 Keairns, L. M., 193
Hart, G. K., 316, 325 Kehat, E., 91, 96-97, 100, 116-117, 120, 123,
Hartland, S., 345-346, 351-354, 368-369 125, 315, 324, 329, 334, 354-355, 368
Hashmall, E, 324 KeIly, N. W., 33, 39, 389
Hasson, D. D., 228, 235 Kim, J. M., 154, 164
Hausbrand, E., 223, 234 Kim, S., 240, 242
Heertjes, P. M., 77, 81, 132, 164 Kirakosyan, V. A., 195
Heinz, J. 0., 312, 324 Klein, D. E., 195
Hertwig, T. A., 39 Knight, J. F., 70, 81
Hertz, H., 222 Knudsen, J. G., 61
Hetsroni, G., 59, 125, 153, 164 Knudsen, M., 240, 222
Hiby, J. W., 117 Kobayashi, M., 152, 164
Hijikata, K., 240, 242 Kochman, J. A., 297
Hoffing, E. H., 311, 324, 329-330, 334 Kolar, A. K., 195
HoIlaway, E A. L., 325 Kondukov, N. B., 134, 164
Hoopes, G. w., Jr., 222 Korchinski, I. J. 0., 77, 81
Horbath, M., 307, 324 Korensky, V. I., 151, 163
Hottel, H. C., 194 Kornilaev, I. M., 133
Houf, W. G., 187-188, 194-195 Korotyanskaya, L. A., 164
Hougan, O. A., 206, 222 Kovensky, Y. 1.,24, 195
Hougan, J. 0., 117,241-242 Kramers, H., 117
HoweIl, J. R., 194-195 Kreager, K. M., 117
How, H., 224, 234 Kreith, Frank, 1, 221, 245, 264, 296, 297
Hruby, J. M., 186-198, 194 Kronig, R., 240, 242
Huebner, A. W., 112, 118 Kruglov, A. S., 164
394 AUTHOR INDEX
Krylov, v. S., 206, 222 Mickley, H. S., 52, 59, 150, 165
Kulik, E., 226, 235 Miller, E, 194
Kumar, A., 222 Mills, A. E, 63, 237, 239-243
Kumar, A. D., 352, 368 Minard, G. W., 306-307, 309, 325, 334
Kunii, D., 78, 81, 127, 146, 150-151, 154, Mirand, G. W., 368
164, 166 Mixon, F. D., 97, 117
Kushnyrev, v. I., 235 Moalem-Maron, D., 22, 24, 224, 230, 232-
Kutateladze, S. S., 224, 228, 234 234,236
Kwank, M., 116 Modest, M. E, 152, 163
Moler, C. B., 222
Molerus, 0., 138, 140, 166
Langenkamp, H., 166 Mori, Y., 242
Lang, W. R., 297 Morley, M. 1., 164
Langmuir, I., 204, 222 Morooka, S., 81
Lapidus, L., 86, 88, 116 Moslemian, D., 154, 165
Laurendeau, N. M., 194 Murray, 1. D., 131, 165
Leal, L. G., 150, 163-164 Murty, N. S., 232, 236
Lee, S. 1., 241-242
Lefferdo,l. M., 194
Lekic, A., 224, 226, 234, 240, 242
Nadig, R., 228-229, 235, 242
Lerner, Y., 236
Navon, V., 235
Letan, R., 42, 83, 91, 96, 97, 100, 116-117,
Neale, D. H., 194
119, 125, 233, 236, 306, 309, 315, 324,
Nichols, K., 368
326, 329, 334, 352-356, 368
Nir, A., 150, 165
Leva, M., 313, 324
Noack, R., 157, 165
Levenspiel, 0., 78, 81, 127, 146, 154, 164
Nobel, P., 154, 166
Levich, V. G., 206, 222
Noring, 1. E., 194
Lewis, 1. B., 81
Nosov, V. S., 154, 165
Lewis, W. K., 116
Lichtenstein, 1., 33, 39
Licklein, S., 154, 165
Liljegren, 1., 163 Oki, K., 133, 165
Lin, 1. S., 135, 137, 144, 163-164 Okumoto, Y., 81
Lobo, W. E., 313, 324 Olalde, G., 118, 360, 366, 368
Lockhart, E 1.,311,324,329,330,334 Olander, R., 118, 360, 366, 368
London, A. L., 283, 297 Oliker, I., 224, 234-235
Lumley, 1. L., 142, 165 Olson, D. A., 242
Luss, R., 235 Onda, K., 81
Lyczkowski, R. W., 164 Oshmyanshu, S., 368
Owens, W. L., 221
Ozisk, M. N., 194
Maa, 1. R., 222 Oznayak, T., 53, 59
Major, B. H., 233, 236
Majumdar, A. K., 218, 222, 241-242
Malcolm, G. E., 222 Palaszewski, S. 1., 56, 59
Malik, M. A. S., 222 Parsons, B. P., 209, 222
Mamaev, V. V., 154, 165 Patel, B. R., 240, 242
Marschall, E., 120, 125 Peck, R., 235
Marscheck, R. M., 132, 165 Peierozchikova,l. P., 164
Martin, 1., 176 Penney, T., 216, 221, 239, 242
Maskaev, V. K., 195 Permyakov, V. A., 234
Masson, H., 134, 165 Perona, 1. 1., 67, 70, 75, 81, 119, 125
Mastanaiah, K., 153, 165 Perry, 1. H., 39
Merry, 1. M., 134, 165 Perry, M. G., 164
Merkel, F., 389 Perry, R. H., 324, 334
Mersmann, A., 120, 125 Peters, M., 280, 282-284, 297
Mertes, T. S., 116, 306, 325, 327-328, 334 Pfeffer, R., 154, 165
Michiyachi, I., 51, 59,79, 81 Pfender, E., 51, 58
AUTHOR INDEX 395
Pigford, R. L., 34-35, 39, 81, 222, 238, 242, Shirai, T., 165
296,297 Shulman, H. L., 81
Pikulik, A., 322, 325 Sideman, S., 18,22, 24, 120, 123, 125, 224,
Plass, S. B., 99, 117,315,319,324-325,356, 230, 232-234, 236, 314-316, 325, 356,
363, 367-368 369
Prausnitz, J. M., 117 Siegel, R., 194
Price,8. G., 116 Silver, R. S., 239, 242
Proulx, A. E, 81 Simms, A. E, 316, 325
Pyle, C., 34-35, 39 Singh, A., 150, 165
Singhal, A. K., 222, 241-242
Skachko, I. M., 164
Quader, 154 Skelland, A. H. P., 346, 369
Slattery, J. C., 49, 59
Smith, B. D., 315, 324
Ramaswarni, D., 164
Sodha, M. S., 222
Rao, V. D., 232, 236
Sohn, C. w., 150-151, 165
Redish, K. A., 152, 163
Sonn, A., 117
Rhodes, E., 226, 235
Soo, S. L., 131, 150, 152-153, 157
Rhodes, H. B., 116,306,325,327-328,334
Spalding, D. B., 57, 59, 222, 241-242
Richardson, J. E, 87-88, 116,306-307,321,
Stamatoudis, M., 78, 81
325-327, 334, 337, 353, 368
Steele, B. R., 194
Riemer, D. H., 357, 368-369
Steiner, L., 345-346, 348, 351-354, 369
Rios, G., 165
Steiners, S., 324
Rivard, W. C., 53, 59
Struve, D. L., 116
Rivkind, V. Y., 351, 369
Sudhoff, 8., 233, 236
Rocha, J. A., 39
Sundarajan, T., 235
Rohsenow, W., 24
Suratt, W. 8., 315-316, 325
Roscoe, R., 87, 116
Swenson, L., 33, 39
Ross, D. K., 152
Syamla1, M., 53, 54, 59
Rossetti, S., 165
Syromyatnikov, N. I., 165
Rowe, P. N., 77, 81, 131
Ruby, C. L., 77, 81
Russo, R., 194
Ryskin, G. M., 351, 369 Taitel, Y., 229, 235, 325
Takenati, H., 81
Tarnir, A., 229, 235
Sakiadis, B. c., 306-307, 309, 311-312,321, Taneda, S., 117
325-326, 334 Tavlarides, L. L., 78, 81, 125
Sam, R. G., 240, 242 Tayeban, M., 77,81
Sarma, P. K., 232, 236 Taylor, G. I., 147
Sarofim, A. 'E, 194 Tennekes, H., 142, 165
Sastri, V. M. K., 232, 236 Thomas, K. D., 230-231, 235
Saxena, S., 16, 24 Thomas, L. M., 193
Saxena, S. C., 151, 154, 165, 195 Thring, R. H., 151, 165
Schlepp, D., 221 Tien, C.-L., 24, 152, 154, 163, 165, 194
Schlunder, E. V., 53, 58, 151, 166 Timmerhaus, K., 280, 282-284, 297
Schlunderberg, D. C., 154, 165 Tiwari, G. N., 222
Schrage, R. w., 205, 222, 239, 242 Toei, R., 127, 154, 164
Schuster, A., 194 Torrey, M. D., 53, 59
Schuster, J. R., 51, 59 Tortorelli, P. F., 286, 297
Seador, J. D., 154, 164 Treybal, R. E., 77, 81, 316, 325, 331, 334,
Seban, R. A., 239, 242 346, 359, 369
Seo, Y. C., 59
Shavin, E, 118
Sheridan, J. J., 81
Sherwood, T. K., 30,33,39,74,81,205,238, Ullrich, L. L., 81
242, 296-297, 313, 325 Upadhyay, S. N., 78, 81
Shipley, G. H., 325 Urbanek, M. w., 112, 118,315,325
396 AUTHOR INDEX
Vallario, R., 22, 24, 234 Whitehead, A. B., 139, 166

Van Heerden, L., 154, 166 Whitelaw, R. L., 154, 165
Van Krerilen, D. w., 166 Wilhelm, R. H., 116
Van Velzen, D., 134, 166 Wilke, C. R., 39, 81, 222, 238, 242, 296-297
Vasalos, I. A., 194 Williams, J. R., 152, 163
Verioop, J., 164 Williams, R., 164
Vermuelen, T., 77, 81, 222 Won, Y. S., 239, 243
Viko, J., Jr., 176 Woodward, T., 99, 117
Vitt, o. K., 195 Wright, J. D., 234, 299, 300, 302
Vohra, K., 368 Wroblewski, D. E., Jr., 194
Von Berg, R., 77, 81 Wunschmann, J., 151, 166
Vortmeyer, D., 152
Wall, D. A., 118 Yagi, S., 150-151, 166

Walter, D. B., 357, 369 Yang, Y. S., 195
Walter, L. E, 324, 334 Yeager, K., 219, 222
Walton, J. S., 154, 164 Yong, J., 166
Wang, T., 39 Yuen, w., 172
Wasen, D. T., 154, 166
Wassel, A. T., 63, 239-243
Watt, J. R., 219, 222 Zabrodsky, S., 16,24
Weaver, R. E. C., 116 Zabrodsky, S. S., 151, 165
Weber, M. E., 120, 125 Zaikovski, A. V., 164
Wei, J. C., 205, 222 Zaki, W. N., 87-88, 116, 306-307, 321, 325-
Weinbaum, S., 59 327, 334, 337, 353, 368
Welty, J. R., 149, 163, 197, 238 Zenz, F. A., 116, 324
Wen, C. Y., 52, 59, 128, 130, 146, 166 Zhang, Li., 166
Werner, D., 368 Zhanwan, W., 166
Werther, J., 138, 140, 166 Zheging, Y., 166
Westermann, M. D., 117 Ziegler, E. N., 154, 166
Westwater, J. W., 81 Zimmerman, J. 0., 81
Whitaker, D. R., 117 Zmora, J., 117
Whitaker, S., 49, 52, 59, 149, 166 Zubir, N., 87, 116
SUBJECT INDEX
ACDW vortice, 141, 142 phenomena, 120

Ackerman factor, 30, 39 size, 321
Adiabitic: trains, 232
humidifier, 31 -type condensing, 232
saturation temperature, 31 Bulk properties, of near-equilibrium conditions,
Agitated column, diagram of, 4, II 149
Air:
bubbles, 20
flow, 5 Calculations for mass transfer effects, 253
temperature, 56 CAPTF (computer-aided particle tracking facil-
Antonoffs rule, 345 ity), 134, 142, 147
Argon, 173 Carbonate, 68
Axial mixing, 353 Carnot-cycle efficiency, 301
Centrifugal forces, 9
Ceramic Intalox saddles, 33
Backmixing, 93, 108, II4, 309, 353-354 Claude cycle, 209
Baffle(s), 4, 7 evaporator, 216
column(s), 34 Closed cycle cooling, 4
disc/donut type, 34 Closed heat-exchange devices, 1-3, 9
Barometric steam condenser, diagram of, 4, 10 Coal:
Bed geometry, 198 combustion, 23
Behavior of drops, 348 gasification, 16
Ber! saddle(s), 75, 266, 288, 311, 314, 329 Coalescence, 93
Bernoulli equation, 306 Cocurrent flow, 9-10
Bicarbonate, 68 Coils, 16
Brayton cycle, 260, 290 Colburn (i-factor), 254, 283
diagram of, 263 Colburn-Hougen analysis, 207
Bubble(s), 49, 74, 146, 151, 198,215,304 Cold fluid streams, I
column(s), 31, 20, 76 Collective solids, 154
diameter, 311 Column cross-sectional area, calculation of, 304
flow behavior, 140 Column sizing, 272
heat transfer in, 14 Combustion, 16
-liquid heat transfer system, 43 of coal, 23
motion, 146 Commercial packings, diagram of, 5
397
398 AUTHOR INDEX
Computational models, 51 Drop(s), 74, 77

Condensation, 18, 68, 224 characterization map, 350
heat transfer, 23 phenomena, 120
Conduction, 20, 21, 197 size, 93, 124, 307
Continuity equation, for fluid phase, 50 -type condensers, 224
Continuous fluid, 19 velocities, 97
Convection, 21, 197 Droplet(s), 83, 212
Convective heat transfer, 46 -droplet radiation, 47
Convective mass transfer, 72 temperature, 56
Cooling: Dry cooling towers, 5
pond(s), 4-5, 21 diagram of configurations, 11
diagram of, 14
towers, 4-5, 32, 204
with dehumidification, 30, 33
Copper, 171
East Mesa plant, 253, 354, 356-357,360,366
Countercurrent flow(s), 3, 9-10, 88
Economic analysis, 281
Coupling:
Elgin tower, 106
one way, 47
Ellipsoidal cap:
two way, 47
dimpled,74
Crossflow tray(s), 4, 7, 36
skirted, 74
columns, 34
Ellipsoidal regime, 348
Crossflow cooling tower(s), 5
Energy conservation, 224
Crystallization, 113
Energy transport, 47
Entrained particle flows, 168, 192
Entrainment, 61
Dankwerts model, 75
Darcy's law, 51 EOtvos number, 74, 79, 121, 233, 348, 355
Dehumidification, 31 EPRI,62
Dirac function, 180 Ergun's equation, 53
Direct-contact: Eulerian basis, 250
boiling heat transfer, 357 Evaporation, 21, 23, 206, 208
computer code, 241 application, 208
condensation, 223, 237-239 coolers, 23
crystallization, 113 effectiveness, 210
desalination process, 111 for vapor production, 208, 239
devices, 3, 9 fundamentals of, 23
equipment design and operation, 124 heat rejection, 218
evaporation, 203, 237-238 limiting rate, 238
heat exchange, 25, 27, 33, 41, 68, 69, of a liquid, 203
245,259 rate of, 204
heat exchanger(s), 261, 263, 294, 302 Evaporative:
heat transfer, 1-4, 7, 9, 19,22-23,26, air cooling, 219
61,67, 119, 197 cooling principal, 204
analysis, 257 External heat transfer mechanisms, 19-20
high temperature applications, 6, 23 Extraction, 114
liquid-liquid heat exchanger, 335 spray columns, 99
preheaterlboiler, 303
process, 20, 114
solar flux receiver, 16 Fan entrance, 381
solid particle, 168 Film theory, 72
spray columns, 343 Finned-tube heat exchanger(s), 264, 283-284,
spray tower, 347, 349 292,294
Disc-donut-type baffles, 34 economics of, 282
Dispersed: illustration, 260
fluid, 19 layout of core, 285
phase fluid injectors, 345 Fischer-Tropsch synthesis, in a fluidized bed, 70
phase system, 50 Flamant solar fluidized bed, 170
Distillation, 67, 220 studies, 183
columns, 23 Flash evaporation, 218
Documentation, 63 Flashing jet evaporators, 239
AUTHOR INDEX 399
Flooding, 266, 272, 277, 305-307, 324, 327- fluidized bed, 171
329, 338, 358, 362 reactor, 172
Flow diagram of DCHX test loop, 268
Flow direction, 9
Flow rate ratio, 336 Heat exchanger(s), 1
Fluid(s): applications, 4
continuous, 19 direct-contact, 25-27
dispersed, 19 enhanced fluid connection, 153
dynamics, 19 fluidized bed, 16, 18
particle systems, 78 fluid mediated particle-surface, 153
phase, 50 heat-phase, 16
Fluidized: particle-particle, 46
bed(s) , 127-128, 168, 192 three-phase, 18
heat exchanger, 16 Heat tracing, 273
solids-gas/liquid, 16 Heat transfer, 1, 3, 14, 21, 98, 120,206, 314
Fogging, 61 analysis of air/molten salt direct-contact,
Forced gas flow, 169 257
Fractional distillation, 220 analogies, 29
Free-falling: between fluidized beds, 16
flow, 192 coefficient(s), 1, 54, 264,278,283, 303,
particle films, 168 322
French fluidized bed, 169 computational techniques, 41, 63
Friction factors, 124 condensation, 23
Formulation of rate equations, 71 convective, 46
Fouling, 1 correlations, 355
Fourier's law, 51 design, 25
direct contact, 1-4,7,9, 19,22,61,67
direct particle layer models, 151
Gas: due to particle impact, 152
absorption, 67 in a particle system, 10 1
cooling with humidification, 31 in bubble columns, 14
-droplet systems, 57 in bubble-liquid systems, 57
heating with humidification, 32 in gas fluidized beds, 127
-liquid: in liquid-liquid systems, 57
contacting, 25 liquid-liquid direct-contact, 119, 355
packed tower correlations, 313 macroscopic modeling, 151
separator, 68 /mass transfer analogy, 264
systems, 42, 76 measurements, 151, 154-155,276
-particle systems, 57 mechanisms, 19
-phase coefficients, 34 method,132
-solids fluidization, 23 multi-phase systems, 42
-solids fluidized beds, flow regimes of, nonmechanistic continuous models,
129 152
streamlines, differences in, 130 microscopic treatment, 152
turbine generators, 2 radiation, 23
Gauss quadrature formula, 186 salts, 270
Geometric: stages, 36
description of contact equipment, 124 system design, 317
parameters, 339 to immersed surfaces, 148
Geothermal : to stationary particles, 152
brines, 42, 343-344, 358 Heavy hydrocarbon fractionator, 4
energy, 224 Heliostats, 7
power cycle, 67 Heller cycle, 224
Glass, 172 Henry's law, 72
Global models, 43 Higby's equation, 77
Graetz solution, 228 High efficiency packings, 37
Gravity,9 High temperature:
Grid packing, 309-310 (radiative) heat transfer, 191
GTRI: solids-gas interactions, 167
entrained flow, 171 High solar emissivity, 191
400 AUTHOR INDEX
Holdup, 91, 124, 358 Local models, 44

of the dispersed phase, 352 Log mean temperature difference (LMTD), 2-3,
Hot fluid streams, 1 276, 290, 303
HVAC applications, 204 Long columns, 104
Hydrodynamics in liquid-liquid systems, 121- Longitudina1 dispersion, 98
122 Los Alamos, 53
Hydrogen ion concentrations, 68 Low infrared emissivity, 191
Indirect dry cooling system, diagram of, 13 Magnesium chloride, 114

Industrial practices, 25, 61 Maintenance costs, 282
Inertially dense systems, 51-52 Marangoni effect, 206
Intalox saddles, 288 Martin Marietta, 286-287, 297
Interfacial: Mass transfer, 61, 98, 116, 120,205,261
area, 74 coefficient for Raschig rings, 289
processes, 204 coefficient for types of packing, 290
turbulence, 205 contaminants, 247
Internal, heat transfer mechanisms, 19 effects, 119, 253
IPSA code, 57 in heat transfer processes, 67
Isobutane, 253 inert gases, 247
-water systems, 360 of additives, 247
phenomenon, 67
Measuring solid particles velocity:
Jacob number, 231, 319
Jet condenser, 229, 238 cross-correlation method, 133
drag force method, 132
j-factor, 73, 78-79
heat transfer method, 132
laser method, 133
Kerosene, 114 tracer method, 132
KFIX numerical code, 53, 57 Mechanics of vertical moving systems, 85
Kutateladze's solution, 229 Merkel:
approach, 219
equation, 372
Laboratory columns, 112 Metal:
autocorrelation(s), 147 Hypac rings, 33
fluctuating velocities, 142 Pall rings, 33
function, 147 Methyl salicylate, 75
frame, 250 Minard-Johnson correlation, 306
integral time scale, 147 Minimum-fluidized velocity, 127
LBL: Mist eliminators, 377, 386
absorber, 173 Mixing effects, 95
entrained flow, 172 Molten saits, 260-261
Legendre polynomials, 186 Momentum equation:
Lewis number, 32, 39, 219 for continuous phase, 50
Liquid coolant, 16 for discontinuous phase, 50
Liquid fluidized bed(s), 95 Monoethanolamine, 261
longitudina1 dispersion in, 96 Morton number, 74, 79
Liquid-gas processes, 22 MSF (multistage flash desalination system),
Liquid-liquid: diagram of, 217
direct-contact heat transfer, 119, 122 Multiphase momentum, 47
extract, 120 Murphree tray efficiency, 37, 220
flooding, 311, 329
heat transfer, 316
packed columns, 310 Napierian logarithms, 37
packed towers, 331 National Bureau of Standards, 297
separator, 68 National Science Foundation, 22
-solid suspensions, 154 Natural draft cooling tower, 382, 385
spray, 100 Newtonian:
spray column(s), 89, 304 fluid, 121-122
system(s), 42, 75, 77, 91, 315 liquid, 121
transport, 119 Nitrate salt, 7
AUTHOR INDEX 401
Nodal points, 55 Phase function, for opaque geometric spheres,

Non-Newtonian fluids, 121-122 180
Nonuniform distribution plate, 141 Physical models, 44
Norton Chemical Company, 269, 297, 309-310 Pilot plant columns, 112
Process Products Bulletin, 313, 325 Plate towers, 67
Numerical models, 43 Plug flow, 95
Nusselt number, 46, 115 Point models, 44
Nozzles, 34 Porous material, 48
Power spray canal system, 14
Prandtl number, 38, 80, 115, 264, 319
Oblate, 74 Pratt's method, 319
Preheating, III
One-way coupling, 47
Open circuit water cooling, 4 Pressure, 9
feedwater heater, 10 Property values, 296
turbulent wake, 74 Pulverized coal combustion, 6
Operating curves, 92
Ormat turbines (Israel), 302
Quartz, 172
Oscillating chops, 121
exit tube, 173
OTEC (Ocean Thermal Energy Conversion),
208
falling fUm exchanger, 239
Radiation, 6, 14,20, 197,200,249
systems, 224
droplet-droplet, 47
heat transfer, 23, 149, 151
solar, 7
Packed bed(s), 73, 89, 95 Radiative:
diagram of, 4 modeling techniques, 175
condenser, 231 single particle model, 175
Packed bubble columns, 76 transfer equation, 176-177
Packed column(s), 28, 33, 75-76, 84,259, 261, two-flux model, 180
265,311 transport, 189
diameter(s), 314 Radioactive particles, 147
exchanger, 322 Rankine cycle, 208, 299, 300
table of parameters, 266 engine, 301
Packed towers, 67, 313, 317 organic, coupled to a solar pond, 301
Packing(s), 261, 309 Raschig ring(s), 33, 75, 231, 266, 269, 272,
density, 142 278-281, 288, 293
diagram of, 5, 262 Rayleigh scattering theory, 173, 192
effect of size and type, 288 Reynold's number, 48, 74, 76, 80, 105, 109,
Pall ring(s), 269, 279-281, 286, 288-289, 293, 115, 121, 149-150,282-284, 315, 326,
309-310, 314, 317, 329 337, 348, 352, 355, 365
Paraffin, 19 analogy, 264
Park Chemical Company, 297 Rigid drops, 77
Particle(s), 19 Rod bundle, 143
circulation pattem(s), 138-139 Roscoe function, 110
diameter, 46
-fluid temperature difference, 46
-gas flow, 167 Salt film, 7
injeeted into a gas stream, 45 Salt-gradient solar pond(s), 299, 300
motion in a packed bed, 46 illustration of, 300
-particle heat exchange, 46, 53 Sand, 171
properties, 198 particles, 174
scattering geometry, 179 Sandia:
surface impact, 153 free-falling flow, 174, 187
Particulate vertical moving systems, 85 free-fall radiant heating test, 175
Peelet number(s), 80,96, 97, 146, 150 Scale-up, 108
Perforated plate: Scandium-46, 134
column, 6, 84 Schliinder model, 53
contactor, 4 Schmidt number, 76, SO, 264
Phase change, 20-21 Schmitt trigger, 137
402 AUTHOR INDEX
Sea water desalination, 42 Tchebycheff method, 372

Semi-cylindrical bed, 131 Thermal:
Shear stress tensor, 51 conductivity, 46
Sherwood number, 79-80, 264 equilibrium, 25
Simultaneous direct-contact heat and mass trans- resistance, 7
fer, 123 storage, 23
Single particle tracking, 135 stress, 1
Single tubes, 16 Thermally dense flow, 46
Skewed collimated incidence, 181 Thermally dilute flow, 46, 52, 57
Slip velocity-holdup, 86-89, 93 systems, 54
Solar: Thermodynamic performance, 3
cavity receiver, 188 Three-phase heat exchanger, 16, 18
central receiver systems, 6 configurations, 18
direct-contact flux receiver, 7, 16 Tower inlet, 377, 384
Solar Energy Research Institute (SERl), 22 Trajectory approach, 55
Solid: Transfer models, 30
-fluid systems, 89 1\1be(s) bundies, 16
-gas transfer systems, 22 Two-and-a-half-dimensional bed, 131
Solids: Two-dimensional bed, 131
circulation, 146 experiments, 312
in gas fluidized beds, 127, 131 Two-fluid model, 57
velocity distribution, 144 Two-phase flow, computational techniques, 41,
Solute extraction, 220 63
Solvent extraction, 67 Two-phase transport, 61
Soo's theory, 150 Two-way coupling, 47
SPHER (small particle heat exchanger receiver), Type J film fill, 374
172
Spherical regime, 348
Spout evaporator, 214
Spray: Vacuum evaporation, 204
chamber(s), diagram of, 4 Vacuum steam fractionator, 4
channels, 6 condenser for, 9
columns(s), 13, 33, 84, 95, 102, 104, 108, Vapor bubbles, 44, 204
114, 306, 308, 310, 344-345, 361 Vapor/liquid spray, 311
distribution time in, 97 Velocity of drops in swarms, 351
heat transfer, 34 Vertical-holdup, 86
local drop size in, 93 Vertical moving systems, mechanics of, 85
one-meter, 35 Volumetric heat transfer coefficient, 254, 265,
droplets, 34 267, 276, 278, 292-293, 314-316,318-
nozzles, 33 319, 322, 340, 365
pattern, 34 experimental measurements of, 267
quencher, diagram of, 8
systems, 4
tower(s), 3, 76, 321
diagram of, 1 Wake model, 100, 104, 114
-type condenser, diagram of, 225 Water, 19
Stanton number, 264 -cooling towers, 371
Stirred tanks, 76 spray, 380, 386
Stoke's: stream, 31
drag coefficient, 45 Weber number(s), 121, 348, 358, 362
flow, 172 Wet-dry cooling tower, diagram of, 13
region, 87 Wetter-wall column, 84
Stratified bed, 147 Wobbling, 74
Sublimation, 204
Sulfide, 68
Surfactants, 206
Zehner model, 53
Swarm, 147
Zen's graphical correlation, 87

System Calculation

Uploaded by

Copyright:

Available Formats

System Calculation

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

System Calculation

Uploaded by

Copyright:

Available Formats

What is the document discussing?

What is the document discussing?

What assumptions are made in the example calculations?

What assumptions are made in the example calculations?

APPENDIX

EXAMPLE CALCULATIONS FOR

Consider a direct contact boiler in which isobutane is vaporized by contact with

B = brine flow rate, lb/hr

Of course, Xo is zero. Observation of experimental direct contact boilers indicates

The isobutane phase is essentially pure isobutane, and there is no resistance to

where U. = volumetric heat transfer coefficient, Btu/hr ft3 F.

A well-accepted general correlation for predicting values of U. is not yet

Using a value for U. of 35,000 Btu/hr ft8 F, we obtain

Am/MOLTEN SALT DffiECT-CONTACT

0, parameter in HI. correlation

D mass diffusivity, m2/h (ft2/h); 10.76

k thermal conductivity, W /m C (Btu/h ft F); 0.5777

NOTE ON SYSTEM OF UNITS

Figure 1.1 Conventional finned-tube heat exchanger.

Wire Mesh Mist

Figure 1.2 Direct-contact heat exchanger.

concept it is necessary to transfer heat from the molten salt to air.

Figure 1.3 Three types of packing for a direct-contact heat exchanger.

Salt Salt Air to Process or

t:t..,..,......JL-...:~_--4=----::::I--_ Air to Process

Figure 1.4 Applications of direct-contact heat exchangers.

It is possible to use mass-transfer data by invoking the mass-transfer/heat

2 THE HEAT-TRANSFER/MASS-TRANSFER ANALOGY

A= [;J -.--_G_O.::..,---:- (2.4)

ha = [ Sc JfJ GO, . (2.5)

hla = [sc l 1fJ

Table 2.1 Parameters for packed-column heat and mass transfer

m 0.6 0.6 0.5 0.5

Experimental conditions can then be used to calculate the heat transfer

O~ ______ ~ ______ ~ ________ ~

3 EXPERIMENTAL MEASUREMENTS OF VOLUMETRIC HEAT

Air Flow Meter

r" ~~=::[==J::==== "-Compressor

Figure 3.1 Flow diagram of DCHX test loop.

Air Outlet ~-- Salt Distributor

1122'i1!i'lM8-t--- Packed Bed

Figure 3.2 Details of the DCHX packed volume.

Figure 3.3 Photograph of the DCHX test apparatus.

out (outlet pipe) temperatures, respectively.

3.4 Column Sizing

Figure 3.4 Generalized pressure drop correlation.

3.5 Operational Problems

6-in. diameter column

1 in. 1.5 in.

Table 3.1 Nominal column operating conditions.

Salt inlet temperature 350C

Table 3.2 Analysis of upper tank residue

The second feature we abandoned was a pneumatic system for transferring

3.6 Heat Transfer Measurements and Procedures

(T.. - Tao) - (T.o - Tai)

Table 3.3 Measured data showing effect on Q. and Ua

Tai Tao T,; T.o m. ma Q. Qa Ua

3.7 Results and Discussion

Table 3.4 Heat-transfer data and comparison with mass-transfer calculations

O~ ~ ~ ____ ~

O~ ~ __ ~ ~

O~ __________ ~ _ L ~ __________~~