MONROE et al.

-HEAT TRANSFER TO BOIL/NG L/QUIDS 613

HEAT TRANSFER TO BOILING LIQUIDS AT LOW

TEMPERATURES ANO ELEVATED
PRESSURES*
By A. G. MONROE, H. A. S. BRISTOW and J. E. NEWELL

Previous work on heat transfer to boiling liquids is reviewed, and the variations in the boiling
heat-transfer coefficient with temperature difference across the film are described with reference to
the boiling mechanism and the physical properties of the fluid.
Heat transfer to oxygen and nitrogen boiling inside tubes at pressures from atrnosphcric ro
2000 lb.jsq. in. gauge has been studied experimentally, Air was used as the heating medium, and on
account of the large temperature difference between the heating medium and rhe boiling liquid all the
measurements fell within the metastable film-boiling range. Within this range the influence of
temperature difference, pressure, and mass fiow rate werc examined. The higher pressures used
exceeded the critical pressures of the fiuids, and under these conditions the transition from liquid to
vapour occurred without separation of two phases. Behaviour under these conditions is comparcd
with thar at lower prcssures under which phase separation occurs.
Under all conditions the boiling coefficient was found to diminish with increasing ternperature
difference, and to improve with increasing fluid fiow rateo At pressures aboye the critical the rate of
heat transfer is infiuenced by the absolute temperarure of the fluid.
The results of rhis work have been used as a basis for the design of atmosphericalIy hcared
evaporators for liquid oxygen and nitrogen.
Introduction
Heat transfer to boiling liquids is a form of heat transfer with change of phase. The appearance of a
second phase greatly complicates the mechanism of both mass and heat transfer and it is not
surprising that the study of boiling liquids is still the least developed sphere of all forms of heat
transfer. The same applies to heat transfer at high pressures in the super-critical region. There
is no visual change of phase in this region and the introduced heat increases only the enthalpy of
the fluid. However, the- rapid changes with temperature of the physico-chemical properties of
the 'boiling' fluid do not make the problem of heat transfer studies any easier. So far there
is no satisfactory general quantitative expression which would correlate the physico-chemical
properties of the fluid and the geometrical parameters of the apparatus with values of the
corresponding heat flow rates. In the study of boiling, the nucleate boiling region is of primary
importance to the designers of evaporators. There are, however, cases where film boiling of liquids
is inevitable, for instance steel quenching baths, rocket-motor chambers cooled by liquids, and
thermal cracking furnaces. In this paper another case of industrial importance is added, namely,
the evaporation of liquefied atmospheric gases by air heating.! Liquid oxygen and liquid nitro gen
can be evaporated by using atmospheric air as the heating medium. The boiling in this case is of
the metastable and stable film types.
Film boiling: general
Investigations of the mechanism of boiling have established the existence of four distinct regions
in which the boiling possesses different characteristics, according to the temperature difference
across the film. These are defined respectively as the convective, nucleate, metastable and stable
film-boiling regions (Fig. 1).
The last two types of boiling are characterized by the presence of gaseous phase in appreciable
quantties within the boiling liquid, whereas in the first two types, convective and nucIeate boiling,
the mechanism of heat transfer is governed by the properties of the liquid phase. In the case of
nucleate boiling inside a tube at small ternper-
ature differences Boarts- and his co-workers 100,000
50,000 -- -_ ---_-
_
found that the boiling coefficient varied with

:;; g: --

.~
(;' ~- 1--'"

10,000 1--- -- f--- -- _- ----;..,..

the o 8 power of the mass flow rate, i.e. hB ce ~-f[

- --
(Re)O.8. One suspects that at lower Reynolds ,
numbers this relationship is not valid because 5000 t---- --_ --
"f~-:- ~
1000
of rhe larger proportion of gaseous to liquid _ 500
- ~~ Io..Y(~
phase. The presence of gaseous phase is as a
f- NO
m
~
~~
.s:
100
SO ~7---- ---_ - "'~
'<-_
rule detrimental to heat transfer, except where 10 /
it induces or increases turbulenee of the liquid 5 ~- ~
..,.0
.
phase. The result of these two diametrically 1-0 C,

opposite effects is fortuitous and depends, 0-5 ~~<';-f-- ----- --

, lA cP -1
o-
 0-51-0 s-O10 50100 500 5000
among the other variables, upon the 0-' o r 1F.
geometrical arrangement of the equipment
concerned. There is a maximum value of the FIG. 1_ Coefficient of heat transfer (h) [or water boiling
heat flux, known as 'critical value', beyond on a chromel wire'4
* Read before the Chemical Engineering Group on 8 April, 1952.

J. appl. Chem., 2, November, 1952 L

614 MONROE et al.-HEAT TRANSFER TO BOJLING LIQUJDS

which the insulatng properties ofthe gaseous phase impair the heat transfer and cause a diminution
of the film coefficient. The gaseous blanket, however, is not stable; it collapses and grows
alternately on the heating surface, hence the term ' metastable ' film. As the amount of vapour
phase increases, less heat is transferred, until a minimum is reached where the whole heating
surface is covered by gaseous film. Ths point corresponds to the ' spheroidal ' state of boiling
(Leidenfrost phenomenon) and a further increase of ternperature will cause an increase of heat
transfer owing to the increased convection within the gaseous film. We can say safely that the
gaseous phase is the controlling factor in this case. The behaviour in this region can satisfactorily
be expressed by mathematical relations," This is made possible by the faet that heat transfer
occurs almost entirely through the gaseous phase, and thus satisfes the usual correlations for
natural convection. The characteristic relations between the boiling coefficient on external
surfaces and tbe temperature drop have been established by a number of independent studies, and
most investigations have been concerned with the determination of the critical-temperature drop
and the maximum heat-fux value. It has generally been assumed implicit1y that the fluid flow
was governed entrely by natural convection, The rate of formation of vapour and the resultant
conveetive lift are intimately connected with the heat input. As the diameter of a horizontal
heating element is eonsidered tO be immaterial," it may be concluded that there is a fixed relationship
between he and ~ T at constant pressure, independent of the size and shape of the heating elemento
Boiling inside tubes
The picture is more complieated in the case of flow under foreed convection. Here the
blanketing effect of the gaseous phase can be overcome by the shearing action of the flowing liquido
Thus the rate of heat transfer depends upon the rate of liquid fow, so far as it promotes mixing
of tbe liquid and vapour phases. Stroebe! and Bergelin" distinguish three basic types of liquid-gas
fow: bubble fow, slugging, and ring flow or jetting. In the first type offlow both phases are fairly
intimately mixed and the whole volume of the liquid is permeated byrhe gas bubbles. In the
second type of fow the liquid phase moves in bursts and occasionally mixes well with the gaseous
phase. The third type of fow can be represented by liquid flowing round the periphery of the
tube and gas moving in the centre, or by an uninterrupted jet of vapour through a layer of lqud
as on a rectification sieve tray. The' bubble' flow takes place mostly during boiling on the outside
surface of a heating element whereas boiling inside tubes produces most1y the 'slugging' type
of fiow, particularly when the tubes are vertical. The posirion of the tube is also an important
factor, and its influence is in opposite directions according to whether the boiling occurs inside
or outside the tubes. It is generally recognized that boiling inside tubes gives coefficients about
25% higher than boiling outside under similar conditions.? The effect of tube orientation is such
that boiling is much more effcient inside vertical tubes than inside horizontal ones. When boiling
occurs outside the tubes the reverse has been found. R This can be explained by the meehanism of
removal of the vapour blanket in the two cases. When boiling occurs inside the tubes a vertical
position ensures good mixing of both phases during the slugging fow, which is usually eneountered.
In the horizontal position liquid flows along the bottom of the tube and vapour occupies the space
aboye and acts as a blanket to the upper surfaee. Only occasional waves and ripples occur in the
liquid surface to promote more intimare mixing of the two phases. When boiling oceurs on the
outside of a horizontal tube the vapour bubbles quickly disengage from the surface and the rate of
heat transfer is largely independent of tube diameter. In the case of vertical tubes the gaseous
phase accumulates along the outside of the tube, and its performance is consequentIy inferior to
that of a horizontal tube.
Flow rate
When the liquid motion inside the tubes is assisted by pumping, the rate of heat transfer
improves with increasing liquid-flow rateo In the past, opinions concerning this effect were
controversial. Recent papers 9 confirm the fact that the heat transfer improves as the proportion
of liquid in the bi-phase increases, and this effect is achieved when the mass flow rate is increased
by pumping. As a result the heat-transfer coefficient inereases as a fractional power of G, the
mass flow rate, This has been confirmed in the present work on metastable film boiling at low
temperatures. Once the effect of mass flow rate upon the boiling rate inside the tubes is accepted,
the effeet of the tube diameter may be inferred from the Reynolds number. For the same mass
ow (Wlb.jhr.) the rate ofheat transfer will improve as the internal diameter ofthe tubes diminishes.
When there is a constant temperature-gradient between the heating surface and the liquid, the boiling
coefficient may be expressed by the classical equation:
Nu = C (Re)'" (Prt (1)
When the ternperature gradient is constant, equation (1) will be valid for the heat exchange to or from a
bolng liquid, tbe constants C, m and n having their appropriate values. These will depend upon
J. appl. Chem., 2, November, 1952
 MONROE el al.-HEAT TRANSFER TO BOILING LIQUIDS 615
the definition of both Re and Pr, which becomes arbitrary when two phases are present in varying
proportions. The decision regarding the physico-chemical eonstants of the fluid is easier in
extreme cases: these are found in the early nucleate-boiling region when liT is small and Gis Iarge,
and in the very advanced metastable or stabIe film-boiling region where liT is Iarge and G is smalI.
When liT is small and G is Iarge, it is fully justifiabIe to disregard the vapour phase. Results
obtained by Boarts'' in this region were expressed by the equation:
Nu = 0'028 (Re)o.8 (Pr)O'4 (2)
in which the values of Re and Pr are calculated from the properties ofthe liquid phase. This equation
differs from the classical form onIy in the respect that the constant on the right-hand side is about
25% larger than normal. In the case of vgorous metastable fiIm-boiling the effect of the gaseous
phase is predominant and one would expect that the heat transfer rate could be expressed by the
classieal equation, provided that Re and Pr are calculated from the properties of the gaseous phase
at the temperature of the bi-phase and at the mean film temperature respectively. The range of
validity of such equations is very limited, as the increasing proportion of liquid phase in the boiling
bi-phase renders the assessment of Re and Pr entirely on gaseous properties very inaccurate.
In fact the Nusselt number varies with the o 8 power of the mass flow rate only at very low vaIues
of Re. The straight line of the logarithmic pIot gradually changes into a curve with constantly
decreasing slope. The Nusselt relationship can be applied only to the metastable film-boiling
region where there is a large temperature-drop across the boiling film. The development of a
general equation, which would cover a large range of flow rates, is extremely difficult as it requires
the definition of conductivity and viscosity of a bi-phase mixture of constantly varying composition.
It is not possible, either, to correIate the Nusselt number with the o 8 power of the mass flow rate
when the temperature drop across the boiling film is small, because under this condition the
proportion of liquid to vapour is large, and the effect is similar to that of increasing the mass flow
rate.
To sum up, we ean say that when a given liquid boils on the outer surface of a tube or other
heating element under constant pressure and temperature difference, the rate of heat transfer is
governed entirely by the properties of the fluid and the state of the surface. This is not so
when boiling occurs inside the tubes, as in this case the flow rate or proportion of liquid to vapour
must also be known.
Pressure effect
In the case of nuc1eate boiling, the rate of heat transfer inereases with pressure. The film
coefficient is usually expressed!"- 12 as an exponential function of pressure to a power varying
between o-zy and o 4. It is claimed, further, that in the region of nucleate boiling the maximum film
coefficient increases appreciably with pressure up to one-third of the critical value.P Above this
pressure the maximum coefficient increases slowly to a practically constant value, independent of
pressure. The critical-temperature difference which marks the transition from nuc1eateto film boiling
decreases gradually to zero as the pressure approaches its critical value.P This is perhaps not
surprising, as at pressures aboye the critical value the sharp distinction between liquid and gas
disappears, and the difference between nucleate and film boiling ceases to existo It is therefore
understandable that the critical-temperature drop diminishes to zero and the maximum coefficient
assumes a practically constant value at pressures aboye the critical,
Within the film-boiling region the effect of increasing pressure is to reduce the rate of heat
transfer across the metastable film and to increase the rate across the stable film. Both insidel!
and outside boiling'" show similar behaviour. At very high pressures, both amplitudes of the wave-
shaped curve ha versus LlT (Fig. 2) became smaller and one would expect that at extremely high
pressures the curve would become a straight line, in other words the film coefficient would be
independent of the mean temperature-drop across the film. However, this ideal behaviour can
never be realized because of the great variations in the physico-chemical properties of the fluid
with temperature and pressure.
Effect Di physical constants
The variations of the specific heat of oxygen with temperature at several pressures below and
aboye the critical value are shown in Fig. 3. There is a striking effect of temperature on specific
heat in the vicinity of the critical pressure and temperature. Unfortunately it is not possible to
construct the corresponding curves for the viscosity and conductivity of oxygen and nitrogen
because most of the necessary data are lacking.
Fig. 4 gives the available values for the thermal conductivity ofliquid and gaseous nitrogen'< 16
and Fig. 5 shows the inftuence of temperature on the viscosity of oxygen.17, 18
It can be predicted that the Prandtl number will have a decisive influence upon the value of the
heat transfer coefficientat high pressures, particularly approaching the critical value. Above this value

J. appl. Chem., 2, November, 1952

616 MONROE el al.-HEAT TRANSFER TO BO/UNG LIQUIDS

100,000
0'S5 L9 l.l IJI. .:= f-.
~~('~><.
' e..: r . .

-
.._ .~~ "
as _ -
-C-

-~
7
r< 1\
~l\. -:
I 1\\\

I
I 1\ ~- . C'

f-
Z 1
10,000
\.\ ~.....
w
.V ~~ ~\
\ \. ,~ _ ...
U _. \
'" IL
.;

..
&\
LL - \,'Io r'"
~.-

W _ .. --V-
-/ 'S
~
::!
ti. 1000
I .... >

.'\,:\i
LL
- - t>

V7 - l
~\ V v~
ou
f
.. -
... ~.i r' = f:s;-:'
._--

I it~=-- -'-,-- . ----

~ ::-:'-c
~ .
;7 / }; li
- -

._.. -
...
_. r-
....,

100 /
/y /1 11 '1
01 10 10 100 o 1000
TEMPERATURE DROP ACROSS FILM, C.

FIG. 2. ldealized boiling curves at uarious reduced pressures

we can no longer refer to the boiling
.\ MAX = 12 at -117.50 e coefficient, as the heat transfer takes place
-
-
- 1,-
-1 , i
.. c-- .
\ -
I.~
------ - r-- ._ .. f-.
without change of phase. The enthalpy and
temperature of the fluid vary continuously
'0
'+,,, ....1 -
- -f- f-- in common with other physico-chemical
1" ,
1 properties. The values of these properties,
o
u .
. 1 - f-- _ .. ... at a given moment, determine the rate of
.i 15 1 ,. - . _ ..
- f-- -f- heat transfer and consequently the magni-
o
---
. .
. --+ e-f- tude of temperature drop across the ' film'
u
t- - .... for a definite value of the heat flux. The
~ .... -- -f--I--f- disappearance of a second phase introduces
W
I .- -- .1-. _ .... _..1-- f--
another degree of freedom and it is no
~ 10 - ...... .. - . - 1--1-- 1--
longer possible to determine the heat trans-
Q 09 .-'"
r-~2000 Ib..q.ln.gQug<
fer coefficient solely from the temperature
~ O,B \;0' 1500" "
lJl 0.7

06
- -

'IV
I
,~
'X 1 OOO"
750" "
drop and, for inside boiling, the mass flow
rate. The average temperature of the film
05 ~V \.JI ')( r---O" "
----
must also be taken into account. It is thus
" l,..,::~ .._~ t-,)5 ~
04
-
03 f-
- -- "'"t-- 1:::- f-- .- -
possible ro have several values of the
02 .. , boiling' film coefficient at high pressures
o ,1
-200
- -1-.

-150
H- -100 -E O -O -40
I l
-20 O
- for the same temperature-drop across the
film, when the average temperature of the
TEMPERATURE, =c. film is variable. This fact was not apparent
in the sub-critica! region because the basic
FIG. 3. Specific heat of oxygen at high pressure : critica!
tempera tu re, 1r8 8 j critical pressure, 735 lb.l sq. temperature, i.e. the temperature ofthe boil-
in. gauge ing liquid, was automatically fixed by
pressure. Values of the temperature drop
across the film were always related to this basic temperature. There is no fixed temperature in
the region of pressures and temperatures aboye the critical, and the temperature drop can
occur between any two temperatures. Although, in the super-critical region, the physico-chemical
properties vary quite smoothly with temperature, i.e. without any sudden transition between
liquid and gas, their rate of change can be extremely rapid. As heat transfer aboye the critical
point takes place without change of phase, it may be correlated by the use of the normal equations
for viscous or turbulent flow. This may be inferred from the influence of the mass flow rate
(G) inside the tubes upon the film coefficient. However, the uniformity of the relationship
is again affected by the physico-chemical properties of the fluid. Let us consider a case
where the fluid-flow rate inside a tube has been increased. This tube is capable of a certain
rate of heat transfer when containing a fluid of fixed characteristics, but the increase of fuid-fiow
rate alters the mean temperature. This has the effect of increasing the viscosity and conductivity
of the fluid, and reducing its specific heat. The result may be a decrease of the slope of the
J. appl, Chem., z, November, 195Z
 MONROE et al.-HEAT TRANSFER TO BOILING LIQUIDS 617
Q..
_-40 ,
~u30 r- ~l"LIOUIO
r11
M 20 k,
~,
'ti
]. 10 I
>- 9

!: 8 ~fS (Atmospherlc PrCUl.lflll)
> 7
~ 6 X
'" S
e
........ ~
~ 4
U .-<r'"
,.-""":P' I --
3 - -- 1-- J.-- J.---
..J ......o
t--'
r,ll V
...:z:~ 2 , .,P
1-

I
7O 60 90 100 110120 130 140ISO160 170180 190200 220 240 260 270 300 320 340 360 38 o
TEMPER ATURE, K.

FIG. 4. Thermal conductivity of nitrogen and oxygen

Nitrogen, x ; oxygen, O
hB versus .6. T curve, in spite of the increased G.
This would indicate that the effect of Pr is
stronger than that of Re in this particular
region. Finally a comparison of the respective
hB versus .6.T curves shows that the film 'Q
boiling coefficient inside the tubes reaches its
maximum at a smaller critical .6. T than when
the boiling is outside the tubes. This con-
clusion can be of only qualitative value
considering the fact that the film coefficient
inside tubes is strongly affected by the flow rateo - . ..1_ ~+_.-
"-1--~-+-I- -'--~~-+-I-I
I-H-
f- -
In fact we are comparing a family of curves 1 GAS (Atmospherlc Pressure
for various mass flow rates with one curve of .I - -- /1_LLL~ l-!-L '
natural convection boiling outside the tubes.
10'rn~~~-~_~d::A~-g-~n-m~~n~
Experimental
60 eo lOO 120 140 Tt50 leo 200 220 240 2 O 280
It is an established practice in the industry TEMPERATURE,OK.
to transport oxygen in bulk as liquid, which
is evaporated under pressure, into storage FIG. 5 Infiuence of temperature on the viscosity of oxygen
and nitrogen
vessels on the consumer's site. The pressures Oxygen, X; nitrogen, o
used vary from 300 to 2000 Ib.jsq. in. gauge.
The heat of evaporation is usually supplied by steam or electricity, but it would be an obvious
economy to obtain heat from the atmosphere, which is normally at a temperature about 200 aboye
that ofliquid oxygen. Not all ofthis avaiIabIetemperature difference can readily be used, however,
because ice is deposited on the surface of the evaporator tubes and makes rapid heat-transfer
impossible. To enable air to be used as the heating mdium it is necessary to maintain the air
temperature and also that of the evaporator surface aboye the freezing point of water. This
imposes rather unusual problems in heat transfer which have led to the study described in this papero
It was found that published data on heat transfer to liquefied gases were few. They related
almost exclusivelyto measurements ofheat transfer to boiling liquefied gases, at atmospheric pressure,
frorn a heated eIement immersed in the liquido The vapour produced during boiling was allowed
to rise freeIy to the surface of the liquido
lt was not known if these data wouId be applicable to boiling inside tubes, for in addition to
the well-known effect on heat transfer of the temperature drop across the boiling film, the continued
presence of the vapour phase and the forced convection of the liquid-gas mixture past the heating
surface considerably alter the conditions under which heat is transferred. The evaporator was
required to oprate at a pressure of about 2000 Ib.jsq, in. gauge, corresponding to a reduced pressure
of about 3. For these conditions no data were available, either on heat transfer or on the physical
properties, such as viscosity and thermal conductivity, from which coefficientscould be qualitatively
predicted. Estimates of these physical properties could be made by the method of Gamson.P who
used the principle of corresponding states but only at temperatures aboye the critical, which for

J. appl, Chem., 2, November, 1952 La

618 MONROE el al.-HEAT TRANSFER TO BOILING LIQUIDS

oxygen is - lI8 8. The uncertainty ofthese estimates increased greatly as the critical temperature
was approacbed.
An experimental investigation was therefore undertaken to determine the values of the
coefficients of heat transfer under the required conditions and to investigate the effect on tbese
coefficients of variations in the following parameters :
1. At atmospheric pressure.
(i) The temperature drop across the boiling film; (ii) the velocity of the liquid through the
tubes; (iii) tbe diameter of the evaporator tubes.
n. At high pressure.
(i) The flow rate of the fluid through the tube; (ii) the pressure; (iii) the temperature drop
across the 'boiling , film; (iv) the mean temperature of the fluid.
The behaviour on evaporation of both oxygen and nitrogen was studied at atmospheric pressure.
At high pressures measurements were made only with oxygen.
When two phases were present, at low pressure, evaporation was studied only in vertical tubes,
as the presence of an appreciable amount of vapour in a horizontal tube would render a large part
of the surface inoperative for heat transfer.
Measurements at atmospheric pressure
Measurements at atmospheric pressure were made with a bank of four copper tubes, each
seven inches long. These tubes were assembled vertically and in line in an alumnium air duct
as shown in Fig. 6. Liquid oxygen or nitrogen was supplied from a metal Dewar vessel, the rate
of flow being controlled by varying the applied pressure. The mass flow rate (G) varied between
20,000 and 264,000 lb.jsq. ft.jhr.

G"'$METER

FIG. 6. Liquid-oxygen euaporation rig

The liquid-gas mixture emerging from the tubes was separated, the liquid was weighed, and
the gas was metered after being warmed to room temperature.
The external surfaces of the evaporator tubes were extended by means of fins, and they were
heated by blowing across them a stream of airo The air rate was controlled by a damper on the
fan inlet. The gross air velocty in the duct was varied between o and 65 ft. /sec.
The quantity of heat passing across the tubes was calculated from the volume of gas produced,
making allowance for the 'dead losses' of the system, Le. the heat leak into the Dewar vesseis,
the separator and the connexions, which was measured separate1y. These dead losses, although a
fair1y high proportion of the total evaporation (generally about 30%), proved remarkably constant,
a redetermination after an interval of months showing a negligible difference from the original
measurement.
The temperatures of'the tube walls were measured by attached copper-constantan thermocouples.

J. appl, Chem., 2, November, 1952

 MONROE et al.-HEAT TRANSFER TO BOILING LIQUIDS 619
619
The liquid stream was assumed to be at its boiling point as determined by the pressure on the supply
vessel.
Most of the measurements were made with tubes of t in. internal diameter, but to determine
the effect of tube diameter, a number of measurements were also made with tubes of internal
diameters of 0.67 in., o' 404 in. and o' 240 in. respectively.
Measurements al high pressures
Experimental runs at pressures aboye the critical were made with oxygen flowing inside a
horizontal tube t in. in internal diameter fixed inside an aluminium air duct (Fig. 7).

,.

,_ SUPER
; - HEATER

HIGH PRESSURE LlqUID OXYGEN

OXYGEN FROM ElATH
COMPRESSOR
COOLER 8LOWER GAS METER

FIG. 7. High-pressure oxygen-eoaporation rig

Oxygen from a gas-holder was compressed to the required pressure and cooled by passage through
a coil immersed in a bath of liquid oxygen before it entered the measuring tube. The rate of flow
of the fluid was regulated by the high-pressure inlet and outlet valves. The mass fow rate was
measured by passing the oxygen finally at atmospheric pressure and room temperature through a
dry gas meter. No separator was required as all experiments in this series were carried out aboye
the critical pressure, and the fluid was therefore invariably homogeneous.
Heat was supplied, as before, by a stream of air passing through the duct. The amount of
heat was determined from the oxygen fow rate and the enthalpy increase. The relations between
enthalpy, temperature, and pressure were obtained from the thermodynamic diagram for oxygen,
prepared in 1940 at the Kammerlingh Onnes Laboratory, Leiden.
The inIet and outlet temperatures of the oxygen stream were measured by means of specially
constructed high-pressure copper-constantan thermocouples (Fig. 8).
The tube-wall temperature was measured with
attached copper-constantan thermocouples. A
graph of the temperature distribution aIongthe tube
was then plotted and an ' equivalent straight line '
was drawn. This' equivalent line' was used for
calculating the temperature drop from the tube
wall to the inside fluid, and was constructed as
follows (see Fig. 9) :
The actual wall temperature at the outlet of
'150

o
CJ
~.~
-156
';
-152

0-154 '---- - t~- i

~.~
7'
..\...
~,\t<;
I '~

.'.-.,.\(0.-

ffi -156 -- --- r- _._._--- -(o.?\)~~l-~f-- ,----

?3 l-- -- --
..~i
Q_

~-160 /'"
-162 /'"~ V
V
-164~
o 10 20 30 40 SO 60 70 SO 90 100 110
DISTANCE ALONG TUBE, in.

FIG. 8. High-pressure thermocouple FIG. 9. Typical tube-wall temperature-distribution during

rype PY high-pressure run

J. appI. Chem., 2, November, 1952 i,

620 MONROE et al.-HEAT TRANSFER TO BOILING LIQUIDS

the tube was taken as the end point of the line and the equivalent Iine was drawn so that the
areas between it and the original wall-temperature line lying on either side were equa!.
A11tubes used in these measurements were of copper and had an internal diameter of o 5 in.
For different experiments, lengths of 112 in. and 28 in. were used.
Results and discussion
Boiling under low pressure
l. Temperature drop across the boiling film
The relationship between the film-boiling coefficient and the temperature drop across the film
is given in Fig. 10 for liquid oxygen and nitrogen boiling at atmospheric pressure. The shapes of
the curves are typical of metastable film boiling. They range between a minimum value of he
at a temperature drop of about 200 and a maximum at a drop of about 5. The internal diameter
of the tubes was o 5 in. and the flow rate was maintained at I lb. of liquid per minute, corresponding
to a mass flow rate (G) of 44,000 lb./sq.ft./hr. The similarityofthe two curves is obvious and can
be explained by the fact that the rate of heat transfer is strongly controlled by the gaseous film.
The physical properties of gaseous oxygen and nitrogen are not very dissimilar.

I~ _..- --

~
\\
a. \\
UJ
\)
LL
ti)
t\..
Z \.\.
~~
a:
1-

~ ~ ~
UJ
:z:
'" ~ ~
"
::::!
LL
~
o
~ ~ ::o.:..P
,]N: r-o o
~
2 ~ b:>= lo'-' o ',.::;
::::! 10
o " ~ n
ID
o 50 100 150 200
TEMPERATURE OROP (aT), oc.
FIG. 10. Boiling film heat-transfer coefficient (hB) at varying temperature drops CllT), for
liquid oxygen and nitrogen. Flow: 1 lb.lmin. through vertical copper tubes of internal
diameter ! in. G = 44,000 Ib.jsq, ft.jhr. Range of heat flux : oxygen, 12,000-22,000
kcal.lsq. in.lhr.; nitrogen, 10,000-20,000 kcal.isq. in.lhr.
Oxygen, O ; nitrogen, ~
2. Flow rate
Fig. 11 shows the effect of the flow rate upon the film coefficient. In order to eliminate the
effect of varying temperature drops across the film, ~ T was kept constant at about 200 This 0

temperature drop corresponded to the minimum rate of heat transfer by metastable film boiling.
Four tubes were investigated of 0'240 in., 0'404 in., 0'500 in. and 0.670 in. internal diameter
respectively. It is noticeable that the coefficient improved with increasing liquid-flow rate and
diminishing tube diameter. An attempt has been made (Fig. 12) to correlate the results for a11
tubes by plotting the Nusselt number against the Reynolds number. Nu was evaluated for gas
at the mean temperature of the film and plotted against Re, which was calculated on the assumption
that the fluid consisted entire1y of liquido The method of assessing Re for this plot was immaterial
and the liquid basis was chosen arbitrarily.
The usual relation Nu o: (Re)o,s appears to hold at very low Re (up to 3000), but at higher
values the slope diminishes. This is explained by the fact that at low rates the fluid is
predominantly gaseous, but as the flow rate is increased, a greater proportion of liquid persists in
the mixture. If Nu were plotted against the true Re of the mixed phase the effect would be to
increase the slope of the curve at the higher liquid-fiow rates, and the normal relation might then
persist over the whole range,

J. appl. Chem., 2, November, 1952

 MONROE et al.-HEAT TRANSFER TO BOILING LIQUIDS 621

?
::!: 200
w

e
O
[ O

6
O
Q)
'O
O
f-

'V u~
+ ~ii
J~
6
o

-o O
;'V
6
Itl 'V
O
r-------
+1I'Jo;"
'J V L I.J

IV lA"" O
IJJ 50
O
i 40
3 4 5 6 78910 2 3 4 5 6 7
V
/
~ LlQUID FLOW RATE, Ib./min. V
2 3 4 5 6 7 8 9104 2 3
FIG. 1I. Effect oj liquid-flow rate on boiling coeffi- R~ BASED ON LlQUID FLOW RATE
ciento Constant temperature-drop, about 200 0 ;

fiow rate, I lb.lmin., therejore G = 44,000 lb.lsq. FIG. 12. Relationship betuieen Nusselt and Reynolds num-
ft.lhr. in tubes ojinternal diameter t in. [Thefiow- bers jor oxygen and nitrogen boiling inside tubes at a
rate units 3, 4,.... 10 are to be read as O' 3, temperature drop oj about 200 0

O' 4, .... 1'0)

+ nitrogen, o 5 in. tubes
o internal diameter o- 240 in. O oxygen, 0"24 in. "
6 0'404 in. a 0'44 in. "
O o- 500 in. O o'S in. "
'V 0670 in. V 0'67 in. "

9"
;
0.1
~ O
~ ~ ~ m BOILlNG OXYGEN.

i '" BOILlNG NITROGEN.

V N.~!
I
0
- INSIDE TUBES.

L~ t
P\

/"
VI OUTSIDE TUBES.
I ;;X
I ! ~/
1
HASELDEI ,~ ~,

/ ~'
,,.
q"
19~9 Q..~I

, d" ~'"
,, , " ~D
OC' . /
700 , f)
Kn
600
.sO ~~~ N. SLOPE - . 96

!Il 400
~ O. SLOPE- . 95
s: "\::~ f'L,
30 O
RANGE OF HEAT FLUX
~tO)~
OXYGEN 12.000-22000
20 O
K.CAL/m/hr ~J~~~
~?)O
NITROGEN 10.000-20.000 ~
K.CAL/m~r ~ -",cOI E"IJt9~0N1.
10O
~ --
) 4 5 10 20)0 40 50 lOO 200 400 1000
t.T Oc
FG. 13. Surface boiling heat-transjer coefficient (hB) versus temperature drop (f1T)

3. Boiling of liquid oxygen and liquid nitrogen : review and comparison

The published data on heat transfer to liquefied atmospheric gases are plotted together in Fig. 13.
A distinction has been made between boiling outside and inside tubes. Haselden'" and Giauque=
studied the nucleate boiling ofliquid oxygen, and Wei!22that ofliquid nitrogen during outside boiling.
Weil also studied the boiling of liquid nitrogen in the metastable film-boiling region 23 and Bromley"
investigated stable film boiling. The final stage ofthe metastable film-boiling ofliquid oxygen has
been studied by Giauque-' and his results form an isolated piece of information on boiling ofliquid
oxygen in this region.
It can be concluded that there is sufficient information available to construct almost the
complete curve for liquid nitrogen boiling outside at atmospheric pressure, but only partial
information regarding liquid oxygen. The only published data on inside boiling consist of two
curves by Haselden' for liquid oxygen and nitrogen boiling under natural convection conditions.
Our work supplies two families of curves, one each for liquid oxygen and nitrogen, two specimens
of which at the constant mass fow rate, G = 44,000 lb.jsq. ft.jhr., are included in Fig. 13. The

J. appl. Chem., 2, November, 1952

 6zz MONROE et al.-HEAT TRANSFER TO BOILING LIQUIDS

close approximation of Weil's curve to that of Giauque is significant. No data are available for
stable film-boiling of liquid oxygen.
In general the shapes of the curves for boiling inside and outside are similar, although the
critical ~ T seems to be smaller for inside boiling. This can be exp1ainedby the gradual accumulation
of gaseous phase during boiling inside tubes particularly in the vertical position, even when the
temperature drop is relatively small. This results in a kind of metastab1e film boiling. The effect
of flow rate in the nuc1eate region is also pronounced, but information is far from complete. Very
high values of boiling coefficients can be obtained by strong1y increasing the 1iquid-flowrate, whi1e
keeping ~T constant. It wou1d be perhaps more satisfactory for the purpose of comparison to
relate the boiling coefficient to known rates of formation of gaseous phase." The present choice
of datum, i.e. Re assessed on either liquid, gaseous or mixed phase basis, is arbitrary and does not
describe precisely the conditions of boiling.
The slopes of our curves in the metastab1e film-boiling region are very close to - 1, which
means that for a certain set of conditions the heat flux acceptable by the boiling liquefied gases
appeared to be constant over the metastab1e region. In the stable fi1m-boiling region the slopes
should be different for outside and inside boi1ing. The first can be seen from Brom1ey's curve
(Fig. 13), the small slope of which indicates natural convection conditions. At a high temperature-
drop this slope will increase owing to the effect of radiation, For inside boiling with forced convection
it is to be expected that the slopes will be higher than those for outside boi1ing.
Boiling under high pressure
The effect of pressure upon the boiling of 1iquid oxygen was studied in the region ahoye the
critical. As there was no change of phase, true boiling did not take place but the enthalpy of the
fluid increased uniformly within the range of temperatures investigated. Fig. 14 illustrates the effect
of fiow rate upon the film coefficient at different pressures. The investigated pressures were 2000,
1500,1000 and 750 lb.jsq. in. gauge, the last being iust aboye the critical pressure ofliquid oxygen
(735 lb./sq. in. gauge). The fow rate varied between 9000 and 111,000 1b.jsq.ft.jhr. and the
temperature ofthe fluid from -180 to --10.
The favourable influence of pressure upon the film coefficient in the region aboye the critical
can be readily seen from Fig. 14. Heat transfer a1soimproves with increasing rate of flowand the slopes
of the individual curves suggest the flow was viscous. Verification of this is not possible because of
the scarcity ofphysical data. The curve for 750 lb.jsq. in. gauge pressure covers only a small range
of flow rates as it was impossible to increase the flow rate (G) beyond about 44,000 without
unbalancing the pressure and causing separation of the liquid phase. The remaining curves show
an infexion point, whieh could be reproduced at every pressure investigated, in the vicinity of
G = 44,000. This inflexion, however, is not attributed to the effect of the flow rate but to the
fact that the average fluid temperature in these cases was not far from the critical value, and near
that point the specific heat of a fluid varies very considerably with temperature. By varying the
flow rate only, it was not possible to maintain
constant either the temperature drop across the
"- , - _._-,..._--_.,._-".-
oU 1000 a:: a..:" soo
a:i eoo "
_ ..tg 'O
1.11'"
",,-
- --
1.11 o 700 ",E
U. " 600 ."'"

u. ~ 600
_ ~.,\',,'l,
Zcr 500
"Z'~fT /!.-;~ f~ a<'"
:""-:- ~
<...:
500
1;:: '~c rov 400
~ D
a: "
b<:e ~:::::p..
400 t- ....
o - 300

--~
t-~
t-':::o 300 1.111-
<~ ~ ~ :rz1.11 ~j
LIIt- ,,_....:.~ o 200
:rz 200
1.11-1.11 O'
~ ~ 1.11- -
~
~
oU u.
(U: tJlu. ~
ZUJ
~~ 100 -O 100
3 4 5 6 7 e 9 10
8 I 2 3 4 5 6 7 89 10 2 3
u 2 20 30 40 50
TEMP. OROP FROM WALL TO FLUID oc.
llQUID OXYGEN FLOW RATE (G), Ib./min.
FIG. 15. Effect of variaton in temperature drop on the
FIG. 14. Coefficient of heat transfer to liquid oxygen al high heat-transfer coefficient 10 oxygen flowing in tubes
pressures. Ranges of other variables, low rate to high 92 in. long, with an internal diameter of o 5 in.
rate: Heat fluxes, 2500 ro 110,000 kcal.lsq. m.lhr.; tem- Flow rate constant at o- 5 lb.jmin., therefore G -
22,000 lb.lsq. ft.lhr. Ranges of other variables: mean
perature - 160 to - ro" and - 180 to -- lOO"; tem-
0 0
temperature, - 120" to ISO"; temperature illcrease
perature drop, 80 to 140" [Theflow-rate units 1, 2, .... of fluid, 50 to r ro"; heat flux, 2700 to 7300 kcal.lsq.
10 are to be read a.' o' 1, O' 2, .... 1 . O] m.lhr.
V 2000 lb.jsq, in. gauge
O r soo v 2000 lb.jsq. in. gauge pressure (137 arrn., Pr ~. 2' 74)
O 1000 O rooo Ib./sq. in. gauge pressure (69 atrn., Pr = 1 38)
A 750

J. appl. Chem., 2, November, 1952

 MONROE el al.-HEAT TRANSFER TO BOILING LIQUIDS 623

film or the average temperature of the fluid. The system was not monovariant and a further proof of
this was obtained during the study of h versus LlT relationship (Fig. 15). This was conducted at two
pressures.zooo and 1000 lb.fsq. in. gauge respectively. The two curves in Fig. 15 for
differentpressures
at the same fow rate (G = 22,000) conform generally with Fig. 2, but the individual points show
a considerable scatter which could not be justified by experimental errors, and was caused by
variations of the mean fluid temperature. An attempt was therefore made to separate the effect
of LlT from that of the fluid temperature. Fig. 16 shows the variations of the film coefficient
with the fluid temperature at constant ternperature differences of 10D and 36 respectively.
Assuming the mean fluid temperature to be -1400 throughout and introducing the necessary
temperature corrections, a better correlation of hs versus LlTis obtained in Fig. 17. It will be seen
that the curves are not horizontal and consequently there was still a film effect at pressures of 2000 and
1000 lb.fsq. in. gauge within the range of temperature drop covered (3 to 35). The two curves
intersect when LlT is about 7 and at higher temperature-drops, corresponding to the metastable
film-boiling region below the critical pressure, the curve for the higher pressure (2000 lb./sq. in.
gauge) les below that for 1000 lb.jsq. in. gauge, in agreement with results published else-
where.P The shape of the curves is due to the rapidly changing physical conditions of the fluid
with temperature. To study high-pressure boiling as a monovariant system, it would be necessary
to fix all but one of the following parameters: fluid flow rate, temperature and pressure of the fluid
31,)0 ~ 2000r--'--~~~.-n------.---r-'\-.
1-'
Z
loLI
V
:
u,
us
OV
uo!,
-- _.

,,$/'
,,/'
~
7 -
I I I I
Temperatur.
constant at
.1
drop
10C.
~
~
O.
1000l=I;(,~k=::t=::t=+i::ttl====t.=:
BOOI---+-',,-+--~'
v~ 700f----~'Z
r-- ---f--. - -
i=
-

..'V.
-_. - 1-
-
--- _. ------- f--- a: 600
0:'::
':":~;" 200 ~ 500~-~--+-~~~r~-----+---r~-~
--f--- e- 400f---' ~~D I _.-.-t---+--t--i
z:f
-
0:-' -c cT
a:':: 300
----1-I--I--t--t-+--H''fui'~----j---+--+--j
~ ~ o
1-3
.>/. Ternpereture drop ~
~
us - _. -..
constant at 36 oc.
1---" ;z, ;~ 200~~ir-+-+-~+~~~~-~~~~~'V--+--~~
J:
J
'o"
Vl
z
100
-170 - 160 - 150 - 140 -130 -120 -11
O
~
~
~'1
I092~~3~~4~5~6~7~B~9~1~0~--~2~O~~3~0~4~0~50
MEAN FLUID TEMPERATURE, oc.
TEMPERATURE: DROP, OC.
FJG. r6. Effect oj mean fluid temperature on heat- FIG. 17. Fig. 15 replotted with al! coefficients corrected
transfer coefficient, Constan: fiow rate of o 5 10 a temperature of 140'" by empirical correction
lb.lmin., therefore G = 22,000 lb.lsq. [t.ihr. con- jrom Fig. 16
stant temperature-drop

and temperature drop across the film. No attempt has been made at any mathematical treatment
of heat transfer to fluid at high pressures and low temperatures because of the scarcity of physical
data relating to these conditions.
The accuracy of experiments throughout this work was determined by two main factors :
(1) the determination of the enthalpy changes, and (2) the precision ofthe temperature measurements.
In the experiments on low-pressure boiling, mass changes of liquid oxygen or nitrogen were easily
determined and the heat balance was checked against similar enthalpy changes on the air side.
At high LlT the inherent errors in measurement of temperature by thermocouple were immaterial ;
at low LlT great care was taken to avoid errors due to thermal conduction through the leads.
Although the lack of vertical scatter of points in Fig. 14 indicates that the experimental results are
reliable, no weight is attached to the results at values of LlT below 10. As already mentioned dead
losses of the apparatus constituted an appreciable part of the total heat load, but they could be
deterrnined accurately with a very good reproducibility. In the work at high pressure there were
no dead losses as the whole evaporator tube was inside the duct. To determine the average LlT
an equivalent wall-temperature line was produced for every measurement and gave results comparable
with those obtained by graphical integration. There again the accuracy of temperature measure-
ment was good at high LlT but diminished at low values, for which the log (mean temperature-drop)
was used.
Nomenclature
D Diameter of tube, in.
G Mass flow rate, lb'sq.ft./hr.
h Coefficient of heat transfer between fluid and surface, kcal.jsq. m.fhr. c.

J. appl. Chem., 2, November, 1952

 CUMMING el al.-CHROMIUM HEXACARBONYL. 1
hB Boiling surface coefficient.
h; Inside film coefficient.
Nu Nusse1t number.
Pr Prandtl number.
Re Reynolds number.
W Mass flow, lb.fhr.
IlT Temperature difference between the bulk temperature of fluid and temperature of surface, oc.
Acknowledgments
The authors wish to thank the Directors ofThe British Oxygen Co. Ltd. for permission to publish
this papero Special thanks are due to Mr. A. M. Clark for his constructive criticism and kind
assistance in preparing this papero
The British Oxygen Co. Ltd.
Research and Development Department
Morden Factory Estate
Morden Road
London, S.W.19
Received 7 April, 1952
References
l B.P. Appl. No. 10227/50. l. Borovik, E., Matveev, A. & Panin, E. (1940). J.
I Boarts, R. M., Badger, W. L. & Meisenburg, S. J. techo Phys., Moscow p. 988.
(1937) Trans. Amer. Inst. chem. Engrs 33, 363; la Eucken, A. (19Il). Ann. Phys., Lpz. 34, 185.
Industr. Engng Chem. 29,912. Dickens, B. G. (1934). Proc. royo Soco [A] 143,517.
Bromley, L. A. (1950). Chem, Engng Progr. 46, 5.
McAdams, W. H. (1942). 'Heat Transmission ', l' Gerf, J. F. & Golkov, G. 1. (1941). 'f. techo Phys.,
p. 314 (London: McGraw-Hill). Moscow r r, 613; Rudenko, N. S (1939). J. exp,
Stroebe, G. W., Baker, E. M. & Badger, W. L. theor, Phys. 9, 1078.
(1939) Trans. Amer. Inst, chem. Engrs 35, 17. 18 Johnston, H. L. & McCloskey, K E. (1940). J.
I Bergelin, O. P. (1949). Chem, Engng 56, 104. phys. Chem. 44, 1038.
7 Jakob, M. (1949). 'Heat Transfer', p. 651
19 Gamson, B. W. (1949), Chem, Engng ~'rGJi;.45, 154.
(London: Chapman & Hall).
20 Haselden, G. G., & Peters, J. 1. (1949). Trans. Instn
6 Haselden, G. G. & Peters, J. 1. (1949). Trans. chem. Engrs, Lond. 7,7, 201
Instn chem. Engrs, Lond. 27, 205.
n Giauque, W. F., Rcport on Heat Transfer and
e Yoder, R. J. & Dodge, B. F., General Discussion
on Heat Transfer, Section 1, p. 1, Institution of Design of Refrigeration Evaporarors for the
Mechanical Engineers and the American Society National Defenc- Research Committee, 1 Mar.,
of Mechanical Engineers, Il-13 Sept., 1951. 1942, Wash!!lgton, D C.
l. Bonilla, C. F. & Perry, C. W. (1941). Trans. Amer. 22 Weil, L., 8th International Refrigeration Congress,
Inst, chem. Engrs 37, 685. 1951 (London: Institution of Refrigeration),
11 McAdams, W. H., Addorns, J. N., Rinaldo; P. M. 23 Weil, L. & Lacaze, A. (1950). C.R. Acad. Sci., Paris
& Day, R. S. (1948). Chem, Engng Progr. 44, 639. 230; J%.
lO Cichelli, M. T. & Bonilla, C. F. (1945). Trans.
Amer. Inst, chem. Engrs 41, 755. 21 Haselden, G. G. & Prosad, S. (1949). Trans. Instn
u Lukomski, S. M. (1947). Izvest. Akad. Nauk chem. Engrs, Lond. 1.7, 200.
S.S.S.R. Otdel Tekh Nauk p. 967. ss Verschoor, H. & Stemerding, S., General Discussion
1< Farber, E. A. & Scorah, R. L. (1948). Trans, Amer. on Heat Transfer, Section 2, p. 27, Institution of
Soco mecho Engrs 70, 369. Mechanical Engineers and the American Society
of Mechanical Engineers, 1l-13 Sept., 1951.

CHROMIUM HEXACARBONYL. l. THE PREPARATION AND

STABILITY OF CHROMIUM HEXACARBONYL
By W. M. CUMMING, J. A. HORN, and P. D. RITClllE
I. Recorded methods for preparing chromium hexacarbonyl are reviewed and critically compared.
Previous theories on the mechanism of formation from carbon monoxide, chromic chloridc, and a
Grignard reagent are not wholly satisfactory, and an alternative rnechanism s proposed, entailing the
formation of an intermediate organo-chromium-IlI dibalide.
11. Hydrocarbon solutions of chromium hexacarbonyl are decomposed on exposure, in the
presence of oxygen, to radiation in the near ultra-violet, forming arr insoluble gelatinous precipitare.
In the presence of a molar equivalent of a long-chain fatty acid, such as stearic acid, precipitation is
inhibited, though the carbonyl still decomposes to hydrocarbon-soluble products. The mechanism of
decomposition and inhibition is still obscure.
Introduction
The effect of chromium compounds on the combustion of hydrocarbon fueIs in internal
combustion engines has been only incompletely studied, and the few recorded results contain a
number of discrepancies (Sims & Mardles I926; Charch, Mack & Boord, 1926; Weerman, I927;

J. appl. Chem., 2, November, 1952