GENIE
)L. 9
195S
The influence of surface phenomena on the performance of
distillation columns*
F. J. ZuiDEii\vEr, and A. HAUMKXS
Koniiiklijke/ShcH-Liiboratoriuiii, Amsterdam (X.V. l)e Bataafsehe IVtroU-um Maatschappij)
(Received 4 March ll)."»8 ; in ririfieiljornt "£~ March 10.18)
Abstract—A study lias been made «m the inHiifiur of changes in surface tension on the formation
of iutcrfacial area in different types of distillation, equipment. It has been found that liquid
films arc stabilized when the surface tension of the reflux increases down the column, whereas
with decreasing surface tension liquid films break up into narrow rivulets or droplets. Intcrfneial "
area in sprays, however, is not much influenced by changes in surface tension.
For :i given liquid mixture, break-up or stabili/.iit ton of films can be demonstrated by reversing
the direction of mass transfer ; the magnitude of the effects depends distinctly on. the magnitude
of the surface tension gradient developing in the reflux.
In apparatus where the mterfachil area is mainly of the film type, mass transfer rates for
mixtures causing the surface tension, of the reflux to increase may be twice as high (or more)
as those in systems in which the surface tension of the rellux decreases. In the case of technical
equipment with intcrfacial area mainly in the form of droplet sprays, this influence of the system
on mass transfer is but small.
Resumed—Les auteurs out ctudic 1'effct dc changements dans la tension superlicielle sur la
formation de Tinterface dans differcnts types de colonnes de distillation. Us out trouve^ que les
films de liquide sonte stabilises quand l;i tension superficielle du liquide de reflux augmentc,
tandis qu'une diminution de la tension supcrficielle cause la, rupture dc ces films en formant de
petits ruissenux ou gouttes. Cependant, dans des systcmes de gouttes ('interface n'est pas
influence'e beaucoup p:ir des changements dans la tension superficielle.
Tour un melange dc liquidcs donne, on peut demontrcr la rupture ou la stabilisation des
films en changeant hi direction du transfert de masse ; la grandeur des effets depend nettement
de la valeur du gradient de la tension superficielle qui se forme dans le liquide en reflux.
Dans les colonnes oil Tinterface a principalemcnt la forme d'un film, les vitesscs du transfert
de masse, pour le cos de melanges qui provoqucut ttiie augmentation de la tension superfieielle
du liquide de reflux, peuvent ctre deux fois (ou plus) plus clevees que celles qiii se presentent
lorsque la tension superficielle du liquide de reflux dimmue. Dans le cas d'installations industrielles comportant des interfaces principalement en forme dc gouttcs, cette influence du systeme
sur le transfert de masse n'est que faible.
Zusammenfassung—Eine Untersuchung iibcr den Kinfluss von Anderungen, der Obcrflachenspannung auf die Grcn/flachenbildung in Dcstillationskolonrtcn verschiedcner Art hat ergeben,
dass Klussi^keitsfiluic stabilisicrt wcrdcn, wenn die Obcrflachcnspannung des Riickflusses
zunjmmt, wahrcnd bi'i abnehmender Obcrflachcnspammng solchc Kilme zu schmalen Stromen
oder Tropfcn zcrrisscn wcrden. Jedoch wird die Girnxfliiche in Tropfensystemen von Anderungen der Obcrflachcnspannung nicht stark beeinflusst.
*The contents of this pajK-r were read at an " Interne Arbeitssitxung des Kachausschusses Destination, Rektin
;ation und Extraktion der V.D.I. Fnchgruppe Vcrnilirciibtechnik," 21- April, liijfi, 15ingcn-on-Hhiue (Germany) [13]
F. J. ZUIUKRWEG and A. HARMKNS
Fiir em t»e*timmtes Kliissigfceitsgemiseh liisst sirh *l:is XrrrcissenoderStaliilisieren drr FHme
nacfmeiseii, indent die Ilu-htun™ ties StnfTaii.statischcs umpckclirt wird ; die Effekte wenlen
deutlieh von der Crussc dcs sich im Iliw-kRuss entwickclndcn Oberflaelifn.spunmm<:sjjeR*Hes
bedingt.
In Apparaturen, wo die Grenzttiiehe hatiptsarhlirh in Form ernes Films nuftritt. kunn bei
Gemischen, die die OberiHU-henspannuny des KiiekHusses stcijrern der Stoffaustatiseh zweimai
so schnell foder schneller) scin als der bei Syslcmen, wo die Oberflaehenspannun" des Kiu-kflusses
abnimmt. Bri technischen Apparaturen mit finer Grcnzfliiche haupEsacIUich in Form eines Sprudelbetts ist dieser Einfltiss dcs Systems auf den Stoffatistausch nur gering.
gradients cause rapid movements in the surfart
1. I N T R O D U C T I O N
IN operations of mass transfer between a gas and may lead to a spreading or a contraction of
or vapour and a liquid the transfer rates obtained the surface [7], Such contraction phenomena
depend on the rates of diffusion in the two phases have indeed been observed under certain conand the magnitude of the contact area between ditions in the absorption of ethyl alcohol vapour
the phases. These quantities are influenced by the [8], and ammonia [1] into water films, It is
type of equipment in which the operation is readily understandable that the contraction of
carried out, the physical properties of the phases the liquid surface into small rivulets will have a
and the operating conditions applied. Frequently profound effect on the overall mass transfer
the effect of these factors is mutually dependent rates. Presumably many hitherto unexplained
and cannot clearly be separated. As an example, anomalies in mass transfer rates reported in
the change in operating conditions may affect the literature [3, 10] have to be ascribed to this
mass transfer rates on a bubble cap tray quite type of surface effects.
It is noteworthy that the relation between
differently from those in a packed column. This
mass
transfer and interfaeial area has recently
is particularly true in connexion with the subject
received
attention in the ease of liquic
uid
of the present study, which discusses the influence
contacting
[n].
The
stability
of
dispersed
droplets
of surface forces on vapour-liquid interfacial
has been shown to depend on changes of surface
area and the resulting mass transfer rates.
In general, little attention has been paid in the tension caused by the transfer of a solute through
past to the formation of interfacial area in actual the interface.
Complex surface phenomena as quoted above
vapour-liquid mass transfer operations. Most
are
in general very difficult to analyse theoretiof the work done concerns the study of bubble
cally.
The accumulation of a general insight into
sizes emerging from submerged orifices, the degree
these
problems
is a first objective that is best
of wetting of packing units in a packed column
obtainable
by
carrying
out adequate experiments.
or the characteristics of liquid sprays. The
It
has
been
found
that
in the case of vapourformation of interfacial area under conditions of
liquid
contacting
(distillation)
the latter can be
mass and/or heat transfer is not covered by these
previous studies. It is found that the interfacial realized relatively easily. The first purpose of
area depends primarily on the How conditions of this paper is to describe such experiments, and
the phases and mass forces. Surface forces to give a discussion and possible explanation
influence the interfacial area to a minor degree, of the results.
the general effect being a smaller area with
2. S u u f A C K TKXSIOX C H A N G E S IN
increasing surface tension.
DISTILLATION
This situation, however, may change considerably if the formation of ititerfacial area takes
The surface tension or intcrfaeial tension of Huplace under mass and/or heat transfer, particu- n-flux is subject to changes during the downward
larly if owing to these phenomena .surface tension (low of the liquid in the fractionating column.
gradients develop along rhc vapour-liquid inter- This is caused by the changes in composition and
face. It has loiiff been known that surface tension the increasing temperature. It is possible that
90
•*
S
1 lie i m i U C l U l - til > m T < t i i
«Tf..rm--m.-.' ..f i f k t i f f . i f HIM t ((/iiiim-i
jrtu mmivn
mperature and composition effects reinforce or
•utralizc each other in this respect. Since in
ipour-liquid systems temperature ;md composi>m are closely interrelated, the possible variations
surface tension can easily be calculated from
ie equilibrium compositions and temperatures
vd the surface tension of the pure components
: the same temperatures.
It is then found that for systems composed of
impounds of homologous series, the possible
xanges in surface tension are about 2-3 dyn/cm.
or mixtures of compounds of a less related nature
inch larger variations in surface tensions arc,
owever, possible. Mixtures of hydrocarbons and
leohols, for instance, may show surface tension
mges of about 5 dyn/cm at their boiling points.
'ure hydrocarbon systems can also show large
urfaee tension gradients, if they are composed
f paraffins and aromatics. For instance, the
-irface tensions in the system benzenc-;t-heptanc
uige between 21 and 12 dyn/cm, those of the
lixture H-lieptanc-tolucue between 12 and
S-5 dyn/cm. From these data it is seen that in
normal distillation the reflux running down in a
rac
iting column decreases in surface tension
i the urst case, whereas with the second mixture
n increase occurs. It will be clear however, that
hese changes in surface tension only develop
." the operation of- the column is such that a
radient in concentration and temperature occurs
long the column. In the case of operation close
o minimum reflux the changes in composition,
nd therefore also the gradient in surface tension
n the reflux stream are smalls even if there is a
typos of systems can be distinguished with respect
to the changes in surface tension developing in the
rcllnx flow. For convenience these types will be
denoted as negative, positive and neutral respectively. In negative systems, the reflux
decreases in surface tension ; in positive systems
it increases. In both cases the relative volatilities
in the system have to be large enough to allow
the development of appreciable gradients. The
term neutral is given to those systems in which
either the components do not show a difference
in surface tension or in which the relative volatility is very low and the gradients in surface
tension are consequently always small.
The mixtures employed in the present study
are listed in Table I.
Table 1. SKITC.IJ of binary systems used
., „„,.
jt-heptanemethylff/cMiexaiie
H-heptanetoluene
bcnzcneH-heptane
benzenetoluene
benzenecycfobexane
(azeotrope)
ethanol2.2.4-trimethylpentane
(azeotrope)
Surface tension
Hailing point at boiling point
(dtjn/cm)
98-4
101-8
WS-4
110-7
SO-2
08-4
80-2
110-7
SO-2
80-8
77-5
78-3
99-1
71
12
15
12
18-5
21 ~ v ,^
-12 ^^
21 Y'
18-5--'
21
17-5
20
18
11-S
16
-"ide gap between the surface tensions of the pure
omponents of the system. This applies in a
lore general sense to all mixtures where relative
olatility is small. In that ease, even at total
cfiux, the concentration gradient in the reflux
:»ay be negligible. An example of such a system
* the widely used test mixture ?i-heptanelethylq/efohexane. Though the surface tensions
f the components (12 and 15 dyn/cm) differ
onsiderably, the low y. value (1-07) does not
How of the building up of noticeable surface
ension gradients in the reflux under normal
onditions of distillation.
From the above description it is seen that three
01
Most of the experiments were done with
the negative benzene—n-heptane, the positive
n-heptane~toluene and «-heptane-methylci/clohexane mixtures. Because of the low a value,
the latter mixture will, however, be taken as a
representative of a neutral system.
Two systems are of particular interest—viz.
the mixtures ethanol-2.2.4-trimethylpentane and
benzene-c#o/ohcxane. These mixtures both show
an azeotrope at about cquimolar compositions.
As a result the changes in surface tension are
different in sign, according to whether the mixture
££?. • ••- r^;^;?"v>;jO^<*-'^?^^^
;'•-*;•; H'v'V V':"
—
:
•
:o and A. UAK.MKNS
distilled has a low or a high content of the lowboiling constituent.
3. E Q U I P M E N T U S E D
The formation of interfacial area differs
according to the type of equipment used.
Roughly, two groups can be distinguished. In
the first, interfacial area is obtained by spreading
the liquid in thin films over area already present
in the equipment. In this case therefore, one can
speak of supported interfacial area. A prototype
of this group is the wetted wall column. Packed
columns and modified wetted wall columns such
as Yigreux columns also belong to this group.
In the second type of equipment, the interfacial area is obtained by mixing the liquid and
vapour phases. As a result, the formation of
vapour bubbles and droplets occurs ; the interfacial area in this case is not directly supported
by the equipment. Examples are all types of
tray columns which may operate either with
predominantly liquid film intcrfaeial area (foam
and bubble systems) or droplet interfaeial area.
Pure droplet interfaeial area is obtained in spray
Table
M
2.
Survey
of equipment
columns in which the liquid is dispersed by
mechanical atomizing device.
It is found that, with respect to the effect <
changes in surface tension, supported interfaci;
area behaves quite differently from unsupportt
interfacial area. Therefore, representative type
of equipment of both groups defined above wer
used in the present study. Particulars about th
equipment are listed in Table 2.
4. SURFACE TENSION EFFECTS WITH
SUPPORTED I N T E R F A C I A L AREA
(1)
Wetted wall type columns
A most striking example of the influence o
surface phenomena on the performance of db
tillation equipment is obtained when opcratiti:
a wetted wall column. IIETP data measured fo
the small concentric tube column at total rellu:
are given in Fig. 1. In the case of the u-hcptanemethyle#f/ohcxane system, the liquid phas
200
used
(1)
Concentric tube column
Length 200 mm, intcriuil diameter (i mm, width of
annular space 0-GO mm. Only the outer wall wetted,
inner wall dry.
(2) Vigreux columns
Varying length, diameter 2J mm, distance between
similar indentations 40 nun.
Packed column
(3) Length 250 mm, diameter -M» nun. I'arking : porcelain llaschig rings, 6 mm diameter.
(•!•) Packed column
Length 100 mm, diameter 25 mm. Pat-kin^: metal
Fenske helices, 3 nun, or JDixou gauze rings, a nun.
(5) Spray column
IS rotating discs, 40 nun diameter, spaced :;o mm
apart on axis. Column diameter 100 mm.
(G) Oldershaw sieve plate column [2]
Diameter 23 mm, plate spacing: modified to ion mm,
2 or 5 plates.
(7)
Perforated plate atlttintt
Diameter 4.>0 mm. If; per rt-nt open iirra plate,
10 mm hole .si/e. IMnte spacing -MHi mm.
6-2O
O4O 05O
Vapour velocity, m/r«c
FIG. 1. Separating pmver of euncentm* tube column
with different test mixtures.
^ «-hcptane-niethylf//rfohexane ,..-. '
.m~]>iK.'ii/.ene-H-heptane
-<•?.- -. -
shovvctl complete w e t t i n g of the column wall am
the separating p*»wor observed agreed very we!
with the theoretically prrdictcd values [1^]
The influence <>r surface phenomena on the performance of distillation columns
However, with the negative bcn/.cno-H-hcplane
system very bad separation was obtained, the
HETP being about ten times the theoretical
value. It was observed that in this ease the column
wall was poorly wetted ; the liquid flowed down
in narrow rivulets. Improvement eould be
obtained by fixing a wire coil of low pitch (about
•2 mm) along the column wall. Because of
capillary forces, the liquid was now retained and
spread out on the wall and nearly the same
HKTP was observed for the benzenes-heptane
mixture as for H-heptane-mothylcf/r/ohexane
(Fig. 1).
A similar difference in separating power
caused by a different degree in wetting is observed
with the Vigrcux columns. Total reflux data
were obtained for a variety of systems under a
wide range of load conditions. In the range of
vapour velocities of about 0-4 to 0-G in/see the
separating power was approximately constant.
The average HETP values observed in this
range are plotted in Fig. 2 against average
mixture composition. As shown, the HKTP
t
for the negative mixture are nearly twice
Cho •mailing
£ 4OO
T
|
3OO
oT
ui 200
I
100
"
O
2O
Mole /o
Wetting
-, ^ o
4O
V
6O
volatile component
in
1
8O
10O
mixture
FIG. 2. Separating power of "25 mm Vigreux column with
various test mixtures. Vapour velocity 0-4 0-tJ in/sec.
O
A
w-heptaiie-loluene c" ^->>
7i-hcptane-mctliyJcf/c£olifXJine
those observed for the H-hcptanc-metliylr#cfolicxane and the »-hcptanc-tolucne systems. The
wetting of the column wall observed visually was
in accordance with these results : iu the case of
bfiizene-»-heptanc partial wetting and the formation of rivulets, as illustrated iu Fig. 8,
occurred, whereas the other two mixtures showed
almost complete welting. The similar behaviour
of the latter two explains the almost identical
I1I"IT values for these mixtures.
The observation that particularly the benxcne»-hcptanc mixture gave poor wetting only seems
explicable by surface effects. Since there is no
major difference in the " static " surface tensions
of the different systems employed, it was believed
that the sign uf the changes in the reflux surface
tension might perhaps be responsible. In order
to test this supposition, some more experiments
were done with the Vigrcux column. First,
total reflux distillations were carried out with
the mixture bt-n/cnc q/r/ohcxane.
At high
benzene concentrations, tins constituent is the
less volatile because ot the azcotrope in the
system. Therefore, reflux surface tension increases downwards in the column ( lt positive "
mixture). Complete wetting was observed in this
case. AYilh a low ben/cnc content of the mixture,
(N/e/e/hexanc is the less volatile component and
the surface tension of the reilux decreases
("negative" system). As with benzene-Jt-hcptane
channelling of the reflux was now observed.
The separating power measured for the two
mixture compositions was in accordance with the
wotting characteristics found (see Fig. 2).
In a second series of experiments it was attempted to measure mass transfer rates with the
normally " positive " systems Ji-heptanc-toluene
and fi-heptane-methyln/dohexane under " negative " conditions. This cannot be realized in
normal distillation ; however, by conntcrcurrent
absorption of vapour rich in volatile component
into a liquid rich in the heavy component the
direction of mass transfer may be reversed and
thus also the sign of the surface tension changes
in the downllowing liquid. Moreover, by changing
the length of the Vigrcux column or by adjusting
the compositions of entering liquid and vapour,
the surface tension gradient in the liquid can
easily be varied and its influence determined.
Data on the absorption experiments with
Vigreux columns of various lengths arc collected
in Table 3. Besides top and bottom flow rates
and composition, the average driving force,
i.e. the average difference, between the operating
^£i&
SSSSKSfc
*':3^f'
1^
F. J. ZUIDERWKG and A. HARMON'S
line and equilibrium line is given (in terms of
liquid composition). The latter is representative
for the average changes in concentration (and
hence for changes in surface tension) in the reflux
over one theoretical tray. At a high driving
force, mass transfer is rapid and a large gradient
in the reflux surface tension results. At low
driving forces, the reverse is true.
As expected, under alt conditions where
reHux surface tension decreased in the absorption
experiments, channelling of the reflux occurred.
This became more severe the higher the driving
force applied. Consequently, the HETP also
increased with the driving force. This is clearly
illustrated in Fig. -t, in which the HETP data are
plotted. It is seen that for both systems the
data fall on one single line. Furthermore, the
HETP value obtained in normal distillation with
the " channelling " benzene—n-heptane mixture
correlates well with this line if plotted at the
average driving force applied in the distillation
experiments.
Fig. 4 also shows the data obtained under fully
wetting conditions with various mixtures. The
value for the n-heptaiie-mcthylrz/ciohexanc system at high driving force was obtained by the
Table 3.
o wm/i
ten^itt
(in \
T-i^a^jrii-aK;
^-^1
n-heptiine-toluene
,,
fj
J}
„
n-heptanemethylq/nff/hexune
„
1-4S
1-00
0-88
0-65
1-78
SJ
J(
o-ss
,,
065
048
0-48
»
3§
S
m
1-78
1-4S
1-OG
tl
ouu
U-Chonneirn
""* A
3
---"*''
_-^£"
a.
U
2OO
X"
w •tfcr-g
0-10
0-2O
O-30
0-40
Driving force, mote fraction
O-50
FIG. 4. Effect of driving force on separating power o
Vigreiix column, tested with various test mixture*
Vapour velocity 0-3-0-4 rn/see.
£3 benzt-ne-H -heptane
A A w-heptnne-methylt'i/cfc/hexiine
O • w-ht-ptane-toluene
Absorption experiments in Vtgreux columns. Average vapour velocities 0-3 m/sec.
Concentrations in mole per cent n-keptane
jVy.v/fw*
ffis
absorption of methylq/r/yhcxane into n-heptm
(see Table 3). It is seen that under coiulitiot
of positive changes in surface tension there is r
influence of the driving force oh the HETI
presumably because the wetting is almo
complete. Further, it may be concluded froFig. 4 that at low driving forces, the HETP t
negative systems seems to approach that of tl
positive mixtures, i.e. wetting becomes
and channelling disappears.
Feed rates
(mnlc/hr)
top
Product rates Compositions, (mole per cent) Average
(turtle /hr)
Feeds
Products
driving Tttforetictii
,
fore? j plntt.v
buttvni. tup bottom top botfnm
top bottom trnole %) i
lti-0
10-0
22-8
0
10-1
23-2
22-3
20-5
20-8
0
17-G
1G-G
18-1
15*5
1G-G
1G-7
17-4
21-9
20-9
16-5
21-1
15»
16-3
1D-7
1G-G
15-1
1G-9
15-7
1G-3
18-1
IG-8
17-S
15-5
11-9
12-3
14-7
11-S
12-1
11-G
14-G
13-0
14-1
16-2
•J5-G
25-4
20-0
23-4
18-fi
IS 5
lit G
~M~~
4-9
4-3
•MO
;uo
2-;i
4GO
420
500
5i*
Gl
145
5-1
3S
€3
57 -5
54
2G-5
27-5
44
58
0
0
14
14-5
30
34-5
44
o
100
0
100
3-0
8-5
o
o
0
0
100
100
100
100
100
0
23-5
23
43
55
71
(mm)
1G-5
17
25-5
2JS
37
100
100
100
KJO
100
0
HETP
GO-5
GO
52-5
19
2-1
1-3
2-3
4G
55
2-1
1-3
ll-S
as
33
2-0
.150
3!»0
4GO
420
500
cuo
240
•5»«
•£z$is«
•\\$m8%
' .'i/'*A& -^-v
;.r-?;Ate
$»•'•*•»*
.f*.-'.V*itfrS*
The influence of surface phenomena tin the performance of di>t illation columns
There is one remarkable feature in the eorrclaiuii of the HETP data in Fig. 4 which should be
icntioncd. The difference in surface tension
•ctwcen n-hcptane and methylr^cfohcxanc is less
han half the difference in surface tension beween w-heptane and toluene or benzene. This
:ieans that since for all mixtures the range of
iqukl concentration under channelling conditions
-; about tlie same, the surface tension gradient in
he reflux during the u-heptane-aromntics tests
>/as about double the gradient in the «-hcplanencthyltt/cfohexanc experiments. In spite of this
act the intensity of channelling in the formation
>f interfacial area seems to be the same.
The observations of channelling liquid phases
cported in literature are also to be interpreted
n the basis of decreasing surface tensions.
IOLSTAD and PAH SLY [8] found that the contracion of water films in grid type tower packings
ccurred when ethanol vapour was absorbed. In
his case surface tensions of the liquid film decrease
>ecausc of the increasing ethanol concentration.
BOND and DOXALD [1] observed the break-up
f water films when hydrogen chloride or ammonia
••as absorbed in a rippling film in a wetted wall
olumn. Owing to the heat of absorption, a
.ecrease in surface tension also develops in these
ases. It is interesting that the break-up did not
ccur when rippling of the liquid film was
trevented.
2)
Packed columns
tanc-toluenc mixtures are about the same,
whereas for the negative benzene—n-hcptane
mixture about the double HETP values are
recorded.
JOU
i
-
a
•'• i
oT
0
<*
!
•
X
O
20
4O
Mole °/o volatile
6O
component
SO
in
1OO
mixture
Kic. 5. Separating power of 6 mm Raschig rings, with
various test mixtures. Vapour velocity 0-2-0-3 m/sec.
£3 benzciie-H -heptane
A ?^-heptane-toIuene
O M-heptane-methylci/efohexane
A similar difference between HETP values
observed with positive and negative systems was
found for the fine metal packings such as Fenske
helices and Dixon gauze rings. Table 4 shows the
data obtained :
Table 4. HETP data for fine metal packings*
Average liquid composition 65 mole per cent volatile
component
Packing
In packed columns the interfacial area is sup>orted by randomly distributed packing elements.
u view of the results obtained with the wetted
/ail columns, the effect of the direction of surface
ension changes on the HKTP was also determined for this type of columns. Two types of
lacking were investigated in relatively small
olumns : 6 mm porcelain Raschig rings (in a
0 mm diameter column) and two fine metal
Backings (in a 25 mm diameter column). In the
ase of the Raschig rings, efficiency varied but
ittle with the boil-up rate at vapour velocities
>f O-2 to 0-3 m/sec. HETP data measured under
hcse conditions are plotted in Fig. 5. It is seen
hat just as with Yigreux columns, the HKTP
or the rt-hcptanc-methylfj/t'/whcxane and ii-licp-
•—
Fenske helices
Dixon rings
Si/stcai '
n -heptanetoluene
benzeneH-hcptane
M -heptanetoluene
benzeneTi-heptane
Vapour velocity HETP
(m/sec)
(nun)
0-OG
15
0-06
30
0-12
20
0-15
35
Though in the cases of the packed columns the
actual flow of liquid over the packing elements
was dillicult to observe, it may be safely assumed
that the differences in separating power with the
different systems must again be ascribed entirely
to a different decree in wetting.
1*5
• *sS' "<sXri'-<fi
$,!%&$¥&
•&iv$tett
tefef3
->
C
P
O
SI
9 3*
3
o
C*
3< r*. 3 X
cr
C C if.'
p
£. O
£. "1
o fr 3
3"
r; rt" s C
3 O 35 r;
V)
O
c
p^
(T
[y,
O
•^
O
"J
c
V)
^
r?
rt~*C"
S
O
sf
P
>C
T3
P
S P oo
c/:
rt-
3
C
in
£
£
X
•—•
"
MM
-i
P
T
3 rf
a
^
O
JL
O
c- -
-> -i'vwa
v^^?i
-C^f
•,-.•: .^tr
O
£
2
P
rt
r:
Ji
O
3
L
C
S
„
35
£,
""' C
=
""
/
o 3
2
m n>
£. 0
*i
A-
w ja
—.
3d *•*
X. s
a ^ s ?5 3
§ &c3 o
T "-
c
= ' 0 £ r t .
1
S
-: p
tgemole
o
c
_
yi
^
~ cr
y, '
3
S
ft>
3
c 2
3 *<
-;
Co
O
P o 5"
s.
7rt- ,
3
3
o ~
X
O
• ?:';?h5>;
ro
O
-.-
^ rt-
^ 2 2
Number
of theoretical trays /spray
o O O O o
=
^P
~J
(T
O
X
£
"
'
ji«• c
!•* . . ci-
<ji
<—•
7 P
—i
1
I
r M
K-|
o "^
P L
>-.
o
M
o"
£
2.•- f
•—. 0
o >—•
P
H^
*^
3
O
O
CT1
*"
H
"^
»^
ta-I
•'"pi 'fl.fll
seem
.'•'*$£$
(:;>v4.$"
^^|i
C »>C
Ej p"
^
«-»•
J^ _
3" rf
™
""
^*
-«J
^
5"T
~ * c 3 _; 7?
3
i-.
"
^>
T 75
P|.
1—1 ,-».
C
i_ p"
£ ^-
i—
v"
P
"• rt-
0*
r-t-
C
3^ ° (?
3/ o
'J3" 2 •"
N
i*
V)
x y ? 1 "H.y; tr rt"n '< 2
C
>V-''rt<
i!ii
•*"
1' £.
^" ~ ?* ••£ "
Q
?•
* 3
fj"X
3
C
3. £
X
£j
C
3
""
^;-ra
il, w
*"
o-- B
O
rt ^
C"
2 "?
•f7
5-
rt-
75
i3 ?*
£, P"
r^-
71
2
rt-
2- %"
2
^ ^~'
C
y;
3" p- 3' p
HT. v;
o
o 1;
2 ?
< ~"
2 ?
PM
0
P
c"
:'-
~ i^T"
J*5
^
0-.
P
"^
•"
3^ (•*•
r> r- ^' r* n-i
—
—
S ^
C
rt
—
X A- i
C n
f? 5
^ i
2 2 t3* ET o
3
f? ra •-• re
T
p
i
•5 "3
5"
2
e -i
magi
* S
<•*
? o
-.
-r,-
Hi
n>
*1 -«
ti ti
vp vp
0s* 6s-
!Ji
C
vP
^«
ssi
s?^|
;
\
?" s
^
3
5 >$.
"* o^>
_a
s-o
5*
f?
/
'
£
^
'/I
r-t-
«-S
"TJ
o —'
1 1 1 ai
2 "&P 2 *~ p
i_.
°
1
m
«.]
rr: t^-j *
--i
^
C-
£J
—
^
_(_I_i_
— /— •
1 i 1' $
i 1 1f1
If 3
S
»"
';
•
S
oo
y. Q >-<
* o
4
^O
-8
16
o
-*•
i=HoEr
** tl^«
i
i
< 'Jn
X
3 tn
rt- '•*•;
v.
»' J
2- 3?
o S ^ g g R
^ o 1*
(7
g
Awwuge iray CTTiCiency./o
*l
-»'sr ^'PI-
N
^ ?
^ 5
rr 2.
§
3. »*
S
3 dc
=: ^ |
N
"
C
_o
5 2
's FS c
w
iJ C #
'•.'--V^v
• t i-~ .-*ii
—.
g
i
«f
i
'
r
j
" 0
£_
«'
?" rt- ?D"
••5 —* CT"
p >-•
O^
r?"
I? ^ a
S n 3-
^
"
*~~*
f-
rt-
g-jo o o g- d. o
^3-
^.^•mws^fr
r--"v• s-'wsaSa eg&~5s
!MK&t?3M*3t5r'aa*-&3eStib,
S^^ »?^^» w^^^^A^i-sa -liel
*^a-»? S J».^JC±i_-w-«M.-wsKs^5C?rK^,>isi
ssKr^Sff^ r^KBagJS^^s^^a^
54.2^; r.'*^,-.-»-?,
^^?c%-<r;
Hi ii*3-&£3
Fin. H. Tray art ion mi OMersliaw sieve plates witli
ben/.cne-H-hcptanu (left) and »-hcptanc-toluciMr system
respectiwly. Vapour velocity «1»i»iit »•« MI'sec.
;•;,-.: .v*y:S
-'"•' .;'-'-• .^/P"
' I'A
fr'i'.'bl
-.ir^'^
v'-.V:;. •',•«!
!-^-^
- '• j'' l*.^-'^
•£$?&$
;CS^-^|
iWTO'PW^JW^^
Table 5.
Syxtrm
...h<MiliHu?-(fiiptliy1<v/r/<>lioxiiiH"
Efficiency
of Olderskay: sieve plates
.•hYT«!*r mofr %
rultitilt' runtp.
50
I'apanr velocity
(in/sec)
0-32
o-ao
O-4S
•eiizene-M-licrptane
50
85
40
(-heptane-toluene
.15
40
87
92
0-15
0-30
0-48
0-13
0-30
O-47
0-35
0-29
0-30
•tIianol-2,2.4-triinethylpentjme
7
Foam Height
(itim)
51.
10
54
54
52
20
o-o.s
O-21
O-32
0-40
0-40
0.52
0-30
0-.1S
0-47
0-55
0-GS
0-14
0-23
0-30
0-33
0-50
0-G2
0-13
0-2S
Q-44
0-G7
0-13
0-21
0-2»
75
efficiency
' /of
57
0-10
°"f
tenxene- toluene
Trtitj
0-48
0-20
0-25
53
51
40
4S
40
50
50
43
4G
4ft
55
53
55
•
55
53
52
83
S3
85
78
97
100
100
88
00
10
20
10
20
10
20
40
fiO
40
01
73
80
75
05
G3
G3
64
30-40
7O
20
15
5-1O
50-100
>y carrying out absorption experiments. The
tzeotropic mixture cthanol-2.2.4-trimcthylpcnane was used. At cthanol concentration** below
33 mole per cent, this component is the most
volatile. Since the surface tension of -J.2.4—
" negatively *" i?» this region of concentration.
At ethnnol concentrations above 53 mole per cent,
the octane is the most volatile component and
the mixture shows " positive " properties. This
conclusion is confirmed in a spectacular manner
rimcthylpentanc is lowest, f h c system behaves
by the distillation tests. Fi«r. 0 shows
ife;- - - .^^^fS^r'^^K^Sr^S^-^
F. J. ZvuirctwEG and ;\
^^sf^-4
ig3?&S3
a»
of the liquid-vapour behaviour for low and high
ethanol concentrations at about the same vapour
velocities. The tray efficiencies (30—H) per cent
and about 70 per cent respectively) arc in accordance with the formation of mtcrfuciul area
observed.
Absorption experiments with the Oldcrshaw
sieve plates were carried out with //-heptaneniethylczfc/ohexane and ^-heptane-toluene mixtures under conditions of increasing surface
tension changes in the reflux flow. Table G gives
tbe collected data. Tray efficiencies and visually
observed foam heights of tbe absorption and total
reflux distillation experiments arc plotted against
liquid driving force in Figs. 10 and 11, which are
hence similar to the 1IETP graph given in Fig. 4
for the Vigreux column.
A distinct similarity is noted between the
dependence of tray efficiency and foam heights
on the average driving forces applied. The degree
of foaming and increase in tray efficiency with
the two mixtures, however, is clearly different and
in line with the surface tension difference's in
both systems. This is in contrast to the Vigreux
column, in which—with decreasing surface tensions and using a channelling mixture—no such
influences of the system could be detected.
Figs. 10 and 11 show that at low driving force,
the foaming effect seems to disappear and the
50-55 per cent tray efficiency is approached.
In normal distillation practice these low driving
forces are obtained at low or high concentrations
or under " pinched *' conditions. In these circumstances, a loss in fray elliciency is therefore
^
u
5 eo
"D
w 60
Foam
1 *
O^j-
/f^^
&.'
«
i
Spray
• a
-a
,
?40
$20
(_
V .
3
&
O-1O
Driving force,
O20
mole fraction
l''io. 10. Inlluence of driving force on tray efficiency
Oldersliavv sieve plates for various test mixtures. Vajto
velocity about 0-a m/see.
benzenes-heptane
benzene-toluene
»-heptane-toIuene
»- heptane -met hylci/cfohexane
Open points : normal distillation
Closed points : absorption experiments
D
V
o-io
Driving force,
O-2O
mrfe fraction
•Kic. 11. Influence of driving force on foam height on
Oldershaw sieve plates with increasing surface tensions
of reflux.
O • w-heptaiie-toluene
A A w-neptune-methyhv/f/ohexaiie
Open points ; normal distillation
Closed points : absorption experiments
Table G. Absorption experiments icilft Oldrrshctw sieve tray column. Column zcith 5 plates, average
vapour velocity 0-3 m/sec. Concentrutiutts in mole per cent n-heptune
Product rates ComfKtsitinmt. (mole per cent) Average
driving Theoretical
Pfoditclx
JPeeeLt
(intile/hr)
force
plates
bottom ton buttani (mole %)
bottnnt tup j bottom fo/i
Feed rates
(tnole/hr)
System
top
-V!3s
M
n-heptanemethvlq/ftohexane
l.'I-S
I7-S
14-y
1S-2
1H-1
1«-1
].»!.
w-heptune-tolucne
I«-H
14-1 1 17-.1
I
WO
lit. ; i7o
7S-3
10- 1 | 17-5
!;>•» f :-'!•!
5S
KM)
Plate
efficiency
\
Foam
Itt-ififit
(turn)
84
;j."i
4-t
SS
40
:j-7
71
;'.<)
4-:s
Sfi
50
0
78-5
U-j
L'{)
4-2
*-!»
'_'«
0
07-0
5I-.1
11 1-0
2«-0
;!!•()
U
7
il-0
!iO
Tray efficiency maxima at mid-concentrations
as found by TJIUSSEM and by the present authors
for hydrocarbon systems have also been observed
for ethanol-water mixtures [4, 5], ttecuusc of the
low surface tension and high volatility of ethanol
this system should also be identified as positive.
Further, some interesting cases are reported
for the distillation of nitrogen-oxygen mixtures.
Nitrogen is the most volatile component and has
a low surface tension at the boiling point (about
9 dyn/cm), whereas the heavy component oxygen
has a high surface tension of about 13-5 dyn/cm.
Accordingly, GKASSMANX and FUANK [3] found
that when transferring oxygen from vapour
bubbles into oxygen-nitrogen liquid mixtures,
higher transfer rates were observed than for the
opposite case in which nitrogen vapour was
transferred to the liquid. From the surface
tensions of the pure compounds it is clear that
in the first case positive surface tension changes
develop in the liquid, which changes stabilize
the bubble formation and hence favour mass
transfer rates.
The positive characteristics of the nitrogenoxygen system can also be derived from the
distillation experiments with perforated plates
earned out by Lixmz [6]. In the middle concentration range foaming was observed, whereas at
extreme mixture compositions only a low spray
bed developed. The same was found for the system
nitrogen-argon, but foaming was absent with the
mixture argon-oxygen. These results are in
accordance with the surface tensions and relative
volatilities in the respective s}*stems (Nitrogenargon : boiling point difference 10°C, surface tensions 9 and 12-5 dyn/cm respectively; argon-oxygen : boiling point difference about 1°C, surface
tensions 12-5 and 13-5 dyn/cm respectively).
The effects reported above with perforated
plate columns in general pertain to plates with
very small open area and small perforations,
conditions which are favourable for foam formation. In technical columns, however, a higher
free area and large perforations are used. Therefore, and also because of higher vapour velocities,
a relatively larger amount of spray is formed and
foaming diminishes. The results obtained with the
sprny column give rise to the expectation that
o be expected when a positive mixture is distilled,
i'or total reflux distillation this is shown explicitly
-, Fig. 12.
The concentration effect on plate cOU-ieney as
iemonstratcd in Fig. I'-i has also been reported
:a the literature ; in all eases in which noticeable
t.
%
O
2O
Mote
fa
4O
6O
volatile component in
SO
1OO
mixture
FIG. 12. Effect of concentration on tray efficiency of
Oltlershaw sieve plates in distillation with n-heptanetoluene. Averaj^e vapour velocity 0-3 in/sec.
effects were recorded the binary system used can
he identified as positive. Fig. 13 shows data of
TUIJSSKN [10, 11] for the distillation of mcthylryefohexane—toluene and n-heptane-methylf#Hohexane in a small perforated plate column (:J8 mm
_o TOO
SO
i
f~—
i— — —
-^ Jiylcyclctw ane— tolufrr
k-^
A
h*ptane-rt rthylcyelc* exane —i-
f 60
'O
fc *>
1
i-
20
0
2O
4O
6O
3O
TOO
Mole /o volatile component in mixture
FIG, 33, Concentration effect on tray efficiency of
perforated plates observed by THIJSSKN. Average vapour
velocity about 0-2 ni/see.
diameter). The behaviour of the first system was
to be expected because of its high driving forces
and large difference in surface tension
(3-5 dyn/cm). It is of interest to note that also
the 7t-hcptane-methylci/d0hexane mixture, in
spite of its low relative volatility, shows the
concentration effects under extreme compositions.
This mixture should therefore be considered as
slightly positive instead of neutral.
99
F. .T. ZrinF.invEG ami A.
n 100(
£ 60
'y
£ 60
20
O4
O-8
Vapour velocity.
1-2
m/sec
1-6
Fro. 14. Tray efficiency of a perforated tray. 1C, per cent
open area, 10 mm holes.
* n-heptane-toluene, about 50%
^ ?i-hept:nie-niethylry/f/ohexane, about 50%
V benzene-toluene, about 50%
there would therefore be less difference between
positive and negative mixtures with respect to
tray efficiency. This is indeed the case and can be
seen from Fig. 14 in which tray efficiencies
observed for a perforated plate with 16 per cent
open area, 10 mm hole size, (diameter 0-45 m) arc
plotted for three different systems.
G. D I S C U S S I O N *
The experiments reported in the precedingparagraphs have shown that, depending on the
type of equipment used, decreasing or increasing
surface tensions of the reflux may influence
interfacial area and • separating power quite
markedly. A survey of the main trends observed
is given in Table 7. For convenience, the properties pertaining to systems in which no surface
tension changes occur (neutral systems) have
been denoted in this table as ** u n i t v "*.
In order to give an explanation of these pi,
nouicna, it should be recalled that liquid surfac
of high interfacial tension contract when contactwith a surface of lower surface tension. Tl.
is known as the 4t Maragnoni effect ". MARA*
[7], however, did not consider the influem
of vapour -liquid interaction in his studies.
A very well-known phenomenon in whic
liquid film contraction occurs as ,1 result of matransfcr effects is the so-called " wine-gin^
effect ". Wine creeps up the wall of the glabccause of its wetting tendency. When a filr
has been formed, alcohol preferentially evaporates
leaving a residue of higher water content an<
therefore higher surface tension. When fresf
wine creeps up the wall, and comes into contac
with the aqueous film, the latter contracts inti
small rivulets.
A similar explanation can be given for tin
channelling phenomenon taking place whet
distilling a negative mixture in a wetted wal
type of column. Liquid films descending in sucl.
a column are never of uniform thickness, and
hence, locally, thin places will become more
rapidly saturated with heavy component than the
remainder of the film. Since the heavy component
has the lowest surface tension, areas of lower
surface tension develop locally; the remaining
parts with higher surface tensions therefore
contract and the liquid film breaks up into
rivulets.*
*K«i*entially the same explanation lias been given by
U<>\*[» and DON.VI.IJ [t ] for their fibservatiotw of the brt-;ikdown of water (Urns in the c-:i-;e of ammonia absorption.
Surrey of surface phrnvmena in distillation equipment
Supported interfacial area
SSI
Surface tension
changes
no changes
(neutral)
increasing
(positive)
tit-creasing
'
Wetting
Separating
ptwer*
complete
\
Unsupported iaterfitciitl area
nf
driving force
Tt//tf of
intfrfacial art'it
I
none
spray
1
complete
^I
none
foam
>1
.severe
partial
< I
severe
spray
~1
none
(negative)
*Srparating power denotn! u n i t y fur " no
100
Separating
poicer*
Effect of
driving force
none
The influentr nf surface
hH mi OH tin1 prrroriiiimn-
gives a graphical representation of the relation
between the ratio of IIETlrs with channelling
and complete wetting and the UETP of the
columns with complete wetting. Furthermore,
the same HETP ratios observed for the packed
columns arc plotted in this graph.
5).
It is easily understood that the differences of
It is thus seen that at the same HETP, the
.nicentration in the liquid surfaces will become packed columns suffer a smaller loss in cllicieney
.iqhcr and contraction will become more severe than the wetted wall type of columns. This is
f the mass transfer per unit area is higher. easily explained by the fact that in the packed
Hie latter is produced either by an increased columns liquid is retained by capillary forces in
interstices between the packing elements. The
contraction of the liquid surface in this case
of liquid film into rivulets
will be less effective and the loss in intcrfacial
area will therefore be smaller. Reduction of the
channelling effect in this sense has also been
found when providing the concentric tube column
with a thin wire spiral.
The distillation of " positive " mixtures in the
Vigrcux column and the Rasehig rings column
Stabilisation of rivulet
did not reveal any significant differences with
Fio. l~t. Schematic illustration of break-up of a liquid
:ilm in a " negative *' system. Shaded areas denote respect to the distillation of a. " neutral " mixture
in the same equipment. Considering for this
liquid of lower surface tension.
case again the non-uniformities in the liquid
driving force or because of higher mass transfer films, it is now seen that thin and weak spots
coefficients (i.e. smaller HETP values). For the become preferentially saturated with the comYigreux column, the driving force effect has ponent with the highest surface tension. The
already been illustrated in Fig. 4. The influence weak spots are thus reinforced, -and as a. result
of the HETP on the severity of the channelling the film tends to become of more uniform thickness
phenomenon is recognized when comparing the and break-up will be prevented (see the sketch
results from the small concentric tube column in Fig. 17). Wetting of the supporting solid
(Fig. 1) and the Vigrcux column (Fig. 2). Fig. 16 surface with a film of the positive mixture will
therefore be approximately the same as with
o-s
Once the rivulets arc formed, the edges, because
f their smaller volume, will ahvays be more
(turated with the heavy component. Surface
•nsion is thus lower at the edges and this stabilizes
lie contraction of the liquid (see sketch in Fig.
Cor>cerrtr.c tube * wire coil
_6mm raschig rings, _
0-6
Vigr-eux,
Frne wir« packing
KIG. 17. Schematic illustration of the stabilization of a
liquid film in a "positive" system. Shaded areas denote
liquid of lower surface tension.
0-4
0-2
'ccoceitnc tutje
O
100
HETP,
20O
wetting system
FIG. 10. Ht-Litiou In-tween loss in separating power and
HETP when distilling benzene- » -heptane in various
types of equipment.
mixtures in which surface tension gradients are
absent.
The latter conclusion, however, only holds
good for supported interfacial area. In a system
where the vapour is dispersed into a liquid, the
lifetime of the bubbles formed may increase
considerably because of the reinforcement of
101
F. .1. XriiiKttvvKr; ami A. HAKMKNS
ara
weak spots. These weak spots occur between
adjacent bubbles (see Fig. IS); the preferential
saturation of these spots with heavy component
interfacial area in vapour liquid contact I
equipment.
The main conclusion is tint changes in surfc
tension occurring in the liquid phase can havi
spectacular effect on the interfacial area. Brea
up of liquid films results from decreasing surfa
tensions and is of particular importance i
wetted wall columns and packed columi
Stabilization of liquid films occurs when t!
interfacial tension increases ; even foaming m;
develop in this case. The latter effect is of pa
ticular interest for the operation of plate column
Since mass transfer rates are directly proportion.
Fir.. 18. Schematic illustration of the stabilization of to interfacial area, the surface tension effec
vapour bubbles in a ** positive " system. Shaded areas are reflected in the JIETP values of the varioi
types of equipment.
denote liquid of lower surface tension,
The above phenomena become especially notict
causes the surface tension to rise locally. There- able when the differences in surface tensio
fore, liquid is drawn in between the bubbles and between the components of the mixtures at th
coalescence of the adjacent bubbles is prevented. boiling point are higher than about 2 dyn/ci
If the changes in surface tension are large enough, and at driving forces above 5 mole per cent.
even a foam may develop in this manner.
A further conclusion is that equipment witi
Since the degree of bubble stabilization will high mass transfer efficiencies is relatively mor
depend on the change in surface tension of the sensitive to the surface effects than less efficien
liquid on the trays, the relation of foam height apparatus. From a practical point of view thi
and tray efficiency to driving force as shown in means that the effects are of particular importance
Figs. 10 and 11 seems quite logical.
for laboratory apparatus, whilst technical equip
Finally, the behaviour of neutral systems and mcnt will be much less sensitive to the surfaci
negative mixtures in a perforated plate column • phenomena described.
will be briefly considered. For hydrocarbon
Finally, it is of interest to note that tin
mixtures in which no appreciable changes in influence of the changes in surface tension on
surface tension occur, the stabilising factor mass transfer in general is quite severe and
tending to produce prolonged bubble life is appreciably more pronounced than the effect ol
absent. Accordingly, bubbling is poor with such variations in the static properties of the liquid
systems and spray formation occurs preferentially. phase, such as density, viscosity and diffusivity.
From what has been said about the instability TIIIJ>SKN- [10, 11], also MOI.STAU [8] ct al.t have
of liquid films when distilling a negative mixture, in fact argued that variations in these properties
it will be clear that in this case the bubble stabi- could not account for their observations. In this
lizing factor is also absent. Approximately, the connexion it may be pointed out that when
same interfacial area can therefore be expected studies are made in which the static properties
for the two types of systems ; accordingly about of the phases are the main subject, (e.g. when
equal tray efficiencies are observed for the the contribution of the vapour and liquid phases
" neutral *' and negative systems.
to the resistance against mass transfer is assessed)
care must be taken to select experimental con7. CoNCJ.r.sioxs
ditions such that the surface effects do not
The experimental facts observed and the quali- affect the observations. From the results of the
tative interpretation provided give rise to some present study it mav be concluded that such
definite conclusions regarding the formation of conditions cannot easily be realized.
lOi!
jttv.vv.--T •>•? *.«r*W.s yK*,tu.»u..»u. «« * KM j.«Tform:iiuT' of distillation <-<tluinns
T? K1' K K K X f K S
[1] BoM> J. and DONALD M. 15. C'Awii. £»£»£. AYf. 1057 6 237.
[2] COLLINS F. C. and LANTZ V. Jmlusttr. JKngHg. Chem. (.-ttnil.) 1956 IS <>7:i.
|:S] GIIASPMANN P. and FIEANK A. Interne Arbt'itfixitznng (Ms f\«7m«s.sT/tH-sscs Deslilffifwn,
tier V.DJ. Fttchgnippe rerftihrenstfchntk. Gottinyt-n 1JI57.
[4]
Kcktijifffttirtn
witl Extraction
KEYES D. B. and BVMAN L. LT»ic. Jit. JSngng. Erp. Sta. Butt. Scr. »28 7041.
[5] KiRscisitAVM E. Cltcm.-lng..Tcch. 1951 23 213.
[6] LINIIK G. Chem.-l»g. Tech. 1955 27 GG1.
(7] MABACNOXI C. Ann. Pays. Lpz. 1871 143 337.
[S] MOUSTAD M. C. and PAHSLY L. F. Client. Engng. Progr. l!»5t> 46 20.
[9] SicwAan K. Z. IVr. jD/wft. /?/5. 1956 98 453.
10] THIJSSKX H. A. C. Effect of Liquid Conipttsition »n I'lttie Efficiency
JLandbouwk. Onderzoek, The Hague Xo. 61.12. 1955.
in the Kectiftcution of Ithittrif J//.r/nrcs. Vcrsl.
11] WUK W. R. VAN and TIIIJSSEN H. A. (\ Engng. Sci. l»5i 3 153.
12] WILLINGICAM C. H., SEDIJIK V. A., ROSSINI F. D. and WKSTIIAVKK J. /m/rwfr. Engng. Client. 1947 39 706.
13] ZOIDEUWEG F. J. Cltein.-Ing. Tech. 1950 23 587.