'· I l

t
wff0)~~o®ou~Qo®tm
\_)
i!t

.
. . . Destgnea ' ana! co~pl'l e d ror
( use .m
Fradionation and t\bsorp·cion operations ..•

lri Refineries, Natural- Gasoline Plan:Cs,

• ., ' 1 Dl
A~ n d Penocnernica, ~
,. ,ams

Included are
•• > •- . . . . : .. ->.:' .~·
Sec-Nons on ...
lll Instrumentation of
., f '.' ~-.:·..
Fract,ionating Columns
./ ··.
Gl Several new concepts'
of tray design for
Fractionators and
Absorbers

<l> Material on design

and operation of
Absorbers ·

~~ :
.
·.:·.
.··
. ",';'
.
iii·
. f:

' .
·'. A TECHNICA\, MANUAL
Reprinted from

. il•t© u J\ NO G/}'5
wQ_.,
'• I.

. 1~ .:
' ... ~D ® QlJ CJ CJ
........
·i
 .I'
i'
!

A FOREWORD '
I
I
New Concepts in Fradionation and I
('

Absorption for ·rhe Process N\an

'"!
•.. Articles reprinted from The Oil and Gas Journal

o There are many articles and reams and reams of data

avai.lable on instrumentation of fractionating columns. A
great deal of this information has been carefully collected
and intelligently interpreted. However, there has not until
this time been available an interpretive compi.lation of avail-
able information, with the pros and cons of the various
instrumentation methods.
The first section in this manual, Instrumentation of
Fractionation Columns, serves this purpose. While it is not .
all inclusive, it nevertheless does present a complete picture
,.---·'\
\..__ ) of the problem.
o A number of new designs for fractionating trays have
appeared recently, and advances have been made in design
and construction of bubble-cap trays. The second section of
the manual, New Concepts of Tray Design for Fractionators
and Absorbers, is devoted to these subjects.
This compi.lation affords the basis for easy comparison
and summnries of the advantages of each type of trny.
o New da Ia and techniques are 9;vailable for the design
and operation of absorbers. This new information has been
evaluated and is interpreted for Journal readers in the third
and lust section of the man1.1~l, What's New in the Design I
and Operation of Absorbers. i
An outline of the best calculation procedures, illustrated '

with sample calculations, is presented in detail in this

section.
o Here it is-a handy reference manual designed to
bring you up to date and keep you abreast of developments
in the fractionation and absorption fields.

Copydght 1954
The Oil and Go> Journal and The Petroleum Publishing Company
Tvlsa, Oi.lohoma
 .•'

. •-:,:": ., ·.•
... '.,

'
~.._..,
'

INSTRUMENTATION OF FRACTIONATING COLUMNS 2

By John C. Reidel

FORUM ON FRACTIONATING-ABSORPTION TRAY DESIGN ..... 18

l. Development of Bubble Caps ....................... 18
. by R. M. Young
2. Use of Perforated Plates ............. .: ..... .' .· ... ·... 21
by D. C. Lee
3. Recycling Tray in Absorption Columns ....... , ......... 21,
I by E. G. Ragatz
~)
4. Characteristics of Shell Turbogrid Troys,: ...... · ....... 27
by J. A. Samaniengo
5. Efficiency of Float-Valve Bubble Trays .... ·... , .......·. 31
by I. E. Nutter
6. Socony Uniflux Tray Utilizes "S" Member ............. 36
by V. 0. roowl<:s
7. Gulf's Perforated Plate Operating in 35 Towers ........ 38
by A. C. Eld.
8. Flexitray Has Wide Operatin(l Range ........... 41
by G. C. Thrift

HOW TO DESIGN AND EVALUATE ABSORBERS ............... 44

:
 HlACT!ONATOii CO~HROL
§,9 ~>--- ·· ... -'

• •
•
. tn r • q ~
I~JeiCJ sD rocessmg
• Df '
r anrs

1. Control of Fractionating Columns. the subject of fractionator control could be a contro-

versial one.
However, in most instances, a given method is
~ Feed rate. app:icable only under specific conditions. In the final
. 0 Removing products. ru1alysis, the choice of automatic control for a panicul!lr
column depends on the job the column must do and the
o How about heat supply? design of the column and its auxiliaries.
ln this light many basic questions then arise both as
· 0 Finding the flood. point, to the process and the instruments: \Vhat to control?
•. , and how close a control? ... if temperature is the
o Pressure control. , key control, where should the temperature measuring
element be located? ... what method of pressure control
2. Gasoline Stabilizer Control. should be used? . , • is a proportional controller good
enough? . , , or should a proportional·plus·reset be used?
3. A Depropanizer-Debutanizer System. , • , what proportional band should be used? , , . what
about the conrrol valves?
4. Natural-Gasoline-Fractionation Section There is no intention of going into the subject of
control valves or controller selection ar this time.
However, in answering some of the question's, suitability
THIS special Journal section is lor the engineer and of a particular mode of control for a given application
operator of the natural-gasoline and cycling plant, will be noted. In dealing with instrumentarion of n gas·
It outlin~s in question·nnd-answer form some of the oltne stabilizer, some of the factors involved in selecting
basic ideas involved in instrumenting the fractionation the control valves will be summarized.
section.
Why the fractionation section? , , . What about the
. rest of the plant? . , . One could deal with the inlet gas
system, compressor control, alarm and emergency shut- t Control of Fradiona!or Columns
down system, gas-treating, or other sections of the plant. ·
However, some of the most difficult control problems Referring to Fig. I, the vapor furnished by the
occur when dealing with fmctionators. Therefore it h~s reboi!er an0 flowing up the column becomes incre:1singly
been decided to concentrate here on instrumentation of lighter (greater concentration of lower · ~oiling com-
the fractionation section, ponents) after each contact with liquid on the trays. The
Also in approaching this business of fractionator liquid downflow furnished by the reflux becomes
control one finds there are different methods for control increasingly heavier after each contact. This is rd1ected
of temperature and pressure, This i5 apt to lead to as ~ temperature gradient between the bottom and the
confusion unless the process engineer carefully docu- top of tl.e.column. On each plate there is :-~n l!xchange
ments in his own mind which method belongs where. of heat and material. The liquid and vapor leaving c;\ch
It is true that somctim('<: alternate methods may do a tray have a different bubble and dew poinl, ·each tr:ty
given job about equolly well nnd the choice might boil tempt!rature being successively lower as we go up the
down to one of personal preference. To this extent, then, coll!mn.

A special section hy John C. Reidel, of the Journal staff

 temperature--eomposition gradient in mind, the effect
1. St1blo control is <l n~ust ••• of increasing or decreasing the heat supply to the rc ..
boiler or o( changing the reflux rate can be better
appreciated.
COOUNG MEDIUM, Assume that the tower of Fig, 1 is n deisobutanizer
(WATER 01\ OHI ER)
operating on a feed with· an avemg'.: composition of
. ~ t,HOLDS BACK PRESSUf\£ around 90 per cent n·butane and heavier, about S per
,.----("--''' ON COLUMN cent i·butane, and the rest propn.ne. Assume that the job
,.---\i'<l--,.. the column has to do is to make nn overhead product
with minimum of n·butane nnd with p;!tention of .
i-butane in bottoms held to not over say 0,1 per cent.
How well we can meet these specifications wilt
NET OVERHEAD depend, over·all, on how well we can maintain th¢
~--~-PRODUCT OUT ma.ximum temperature gradient between the bottom
and top o ( the column, l n this particular case the
maximum spread available to us will be say around 20°
F. If we were to operate at a considerably lower, gra·
dlent, say at onlY" 15° F., we would be making of(.
specification product with a lot more n-butane in Olo'er-

(STEAM 7
HEAT SUPPLY
OTHER)
head and i-butane in bottoms than can be tolerated.
If the reboiler temperature went up while the reflux
rate remnined the same we would have a smaller temper-
ature gradient and more of the heavier material would
go up the column with a greater concentration of n~
butane showing up in the overhead. But too low a
bottom temperature would be no good either sine~ this
would result in insufficient stripping and more light
material would be condensed and come out in the
Nt:'f BOTTOMS bottoms.
Pf\OOUCT OUT Now if the reflux ratio were increased while every-
thing else were kept equal we woutd hfwe more rt:?ctif!-
o Temperature and pressure •• Tower operating cation. The top trays would be cooled more and we
conditions ''re determined by the composition of tht! ' would get more condensation of heavier material over
feed and the sp~cificat1ons set for the products. Thus, this section, resulting in greater concentration of i-butanc
with composition of the overhead product specified, we in the overhead.
can determine the dew point of the overhead and the· So the cotumn must be run with heat supply set to
tower-top temperatu.re, The top temperature will be give the necessary stripping and with the reflux rare w .
higher than the tcmpcrnture in the reflux drurn, how match to give the maximum temper"ature gradient be· ,
much higher depending on the service (i, e., whether tween top and bottom. Howelo'er, precautions must b~
dealing with a deisobutanizer, for examole, or with a taken that the tower does not get Hooded. For ex0.mpie,
gasolit~e stabilizer), The top temperatur~ is dependent if we are operating close to maximum column capadty
on the temperature o( the cooling medium. and we permit surges in the reflux such th:lt (h~
In some cnses, where water is vert scarce, forced<l.ir maxim1.1n1 is exceeded at times, there may be flooding
cooling might be employed for at least some of the units. of the upper trays of the column with impaired (rae·
In cenain other cases refrigeration might be employed, tionation.
resulting in lower tower operating pressure than is
possible with wnter cooling, o What to contro 1? • , ,
Ordinarilv, however, water is the condenser cooHn? Q. What is the over-~11 problem of fractionator
., medium. The water temperature then determines tower~ control? \Vhat are we trying to do and what must \Ve
top temperature nnd tht! vapor pressure 'Jf the overhead · . control?
product at reflux drum temperature wilt determine the A. Our over·nll problem is to ~<eep the column
operating pressure. operating at equilibrium conditions. In oper~.tor's hn-
With the desired composition of the bottoms product guage we want !he column just "perking" along, mak;ng
~pecified, the reboiler temperature is fixed. Thh; is the specification product. On startup i.t usl!al!y t:1~cs some
tempentture of the particular b o t to m s composition hours for the column to Jttain steadv ooeration. On-::~
whose vapor pressure is equal to the bottom operating it is at equilibrium our aim is to :<eep' .\t lhe:-e.
pressure. t\Hownnce is made in column design for Now there are a number· of things that may hn?pen
pressure drop between bottom and top of column, and which tend to upset the operation. For !!Xamp:~. t!-le
for the drop '>etween the top o( the to,vcr and the steam header pressure may change whlc:-. would tenc:
accumulator. to upset the ftow of stc:.1m to the reboit-=r. \Vhile feed
rate and composition wot!ld ordinarily remain q\!;~c.­
() 0 'rower gradient , •• \Vhat is most important, however,
is the temperature gradient of the tower, and the com·
constant for long periods in natural-gasoline process~ng,
these also might change at one time or :1nothcr.
position gradient that goes along with it. By keeping this A good operator, if he were running the ~oiumn

•
 . rnACTIONATOR CONTROL.

By comparison we have ·a different, and in many

respects easier, instrumentation job to do in the control ; ' ' . 'I
manually, would not try to counteract these changes by say of a distillate stabilizer in a gas·cond~nsate \.__.!
tao sudden or violent action. lf heat supply to the processing plant.
column decreased, for example, he would open up the .
Q, Would you say then that temperature control is
steam valve gradually, just a little at a time to correct
the condition. Similarly, the instrumentution must be the most important item in fractionator instrumentat[on'?
dc')igned to mnk.e smooth gradual ndjtlstmcnts when A. It is hard to say, in general, what is most
necessary in hent supply, feed rate or reflux, or in rate important .. Again, it depends on the given situation.
of product removal. Temperature may, in many cases, be the most difficult
Accordingly, fractionator instrumentntion methods · item to control due to the relatively greater tag associ-
could be classified on the basis of controlling: ated with temperature systems. Heat-transfer lag,
!, Feed rate. capacity lag, and velocity-distance lag, and perhaps
2. Rate nt which products are removed. even some measurement lag must be overcome in frnc·
3. Heat supply rate. tionator-temperature control. This should be remem-
4. Condenser cooling medium rate, bered in part1culor when instrumenting for a close
S. Column pressure. · separation. But in concentrating on temperature. one
A breakdown of control methods in this manner must not forget that pressure control. is important also'
might,' however, be somewhat confusing in that feed due to its effect on tempCfature. Again, undue variation$
rate can be and sometimes is tied in with the rutc of in feed rate and composition or in rate of product re·
withdrawal of products, while pressure control may be moval· resulting in off-specification product cannot be
tied in with flow of condenser cooling medium. Another tolerated either.
approach that has been used is to speak of (1) a One can operate a fractionating column with very
"matcrial·balance'' control system, (2) urectifying-section close temperature control yet have off-specification
temperature" control, (3) ''stripping-section· tempera- product because heat supply rate or reflux r::\te are
- t\lre" control, and (4} nrrnctionator-prcssure" control. varied too much in order to hold a constant temperature.
Another important consideration is column flexibil-
ity. Some columns are designed for product flexibility o Feed rate •• , .
nnd can make n range of products of different specifi- Q, Since feed rate is also important, is it ordinarily
cations. Often we nrc also faced with the problem of necessary to usc cascade control to regulate the feed
designing for flexibility with regard to capacity. In rate?
natural·gasoline processing, for instance, there is a
definite need for such flexibility to provide seasonal
A. Cascade (also known· as "pneumatic set" or . \
.. "interlocked") system would not be necessary ordinarily • ..J
variations in plant c.'\pacity.
Jn periods of high-copncity operation it is possible if the feed comes from storage_ oi' from a large ked
that the column may be crowded beyond its maximum drum. The arr.angement of Fig. 2 would be used.
throttghput rate so that flooding occurs. For this reason,
there is some trend at pres~nt toward the use of differ-
ential pressure recorders for determination of the flood 2. If i eed tan!< is large ...
point so that the column can be operated at the highest . ·
throughput rate possible short of flooding.
Q, It is realized that automatic controls of pressure,
temperature, and flow are necessary to prevent column
upsets nnd keep the product on specification. Dut to
prevent contamination of the product, how close must
the control be, for example, on temperature?
A. That depends on what you are trying to do. lt
depends on the kind of split you nre trying to make.
Temperature control on a butane splitter is an example
of a more exacting fructionntor control requiring th:lt Q. Would a proportional controll'er be suitable in
temperature be held to relotivcly smnll variations. Ex- this applica~ion?
actly how close the temperature ':11Ust be held would
depend to some extent on the feed to the dcisobutanize1·, A. Yes. Assuming a lnrge enough feed drum (jnd
that is how much normal and isobutane (and propane that no other units nrc drawing simultaneously· at vary-
carried over) arc in the feed, and pnrticularty on how ing rates from this feed tank) a proponional controller
pure an isobutane will be taX.cn overhe:ld nnd what would do the job. You could use a ret>1tively high
retention of isobutane is to be permitted in bottoms. sensitivity (narrow proportional bnnd). However, con-
On the other hand, it must be remembered that we sij:ler also carefully the usc of a proportional-?iu~-reset
must hnve eno·ugh temperature change to actuate the instrument, particularly when there is possibility .Jf I0ad
controll~r and so nlUst locnte the temp!.!rature-mc~:mring
changes on the feed side.
element at some point in the tower where there is a Q. Suppose now that we are feeding the intermediate
sufficient change in temperature, between one trny and column of a fr::tctionation train, where the fet>d tank is
the next, relative to a small change in the composition the accumulator drum of the preceding column ... ?
of the ;:-reduct. A. That would be a level control, as in Fig. J .. ,
 .·.

() 3. Fec1! is from nccumulator ••

have as explained before to permit a small change in
outflow from the feed drum, even. though there "'"Y
be n Jarg·~ inflow, and will reset the control point on
the flow-recording controller accordingly. The latter
then acts to hold flow stable regardless of any changes
in pressure drop.
In other words, you may have a situation where the
control scheme of Fig. 3 won't quite do the job, t\s
for instance where pressure on the reflux drum is. not
conStant. Since in natural~gasoline processing, fmc ..
tionating columns arc operated under pressure, the
head due to the liquid level,in the ·feed drum would be
but a small purt of the total bend, Therefore changes in
Q. ls a constant level held in the accumulator drum? pressure woutd alter the relationship bet\veen level and
outflow, hence the interlock.
A. No. we would use an averaging-ty-pe liquid-level ·
control. No attempt is made to hold a fixed level but Q. It was stated that a wide proportional band would
it is allowed to vary above and below some middle be used for the averaging liquid-level controller. Would
position within limits, The control valve in the feed. you ever use " proportional-plus-reset controller? '
line will open or close in accordance with the rise or
A. Yes, a proportional-plus-reset instrument would
fall of the level above or below the set point. The whole
be indicated where surge capacity of the feed drum wns
idea is not to permit any sudden or large change in the
flow to the accumulator to be passed on to the flow too small. With proportional nlone you get deviation
from the control point and so would tend to have wider
fn;nrt the accumulator.
variation in level. By add\ng reset you. tend to bring
In other words, we make use of the surge capacity of
the level back toward the control ..point so that over . . ._
the accumulator drum by using a wide proportional band
'
[. or throttling range, With n· band o( around 100 per period of time yo\\ stay more in the middle of the banJ.
cent we can get o. smnl! change in vntve opening for a Q, Suppose a steam pump were used for handling
large change in liquid level, The system of Fig. 3 then the feed , , , ?
opcrate3 on n definite relattonshlp between change \n
liquid level and change in outflow from the drum. · · ·A, That is not so common a case but is encountered
(./ Q, Since the level is allowed to vary, isn't there
now and then in natural~gasoUne processing. \Vhere th~s
OCCI.lrS we m1ght have to use an interlock control as
some danger that the accumulator might run dry or in Fig. 5.
get too fuH when there is some unusu:\1 upset ln the
flow of overhead product from the preceding column?
A. No, Before the instrument is purchased the maxi- 5. !f steam pump is used ••.
. mum change in the inflow for such an operation n\Ust
be determined, and the controller specified and the .
instrument setting made with this in mind. Provision '·
can be made for a very unusual upset by specifying a
·: level controller in which limit or safety stops nrc in-··
corporated,
Q. Now coming back to the original question on
feed·rate control, wouldn't it be desirable to use an
interlock when feed comes from the rcl1ux drum of the
preceding column?

Here the level controller sets the control point of

4. for stable feetl rate •.• the flow controller, This is made necess:1ry sci. thJ.t iL
steam pressure should change, resulting in f•!Cd·ratc
change, the f1ow controller will correct this right r,\vay
and bring the flow back toward the set point. Th~'i
change in steam pressure would be termed a "'3up?lY
side load change." The steam comes from a corn:l\0>1
steam header and a diversion o( ste:.tm to other usc'l-
causes a drop in steam pressure to the pum?. Conv·~rs·~!y,
.i if other demnnds for steam shou!d fall off, ste:'.nt
pressure to this pump wou!-:l be incre::~.sed,
In summary, th~n. any f1uct 1Jations .._-.vay from !:t('
(.~---.l desired feed rntc would be sensed by the orifice .rnet•:;
'-...--··
A. Such a cofitrol system (Fig. 4) will be in order \n the fc~d line nnd the f!O\V contr.ol!er would act w
when it is desired to mftintnin a very strtbll- feed mto reposition the (.;Ontrol vnlve on the steam line. ,\!so, ::s
to the column, The avemging level controller will be- level vnries within limits in the feed drum the k·v•:-i
 FRACTIONATO!l CONTROl
t
Q. What about using a level controller as was dis·
cussed before?
control resets the Oow controller which in turn ooens A. Commonly that is the method used for controlling ·
or closes the control valve to admit more or -less steam . rate of removal of overhead and bottoms product. In
to the pump. The feed rate is thus kept essentially con- many cnses we would use what might be termed "strip~
stant regardless of level changes or changes in steam ping~section temperature control" in which the he;H
pressure. Although this use of the steam pump is not too supply is regulated to hold temperature const:mt :u some
often encountered in the modern gasoline plant it bas selected point in the stripping section. \Ve would then
been used here as another example of whnt is mea11t regulate the reflux to the column with a flow controller
by cascade or interlocked control. and use level controllers to regulate flow of the over ..
The same reasoning can be applied in another case head and bottoms product. If the product went to
that is not so common hut docs exist, that is, where srornge the level controller would directly operate the
reciprocating pumps arc used but these arc powered by control valve in the product line. If the product were
ntHUrnl gas instead of steam. Here again we could use going as feed to the next column and a very constant
thi! level controller to opemte directly a control valve feed rate were essential the level controller would reset ·
on the inlet gas line to the pump or, if a very constant a flow controller as has already been explained.
feed rate were critical, we could reset a flow controller Referring to Fig. 6, rate of bottoms-product removal
on inlet gas. must balance rate of liquid downflow in .column and
Many applications do not need interlocked control this can be accomplished satisfactorily as s how n.
and it may be well to point out here .. , keep it simple Ordinarily a proportional controller will do the job here.
and economical whenever possible ..• don't use cnscade
control just because it may look good 1 use it only when Q, What's wrong with regulating the flow of over·
there is a definite need for it. head product directly?
o Removing the products , , • A. This scheme is shown in Fig. 7 ...
· Q, What is involved in control of product removal
rnte?
A. LeCs say we have an operation where we are
7. No ieecl change compensation •••
making specification overhead product. Our chief con-
cern then is to keep the product 11 0n spec." Off~specifi- ·
cation product is an economic loss. Therefore we must
regulate rate of withdrawal of the overhead to maintain
a constant composition. If we remove the product too '·.,../
fast 1 concentration of lighter components in the over-
head will decrease. Too slow removal will increase
lighter-components concentration. This will be immedi-
ately apparent to anyonl! who has ever made a material
balnnct on a column. Such a control system might not be so satisfactory
Q, From this viewpoint, then, what is the best way if feed rare and feed composition change.
to control the removal of overhead product? Q. This brings us back apparently to the problem
A. The ideal way would be to control directly on of feed-rnte control , , . suppose there were irequent
the composition, There has been some development changes in feed rate ••• ?
along this line, using UY (ultraviolet) or IR (;nfrarcd) A. A feed-to-product interlock co.uld be used ..
analyzers to measure composition directly. However,
this is not yet universally applicable or practical 1 so in
most cases we still control on a variable other than '8. f-eed a11d tops interlock •••
composition, ~uch as temperature, since this can be
taken as a direct measure of composition.

6. Bottom proclu<t to storage ••

In Fig. 8, FRC-1 is a controller~tr"ansmitter which

sets the coritrol point of FRC<?:. If feed f!ow incre~ses
 then product flow is automatically increased,· Another between components with. a boiling range spread of less
flow interlock is also possible, such as between feed than 10° F. where temperature control might· not be
and reflux, for example, feasible, The schemes of Figs. 3 and 9 have been
~.•..· Q, Suppose an averaging liquid-level control is used presented to show some of the possibilities in what
on the feed drum,· how might an interlock be arranged? might be termed "matcrial~balance control."
A. We could use ratio now control . , , Ordinarily, in nutural·gasoline processing, the feed
rate does not change very much during many hour< of
operation and such interlocks n.re not justified.
9. Another caseadc control ••• Q. How about a tiein between rate of feed and rate
of bottom-product withdrawal?
A. Yes, this would be an example of materiol·b•dancc
control. The following points as given in more detail
elsewhere' briefly summarize this type of control: (I)
it would be used where temperature <.•ontrot is not fen.
sible, (2) feed composition must be checked periodically
and adjustment made for any change, (3} feed cirurn
should be large and tower provided with plenty of trays
to allow for the feed composition changes, ('~) troub!e
might be experienced. due to the lag in the column,
requiring a resistance and/ or capacitance on the pneu~
malic tubing from the controller-transmitter FRC-1 in
Fig. 10.

10. Feecl nntl bottoms intcrl<Uk ••

In Fig, 9 the level controller is used as a transmitter

setting the control point of FRC-1 and FRC-2, keeping
withdrawal rate of overhead produCt proportional. to
.., rate o( feed. The controls of Figs. 8 and 9 1 as well as
( .) that of Fig. 7, do not compensate for changes in feed
composition. In order to do so we would have to employ
a still more complex system. It has been suggested, for
instance,t that a flow-ratio controller be used to operate
the control valve in the overhead products line. The
ratio controller would act according to the changes
transmitted by two flow transmitters, one in the feed
line and the other in the overhead products line. So far
that would be essentially the method of Fig. 9. How-
ever, in addition, a vapor-pressure controller would
ratio set the flow controller. This system would employ CONTRO~ SET
then a vapor~pressure comparison method to compen~
sate for changes in composition.
In other words, the flow ratio between overhead o How •bout heat supply? •••.
product and feed would be adjusted according to the Q. It is noted that in Fig. 10 the heat supply is on
difference in vapor pressure "of a sample o[ the desired level control. How does th;s function and how else may
composition held in a bulb on a given tray and the heat supply be regulated?
vapor pressure of the material actually on the tray.
A. In the method of I'ig. 10, heat is supplied at the
This system has been used, for example, on a dciso· rate required to vaporize whatever material is not re~
butanizcr. A <~differential vapor pressure" controller
moved as product according to the dictate of FRC<.:.
resets the flow control on reflux to compensnte for
Another method would be direct flow control of steam
changes in feed composition. or other heating medium as in Fig. 11. If steam is lht:
Q, The interlocks just discussed seem to be getting heating medium and is drawn from a common s1e<tm
into the realm of "fancy" instrumentation. Would they header supplying other services it would be advi·mble ~o
be normally used in the field processing plant? install a p•.J:ssurc regulator in the _steam line, <ts shown.
A. Such control schemes are not ordinarily employe.:! This scheme would again be open to !he ob}cction
"in the natural·g•so\ine or cycling plant. They are re- that_ it does not compensate for changes in ft.!ed ratt: vr
served for specia\ siluations, such as might occur in ~omposi::Jn,
petrochemical or perhaps in some refinery processing. Given a constant feed rate the control of F;~. ll
() The usc of ratio Oow control, as outlined nbove, mieht
be resorted h:1 where a very close spHt hns to be made,
will maintain a fairly const1nt internal reflux . . ratio
(ratio of liquid downflow to vapor upflow) in th~ s:rip-
as between normal butane and isobutane or, i11 geneml, ping section, which is what we want. However, Ihe L/ Y

7
 FRACTION/ITO!! CONTROl
Q }
operated so that the reflux would change if• column
11. Heat supply on ilow ... throughput were changed .•. ?
'A. Change in the external reflux ratio would not be
·........ ).
big. Here again you would have to gage by experience
what normat variations in throughput may be expected
and set the reflux. rntio for operation just below the
rry~ximum capacity of the column or for the maximum
. expected during a given period of operation. This means
\ that when you are running below the maximum you
will be using more heat than is nccessar; to strip the
feed to the desired bottoms concentration. Also the
ret1ux ratio will be greater than was shown to be neces·
snry by design calculations for the particular separation,
so that there is ·also some waste iri condenser cooling
duty.
Q. Why operate in this manner then?
. A. First o( all, note that at the lower throughputs ·
ratio should change if there is a change in feed conipo·
with this manner o( operation you don't have to worry
sition. With this method it will remain constant, which
about product quality but should actually get a some-
might not be so good. The method of Fig. 11 could be ·
what better product.
modified by using the level controller on the reflux drum
Specifically. the argLJment for running the column
to reset the flow controller on the heat supply.
this way is that it makes the job easier for the tempera·
Q. What method would be used for controlling the
turc controller. This controller has a very tough job due
heat supply that might be applicable for many cases of
to the many lags associated with temperature control
fractlonutor control'l of fr~ctionnting columns, as was mentioned before.
A. That would be a method as shown in Fig. 12.
'\Vith n constant reflux rate, a given change in feed
rate requires a relatively small change in heat supply to
12. Heat supply on temperature ••• the reboUer, the actual amount depending on the refl\!.X
rntio and concentrations involved.
The changes which the temperature controller has
to mn~<e could be reduced still further by using a cascade ··.... /
control on the heat supply, as in Fig . .13, The f~ow con~
troller woutd take care of changes in steam-header pres-
sures: and other fluctuations on the heat supply side.

13. For stable heat ilow ••. ·

..
In this case we are taking the temperature at the
pain~ of measurement as indicating the bottoms com~
position, Jssuming pressure is constant at the point of
rn~nsurement.
Change in heat supply to the column will have
the same effect on the stripping section ns change in
reflux has on the rectifying section. Both heat supply
and reflux mte must be held constant, especi:.llly when
trying to ho!d specification on product. \Yith fixed re~
nux, v~uying the heat to the reboilcr will affect both the Q. It is noticed that in Fig. \2 ·the temperature-
top and bottom product in quantity and conseq1.1cntly measuring element is located on the bottom trav but
in concentration of components. in Fig. 13 it is located some trays up the cotumn, \Vhat
\Vith rdlux on flow control and temperature con· is involved in location of the measurement point?
trolll!d at one tray or point of the column, temperature
and composition graJient is then fixed for the column.
A. This brings us back in a sense to the question of
how close a temperature control can be ma:L!t~ir.cd,
This methQd of control might be termed as stripping·
Actually we must have some chunge in temperature to
section temperature controL This mt:thod th~n co111pen-
actuate the controller. Thl! method shown in Figs. 12
sutes for changes- in fc~.·d composition. A proportional-
and 13 generally should be wor~<ub!e on ch~\n,gt:s in
plus-res~t tempt!raturc controller is a must here. \Ve
tcmpt!rature of l (I ::.or more. \n otht:r words, yo 1J
woulU not uttempt to control with proportionul nlone.
must al\ow for changes in composition corresponding
Q, In thi~ control system th!!n the colllmn would be to temperature changes of not much lt•ss. than t ~ F.
 Say, as an example, that the temperaturt!~nieasuring stripping section temperature control. \Vhat about rct.:ti
4
4

,. 1 element is located below the boltom tray and that we fying-scction temperature control .. , when would that
............./ are trying to take a very ·narrow cut off the boltom . be used? ·
Say that variation in boltoms·product purity is strictly A. That might be used if we were particululy inter-
limited, corresponding to chungcs in bottom temperature ested in the overhead product or if we could not find
of only 11.! • F. Then we must either give up this method a point sensitive enough for temperature control in the
of control and perhaps usc the method of Fig. 11, or stripping section. We could then gn to a system as in
'.we must locate the measuring element at some other Fig. 15.
point in the stripping section where there is a greater
change in the temperature between two plates for the
small change in bottoms composition permitted to us
by product specifications. Fig. \4.
15. Rectifying section control ...
14. Find best temperature point ~ ••

----me

HEAT SUPPLY 'I

I

"----
LR:::r:: CONTROL

In this m~thod a flow controller would be used on

steam to reboiler, so that heat supply is held essentially
constant, while reflux rate can be varied somewhat
according to temperature changes. Variation in reflux
rate will reflect in amount o{ product made and hence
in composition of product. \Ve must again find an
optimum point for temperature control, one where we
get sufficient temperature change to actuate the con~
. () Sometimes the best point for control can be deter- troller properly relative to a small change in overhead
composition.
mined from the design calculations, but in other cases
the couple must be relocated after the column has been
in operation,
:V1ore detail on finding the optimum point for 16. Optimum temper<1ture point.
measurement can be found in the Htcruture. 3 -\
A few more comments are \n order h~re. First, one
must also think about lag when locating the measuring
element. From this standpoint it would be better to
locate at the bottom of the column or perhaps even in
the vapor return line from the reboiler. In practice, a
compromise will probably have to be made as between CONTROL

L~s:
lag on the one hand and temperature gradient on the
other. In very special cases use of rate nction may be
necessary or n differential temperature control. Second 1

~-
in the mechanic~\ design of the column it mav be wise
to provide for addition:~! fittings so th'lt a therffiocol!plc
or other temperature-measuring clement c~n be ~asUy
installed at one point or another as necessary. Third,
remember that a definite rebtlonship must exist between '"'
temp~rature and composition and that this cat) be dis~
rupted by pressure changes. In the system of Fig. 15, as in thato( Fig. 12. a
To minimize effect of pressure change one thing .proportional-plus-reset t~;mpernture controller rnt1.st b~
that might be done is to locate the pressure measurement used. Also 1 simibr to the operation of the Fig. 1:.
as c!ose as possible to the po1nt of temper~ture nl<2a~ure 4 system 1 heat to reboiler is set to give reflux rntio f\.H
ment. This i~ n must if extremely high-purity product is mnximum throughput 1 le3d~ng to the same con:-;iJcr~l·
dcs\rt!d, H pressurt! fluctuations affect the temperature- tions as to reboi!er and condt!n~er duties and prot.!uct
composition rclativnship, p,cssure compensation, at !east p\.aity at lower feed rates discu~sed before.
over a limited range, may be worked inlo the control, Q. Would you consider the use of a proportional-
the temperature control being reset in ratio to the resct~rate
temperature controller in scheme of Fig. 16?
absolute prcss(tre. ,
A. There is some pro ar.d con on this question. Our
Q, We h;we been discussing different aspects of first thought would be to conslder it because of the
 FRACTIONATOR CONTROl.
t? t
in such a system. It has been done on depropanize.rs, ·
for instance. lt would be better, from the standpotnt ·--.~
gre:ller lug involved in the scheme of Fig. 16 ns com· of Jag, if you ·could locate quite low down in the rec.tify·
pur<d with that. of Fig. 15. On seco.nd ~hought, however, 'ing section, but you cou.ld get by, say o~ a.de.Propan1~er,
if WI! used an mstrumenl With dcrrvauve mode, change even with a location h1gh up m the recnfymg section.
in rdlux rnte would be more rapid \vith temperature However, you would not want to try such :~. contr?t
chunge. It might be too rapid when •operating close to scheme on n de\sobutanizer, Of course, remember ng:ttn
the maximum capacity Of the COIUmO nod reSUlt II\ that there are other factors besides the lag.
flooding the top plates and off-specification product.
Q, What other factors:
One thing tl><tt might be done. is to have the
tcmpentturt! controller reset a flow controller on the A. Let's take n specific case such as mentioned early
r<fi\IX, · tn this discussion. This was the deisobutanizer • with
average feed composition of some 90 per cent n·butnne
Q. Suppose we modified the method of Figs. 15 nod and heavier, .?.bout S per cent i-butane, and the rest
16 by having the temperature co~trol!er operate the propane. A temperature contr~!\er w.as u.sed t~ operate
cuntrol vulvc in the overhead maKe ltne and sethng the control valve in the reflux hoe. as 10 F1gs. 1) and l6.
the r<flux rate on flow control. Would there be any.
In such a control (operating to hold tempernture'
advantage in this?
constant by varying the reflux) the time lag is certainly
A. Jn the scheme of Figs. 15 and 16, we have an ·important in view of the close separation being made:.
·averaging kvel controller resetting a flow controller on However, another factor to be considered is the effect
the uv<rhead products line. This interlock takes pretty of varying feed composition. In . th!s case •. wit~ the
good care of uny Ouctuations in liquid flow from the temperature contro\\ed in the recttfytng scctton, J( we
condenser. Now if we decided not to have such an locate too close to the top, variations of propane content
interlock, use of the temperature controller to operate in the feed, say even from ·1 to 2 per cent, woul_d have
th~ valve in the overhead make line would be an alter- nn effect on this temperature control, enough etfect to
, nntive and acceptable method. We could then argue that worry about.
1 ftuctuutions in liquid flow from the condenser wou!ct
I
In one case, by locating on the seventeenth tray.
bt! better tttken care of than if the valve were in the
reflux lin<. We might be able to get by then with a from the top in a 70-tray co!urnn a location was found
where: (1) effect of changing propnnc content was
constant enough internal reflux ratio in the rectifying
minimlzed, (2) temperature still varied suff.idcntly to
section, even without the use o[ a flow controller on the
r<f\\IX, ooive control with changino 0
n·butane content in over-
head, and (3) since this location wus not too far down ·-._
Q. 1t i< of course apparent, generally speaking, that from point of entry, of reflux, no. serious l.ag was en-
a rectifying·section temperature control would be em- countered.
ployed 'when taking specification product overhead and
It is important to note that for this particular opera·
that a stripping·section temperature control ':~uld. be tiofi, with an over·all maximum temperature gradient
U'5ed when making bottoms product to speclflcat1on.
of only some 20., F. available; close pressure control
However, suppose it is the top product we are parti· was also necessary. In this case the pressure controller
culnrly concerned with. Couldn't we still usc a temper-
reset a flow controller on condenser cooling water.
aturt! controller operating the valve in the steam line
to the reboiler, but with the temperature measuring o Finding the flood point •••
clem~nt located in the rectifying section? Q, Pressure control was just mentioned and this
brings up another thought ... couldn't the heat supply
be regulated some way on column pressure?
17. l.ag is important •••
A. Yes, iecan be done and there ls some trend cur~
rently toward use of this method, altt,ough it has not
been ordinarily employed in natural·gnso!inc proc~ss­
ing. \\'hat you have reference to is a system such as that
of Fig. 13 ..•

----~ What this setup does is measure the vapor flow up

the coluu.n although we are nctu~lly controlling on th~
pressure Jrop. \Ve could use ::t pressure controller if the
overhead or condenser pressure were at atmos!'heric,
In operating at pressures above atmospheric it would·

I be a differentlnl-prcssure controller ns .<>hown. The dif-

ferenti~; pn::ssure actually might he measured ncross
some portion of either the strip!Jing or re:::ifying s·~c­
tion. Since the liquid static heads on the pl~~tes arc rc!a-
tivcly constant, nny cho.n.gcs .in these being insufUcient
to base control on,. it is the vapor friction drop across

•
L_ the column th:1t is being measured.
{n column operatbg at close to maximum through-
A. Thnt could and has been done, but is of course put thiz might bl~ a. quite sntisf:.u.aory method. Lc;'s
open 1~1 thl.' objection that there WOllld be a 1ot of l<tg look at this a littk closer. It could be saos-
 close to the flood point ilntl can be trimmcJ hack \\hen
18. Differential pressure control ••• necessary if the record shows tendency to flovu.:•

G Pressure control • , •
. Q. The effect of pressure on temperature and the
HEAT SUPPLY need for pressure control has been mentioned several
times. Pressure control could probably ba lieU in \\'ith
regulation of condenser cooling mcclium. But ... tHl-:ing
the most common case, where wtdcr is the C'l.'L'Iing
rriedi"um , •. why worry about automatic contwl l"'lf
condenser cooling water? Couldn't you just control
manually with a hand valve?
A. In many cnses that is just what you wuuiJ Jl1.
Simply Sl!t the hand valve and pos'iibly adju"t it ~~nee
in a while. Of course there would be more chance wi1h
the manual control of having not enough or too much
cooling water at any given time. So at one time Yl'U
might have enough water to condense the ovcrhcaJ
vapors and cool them: down to the temperature dcsireJ,
At another time you might be wasting water hy h<IVing
the valve too wide open.
factory if feed rate and composition did not chang". You can do better with automatic control, sa\' as
Also, there would be t\ question ns to the limits in pres- in Fig. 19, regulating the flow of water acct,rJin'g to
sure range for which it would be operable. Since in this the temperature of the overhead condcn'iatc.
.. method vapor velocity is measured indirectly it might
b~ most applicable where operation is close to the
nmximum Y<.tpor velocity for 11 given column. Of col.lrsC, ·19. Regulating condenser water •
there is no guarantee tht.H operation would alw<1ys be
very ne:.1r the maximum, ,
Q, Thi:> brings up something else . , . something .that
appeared in the technical liternture a few years back
about the ust! of a differential~pressure recorder to
determine the flood point ... what use is that?
A. You can use a differential pressure controller as
just c.\pluined or as is done in a number of installations
using differential-pressure recorders and indicators to
determine the column flood point. Then the operator
can be sure to stay just below flooding when running at This might be done in those areas where watL'r b a
m~ximllffi throughput.
very scarce commodity and we qnnot <-~fford to Wa:\tc
If the column is to operute at close to maximum it. Note this temperature controller (in Fi£. 19) \\"l'lliJ
throu£hput all the time it might ht! preferable to emp!oy be installed in addition to the tempcr".ture L)r ~Hht..'r
a differential-pressure controHer, as outlined bdore. controller required for primary column contru! as !n
Despite thcoreticu\ comiclerotions as 10 economical regulation of heat medium to the rcboikr. Bc:--t n.•stdts
usc o( heat, where nccess::~ry stripping con be obtained .with this system would be obtained when the inkt ·
while operating below !he. maximum vapor velocity, cooling-water temperature is c!c~c · tu the pr\1duct
there nrl! times when we must push the column for temperature. Proponional·plus~rcsct shotlld hc u:-;cJ.
maximum throughput. lf the maximum vapor velocity This method will hold outlct-wutcr tcmpt.:r<iltm: at ''
is exceeded, sometiml!s due to a sudden surg(', the: maximum for ·given overhend V~tpor, inlet water. itlld
column will "puke" (:sudden surge of vapor "c:-trrks product outlet temperatures.
liquid over with it). This can be very embarrassing, it Beyond what has just been ~aid, there arl..' \Hhcr
i~ true, but is not the only thing that can happ~n if tit.!in~ involving condenser water rate, \Vi!h prc..,surc nnJ
vapor velocity is. t.!:<cccdcd. Tht.! column can flood to the even level control, but the. primary purr('"~ in these is
point where it becomes tot<11ly inoperable, or operation to aid in control of the tower pre:-;~ure.
Ci.ln continue with a partially flooded tower, the lower
tra)'S in the stripping section or upper trays in the recti- Q, Coming· b"'ck tht!"n to prL•ssurc C(\DII'll\.. this
fying section being flooded, resulting in poor fmction~ apparently warrants some dt!tailcd considt•rati(ln. Th~.·r~..
at ion. seem w he many methods of prcss~trc contfl'!. ~n
It mav not be immediately apparent that flooding. is general, whal determines the use of u particul<tr \ll\1..':)
r~:-.p~)n:.;ib.l.; for thf) poor fractionation but this C<ln be A. A particular pressure control method W(luhl c.c
d~tcrmincd by :neusuring the pressure drop nt dilf~:rent applied depending in gt:ncrul on whether: (!) (Wcrlwa...t
reflux rutes. A significant rise in the pre:;::.urc drop product were taken off as liquid or vapor,·and t~) rcl<l·
0 \\Ould indicate' the ~flood point. \Vith a diffen:nti;.d~prcs~
sure recorder installed, the column cnn h1..' t.."""~pL'HI.tcd
lively lnrge or sm<tll amount of uncond~nsahlcs W<'rc
present.

11
 fllACTIONATOR CONTROl
SJ ·i
.·temperature resetting the flow controller _on the reboiler
steam line or some other interlock. ',
· .....j
Q, Suppose purl of lh< owrhoud is removed us n Q. Tiein of pressure control with condenser cooling
·gus ..• ? water flow was mentioned. Just how would that .work
A. If there were u large amount of uncondensables and when would it be applicuh\e?
(int:.rts or low·boilecs), pressure regubtion would gen~ A. This could be used where there is total or partial
erully be very simple, being accomplished by moans condensation but basically negligible or small amount of
of a bud:·pressure regulator in the vent line. Pressure noncondensables is prc::;ent and product is being taken
control in this m11nner is particularly suitable when off as a liguid.
the overhead is small relative to the feed. We nrc
thinking here in particular of an operation like a
dt!t!thunizcr where we nrc tukin.r; off a substantially 22. Non!ondcnsables negligible • ,
nurrow.boillng~runge product overhead as n vapor,
everything dse being condensed. \..Yith a large enough
flow of vupor we can employ n throttling . pressure
contrnl.

20. Venting the reflux drum ' ..

The' pressure measuring element can be located at
the top or some intermediate point in the column. \Yith
nearly all the overhead condensed, this method consti·
tutes an indirect control of temperature, hence of the
vapor pressure of the overhead product. If pressure
This inUicutes continuous venting o( uncondensables. increases above the set point, flow of conde.nser water
Jf pressurl! increases t\bove the set p'Oint more gas would will increase resulting in lowering of the pressure back
bleed oul to bring the pressure back toward the set toward the set _point. 1n this method then you c~n get
condensntion without having to worry ubout too much . i
point. Since tht!re is very little lag associated with this ·._j
prc~~urt! control u proportional controller with narrow subcooling. .
band would be good enough. The pressure measuring This method presupposes· a tube-bundle' type of
point couiU ubo hi! at the top o( the tower or in the condenser with water through tubes. Some. attention
vent line it~elf, must· be paid to the condenser design,· providing for a
high enough velocity through tubes so that lag is cut
down· permitting use of a proportional controller with
21. Pressure clement in tower ••• narrow proportional band. /

Q, \Vouldn't there be some limitations on how much

water flow could be increased?
A. Settings could be made' so that i[ the pressure
should rise above thut corresponding w the maximum
w~Her. rute available or desirable, the bnck-pressure reg~
ulator in the accumulator vent line .would open.
Q. \Yould that be a good way to operntt! where you
hnd some inerts and low-boiling constituents that c!id
not condense but accumulated a; gn~ in the condenser?
Still nnotht!r pos:-~ibility comes to mind in connection \Vouldn't accumulation of such noncondensables reduce
with venting under the conditions where the method of condenser efficiency and wouldn't it..be best therefore
Figs. :.o ant! 2l :~pply, un:J th:H is to use a temperature to vent continuously?
controlh:r to opt! rate the \Jive in the _vent line. A. That is a good point. 0( course there is really
Q. Suppose there were wi<.lc lo<td changes but pres- no problem with nonconUensnbles when you are taking
sure hull to he held v~ry comtant, would you still use off a vapor product overhead as in Figs. 20 and '21.
a proportionat controller'! Ther~ the inerts simply vent out with the product.
A. ln th:1t cnse a proportional·plus-reset controller But in the method of Fie.. :!2 when control!in~ on
would bl! prdc:rablc. This might be done, for instance, water rate and with ovc-rhend product as liquid, Vent-
on a deciiH1nizer oper~tion when~ we were trying in ing of inerts (\nd low boilers mw.y become a major
p~rticul:..lr to hold pr\lprtnc in overhead to a minimum proble111. lf UJ,IOUnt of noncondensables were ;_;uitc
ami thcrdore wanted to hold a comtant pressure d\!s.pitc small perhaps you would decide to operate r~s just noted,
some fluctu~ttions in other tower vari:.!bks. Of course that is by allowing the pressure to builL! up to the point
it must ht! r~:dized'"'that the pressure centro! i'i no·t the where the control valve in the condenser water :int! is
primury or key contwl here. ln d!!e!hanizcr opcratton wiUe open t1t whkh time tht! wnt v~1lve WOllld opt.:'n
Wt' Wt)u\U have to control the h~at -;upp!y, p~rhHps with to rek<tse tht: non-:-onJemnh1es.

12
 However, it is true that it would be best to vent the lf the pressure went above the set point the Valve would
( •,
inerts continuously if possible before the vnlvc in the open wider. This would result in increased pressure in
'-...--} water line is wide open. \Vith inerts in the condenser, . the accumulator. Condensntion would be increased
instead of controlling just on the vapor pressure of the some, with resultant drop in pressure bqck toward set
. condensate you will be controlling on the sum oi the point,
partial pressures of condensate and inerts. As inerts This application is then for the case of total con-
accumulate they will have n blanketing effect, resulting densation with o. narrow-boiling_.range product being
in reduction of condenser efficiency. So ii you could removed. It ha'3 the disadvnnt[lge that a lo.r,gc control
vent noncondensnbles continuously nt just the right valve must be installed in the overhead line.
rule you may be able to accomplish two things: (l) not Q. Could we use the pressure controller to oper:He
allow condenser efficiency to be reduced too much, a control valve in the overhead products line'?
while (2) using the effect of the inerts to help in con·
trolling the column pressure.'
Q, In the discussion accompnnying Pig,· \6 where 25. Hooding condenser bundle •••
we tnlked about a deisobutanizcr, it wns stated that the
pressure control employed an interlock on condenser
cooling water •.• why wr~s the interlock med?
A. Yes, that was the case where we bad a small
amount of propane in the feed and were worried about
the effect of variations in propane content. Since n
close separation was being m~de, nn interlock was useO,
to insure c\ose control, in such a manner as shown in
Fig. 23.

23. Fluctuations in w~tcr flow ••• ··. A. Yes, the back-pressure control vn\ve can b'e set
in the make line. This is another method applicJble for
· the c.:~.se of total condensation of overhead and is known
CONTROL SET as the Hfiooded condenser" method. The water rr\~c
(water through tubes) is set manually and the rate of
condensation is regulated indirectly by flooding· the
.! condenser tube bundle, more or less in accordance with
pressure fluctuations. If pressure increases th-2" b;::;.c:-:-
.i
pressure valve opens wider and more tubes in the cor:-
denser bundle are uncovered. This results in increased
condensation and lower pressure.
This method will not be Satisfactory unle:iS amount
o(· product withdrawn is large relative to amount of
reflux. Amount of noncondensab!es must be quite small.
\Ve might get more subcooling oi the product thar.
desired by this method with $Orne waste of coo;!ny
The 'lpecific reason for using the interlock is that water.
condenser water came from a cooling~water system
Q. In Fig. 25 a reflux dr\Im is shown. If this reOc>x
simultaneously serving other units. The pneumatic-set
accumulator is run pnrtly full it does not se:!m that the
flow controller took c>1re of orcssure fluctuations in the
water inlet line and so made 'the job easier ior the pres· flooded condenser idea would be workable ... ?
~urecontroller. A. That is right. In the scheme o( Fig. c5 the
accumu1ator is run futl. The condenser can then b~ r~n
Q, How about the case where there is a large volume part full as noted, This idea can be modifieC. for c::.sc"S
of vapor overhead and a\\ o( this is condensecl>?. where the condenser is below· the accumulato:-. Th~
·.i A. The control valve would then be s~t in the over.. flooded condenser method is not applicable w~ero:: the
head line between the column and conden:ier. 1
•
overhead product is the feed to the next column.
The back pressure is regulated in the overhead line.
Q. Suppose thnt instead of the w::ner-...::oo!ed con-
1
denser, a forced-air-cooled unit were emploved, wou:c!.
the scheme of Fig. 25 still be app!ic:tblc? ·
A. Yes, H an aerial condenser Wt::re used we wou!C
proceed in the same way with the pres$Urc con~:-o!:..:-r
operating a valve to reg'J\ate flow lr.. the produc~ ::r,c
,. ~nd with the accumulator running full. \Ve. mi~-:hl ~-.cc:.
! temP,eruture indicating points at~ the inlet anJ o11•.t~r.
of the aerial condenser so that we would have ~n ext~:!
'
·''
. '~ . check· on the operation.
i Q, ·The po5'ibility o( too much subcool'ng wa>
! I mentioned in cunnection witi1 the method of Fi'!. ~).
! How c::tn this be avoided'? -
 FRIICTIONATO!l CONTROL
Q ·~

A. Suppo~e again you have a l:trge quantity of vapor

'"•
18. Level control on water ..
overhe:HJ. You can. regulate this as in Fig. 24. Then
vou c;m elimin<1tc or <tt !cnst minimize the possibility
of 5Ubcooling through byposs of n small portion Of hot
vnpor around the condenser.

26. Byp~ss pressure control •••

tion it is also important to take steps toward mmtmrztng

any tendencies toward fouling on the water side.

2. Example of Complete Column Control

The column of Fig.' 29 is a gasoline stabilizer such
tiS might be operated.... in a gasoline plant not making
L..P.G. The function of the stabilizer then 1Would be to

29. Gasoline stabilizer control .' .

WATER !N

2.1
This ngclin is for the Clt:-.1.! where there is quite '' small
amount of unconJ~nsables nnd the overhead prQduct is ·.. __.-'
being taken off as liguid. This would be a better method
th:1n th~lt (lf Fig. 22.
If desired the svstem of Fie. 26 could be modified
slightly a!\ in Fig. 27. .. ·

27. Modified bypass control ... E TP~<r--------

F "

! L,,
L-.YS
Summary on conlro! l'Uives for ~nsollne stahilb:~:r. Fi~ • .!9
Minimum
Vah·e Valve Flow Si;e t1n
Numba Applicatiun Action Critic:d'? B~,;~~ of:
Feed to column Air-to·open N"\l T0 p:t'>S two llm¢3
\Wrm-.tl flow from
Ht:rt.! PRC·2 opcrntcs the valve in the hyptl:..S h.) ,drum
hf..1!cJ a :-.mall con'itnnt pn.'!\surc drop betwcl!n the column Watet to Nn- Air·to·l:'lose Nu :-.l::tximum water
hnd rl!flll\ dn1m. This in~trumcnt could ht.! n pill)!·type ' denser f!\1w \\1 condcns<:r
prc-.,mc C.:lH\lrl)lkr nwuntcd rig_ht ·on the CIJntrol valve. 0 ,-\\.'l:'umu Ia tor .-\ir-to·clo~e y, \\\1r.'t w,'\ta cond\-
Q. ~knlion w;~s mndc of pos.o;ible ticin be-tween lev~.·! n:nt line li\11\' 1\l conJem¢t
r~'ulting in rnni·
contr\'1 :1nJ Cl'ndc-nsc:r C!..'oling-wntcr nHe. When would nwm unh'Unl nQn·
this apply'!
A. Thnt \\'I.'Uid apply where there wns n larger 4 Re!lu:v. tl) Air·l\1·do'ie No To h-.mJk ma·..imttm
nm0unt of unconUen~ahlc-s ond 1he overhcnd W;.ls bcin_g column pumr .::1p~tdty with
tnkcn off os vapor. ':.11\ e wide open
In opcn1ting on level control in thb mnnnN, the ! ProJul.'t with. .-\ir·ti.)·\:11\se No T\' r~us twice the
idcu i'i to r~.·g\.ll;.!IC cooling wnter to mnkc j\tst ~.·n~High Ur:1wnl fro10 nurm.tl f!ow from
the re'o1,1ikr the re'o0ikr
contlensntl! for reflux. '1-h:re ns in the method of Fir!. Ht·~uing :-.:o ;..J,\ximum hot oil
6 .-\ir·l\h\'pen
·:!2 it i~ imrortnnt th;H . . \,n1.h:n~cr J~.·~ign he- :-.\lt..'h th~t medium frt'm fh>'" w rl.'boikr
n high wa!l..'r v~..·k,city ctu\ he m:1intaincU. In this C\'nncc- rtb\likr
 ' ...
, i "r " ... ,

take off some light material overhead as vapor in order Referring to the letters in Fig. 29:
to make natural gasoline of spcdflcd w1.por pressure. t\. Usc a 100 per cent proportional band for this
() For control method shown in Fig. 29 it is assumed that
a small amount of low-boiling material is being vented.
level controller.
B. This temperature. recorder provides a check for
· If the stabilizer were followed bv an L.P.O. column the operator to ma:<e sure that heavier components Co
making a propanc butane mix, then' the instrumentation
4
not go overhead.
could be modified say by placing a control valve in the C. A narrow proportional band is a11 right here.
overhend line o( the stabiiizer ahead of the condenser. D. t\ proportional contro11er can be used here. The
We would in this case still be venting only a small coritrot point is set {or reflux ratio according to the
amount of inerts or low boilers from the accumulator analyses of reflux drum samples.
.' drum and would be holding column pressure by throt- · E. This temperature controller should be proper-
' tling the large amount of overhead vapor, virtually all tiona 1-pl us-reset.
of which would be condensed, F, A narrow throttling range is all right here.

' ··-30. Closeup ot a deprop,anir.er-dchutani1cr control system ••

•D!:J'_ROPANIZE£. DEBUTAN I ZEJ3..

GAS TO ABSORBER
Of\ FLARI!

-QVowAO PP.OOVC.T
f'I\Ot.l 5Till..

iu:•3. DepropnniJ.er·Debt~tilnh.er Systcn densation. At this time, with more noncondensables

coming off than was cal\ed for in the original design.
In Fig. JO, instrumentation on a two~column frac- PRC-1 will shut off the valve in the byposs oround th<"
·."7tionating system is shown. This would be for a natural- condenser and PC·l will take over to control column ....
-.-· gasoline or c~ding phlnt in which still-top product is pressure by venting the noncondcnsables. Another situ-
fed-to a depropnnizer and in which no sep~ration is ' ation in which PC·l might take over the pressure con-
made between normal and isobutunes. trol would be in the event that rich-oil stabilization doc''
Some comments on this system are: not take place as per design.
-},"A lnrge depropanizcr feed tank is provided so 4. Deb\ltanizer pres-;ure control shown, with PRC·2
th~rcontrol of feed on FRC·I is adequate. oper::\ting the control valve in the condenso;!r wnter line,
- · 2. 12~propanizer feed is heat exchanged against de- \s for an operation where there. \viH be no uncondcn-
butanizer bottoms nnd further preheated with heating·· sables or a very small amount of them. Altcrn~tivelv,
. -.·oil fro""! hot-oil system. Note provision o[ TRC- t to the debutnnizcr prcs:iure could be controlkd in t:.:,e
regulato amount of preheat. same way as just outlined for the deproy~nizer. su:
•. - 3. Pressure control is with ?RC-t rerulating amount another possibility would be to h:J.Ve the pressur~ con-
· o( hot v:>.por coming through bypass ur~und c-ond.:P.ser troller operate the vn.lve in the overhead product l!ne,
·• to blanket the colder liquid in lhc reflux accumu\<ttor. that is on the pump discharg~.
'· Ordinarily, operation shoutd be tot<~lly condensing in 5. Lccation of relld valves is indicated. How~ver,
the twin parallel contlenser, Under such condition, no attempt h1.1s been mudc to show nit detail'S of instru-
.-. PRC~ 1 is the column pressme control with PC·\ not mentation or process flow. ·Temperature 8.!ld prcss•Jr·>
operating. The latter is set ior some pressure ahove the indicating points and bypass lines hnvc bct.·n left o~~t
''I'
operating pr~ssure o~ thl! reflux drum, at say l 0 psi. for simplicity.
or more ~bove the accumulator pressure. 6. Possible location o( thermowells on depropanizcr,
, . . After the plant has been in operation some time to provide· alternate points for temperatur<! measure·
cond\tion9' m<\y change. For cxomple. thtJ condensers ment, has been indicated to cmphns!ze imponar:ce of
.- . ·may get fouled so they no longer operate ~t total con· providing such thermowells.

1 s
 fRACTIONATOR CONTROL
,.
!
'
,.
' I , ' ' ','1 /

I ,I\. •< ·.-1;

; ','\' ·' ; , . 31. Here is principal instrumentation for Stano plant extension a{

•·' ,·1, ---------·

'OEETHANIZER
.... ..............
t
TO FlARE
_DEBUTANIZiB_ TO f'lAJ\
i
1'0 '.EXIST INC

STill TOP PROO,

OEPI\OPANIZf.f\ rf.E.O
~
'
f. XIS T, PLAN I
',. .
~!ILl.. TOP Pfi.OO.t> A~~~p.7c?en •
Ff\01.1 €XTH:/510N 1N WJC.IIMl
PLAN\
HFLUENT FROM
HfCOI.fPr\€SSER
CISCHAHGE COOLER

OEETH. rEEO
SVRGE

SET

1
4, With overhead product taken off as vapor, .·..........
4. The fractionation Train dcethanizer pressure control is obtained by throttling
noncondensables leaving the accumul?tor vent. line.
Principal instrumentation for a four~co\umn frac- ·
o Depropanizer ...
tionation train in a natural-gasoline plant is shown in
Fig. 31. This is taken from an actual large installation J. Design of this column calls for production of 95
consisting o( an original plant and a plant extens1on per Ct!nt propane overhead. Note that a temperature
built adjacent co the original facl!ities. 7 Here the still- controller, with temperature element located on the
top product and recompression condensat? from· ?oth.. twenty-founh tray from the bottom in a 35-platc
original and newer plant are fed to the deethan11.er. column, regulates the reflux rate. Steam to. the. depro-
Highlights of instrumentation for this particular panizer rcboiler is on flow ·controller. This represents
four·column train are: a revision of original instrumentation in which the
tempera.ture controller regulated steam to reboilcr while
o Dccihnnizcr , , , reflux was on flow control. Operation with temperature
clement in the reboiler vapor line wns also aut!mptcd.
1. Level controller on deethanizer feed drum resets 2. Depropanizer operates with total condensatio~.
flow controller on deethnnizer feed pump discharge to Pressure.· control originally was through regulation of
obtain stable rate of feed. vaoor flow in bypass around condenser as shown.
2. Heat supply to deethanizer is regulated by inter- Change~ have been mnde since thb flowsheet (Fig. 3 t)
lock between !eve:! controller on reflux accumulator was p·repared, with elimination of the bypass co-ntrol.
and flow controller in steam Hne to reboiler. Since no lnstead ihe reflu.t accumul:ltors arc operated liquid full ·
liquid product is tnkcn off O'l'erhend, this control is set and tht!·rcflux from these accumulators is controlkd by
to regulate amount of propane going overhead to furnish use of now controllers. Pressure control is maintained
reflux. on the depropanizer system through the reflux acc:um"J-
3. Deethunizer bottoms product (essentially propane lators by the use of pressure controllers (not recording).
and he~vicr) is split ns lndic::\ted with one streqm .~oing Provision is made for venting by me:1ns of a bac;<·pres-
to the depropnnizcr of the original plant and the other sure control. This will operate, to vent ethane to flare
feeding the depropanizer of the plant extension. This for e:<nmr'le, when deethnnizer operrnion is upset with
is accomplished by utilizing a flow controller on each exce'isive retentio11 of ethane in decth~\nizer bottor.ts
stream, either or both of which can be reset by the product.
dccthf!nizer reboilcr liq\1ld !eve\ controHer. Thus it is
poo:;sih!c to opcr<\tc wit~ either strcn.m on. flm?' control by o Dehutanizcr •••
mnnunl setting if d~sir~d and the othl!r on Jlow control 1. This column is designed for flexibility in pro·
· reset by the level controller. duction of bottoms product ranging from n completely
 A vapor·pressure bulb is located on the. forty-eighth
tray from the bottom 'in a 60-tray column. This bulb
Stanolioo Oil & Gas Co. , • is filled with liquid taken from the forty·eighth tray in
0 DEISOBUTAN\ZER
normal operation. Chunge in tray·composition with re-
sulting change in tray temperature produces a different
~........__._-..----- TO FLARE • . vapor pressure in the bulb and a pressure different\>!
J between bulb and column at point of bulb location.
The differential-vapor-pressure contro\\er then acts
to reset the reflux flow controller to compensate for the
composition change on the forty-eighth tray.
The vapor.• pressure controller has given satisfactory
and very sensitive control. Sensitivity of this method in
deisobutanizer operotion is emphasized by th< fact
thnt a change of t psi. in pressure is equivalent to about
:!;J o F. temperature change which in turn represents
severn1 per cent variation in terms of composition ..
Acknowledgment
ct!soeur
ruo This special Journal section \s an outgrowth o( lee~
EX<tt,
tures on process instrumentation given at Unlvcrsity of
Tulsa.

·.
Appreciation is expressed to industry and instrument
company engineers for critical re:tding Of the manu-

tL.
~=Bfj@'j[TI§I~==========.,_~.
h!lXEO BlJTANt.S 7

------------------1>
~~!SOBUfANES

~-,B,~toA~uEf~~~s
PROPANE
'-' ~

GASOLINE
script.
References
1. Tivy, V. V., "Automatic control of Fractionating
Towers," The Oil and Gas Journal, Novemb.;r 15, 1948, p. S5.
2. Boyd, D, M., "fraction:\tion Instrumentation lnd Con~
trot," P.:troleum Refiner, November l~M3, p. ll·t
3. Uitti, K. D., "Effect of Control Point Location Uoon
Column Control," paper presented tn I, S. ;\, meeting, St. LOuis,
·
.

. September, 1949; Petro\.:um Refiner, Mnrch. 1950, p. UO,

debutnoized materia 1 to 26-psi. R.v.p. gasoline. 4. Bergmo.n, D. J., "Instrumentation for Precise-Yr::tc~io'na~
tion, • ,", paper presented at T exus A. and M. instrum~ntalion
2. Heat supply is regulated by temperature control· symposium, Janunry 1954, The Oil and O.-.s. Jouro.-.1, ,\prU 12,
ler operating control valve on steam line to reboi!er. 19$4, p. 127,
Thermometer bulb is located in reboiler vnpor line. 5, Kaplan, W., ami Bellman, D. J., '.'Differentiui~Pressure
3. Dual pilots on the dcbutnnizer reflux drum level , Instruments for Improving . • . Processing," The Oil and Gas
Journal, September "20, 1951, p. 319.
contro\\er permit: (a) regulation of amount of mixed . 6. Pyle, C., "Control System<> for Distillation Towers/' The
butanes going to storn.ge, and (b) reset of flow controller . Oil and Gas Journal, October 13, 1949, p. 97.
on feed to deisobutanizer. This method of control is 7, Reidel, ). C., "Ulysses Plant ~ow Stanolind's b.rge.st,"
an nddttional measure taken to insure constant deisobu· The Oil and Gus Journal, :-.ray 25, 1953, p. 156.
8. White, R. S., ":-.-to.dern Process Control at El\st Veal-
tanizer feed rate despite abnormal fluctuations in reflux moor," The Oil and Gas Journnl, October 19, \953, p. 110.
drum level. If such fluctuations should occur they wottld · 9. Stark, E, L., "Instrumentation (or Light-Ends Fractiona-
be taken up by varying the amount of mixed butanes tion.'' The Oil nod Gfts Journal, Januo.'ry 19, IS' 50, p. 6:!.
withdrawn to storage. 10. Reynolds, E, H., Troutman, W., and Lawn, G., "r\pplica-
tion o( Automatic Controls to G;tsoiine Plt\r"\t . , • Processins."
o Delsobutnnizer , , , The Oil and Gas Journal, April 5, \947, p. 66.
11. Cicalese, J. J., Davi~i, J. A., Harrington, P. J .. Hough~
L Since production of specification products from land, G. S.. Hutchinson, A. J., nnd Walsh, T. J., "Typic:o.\ Fr"lC·
this column is a tougher job than for preceding columns, tionator Control Svstems, • .", Tht: Oil and Gas Journ:-ti,
a different approach is taken with regard to the instru- ~ovember 30, 1946,·p. 90.
12. Mattb:, E. D., "Iostr\lments in Process Control," The Oil
mentation. and Oai Journal, A\tgust 2}, \947, p. 83.
2. By controlling column pressure with the pressure lJ, Breedlove, H. B., "Revised Depropanizer ln:c;trumenta-
controller operating a control valve in the ovcrhcnd line tion .•. ",The Oi\ and Gas Journi11, .Aug\tit 2.5, 1949, ?· \3?..
between the column and condenser, effect of chan\?.e in 14. Wilson, G. W., "Fractionatin~ Column [nstrumcn~ation,"
T!,e Oil and Gas Journal, December ~2, 194S, p. 63.
temperature or rate of cooling water on the column 15. Lefkowitz, 1., " ... Control Svstcm , .• for Di~ti\latlon
pressure is eliminated. However, the renux drum is ••. ," paper presented at I. S. A. HOuston meetin_!;, Stptcmb!r
press1.tred by bleeding some vapor thro1.1gh a bypass t 95!.
around the condenser to insure suffident N.P.S.H. 16. Tab!:>ert, T. D. "lnstrumentution of Pac~<a~ed Gasoline
Plants," 'fhe Oil and Gas Journal, ~-fP.rch 16, 1953,·p. 11YJ.
for the reflux pump at all times. 17. Cook, T. P., "Pacbf!.l!d Gasoline Plant ... ," '!l"le Oi~ und
3. Dcisobutanizer heat supply is on flow control. Gas ,Tournai, October 15, 1951, o. 96.
Means must therefore be taken to compensate for 18. Reidel, J. C., ''Process 'Instrumentation." The Oi! :tnd
changes in feed composition or pos·.;ibly srr,all changes Gas Journnl manual reprinted from Rdint:r''l ~c!;:booi:, 11.'~0·51,
19. "lnstrumt!nts und Process Control," De\m:tr P'.lb:isha~.
in f«d rate. This is done by means of a differential Albany, N. Y .. 1945.
'''"') vapor pressure controller resetting the flow controller
(.J on the reflux line.
20. Shannahan, C. E., "Refinery Instruments .. ," series in
Petroleum Proces.::ing. July 1951.

:t •t
 Here ore many new concepts for an established oper-
ation. Design and performance of liquid-vapor contacting
devices (as presented before ·~he C.N.G.A. at Los Angeles
in January 1954 and the N.G.A.A. at Dallas in April
1954), are covered in this special Journal section.
1. Bubble caps. 2. Perforated trays. 3. Ragatz cycling
tray. 4. Shell Turbogrid. 5. Nutter float-valve bubble
tray. 6. Socony-Yacuum Uniflux tray. 7. Gulf perforated
tray .. 8. Koch Kaska de Flcxitray.
/

"
THE petroleum industry has used neers concluded that daia so obtained g. p.m. and the tOp vapor rnte is 16,000
bubble-cup trays for the past 34 are much too Hmit~d in· application. cu. ft. per minute. In addition, studies
years. This continuing popularity is due It was decided to continue Braun's re- can be made with a wide v::triety of·
to three things. Columns equipped with scnn;h progrnm wi!h a commercial~size :·hydrocarbon test systems.
bubble-cup uays can handle a wide . test unit cupnble of supplying a wide For studies on bubble-cap trays, the
variety of feed stocks. They give good range of data, column normally contains [hree ~ravs.
performance over wide ranges of liquid Fig. 1 shows a schematic diagram of These trays are fully equipped for.·~\\
and vapor loads. They have relatively the test unit. The principal pieces of kinds of measurements such as pres~
low maintenance cost. equipment are: (1) a 66~in. test column, sme drop, hydro.ulic gradient, t>ntrainw ·
The early use of the bubble-cup tray 18 ft. head seam to head seam, (2) a ment, liquid level in downpipe, and so
was in such applications as the separa~ vapor surge tank which is 7?. in. in on. In addition, the column i:; equipp~d
tion of crude·oil into gasoline, naphtha, diameter and some 12-ft. high, (3) a with windows on the sid(!s :.1nd on th~.~
1

kerosine, and so on. In these early ap· l50~hp. blower for circulating vapor, top to permit visual and photogr:1.phic
plications, very little in the W(\Y of (4) a 15-hp. pump for circulating liq- studies.
fractionation was required of the bub· uid, and (5) n liquid cooler, P erfonnancc Factors
ble-cap tra:. Comparatively little de·
ve!oprnent work was done on fraction .. Operation .•• Vapor passes from the In discussing the factors that in~
ation devices. In recent years, however, top of the column into the surge tank fluerlcc the per!·ormance of bubble-cap
rigid prodt\Ct specifications and rising where liquid droplets arc settled out. trays, we flnd it convenient to think
capital costs have required much de~ The vapors th~n ilow through an ori~ lnierms of a capacity graph, shown- in
ve!opmcnt work to be carrit.!d out, both fice run and back to the suction side Fig. 3. Vapor rate is p!o~ted as ordinate,
by the petroleum industry and by engi~ of the blower. The Houid is circubttd :1nd liquid mte as the abscissa. The
nt!ering companies. from the bottom of the" column, througlt region of satisfactory oper:-ttion ts
a liquid cooler, and to the top tr;y, shown as :1 crosshatched is ! a n d
Test Equipment The liquid cooler maintain'.l consti'.'1t~ bounded by :;ome six cond;tion') that
The reseacch that C. F. Braun & Co. temperature condit!ons in the column. limit satisf::tctory performance: ( 1) Put~
carried out from 1925 to 1950 was done Fig. 2 shows a picture oi the test unit sution, (2) dumping, (3) coning, (~~)
on pnrtinl·trny section~ in the laboratory ~:1kcn during the installation of test blowing, (5) entrainment, and (6) (!ood·
·and on commercial columns in the field. trays. in g.
As. a result of these ~tudics, Braun engi~ The unit is designed to operate over
Author h. with (', F. Bf;lun & Co., a \Vide r:'lngc of liquid and vapor nHes. Pulsation ••• This occurs when the
A\hnmbr<~, ColiC. Thl! maximum liqllid rr~te is some 550 amount of v:-~por passing through a tr~ty
 I •

~',<······· J}~'.;lfl:~:;;~ti,;i;;~
VAPOR
TEST SUI1GE
COLUMN TANK
GG~x 18' 72'tld

~IC\110

.:;:;<...---1J
CQO~ER

.\i\... . . {!J
!:>:·· ..•; .... .,.
~
l'Y.i""::'·'.'':':/r
',(1!!
l~n~~~
·~
. . . . ·:
.(j.·c.···:.•·:: ;·r'·{'-·.c·'.·;·::: •;·:.. · ..
.··.;-~:'·:·.::-·~~-(~:~:~ ·~::~i~~;-. ~·..::·~-~~).
· · · ·:
I'
.::'.
• .. .1~·
.,..
·' :·: ':' I I . .., .· .. -._· . . ;:,\.1,'
\''' '";.. ~. . 'I" ·.~.. ,· ....... :<~- :·~/~!
Fl~. 1-Schemntk Oow dia~rnm of C. F. nraun column '-.y ; ._.; .··.. ; ·:·.
cnp;u:Ity test unit, Equipme11l Included 66·1n, column and
blower and purnp for clrculalln~ n1por :utd !!quid.
(.' _;...," ·.",'·i r+ . .<~::···<~·:" :·::~.~., ~ ~--~·}
• • '.. :;: ::•.: :::. '1 .• •• : • •• ~;· ...
.. ' ..
-,,
··~ ;:.. .
.t,{"
..
... :'"~

.. . •''

1."

r iA~Jsrftt=~ · · .·'\· . '" ··:~

·1 ;~
, ~<~~.~7.;7:"
!:·~1~)~;):•.i_,,:. ·.
. ·.·. . .·: · .· . ~.;·.~~-·~. .:. _.~.·~~.:.·~
. 1 ., • ; •• :. JJ
Wl~~.:~:l4224f~iie:rd~~~~·:~j~:;:~~~~~~~~J~
••. ' u-

() Fig, :3--Capadty grnph shows limiting conditions for bub·

blc caps,
Fig. 2-Here ts the test unit on wh!c:h bubble·c:ap performance factors were de-
termined,

is so small that the caps pass vapor The major cause of dumping is the is often higher than that needed to
intermittently. The intermittent vapor kinetic energy in the liquid stream as prevent dumping. And since we o.rc
flow causes a pulsating condition on the it kaves the downpipe and enters the interested in gett!ng _good vapor.Jiquicf
tray and results in poor vapOr·liquid tray. \Vhen the energy in this stream is contacting, we design ior vnpor r:ues
contacting. A secondary effect is that, greater than the vapor driving force that bring all of the caps into action.
between surges of vapor through the across the caps, the liquid will dump \Vhen this condition exists, we say thnt
caps, liquid dumps through the cap to the tray below. These conclusions the tray is stable.
riser to the tray below. This short· are based on tests of several types of
circuiting of the liquid further de- tray layouts, using different numbers Coning ••• The third factor that limits
creases the tray efficiency. Tests show of cap rows, and a variety of hydro 4
satisfactory performance is conif:g.
that the point at which a bubble.cap cnrbon test systems. Coning takes place when the combi~~t·
tray }Ulsates is influenced by three It has been found tbat an inlet dam tion of weir height and the liqutd crest·
things: (1) cap size, (2) slot area, and ~lpstrcall\ of the first row of caps pre· over the weir does not adcauatelv seo.l
(3) propertks of the hydrocarbon sys- vents liquid from impinging on the the Cilp_ slots. If the, slots are not $e0.!eC.
tem. caps and reduces the amount of vapor properly, vapors can pass from theca:).)
required to prevent dumping. In many to the tray above without cont2<::i!~~
Dumping , , • This resullc; at relatively cases it is nctunlly imposstble to make liquid on the tray. This resu:ts in :1
high liquid loads when some of the up- n tray dump when using a properly marked reduction in trav dflde!~c''·
stream caps do not pass vapor. Liquid designed inlet dam. This is a new Tests indicate that a liqt1id seal of 0.5·
dumps through the risers of the caps appHc:uion of a rnther old device 1 as in. will prevent coning. in most •.:o.sc").
that :ue not operating~ and decr\!ascs shown in Fig. 4,
the tray e(ficiency. The loCation and height of the inlet mo,,·ing ... , Blowing results when :~e·
Heretofore, n widely accepted view dnm must be ~iven card;![ consideration amount of vapor jXl.ssing through t~e
has be<n thnt dumping is caused by· for n ~ivcn s;rvice. H th~se two factors caps is so great, that it Hterally te:~rs
hydraulic gradient. That is, when the are nO't c::>nsidered, the use of the dam hates in the liquiC on !~e tr:1y. T~i-:;

0·-
hydraulic gradient above a row of caps cnn cnuse n poor distribution of liquid, condition gives poor vo.por·liq\Jld con-
• exceeds the pressure drop through the and can dt!crease the capacity of the tact with a consequent reduction !n ~r~~'l
caps, the caps stop bubbling and liquid column. · efficiency. '
will dump through the risers to'the tray The blo\ving limit of a bubblc·c:lp
below. Hydraulic grodicnt has been Trav stnbilirv , •• The vaoor rate needed tray is influenced by four things: ~ 1)
found to have little effect on dumping. to cause all cnps on a 'tray to bubble size of the caps, {2) area of !he bub0:·:-

l 9
TY.lll.Y DX:S!Gl\T

cap slots, (3) skirt clearance between "Saving in capital cost ....
the cap teeth and the t"Y• and (4) the
properties of the hydrocarbon system. and improvement that can be
7 '·

r
BUBSLECA?S\ _ OUTlET made in the performance of ·._/ '
Entrainment • , , Several undesirable '\ · WEIR
effects are associated with the entrain- many existing columns are re·
~:~G!G!e~~~l
ment of liquid in a column: (I) off.
grade overhead product, (2) a reduction
suits of our research on bub-
in tray efficiency, and (3) possibly de· ble-cap trays."
crensed column capacity. · OOWNPIPE/
Entrainment of liquid is of two types:
(1) t\ relatively fine mist is carried to ' -'~--.L.J....,,- velocity. The factor ranges from 0.3
the tray above by superficial vapor· Fl~. 4-llluslrntlnv, uo;c o( lnld clnm In bub. to 0.9.
velocity, and (2) large droplets of liquid hh:·cnp co\unm~. This i'i n new ~ppltcntion
are jetted from on~ tray to the next. of :ln oiU tlcvicc, Results of Test Work
The entrainment due to superficial To illustrate the results of our work
velocity is usually less than one weight edge of such things as pressure drop, on bubble-cap trays, we have used a
per cent of the v:.\por, and is somewhat hydraulic gradient, liq'Jid rate, head loss typical cyclohexane fractionator. \_Ve
independent of tray spacing. Entrain· in the downpipc, and aeration factor. h'uve made computations of colvmn :nz~
m.enr due to a jetting action on the The at!ration factor is used to determine and humber of trays that would have
tray can amount to several weight per the operating level of the mixture in the been predicted over the past 6 yoars
cent of the vapor. This type of en· downpipe from a computed clear liquid according to the best information avail~
lrainment occ1,.1rs at high vapor rntes level. It takes into account the physical able at the time. F!g. 5 gives a com~
and is greatly inlluenced by tray spac· properties of the fluid mixture and its parison o( these computations.
ing, size and type of bubble caps, tray
layout, fluid properties, and liquid mtc. TABLE 1-SAMPL!l PROBLEMS Savings .•. Six years ago the column
Additional results of our work show Cns.e No. ( 2' J 4 would have had 125 trays and would
that air-water tests atone cannot be s~rvice ,-DepropMi7.er"""" ,...Absorber-, ho.ve been 144 in. in diameter. Tod:ty
used to predict entrainment for hydro· Liquid rak, we would recommend a 114·in. column
carbon systems. And in most cases en- g. p.m . ..... 800 50 $SO ll with 96 trnys for the sJme service and
trainment is not a limiting factor in Liquid density, operating conditions. As shmvn in Fi£.
column capacity. lbs. cu. ft. ll Jl 4l 4l 6, this reduction in column size results
Vapor rate, cis. OJ 29 18.l 26 in a savings of .$60,000 in terms of to· ,-··
\
Flooding , , , This is a phenomenon . Vapor density, . day's dollars-about one·third of the. '·.....;_/'
lb. per cu. ft. \.5 1.l 1..5 l.l
well known to all operators of com~ 275. column cost.
Pre'isurc, psig. 275 lOO lOO
mcrcial columns. It occurs when the The saving in capital cost of new
level of the liquid-foam mixture in the TABLE 2-COLUMN SELECTIONS columns is one result of our reseo.rc!J
downpipe builds up and overflows to Case No. 2 l 4 on bubble-cap trays. An equally im·
the tray above. Liquid continues to Diameter, in. , ..... n 60 66 . l4 portant result is the improvement th:u
build up in the column, resulting in Tray spncing, in, " 2·~ 24 . ~4 -::!4 can be made in the performance of
an inoperable condition. In the ex· Entrainment, lb. many existing columns. ·
treme, liquid flows out the top of the per \00 lb .. "",· 1.5 .<1 .. 4.l
Upper opr. limit'", Snmple Problems
column. per cent of design t125 !110 "i'115 :tl\5
The flood point of a column is Lower opr. !imh"', As part of the symposiuni, we were
reached when the liquid and vapor p<!r cent ot design ·l50 ~:!0 . §70 . (25 asked to size four bubble-cap columns
rates cause the downpipe to completely "'Liquid and vapor for conditions specified by C.~.G.A
tF\ooding
fill with a liquid-foam mixture. The §Troy stability The conditions are given in Table l,
tlood point is computed from a know\~ ~Pulsation and the solutions i;-1 Tnble 2.

a
150f-+--i--+-'--+--f--f-j "J 150 180
VI 144r-~'~---L-~-+--r-~~
"':::>0 144 170 "b
CJ VI
w CC I.,J
l5z 13a 1-+.2-:x- T
138 1160 X
,.,,
VI {_)
VI
'_):
-. 132 f--i--;- '~ ·: 132 \50 ·-~
>- _J

"'~w 12 6 1-,.]--i---T\"' u.
0:
~ 12.6 140
_J
0
.., 0
w D
:( 12 0 r--:--: 0:
'-" ;; 120f- >-
a 0)
::.!:
0 if>
0
II 4 f--'--+-1 :::> 1141 0
z
10 8 '-;-19.;•70
i
18,-L-1:;.;9';,50;;--'-;-;19""5"'2-'-~,.,-J
195•1
90 108
1948 1950 1952 1954
110

YEAR YEAR
Fl)!. 5--R~.rulls or column Ulam<'ter und number of Fie. 6--Suving'l rcsulfln~ from r<''cluccd cQ!umn size Jn-
trn)' compulallon~ for c)clohe:wnc frnc!ion:~tor. I 1>5·S for cyclohcxauc fr:Jel\on:.~tor.
 ·::.: Tnll:t DESW!\1

... '

- :by n. c. tee
pERFORATED plates have been used developing sicve-plnte design methods. check the effect of per cent hole area.
'. for many years in distillation and The experimental results, which have and various outlet weir setting-s Wt·r:e
absorption columns, but their popularity previously been reported, 1 indicated that used to investigate the effects ~of c:r¥
has never equaled that of bubble-plate properly designed sieve plates would ferent Hquid depths on the trays. Fig. 1
columns. lt is felt that the greater appli- have n wide range of stable, efficient is a plot of ·typical data obtained durin~
cation of bubble troys hc\S been due operation, and that the use of per~ the ntns, with the curves showin~ th·.!
largely to the lack of available design forated-plate distillation and absorption · actual limits of good operation ...
. data on perforated plates and to the columns would be desirable. Data by
ease o[ "rough design" of bubble-cap ocher authors~ a·~ have also stressed that Orifice cocffkicnts ••. These, ns de-
• trays, sieve plntcs can be designed to give termined from pressure-drop me:::!SU:-e-
. Sieve plates have been reported to a wide range of high-efficiency opt:ra-. ments on the 611! -ft.-<Iiametcr tr:1y w.:re
suffer from a narrow operating range tion. not nlways consis(ent, par!icuh\dy wi~h
w.hil~ bubble-cap trnys opMate well over. the V3 -in.-dio.meter holes, where er':"ors
n f~1r!y wide rnnge of rates. However,·~ · · How. ':rests Were Made were introduced by a nonuniformity oi
! .•.. ·. certain features o( perfornted plntes The experimentar work was per~ hole size and shape bt?causc o( h:-tr.d
such as \ow cost. high efficiency, high formed· on a 61h-ft.-diameter column· drilling and worn drills. Therciore, :1
capacity, and resistance to fouling, ma~~e w\th a sing!e sieve tmy operating on sepnuue pressure drop study w:1-s m:1.d~
their use attractive if reliable dc~ign in- the air-wat~r system. Manometers were inn 6~in.-diameter column. 1n this work
formation is i\VIIil:\blc. used for measuring pressure drop across the holes were caref\llly drilled to ex.Jct-
.. - •. Until the summer of 1949 the experi- the tray. Liquid depths on the tmy . ly !id, f-1, and Y.'l ~in. hole sizes; and v:>.ri·
ence o[ Celanese Corp. ol Americn had performance could be observed from ous tray spacings, plate thicknesses ~m<l
been limited to conventionnl bubble-cap · above the tray and also from below drilling pC~tterns were studied. The eo.l·
\ tmys. At that tim~! an experimental the tray .t~rough windows. Data were culated orifice coefficients for the c'.rv·
( "..;): Sl\\dy of perforated plates was initiated taken on St.1ch operating variables as tray pressure drop were essenda\ly co;1.
_,_for the purpose of: (t) making a vlsurd the weep point, the maximum vnpor stnnt at about 0.85 with th~ pl:W'·
comparison of the behavior of per- mtc ~ssible, an~ yress~~e drop. thickness, drilling pattern, ~nd ~o>~
forated plntes with bubble-cap trays. · ·-'diameter having no appreciable eff-:::c:~
(2) determining the operating range Hole _sizes ••• Runs were first made· on the coeffiCient over the ran q e
limitations of perforoted plates, and (J) on lh -in.-dlametcr holes. These were· studied.
.,' Author is wilh Ce!nnese Corp. of America, later drilled out to ,',-in. holes and
at Bishop, T(X. Pnpc:r pr<:St:f\ted at C.N.C ..-\. then to Y.\ ~in. holes. For each hole sizt! Pressure drOp The studv of the
Tray Syrnpo~ium, Lo5 Ang.ele'3, Junuary \%4. varying numbers of holes were used to presst~rc drop through dry tra>·s in :::e
. .•6-if1.·dinmeter .column was e:..?a:1d£'C to
·investigate the effect of liquid on th~'
orifice coefficient .. 1t was demons:r:'.~eC
thnt the ·drv ·orifice coefficients \'-ii:r-:
. ~'not apprecia'bly ::\ffected by ~~"1·: r:~s<:n~.~._­
·of the liquid above ·!he hoks ;:nC th:\!
for design purposes the toto.! pr·~S')~!l"'.?"
., ! ====r-ctrop\hrough :1 sieve trav CL"~.~:c :,._.
deH~rmined bv addino thC ind(v:cb~;
·· pfess.ure. drop.$ through !he d~y ~.~~y
... -~nd .through the C:.\!cUl.:tt~C C~~t:: "-~~·
liquid. · ·
---OVTL[T WCIR:HEIGHT =4 IN.--.. ··-- -----· : .
!2 1 000 - ~3/Jil-~N:_.HOLE.S -~
Acr.ttion factors ,._ .. In analvzin? !he
0 300 G. P. M, · ·-· ·--\vet-tray Pressure-drop d:.1ta ot::::~:1:c .:-r
0 200 -G.P.M.
100 .G,P.M.
··l .
the 6.5-ft.-dinmeter tray, it w::-; ::0:~~
thnt the observed pressure C:ro? ::--.~.y...:·;:
-· "~"- ·50 ~G~P.~i.
0 · the liquid-on.the· tmy wf':k~ \v;-,5 c~
4/XJO ti,ooo s ,ooo .....:.._. 1o,ooo . 12poo t4poo . _ term\nc_d by subtr:1.cting ~h·~ l~~:·-:::·,
STANDARD CVBIC PT. PER 'MINVTE. - drop 'from· :he observed to::!! ~):-~''ls·.::
' s:--..--,.:=~=-·c;.·"'··-7~·c:-;o-=or-=r=;=c=;C:...-..-~--.---'--'?~i==i~FT= dr.op.,_~£~_0.5_\n some cases grc~~c-:- ~h:1
.•' • .. () ~00 tpoo
LB. PCR HOUR PE:R SQ. FT. CG1..UMN AREA
t,soo -- · 2poo the calc\J!atcd c!e:.\r !iquh:i d.:>?~h I."
the tray and in other co«< '"" ~,·
' : Fl~. 1--iyplcn! pre.nure·drop <bin obt:lin~clln Celan('se Corp. of Antedcu,tests on.6lfz·f~.-d{t~n·,. than th~ c~dculo.ted vn.luc. t\•:r:).:>~)n :·:1
eler column .,..!th sin\:1~ ~·in"t.~ lwy on uir·w;ut•r ~yslt·ttl. . - · :tors, dcfin-~d .:\$ tht~ r:.\~:o l1:· ~:1~· ;
Tlll1Y DES!GJI.T
served pressure drop .through the liquid 0.r-onY TRAY PRf:SSVR£ DROP • !NCHCS WAH:R
'
~
to the calculated clear liquid head, were '
calculated from this data.
A correlation of this nerntion factor r
WATtR
ftAH. C.P,V..
2,000 3/16-lN. 1,9oo Jftt-m.
HOI..t:j HO.l.£5 :J·
as a function of liquid depth on the •o 0 0

tray is shown in Fig. 7., which shows

100 h •
u " I
o. •I- 0

~
200
the general trend of decreasing ncra· ¢, -> a
tion factors with increasing liquid. '"'
I I
~
dePths. This correlation gives nn csti· ' '

mate of the pressure drop through the

liquid on sieve trays as a function of
0
(>
.• Q l
weir and liquid rate.
. In the range of outlet weir heights t}o
. " ~~
~
r
I
normally used (1 to 2 in.) the aeration •
factor Is less than 1.0. Since these data •
are for the system air.. water and the
0 .2 •
I
effect of different liquid and vapor
densities on the aeration fnctor is not }-0 IN, WtiR ..----.4 }-tiN. W(IR-1 }-4 IN. W(IR ---i
I
known, an aeration factor of 1.0 is }---i IN, 'UCIR--f )
,.
normally assumed for ct;sign purposes
so that the actual pressure drop through
the liquid should be on the safe, or 0
0 1.0 2.0 3.0 40
I '
~·
• l
<.0

low, side. Clt:AR LJQVIO Dt:PTH 1 INCHES

Fig. 3--Correlat/on of minimum allowable Yapor rate on the e:'(perimcnlal sieve traY for air·
Wide range , , • The data in Fig. 1 pro- wnter system.
vides an indication o( the wide range of
stable tray operation which was ob· to 36 per cent. Data on plant columns this amount of liquid 'without weeping.
tuined in the sicvc·platc study. The show opcrnting ranges equally wide or· This corrclntion doe'i noc inc!w.!e )UC'h
vapor rnte at the upper limit of good wider. factors as surface tension, k\ne!ic en-
operation in this series of runs was ergy of the vapor, and tray diamet"r,
shown to be abo\lt 14,000 standard Minimum vnpor rnte •• , A simple cor· which quite possibly have some effect
cubic feet per minute, The lower limit relation of the minimum allowable on how far the vapor rate can be re·
on the vapor rate, the point at which , vapor rate on the experimental sieve duced. However, operating data from
liquid began to weep through the holes, plate for the air-water system is shown several nerforated-plate columns ind~~
was found to be about 5,000 s.c.l.m. in Fig. 3 in which the calculated liquid . cnte that' the correlation is a f::1ir meth·
This corresponds to o. stable operating depth on the tray was plotted against· od of estimating the minimum vapor
range of 100 per cent vapor rate down the pressure drop necessary to support rate for design purposes. Addicion::tl ~x­
perimental data arc. needed for 1 more·
r - AERATION FACTOR generalized correlation method whic:1
•4 will permit more nccurate c::tlcubtions
MR A.ATI: 1.5,760 12,000 7,900
s.c.r, M, 1/U•IN. :1/!Q·IN. 3/18-iN. of the minimum rates.
HOl.£5 HOl.ES HOlES

+
•
4,000
7,000
10,000
IJ,OOO
0
•0
()
..
""
.,
"•
•
~
, 'What ·hole size? ••• In the selection
o( the hole size to be u1ed, it wns
concluded thnt ~-in. hoks wer~ zen·
erally the mo't desirable for com;,,.
HS 1000
• • '

1.0
f\ !•
• \._"!'a
I
I
~ ~ J.
0. •
0

'·R~~-z, I
~~~
• J. ' '
~A ~· /?'a ~ f.\
• \5J I
o.
"2l ~ .
o} 0 I •
~J!?-li~2:~o
*() i ,\, ~-,._
• 1JE.l .N'fl +
-.;·.:...o~'-~-----1
04
,
}-OtN, WEIR+IIN. WCtR+2 JN. WCIR1
I ••~4tN.Iwe:m-l .\
I ,
0. 2
0 ! 2 J 4
I
5
CLEAR LIQUID DEPTH, INCHES ' ( - - - - - - - HT. 0 I N . - - - - - - - - - - - - - ' ,

f'lsr. l~Correlotion of n~r::allon hlo.:lor ~·i function of liquid d~!lth on Fig, 4--R~o.:tssed Inlet stat pun:\ nre one o( the tl~·,l~-:11 fculurc~ usct!
!ray. This ihU t'~tlmute of pn."iSUre Jrop lhrnu).:h liquid on sl~·"e truys. lo mlnlndz~ prenuuure \Hcpln~ o( lltJllld on inll'! ~iJe oi !ttl}'.

22
 u

Fig, 5--Mort nrw t:olumns for Celnncse Corp, 01re now sleve.trny Fig, 6--Another Celnuese instnllntlon. More thnn 10 periorah:d tray
tn~ca.ll::al!on'i, H~re l'i on~ e.'(nulple of su<:b an ln.s!n!lntion. columns have be>:n lnst~lled Uy this eom\l<lU>' in th<' pa~t J ~·e:tt\,

() cia\ use. The smaller holes appeared to

give better vapor·liq\lid contacting, with
Some Notes on Design
'The design of a sieve-plate column
ber of holes to be used depends on
the requirements o( the system. If th•:
the i'l·in. holes alniost as good as the fotlows the same line as the design of design vapor rate is very high, say
V• -in. holes. 'The use of V; -in. holes a bubble-plate column. 'The co~lumn approaching the ·limiting rate b;.1sed on
with heavy-gage plate is normally not diameter, number. of trays, tray spac- cross-sect\onat tower area tb.e selected.
1

practicnl when punched holes arc used ing~ and downcomer dimensions are
design pressure drop can be rrbtive:)'
because of the mechanical limitations hi~h in order to allow the tower to Ce
arrived at in the same manner as for
of the hole-punching process. 'The v, -in. bubble travs. oPerated at much lower vapor rat~-5 :t
holes, although being the cheapest to· Allowari'ce can be made for the some· desired. Conversely, if the design \':t?Or
fabricate and the most resistant to foul· rate is considerably less than the ultl-
what higher capacity and higher effi-
ing, appeared to give poorer vapor- mat~ capacity of the tower, then the
ciency of the perforated plate column
liquid contacting due to wider sprteing if sufficient data are available for the selected design pressure drop m:.?ht \::lc
of the holes and to a tendency of the sizing of the column this closely. In de- taken much nearer the weeo ?oint w
vapor to jet through the liquid. signing the proper number of holes for permit higher vr~por rates ';o · bG oO-
the tray, the m:ljor consideration is that tained without a res1..1lting ~xc~ss~vc
}'Ugh efficiency ••• Visual observations
of calculating pressure drop through pressure drop. For norrnai opera~:on, i~
of the nir·water contacting obtained
the dry tray and estimating the desired is cuo;;tomary to Jssume :1. d~s!gn pres-
with the perforated plate indicated that sure drop which witl permit ~l:c t0we~
the tray efficiency should be higher range of pressure drop for the design
conditions. to be op~ratcd both above and be!ow
than for a bubble-cap tray. Soon after the design rate.
the completion of tPe experimental Pr~ssurc drop ••• This is calculated
work, a 60-trny by 6.0 ft. coh.lmn was by a suitable orifice equation using an Vapor rntc changes •••. For cases \:1
designed nnd installed at the cl~lanese orifice cocfticient of 0.35 for 1/8 to which the vapor rate chan~cs cc~s!J­
Chemcel plant. Performance checks ~/.~·in. thick plate. 'This is simplified to
erab\y from the top of the t;\ver ~o •.:~·:
confirmed the expected high capacity a special equation for i'.r-in. holes, in base, it is necessary to cak:..d:1te 1:.'.:
and wide operating range at high efii~ which the vapor flow ra~e per hole is h o 1 e requirement~· at scv~r:1.l no~:l:,;
cicncy. Comparison of over-all tray eifi· t:xpressed as a function of gas gravity, through. the tower. In c::ts.es of thi's '>or:
ciency with that of a bubble-cap column temperature, pressure, and pressure drop where the hole reouircmcots Y'J~Y
.
in th!! same st:rvice showed the sieve in inches of wntcr. T0 obtain the com- throu~.h the tower, it ';s r;ustom:J.r:? :0
'l
plate column to have a flatter c((iciency po!\ent presmre drop through the liquid, drill all trays with th..:: mnxlrnum n•Jrn-
"n aeration factor o( l.O is normally . b~r of ho!Cs and use blank!ng strips
_'__
curve and one which exceeded that
of the bubble-cap column by 5 to 15 assumed for outlet weirs of about 2.0 tack-welded to the tr:ws to obt~in •.;-,r:
per cent. The superior plate efficiency in. in h ... ight. desired number of hole's in ·:~c:1 se-:t:ol:
of sieve plates ha~ also been reported of the tower.
by Jones and Pyle,> Numh<'r of ho1es • , • Selection. of num· Carrying this a step fur!her, it Is ;:ws·
T!l.I!Y DES!Gl\1
sible to perforate all trays with a cer· down, any chnnges in the orifice coeffi. column, even if no reduction in column
tain standard minimum hole spacing in cient arc minimized. Actual measured size is made, Estimates of the co·Hs of
cases where there is a chance that the pressure drops in columns with this type the plates alone. show that sieve plates
tower service might be changed. to re~ tray have chechd fairly closely the cost only 25 to SO per cent as much
quire additional holes at some later values calculated from orifice coeffi~ as bubble-<:ap trays. \Vith the sieve~tr:ty
date. A common hole arrangement in cients for smooth drilled holes. installation, greater capaci(y and efii~
cases of this type is ~1.r-in. holes on a Installation cicncy are realized, nnd the range of
·~~·tn. triangular pitch where l8·in. tray Accompanying photographs s h.o w operation is satisfactorily wide for nor~
spaCing is used. The usc of blanking typical sieve~plate installation in Chern .. ma! operation. Finaily, maintenance of
strips is advantageous in permitting tht: eel towers. the columns are facilitated bccau$e of
use of the optimum number of holes for Most new columns for Celanese Corp. the greater ease of work. inside the
each tower service to maintain high tray are now sieve-tray jnstnllations. More columns.
efficiencies in all cases, thnn 70 sicvc-tr"y columns have been References
installed in the past 3 yenrs, the largest 1. Mnyfield, F. D., Church, W. L., Creen,
Premature '''ccplng • , • A feature of being one of ll-ft. diameter. Where n A. C., Lee, D. C., nnd Rasmussen, R. W.,
sieve-tray operation which was noted in few replacement trays are needed in lnd. Eng. Chern.; 44, 22)3 (1952).
the visual study of the 6.5-ft.-diameter corrosive sections ot bubblc·cap coJ .. 2. Jones, J. 8,, and Pyk, C., "Relative
tray was the premature weeping of lk1· umns, sieve plates nre commonly used Performance of Sieve and Dubble·C;J.O !1 !ntes,"
annual meeting of the :\.I.Ch.E., St. 'Louis,
uid on the inlet side of the tray due to as the replacements. The reason for the (December 1953).
the impact head developed by liquid s!!lecrion of sieve tmys is to make use ), Mnrsh, H. R., Proc, twenty~fifth annu::~l
as it overflowed the inlet wdr. This was . of their advantages when proper design fall meeting, C.N.O.A. '25, S·l (:--fovemba
particularly true at high liquid rates. in 1950).
methods "'e "va;J"ble. ·f. Ragatz, E. 0., Proc. twenty fifth annual
4

which cases the tray began weeping at The first cost of a sievc·tray column fall meeting, C.~.G.r\., 25, 31 (November
relatively high vapor landings, although is substantially les< than a bubble·cnp t950).
in genernl the wcepage was not severe
and would probably h"ve had no serious
effect on tray efficiency.,
Two design features have been in·
corporntcd in prncticc to help reduce
~t)il"
11 ITl~fA'
~ - <-
~f'
1 •ll
•
OO<l
or eliminate this problem. The first is
the usc o( recessed in!et seal pans as
shown in Fig. 4. In this type design
the liquid flowing out of the down-
comer has less distance to fall, and
therefore the tendency of the liquid to
be driven through the holes on the tray
at that point is reduced. Another ad~
vantnge of recessed seal pans is .that it IN the majority of t.oday's commercial being laid on increasing the efficiency
provides n. deeper seal for do\vncomer absorption and fractionating columns, oE today's high·pressure, high~rccovcry
b"ckup. In addition to the use of re- the desired component interchange be~ absorbers.
cessed sea\ pans the trays should nor~ tween. vapors rind liquid is attained
Low Absorption Efficiencies
mally not be perforated closer than 2 by contacting an upflowing discontin~
to 4 in. from the inlet weir. In cnses uous vapor phase with a crossflowing In the late 20's· when absorbers were
of high liquid loadings, this undrilled continuous liquid phase on n series of largely operated at pressures of the or·
space could be made even wider to superimposed, horizontally positioned, der of ..w.so psig. for "but:1nes" rec-oY·
help prevent impact-type wcep"ge. ntravs"' within the column shell. The cries of only some 75 per cent, the rna:
1n the faVrication of new towers, more intimate the contact attained be- jority of such units exhibited re::tson~
trays are deiiberntely tilled a slight tween upflowing va?ors and each par- ably good efficiencies when employing
amount toward the outlet weir (about !icle of crossflowing liquid on each as (ew ns 16 travs. However, such effi.
lfa to V.1 in.) to minimize hydraulic trJ.y, the more efficient is the column ciencies were ah~avs less than those ~t·
gradient effects and to insure thrH work~ operation. tained with simih1r traY desir:ns em~
man-ship tolerances in fabrication do not Under these circumstances, the first ploycd in the plant stabflizers. -
result fn occasional ·trays being tilted objective .of the tray designer· is to As operating pressures h:.we !ncreased
towards the inlet weir. 1 brea:< the upflowing vapor stream into and recovery reuuirements 1aised :o
extremely fine bubbles as it enters the their present ·high levels, employment oi
Punched holes , • , The use of punched column trny, thus providing a maximum as many as 30 to 40 tr:lys finds the
holes has led to one possible inac· oi vapor surf:\ce for the desired vapor- conventionally trayed absorber still ~e 4
curacy in the design methods although liquid interchange contact. If, however, riously deficient in absorpt~on cffectiw·
opero11ing data to dnte have shown no the trav still e.x:hibits an undesirable ' ness. This ddiciencv imnoses an r.:oco-
appreciable errors from this source. low efficiency once this vapor~disper~ nomical!y significant ndc!'ed burden on
The perforating process tends to shear sian objertivc has been properly at~ 1he modern gasoline o!ant's initi:1l con·
or tear the materinl from the tray tnined, then other phases of the tray's struction a~d sub:~qu\!nt opc~ating
after the punch has penetrated about fluid dynamics obviously coli for study · costs, and hence deserVt!S seriou'i \'On·
halfway through. This gives a clcnn and improvement. sidcration by the indUstry.
sharp hole ~nly part of the way through This art!c!e deals whh such traY im .. ·
the tray nr.J n rather ragged, enlnr~cd proverncnt, with particular emPhasis Whnt is the problem? . . • A several-
hole the remnimlcr of the wny. By A\HhM ~~ mannr,o:r of F.Uw. G. Rngatz Co., year study involving some 350 tests o-n
plncing the shnrp edges of the holes Pro~:e~s G.ngint>cr'>, S:~n ~lurino, Crt!il. 100 differ('nt commerci~ll columm r~tn£·
 ( ... ) I
>or------1-------v·~·~·~'''<2'"~-

I( ~
ItO lo!OI.. Wf, l(AN 011.
100• 'I RICH OIL
tffl(l(Hl' C011VC!HIOHAI.. ASSORI!i(iS /_
v
"~~-~ /
~~
"0
"' /
~~----r-------+-------4--------+-------+~~----:--------+-------"
20

1/

..
'I
10

----1
~-..:--~~..I
:7 00'-------,-,-,------,~,-,------~.,~,~~--~.;,.~,~~~.',·,-,------,-,~,-,------,,~,-,------~

Fig, !--This demonstrntu lneren.~c In oll/gn.f ratios In going from enrller low-prcs- ·• · Fig. 1-R..'l.ptz recycling ·tr;ty construetion.
15 p~t' Ct'llt 11 butanes" r~I.'OYl'r)' to \ltt.,'.~l.:'ttHlny hi~h·pres~un: hl~;;h·propune
\11tC
rCt:'OYCrl~!. ~· .,., -. ........,.............
~

ing in duty from 60-75 per cent "bu- 11

effective vapor dispersion" must be at umns. In this fact, it is believed, lie~

·o. tanes" recovery -at 40 psig. up to 80-95 ..

per cent propane recovery at 850-1,500
psig. (coupled with a substantial num-
fuult... . .. .... . . .... ,
t, In analyzing the fluid dynamics of
.. the. '.'k~y~' .. to efficient absorption·tr:\y
design for today's h\gh·pressure, hi3h.·
current absorber operations versus the recovery operations .
.. ber of "dcvelopment 11 tests on several older lower-recovery operations, we find
different experimentallray designs) lead one striking difference, namely, that Increased oil·gas ratios •.• Fig. t clear-
to the following conclusions regarding the f\owing-volume ratio of oil to gas ly demonstrates the increase in "!~ow­
today's absorber-efficiency problem: is muth higher in the present-day col- ing" oil/ ga~ ratios whicb. has occurred
., l. Component interchange on the
. ·absorber tray is the most difficult en-
countered in the natural-gasoline recov- 4

cry operation. This is due to (a) ex-

tremely low concentrarions of the indi~
vidual components in the leaner g:1s
.. slreams, (b) wide "spreads" between
molecular weights of the vapors and
liquids involved, and (G) relatively low-
.4tempernture levels of the operation·.
2, Despite these handicaps, many of
-· the earlier low-pressure absorbers ex-
;_>ibitcd relatively good efficiencies.
Among the more efficient o( these
.. ear!ier units were many perforated--
plate columns-indicating full ability of ·
:·cn·properly designed perforated plate to
cficctivcly disp~rse the upf\owing va-
pors in !he croSiflowing tray liquid.
.i 3. The mechnnics o( duplicating on
a high-pressure perforated-plate tray the
-:: degree of vapor dispersion attained on
the older low-pressure trays is well un-
derstood. Yet when this degree of dis-
('J"
\. . persion is so duplicated a simple high-
pressure perforated-plate tray fails 10
duplicate the earlier low-pressUre tray
efficiencies by a large margin. This in· F!r.. 3-T'erformnnce d:ltn In e.xtrndloo of propnoc, lsobutane, and norm:ll butnn·! hy 1~10
dkatcs that somt! fnctor Qlht:r thnn 70ll-psl. nb:o.otb(f:i, one equipped with n:qclln~ und th~ ot/ur )l!l.h bubO!e·o:np lf:l}"S.
4
 J
~r
COlULtl
"'0 I.(Aio! Oil.. ·· ...)
tU&Sl(~O.P
,.. ,.
/I((YCliHG

,.
- ~· '
OtlU(f(ll
!jQ, flfJ.tS
... u ..~l(li GAS " '
, ~I '

lT''
{l,OOO ~l.~O

'•
I__J~
VA((Y[;!Ii<i lRJ.Y
I(~ J.o)~QI!PliO.i
'
• . \t.I ~~·l '

F
" ~•
I ..........
" '
... ,,,,,~ .;

. ~
IC4 J.tt$0~Pf101f

\1 _,./"" .
'
•
~
~
I ~l •'
t?f
t1
" 0

r~; --- ,..P A(C'I'CI.ING VIA'!'

'
• •
• ;""' c, J.~~OAP1101l
'
10
- I , 0 '
-..tor ___ . . ~-r~------ SUfJ8U•U,, r11u,......,.....-"'r_.. --~l ' •J;
~ ~.
_,....-

__.T _____+ . ~r on. ~;I

------r-'~· ::~:--"-r
ltl(H
0 v": !--' RI(H (iA$

•. , "" . 'I
-to Of WU,CA~AC!TY@ t(l~tt

,,
CiAI. I, O/VCf ltG
•
OV!II 1 T(J.Y ___,

ltl(H 01\, Off TIU:t-'"f

'
rT
70
" '" 10
" " '" "'
Fir.. 4-Crossplotttd d;~la showtns: SU!lerlor perfonu:mce o( re<:yclln'l trny, Flg. 5-Tray-terupernlure g-rndient £or rtcyde-tray absorber,

in going from earlier 75 per cent "bu· gree in the "recycling" tray design 11- · cent of the tray are:\ between the down 4

tnnC's" recoveries nt some 50 psin., to \ustrated in Fig. 2. comer and overflow weir, according to
present-day high-propane recoveries at the trnffic requirements of the particu-
Tray constructiOn ; , • The active sec~
pressures up to. 1,500 psia. nnd higher. lar tray in question.
t\ssuming an equally effective breakup tion of thi"S trny consists of several
of 100 flowing cubic feet of gas, it is appropriately perforated horizontal 14- How tray work.'> , •• \Vith this tray ar~
16·gage plates over which are posi rungement, the dow_n-Oo\ving liqu{d
4
·,,,./
obvious from Fig, l why a 1,000-psig,
tioned a series of "recycling baf(les." traverses alte_rnate "strip" zones of _gus
opcmtion for SO per cent propane re~
covcry would constitute n much more Liquid downcomers with overflow weirs contact and froth coalescence rts it
arc employed on i'\lterfiate sides of the
difficult gas~liquid contacting problem flows across the tray from dmvncomer
than a ·50-psig, operation for 75 per
trays as in a conventional bubble cap to ·overflow weir. Through coalescence
4

structure. The recycling bn(flcs are in- at the recycle baffles, a substantia\ por~
cent butanes recovery.
terposed in several pnmllel rows trans- .· tion of the liquid leaving a gns-contac; 4

The "Recycling" Trny verse to the path of the crossflowing ing zone is recycled back into 1he bo~~
liquid, with t'\lternate strips of "free" tom of said zone, ·while anoth~r sma::e:
Shifting o( the improvement emphasis
pt:!rforated plate being left between the portion flows forward into the 'oottom
from vapors to liquid indicated that, in
nddition to making ccrtnin that all por~
baffle rows, The baffles rest directly of the adjacent downstream contacting
lions o( the vapor stream are equally on the perfora!ed plates, thereby blank· zone. Through such rep"ated liqu\t:
well dispersed and held in !he contact· ing off <"11 underlying perforations. "trapping," coalescence, and r.:cycl!n~.
ing zone for equal periods of time, the Perforated plmes and ba((\cs are the nfon:mentioned liqui<l-mo.nipu\;.~tion
sized to go through convention[~.\ 18-20~
modern absorber-tray design should: requirements for an efficient high-re-
in. manholes, and supported within the covery absorption tray are rea\iv!d.
L Make certain !hat no portion of column on angle 10 r'rids" to which they
the crossflowing liquid can "shoot" are removabti atta~hed. Liquid Jown- As with perforated-plate trayc; g('n~
across the tray through. the upper section comers, liquid receiving plates, nnd per- emily, <llt portions of the recycle-tr~y
ol the contacting zont\ while other par~ . forated~plate and recycle~baffle sup~ gas strenm are equaily well disp~rs~C:.
tions arc trapped in eddy currents on porting grids, are normally welded per- and held in the cont:.lcting zones ~0:­
!he tray and hdd therein for dispropor- manently in plr~ce within the column substantiaHv eouut oortion;) of t!m~.
tionate ·pt.!dods of time. shell. Contact wilh th'e cro~stlow\n'."!: tiqu\d !.:;
2. Insure repeated coalescence and increased in cffectivcne:>s in~ the case
reagitation of the liquid stream as it Duffles , , , Each baffle consists o: an of the recyc\lng tr::~.v, howevoc:r, bv the
crosses the tray, looking to more readily open-top "box'' structure. t\ scric:, of pan1Hel nOw Of v~·por and EL;ui(\ ct~
bdnging an pnrticlcs ')f the liquid holes tangent to the box-bottom arc fccte<.l by the ~bove noted repe:1ted ~n~
i stream into intimate contact with n provided fn the downstn..•nm side o( th¢ traduction of coalcsc~d liquid strc:-t:ns
I' I componcr.t- exchanging vapor surface. box, while an open "slot" \s provided into the bottoms of the cor.~acting ~~L'm~s.
3. Provide sufficient residence time in lhe upstream side o( the box im-
on the tmy to insure th\\t each particle medialely above the box boltom. The Tray rcrformance
of crossflowing liquids is brought into sitles o( n·,._. br~ff\c nrc normJlly the 'The increased effickncie$ attain:'l':)!e
e!fcctivc contact with its r\pproprinte snme height ns the overflow weir, and with this. new tray structure a.re dL'mor:-
share of ~as, l..'tf sufiicknt cxt~nt to provide the nee~ stratcd on Figs. 3 and 4.
These -desirable tray-efficiency dy- ~s~nry liquid residence time on the Fit:· J prCscnts a stra!g,ht·Hnc-tl'>.~l~t
. -- ·'~-- •• 1 : .... 1-.:nh rl,o 4
'"'v. The bnff:cs h!nnk off ·10·60 per analysis of o. serit.~s of s!de-'')y-s\d~ t('·W.
 'fY!IiY DT.SXGl\1
~run on two 700-lb, absorbers equipped extra-high lndivirlunl tray efficiency. that for an equal such cooling one
respectively with recycling and well- lt is generally assumed t h a t a would have to have an equal gas-liquid
designed bubble-cap trnys. present-day commercial column having contact) and hence should attain an
() Fig. 4 is a crossplot of data from
Fig, 3, and presents isobutane ·and pro~
24 or more trays approximates an
14
iafinite tray" unit for component ab~
equal propane recovery. Such is not
the case, however, since it is much
pane yi~ld:3 vs. column loads at se~ sorptions below GO per cent. Under easier to attain 11 thennul" cquilibri1Jm
lccted A 8 's <~bracketed" by the test these circumstances, all such columns on a column tray than "component"
operations.of F_ig., 3, ........ ~-·-·- __ should exhibit the same recoveries for .. equilibrium-panicularly when dealing
the same absorption factors in the range with extra 1arg•! flowing oil-to-gas ratios.
4

Test. conditions , , , Both test columns b~low 0.60. However, this" assumption \Yhile development of the recycling
were equipped with 24 trays located on neglects the effect of the relatively large tray has laid particular emphasi< on the
20-in. centers. The .recycling-tray col- temperature gr<.~dient developed across absorption operation~ it should be cor-
umn was 72-in. in diameter and the the bottom few tr\\ys of most non- responding effective wherever cxua-
bubb\e-cap column 78-in. in diameter. intercooled high-pressure absorbers. high tray ,efficiencies are of process im-
N~. intercooling' was employed, and ......... · :· · ·•· ·-- portance.· · .. · ·
Jean-oil to rich·gas ratios were held be- Temperature gradient , , , Fig. 5 pre· Comparisons such as presented in this
tween 10 and 11 gal. of lean oil per sents the tray-temperature gradient for paper should obviously avoid specific
1\Lc.f. of rich gas. Test loads ranged run .11 F" of the recycle:t~ay_ absorber. __ i_dentification. However 1 we do W;tnt
from· some 40 to 98 per cent of maxi~ The large drop across the bottom three to go on record that the data presented
mum for each column, with the maxi~ or four trays (& o F. across the bottom a~e the re$Ult of fullest cooper:ttion of
mum rich-gas loads established at ap- tray alone) applied to an extra-high several companies ami a number of
proximately 52,500 M.c.f. per day for ·efficiency tray fully explains the marked --individuals for·which we wish·~o extend
the recycling~tray column, and 42,000 incre~se in propane recovery attained our keenest appreciation.
M.c.f. per day for the bubble-cap col- with the recycling-tray column. The various special features of the
.... cunn: Jv!axim~nl ~apacities. f~r ·both ,. ·-.This .. temperature drop_is. the .re.sult____:..~\!l;>Jeclrecycli.n&-Jray_are.fully.. covered
colum~s we~c re~uced substanttally. by of heat interchange with the incoming by United States and foreign patent
_excepttonal troth1ness of the absorpl!on ... colder·.rich·gas. lt mighLbe. assumed-:.applications:·· ... ,.,,.c:• , ........ .
oil employed,
Trny comparisons , , , While the su-
periority of the recycling-tray column
-
;..., ..· ·.

d~..•
is obvious from the plotted data of
Fig. 3, accurate comparison of the ef-
... ; :.-... ~'
'V'o
fectiveness of the two columns is best
obtained from the crossplotted daL1 of
I?'• · ~ · ~ ~· ' ~.,, •
Fig. 4:
!. An effective operating range from
at least 30 per cent to full load is ., "
r?' , •o
;;yrrmm!iG (.(l)~§'aM\\:l.mJl'l,· l!'ila1!l{l \~.i:lr·~ J~l ~ l
=
If ,
' 'l
v
clearly indicated for the recycling tray a· ~ \1":\ r,: ·· • ~ I'!'\ • ·
column, while it is understood that the
effeCtiVeneSS Of the bubbie'·CUp CO\Umn n.
Bt3W:;ur~s~.~WA
_,. .._j, .__. ~ <..ll
'J
wl'©rGJ!
, tt. ~~li'®lrn©rrrllf~1tr~\\:.tvr:'Yt;;~(!~"}fe
·ay ' "' .. ~""·' ~ l .;.._.' A
:J . ~ • •
<l
4

breaks in the region of 45 per cent o(__ .. _

full load. ··· ·r·;-.7"", . hy J . .!'!.. Sruna:rnicgo · · · ·-.
2. For the particular tray traffic in
question, the recycling-tray column has ·· ·
' ' TURBOGR!D' trays (U. s. and for- . turn! support framework for the bars,
a 45 per cent h1gher capnclty per gross
"\Uarc . foot of column cross-section - eign patents applied for)· are the the bottom side of the partly completed
than the bubble-cap column. . result of an extensi'lc research program · tray, and the top side of the f!nish«i
3. Whi!' the efficiency of the re- on distillation equipment within the·:. tray arc shown.
cycle •tmy column increases substan.. Shell companies. They are simple in . ·Fig; 2' shows the stamped .. plate-type
tially in going from 30 per cent to full ·~construction and have high c~.pacity, · Turbogrid tray;· the tray consiscs of
load, its 30 per cent loading efficiency low pressure drop, nnd low liquid hold-· standard·traf$ections with sketch plat~'.i
is only slightly lower than the maximum up, while .providing about' the. same at the tray edge~ In. Ll:le picture, ooc
eft~ciency aw:dned wiLh. the bubble-cap separation, m·e as u r e d as theoi'etiCal:-. of the standard· sections.. is. r~;..·mowd;
column. plates pei' "fOot of c'Olumn, as convcn- the area covered by a sta11dard section
4. In ~oing from 50 per.cent to lOQ ··tional~tray · tyFcs. Their simplicity re~ · serves as. a manway in the trav. The
per cent loading, the recycling-tray cot~ -stilts· in lo\V ·cOst" and ease of maintc-. ·trays are generally. installed \·vi thou~
umn showed highN recoveries of some nance. - .. :--downcomers. for liquid flow. Oown-
7 per cent for isobutanc.\ and from 3.5 The trays consist of a f1ut grid of. · comers may be used if desired, \:Jut ue
to 8.5 per cent for propane, compared parallel slols, the grid extending over usually not recommended s;ncc ~hey
with similar recoveries from correspond- the entire cross-section of a column. reduce the effective tray area, and hence
ing per7enta,gc loadings of thc.l?ubbl¢· The slots can be either .stamped per-·: the tray capacity.-
cap column. foratio~s in a flat metal plate or may· :. -·' -c:.
consist of the spaces between parallel- Ope"rntion: ,· ,. , In normal operatlon of
Incre.iSe<l" Piopane' Recovery bars. . Turbogrid trays, the parallel slots in
Of these four efficiency comparisons~ Fig. 1 illustrates a typical. Turbo grid the trays serve ·as the passages for bo~h
that o' a markedly increased propane tray of the paralkl-bar t~pe. The struc- ·liquid and vapor Lorough Loe troy. The
recove1y in the region of 55 p~r cent Aulhor is with Shell Dev¢1opment. Co., flow of both liquid and vaoor OCC''JfS
absorption is the most indicative of nn Emeryville, C::~lif. as a cycling or 'intermittent 'jet nction,

27
 ........ '". "' ., ' '"," .- '. "• :., .. • ':· ·";1\ ·" ,_, .. ' ... ,. '.,
.., , ''·"p':v;.~ .. ;:•· "''""'· .•·. ')·~ .,_•., ... ..... y-:,~:,:c,J.:.:.,,I
I

;. '"'1<>·'"·;,'"' ...... ·. , " ..... • . ··m . , , e· i'"

T!lll.Y. :OESICJ\1

.·.........1'

Struclurnl frn.me"Wotk (or grld h'ay, Frnme-work cut Into four llottorn side of portly completed h'ny, Weldin~ and trimming of bnrs and
w.:tlom In final rmemh\y, .'iCclioning of tray incomplete. Stomped-p\ulc type.

portnnce of column height and diame-

ter. Tray spacings of 12 to 13 in. nor-
mally are used in Turbogrid installn-
tions, with 18-in. tray spacing being
preferred in large column'i to improv·~
tray accessibility.
How do Turbogrid Trnys Perfom1?
The following data on five Turbo-
grid-tray columns in various hydrocar-
bon distillation services mustrate the
pbnt~scale performance of Turbogricis.
Four of these columns ·are norm<"l~ in~
stallntions of Turbogrid trnvs, and one
is an installation with down'comers. All
of the columns have equaled or be!terec!.
design performance, and they {fave ex-
ceeded bubble-cap-tray capacily and
efficiency for .cases where compnrrnivc-
clata are available.

Column No. 1-various services

This is 5 ft. in diameter with 40 ~rid
trays of the stamped-plate typ·~ .. in-
Top ~ide o( t'l,lll\Pleted trny,
stalled at 12-in. tray sp~1cing. The col-
Fl2. 1-All three views on thl~ puge are Cor par:lllel-bnr type o[ Turhogrld, umn has 19 rectifying and 21 strip-
ping trays. This inst:1llat\on wa<> dl'-
uniformly distributed 'with respect to css requirements are mainly the open- signed to operate in a nur.10er of dlf~ ·
both phase flows over the entire grid slot area in the tr;w, the slot width, ferent services, and this f!exi.biltty in
pattern. 'The interference of the: coun- and tray spacing. the open area is performance has been sub:>tJ.ntiated in
tercurrent flow o( liquid :tnd vapor selected to provide the optimum eco- test runs. "The Turbogr~d trays replaced
through the tray results in n buildup nomic balance bl.!tWeen tray capacity bubble-cap trays which were !im~ted :n
of liquid on !he !ray, !he liquid holdup and efficiency. A lMgc open aren gives capacity and separation. Howev.;:r, other
being opproximo!ely equal 10 !he 1ray highest capacity but :tt some sacrifice plant changes concurrent with r.he Tl,.!r~
pressure drop. When grid trays are in- in tray efficiency ::md operating range. bogrid~tray replacement prech1dc :l.n t'.x~
Siolled wi1h downcomcrs !hey perform Other factors, such ::t~ 'pressure-drop act comparison with the orig:nal b•J:.b;~.
similarly to convention~ I perfomt~d _ limitations, will also affect the selection cap column in this co.se.
trays wieh downcomers. of the trav-free nren. In ·some cases, Table 1 summarizr.s the rate and
holdup rec{uircments govt:,·n the ::.:ruc- performance data at 15 pstg. for oper·
Dcsit~n Varinblcs lurnl design of !he !rays. Where loler· · ation of column No. 1 with 12rid tr~vs
The significant design variaoles of a ancc to corro~\on or fouling is desired, installed for two different fe~d stoc~s.
Turbogrid-tr:ty column are the strt!~\m slot width, nnd to a lc~ser extent bar Feed stock No. 1 was a fttH-boiEn£.-
properties anti mtcs, the structurnl de- shape. in bnr-typc trays, become J:~ctors range strai~htrun gasoline ~nd feed stoCK
tail$ of the tray layout, ::\1\d the specific to be! considen:d. No~ 2 wn'S prcfr~'lC!ionated str:J.i.~h(run
rcquirument.s ~.... ( the p:Hticulrtr scparn· Th1J tr:1y spacing selected represents Q:tso!ine bottoms. R~ore:;er~ta!ivc !e~'d,
t\on prot:c:is, The structyra.~ dctni!s that a compromist: between the factors of lops, and bottom~ A.S.T.:'Yt. di:;tillati0n
·- ... ··--• .... :'-:1:,, ""'I th.-. t.~hliv~ in1~ <'urve~ nre shown in Fir!. 3 t'or runs
 ·--~ _,'
~~~~~~~~~~~~~~~~~~- umn . would not operate satisfactorily
( . unless the reflux rate were reduced,
.... ) thus adversely affecting thc"separntion.
Also, the bubble-cap column would not
perform satisfactory with a feed con-
taining a high percent::\ge of C2 's. The
Turbogr.id installatiot\ operates with de·
sired reflux and reparation for the
higher nverage feed rate now required,
6,000·6,500 bbl. per day, and can ~lso
handle the maximum feed rate of 7 ,400·
bbl. per day ii necessary.
Comparative bubble-cap and Turbo-
grid-trny performance i~ summarized in
Table 2 for flow rates that coutd still
be handled satisfactorily by the bubble-
cap column. It is s~en that the grid~
tray installation achieves a separatiofl
equal to the bubbk·cap trays at flow
rates close, to the maximum for !h·.:-
latter. The separation shown is main-
tained at feed rates of 7 ,t.OO bbl. P"'
day for the grid-tray installation, whi!e-
in the case of the bubble-cap tray it
Fl~. 2-Here h a .stamped#plnt~ type of 1'urbogrid tray, _consisting of stnndnrd sect\ohs wlth deteriorates r~pidly for feed. rates in ex-
5kelch plate~· at tr:t)' etlgt!.
cess of 5,000 bbl. per day.

Nos. 3 and 7, feed stocks Nos. l and 2, tray spacing. The grid trays are of the Column No. 3--depropanizer , .. 'Tur-
respectively. bar·type construction. bogrid trays in this column are in the
Fig. 4 illustrates the separation The column is used to fractionate a . stripping section of a 5-ft.-d(ametc-r C.c·
achieved with both- feed stocks, mens~ feed consisting primarily of C-2's and propani1.er · column operating at 230
( \ ured as initial boiling point of bot~ C3 's, The principal function of the col- psig. ~n this installation, 19 Turbcgrid
\....~...~ toms minus end point of the tops, as umn is not strictly deethnnizution but trays (15 at 12-ln. tray spacing, c, at
a function of the percentage of design rather production of a top product con- 18~in. tray spacing) rephtced lt, bub'ol,~­
rates and the internnl reflux~to~feed taining small amounts of C 5's. The orig- cap tray~· at lS#Ln. tr[l.y spnc!ng. The
rutio. The data show that design sep· inal bubble·cap column performed satis- circumstance requiring the chr.nge from
amtion is signiftcantly exceeded at dc factorily at" feed rate of 4,950 bbl. per bubble-cap to Turbogrid !r~ys w:ts net
4

sign feed and reflux flows and that day and 500 psia. to give 96.3 per cent in this case one of capnc:tv or se'Jar:;:;.-
performance is still satisfactory at \20 propylene recovery in the bottoms and . tion limitation of the bubble·c~P in-
per cent of design feed rates. It appears, 43.3 per cent of C 2's going to tops. stallation.
furthermore, that d~sign separation c:~.n However, at higher ieed rates the col· A routine column inspectton h::-.d in~
be expected for rates as low as 60 per
cent of design, indicating n twofold TADT.E l-PERFOR.\1ANCE OF TUlUlOGR!D TRAYS JN GASOLt;\'E FRACTlO:"'A·
operating range in this service while TION, COLV:-.tN NO. J.
meeting product specifications. The col- (Stamped#pl:J.te•type trays, 12-in. trny spaCing, 5 ft. dinmder, \9 rectifyin~ tr:\ys, ;.1 .stri?-
4

ping trays)
umn bas also performed satisf:\ctorily Feed stock No. '2 (?rdr:J.t·
as a :pres~ure distillate stabilizer nnd as Feed stock No.1 (fun boiling tionatc:d st(a.:~~o-"'.:n
a debutanizer, · range straighu-un gaNiine) bottoms}·
·-~ :· . .Calculated bubble-tray capacities for Run No.- 5 1 s
t 2 3 4 6
.-.:" well-designed bubble-cap trays for the Ftow, bbl. ·per d:.'\y feed 5,300 6,·100 ~:no 4,lS•l 6,000 5,750 5,100 .<,.;! 0
:-::.~.·~est.rates show that at 18-in. tray spl\C· Tops . , ...... , . , , .. , ... , .. , . 4,100 5,100 2,3·~0 J,.JOO 4,350 ~.100 ~.~so 2,6Y')
·r~:-:·.~g,. the bubble cap trays in thi! original
4
Bottoms , , . , , ... , , , . , , , , .. , . 1,200 l,JOO 2,830 9S·\ 1,650 ).6~<) J,.\:.o ;:..~ ·:u·
Imern:.tt r<!flU:< per feed . , .. , . , . 0.512 0.3J2 0.627 O.J52 o.·r/2 0.6~~"' 0,-\CJl (',J.:J
: •.. ~ostallation would be overloaded (e.xcept , I.b.p. bottorm-~.p. tops , . , . , ..
.. Ill Runs 4 and 8), with calculated down "A.P.I. ie:!d .,,,,,,,,,,,,,,,,,,
4
59.2
28 -5
61.5
16
58.0
10
59.4
-,,
50.S
-- tO
5!.5
16
5!.J
,.,
s:.:...
·:; • COmer backup of clear liquid equaling !op:1 , .. , , . , , , ... , .: , .. , , . , • 62.7 66.1 67.0 6).6 53.5 53,·~ 5S.l 57.'2
-· of exceeding the tray spacing in Runs _ Uottom~ . , , .. , ....... , •• , , .. ·\6.8 1,6,3 43.0 45.3 .:..-u ·Ill.~ .\.~.?. .~·; .)
1, 2, 5, :\nd 6, . Per cent design rntes , ...... , .•. 137 13-'2. 101 % \45 1'2S tlO 9'.

r.•:- • T,\lll.E 1-PF.RFOR,\1AN'CE OF TURllOCRID TRAYS IN DEETHANl2AT10~

. Colurnn No. 2~-dccthnni.zcr service , , , fT\lrbo~dct tr:tys-t9 rectifying, 11 stripping, l2·in. truy sp<1cing; b\lbb;l!-<'~? tr:tp-t3 r~df;r·
i
.:
This column is 4 (t. in diam.::tcr in the
! , ·.·_·l ·. rectifying section and 5 ft. in diameter
in~. 9 stripp;.:g, 16 and ::!4-in. spacing; feed r.tte ~.soo bbl. ?a d~ty;
reflux rate 2,000 bbl. p-:r d>'.y)
('-.J Per cent .recovery in
In the stripping section. The grid·tray . Per cent recovery in tops bO!tO!llS
installation consists of l9 rectifying and ~ ~----'---~
11 stripping trays at 12~in. tray S!Jt1Cing; Component Dubble cap Turbogrid D'.lbbk cop
the bubble-cap installation it roplaced c1- ........................... , . 59 ~f)
., c~ . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . 49 .g
had 13 rectifying trays and 16-in. tray CJ"'" ........................... . 9S 95
spacing and 9 stripping trays at 24-in. .. CJ . , .•.•..•••••.•• • ••...••• , •••• 91 'i :'
 TR.IW m:swru
dicated the bubble-cap trays to be bad· umn has 20 rectifying and 14 stripping tion With downcomers was chosen to
ly corroded, with tro.y failure and con- Turbogrid trays at IS-in. tray spacing avoid mechanical complications attend·
sequent poor performance incipient,.· which replaced, tmy for tray, the orig- ing the removal· of_ downcomcrs welded
to the column shell. Performance of
.)
Bubble-cap tray replacement could not inal bubble-cap trays. This column has
·'be obtained in the requisite time, while not been operated at high capacity be- the installation has been satisfactory for
grid trays were designed, fabricated, cause of condenser !lnd feed~supply lim- rates 25 per cent in excess of those
and installed within 6 working days. itations; its tray efficiency at low feed limiting bubble....cap·tray operation, and
The performance of the column was rntes is summarized below: no loss in separation hns been experi~
enccd. The separation achieved in the
improved by the installation of grid 1
trays in the stripping section. Compara· Rectifying Stripping \ original bubble-cap-tray installation \!n·
tivc recoveries before and after the section section der near-capacity conditions which has·
(pe:r c~nl) (per cent) . been duplicated or bettered in grid-tray
change, (or normal (ecd and rcfl\tX Flow rntes, ~~
rates, are: bbl. per cny Trny Tray operation is:
~ Cn· dfJ. Ca· elfi· .--Recovery in~
Turbogrld F~ed Reflu.~ pacity cicncy pacity ciency lops Bottom-s
Bubble· trays 26 30 44 C1's and lighter, per cent 98.3 1.'2.
7,300 2,200 23
cap in stripping CJ'3 and heavier, pc:r cent lO.J 39,1
ttnys section 10,330 2,660 39 ·10 64 62
Propane recovery In tops, This separation has also B e e n
p.:r cent . , , , . , . , , • , , 1 , 98.9 99.J Ordinarily, usc of Turbo grid trays at achieved in the Turbogrid installation
Butane: recovery in bot. 25-40 per cent of capacity is not recom· (or feed rates as low as 75 per cent
tons, per cenl , , 1 , , , . , 97.7 98.2 mended. The separation achieved in of the attaintablc bubble-cap rates. The
this column at low rates, however, has maximum column capacity with grid
The Turbogrid trays gave satisfac- been satisfactory and the tray effi· trays, which has not yet been reachcC,
tory separation for feed rates from 1,900 ciencies obtained in the test runs were is estimated to be 35 per cent gre=t!er
to 4,900 bbl. per day, and did not substnntia.Hy in agreement with those . than the limiting bubble·c:tp column
flood at the latter rates (calculated as used in the design procedure for grid capacity.
80 per cent of capacity), Test-run data trays. It is evident from these performance
gave a calculated grid-tray efficiency studies that in the operations consid~
o( nt least 60 per cent at 2,540 bbl. Column No. S-clceth~nizcr •• , This , ered Turbogrid·tray c:1pacity is sub·H:1n~
per day feed and 2,890 bbl. per day last example features the only plant tially great~r than that of the bubble
reflux, based on the key components installation using trays with down~ trnys replaced, and that in the C:\:'ies
propane and isobutane. The low pro· comers. In this coi Jmn, a decthanizer, wht.::re comparisons were possible, th·.!
1

p~ne content in the bottoms precluded 16 grid trays replaced an equal num· separation efficiency of the Turbogrid
a more exact efficiency determination, bcr of bubble-cap trays at 24 and 30· column is s.uperior. _Their simplicity,
in. spacing in the stripping section. The which generally results in lower-Co$t
Column No. 4--<lcbutnnher , , , This is purpose of the installation was to in~ installations and easier maintenance, are
a 6-ft.·diameter dcbutanizer. The col· crease column C:l.pacity. The installa~ additional attributes' of Turbo grid trays.

50011-----__!,_______

40011-------:------ ,--_~t~,B~·~p~·;B~O~T~T~O~!~=S~--E=·=p~·~T=O~P=S~,_•_f~.--r------
40r

Of,

0~~--~------_h------~,~~--~----~
0(5\GN POINT
(F"£.~0 STOCK NO. 2
RCF"LUX f~[o:::o. 45)
j
I
1oo•l-------1------...: -2ol---+----+---l-----''-\:---
0 ;"{(f'LUX/f'EED::0,66
0 REfLUX/f'EE0::0,•\-8
0 REf'LUX/n::.LO= 0,34
0
o ·50 100
oo 50 !OO - 40o.Lo_:__:..:_.,,:Oo_:_ _..,,c.';o.,;o---;,-l,:!""o---;,~.,:co;---,;l:i0
PER CF.NT DtSTILLED PER CENT Dl"-TILLED PER CENT DESIGN RATES
Fig, 3--Putorm:mce or Turho~rld trays In raso!!ne £rnctlonnt!on, Fig. -!.-Ptrlomwnce of 1'urbo~r!d tray1 In h.ydrocnrb¢n frnct!onnt!on
colu111n No. 11 ,\,S,'f.~f. rlbt!ll:uion dat;l, column, column ;•.;o, l, This Is for ·10-pl:tte .column, 5 !I, t!bmda 1
11-ln. tray sv:~chl~'• fetd 5to~~!i ~o. 1 nnd 1. (.'::<:'~ 1':~ble 1.)
 TTillY DT.SXGI\!
J. .••

pivoting at the heavier edge and final·

Iy the heavy edge is urgt~d :\wny from
its seating, over the opening on the
tray floor, by pivoting at the light
edge. The merit o( this movement will
become evident in the description of
operation.
Float-valve bubble trays are the re-
. . . I 'lAH VIIW 01 ON I HOM VAt.vt UN!l
sult of development and study of bub·
ble·tray performance started in. 19l~5.
1ed~r=>- J The purpose stimulating this develop~
I mcnt was to· produce a bubble tray of .
. $1CliON "A-A'' IN C!..OSto PO Sill ON
high capacity with superior efficiency·
that :vould also operate with good ef·
ficiency at reduced capacity. Durinz
' - !~OMrTJ'.ICvtEW'
IN aAACl<IT AR.f,\ ·
the first severrtl yenrs of development,
two distinct versions, other thnn the
present equipment, were studied. The
first installation o( float~valve trays W::!s
in September 195l. This installation
was in a 6·ft.·diam~ter, 16·tmy absorb·
er on 24~in. tray spacing ope~atiog :tt
. about 240 psia. and equipped for ex·
·perimental testlng and in regular ser'l·
ice at a natural·gasoline plant in C:tr~
l, Fie. 1-Detnlb o! flont-vall'e bubble-tr.ly unit. son Cquoty, Texas. ·
Construction of Float Valv~s
Float valves may obviously be con·
structed in many forms in coniormlt;
to the basic principl~ just omlined. Oi
the many forms invcsdgated 1 the forrn
described in this <\rticlt.! was selectee!
for commerci31 use becaus~ of its stm~
plicity 1 economy of construction, and
adaptability to .variation in design ~o
satisfy performance requirements. A
plan view of one float-valve unit, sec-
tions of the valve in all Hmit positions,
and an isometric view \n the arc~ of
the retaining bracket are shown in
. Fig, 1. ,

Assembly , , ; The float valves and as-

sociated openings in the tray Door nre
arranged in rows, The retaining brac:z.
.::. .. • · - - ---~-· Fir,. 2-An as.sentbled removable tray for 30-ln, tower. ets at the ends receive corre:;ponCir.P
.ends of adjacent valves in the s~lme ro\~;
and the openings in the tr=ty floor ex-
tend the full length oi span betwe·~n
the bracket edges. The tray . <kef.: is
solid undernc:1th the brac~<ets to re-
tain the section of deck in Q. dg:d, in-
•:· ·~ D-esl~ned
,... tegral piece and slots are provided ct·n·
tr<.1lly across the top of the br~C':<.c~s
to permit the escape of g:1s oif th~ •:-nC.:;
~"Which Woulcll\lso \3e of t.he valves so as to purg~. or S\w~:~
away, any solids that m;,~~ht acnmn:.
late under the brac!<ets ~-.ncl ~nterf~re
::: ------·····-·
·-.
with the movement of •.he v:~.tves. Th~-­
brackets are pos:tionec! tran-sverse ::tn<::
THE name float \'alve is l!~scriptive · tray to limit vertical or horizontal ccnrr~Hy to the tray openings and ::1
- of floating pivotal valves that in· movement. As a contnct medium ~s standard assembly are secureci by two
() volVc.Jl_n_c~v.princip\c .. The valves are
hea~i~r 9n one edge than on the op-
passed upwardly through (he tray, a
floating pivotal movement is attained
rivets as shown.
The float valves and b~ackets ~:re
posl!e edge and are retained nn the whereby the light edge of the valve is made of Jight~g.:1gc stain!ess.·stc~l mote-
~ ·- Author Is with Nutter Engi,neering Co., initially urged -;,way from its seating, rial us :\standard. To date ot! job-;, ex.
···· Amnril!o, Te~. over an opening on the tray floor, by ccpt one ench using l6·gng~ and ! 8·
 ·.....

'ICliON"A·A"

Fig. 3--Singlt·flow float·vnlve bubble trny,

gage, have been 20~gage thickness. Re- Oper:>tion of the Trny phase, 'such as in an abSorber, is. turned
ferring to the large scale"sectionnl view on first. With liquid phase in operotion,
A photograph of. ari assembled re-
ln conjunction with the plnn in Fig. 1, the liquid flows smoothly across the
movable·type float-valve bubble tray for
one side of the valve is wider by ex- a JO-in. i.d. tower is shown in Fig. 2. tray with a minimum o( gradient b•.'·
tending laterally outward intermediate cause of practically no obstnJctk""~n ..
Float valves ore applied to the tray in
the end sections and turns up .to form parallel rows po'iitioncd ut right angles Howt!ver, even with· a m\nimurn flow,
. an L-shaped section. This edge of the to the direction of liquid £low across sufficient gradient exists to subm·~rgt­
valve is therefore the heavy edge be- the tray with the light edges facing into the va.lves near the cntr~nce side of tf~,_.
cause of the offset to one side in width the flow. t\\thoug;L other arrangements tray in more liquid than on the exit
and the upward extending flange. As may be desirnb:~ for certain Upplica- side. Before gas flow is st~ned, a mcd~
the opposite, or the light edge of the tions, it is believed this arrangement is crate weep prevai!s through the va~\·cs
valve, which raises initially, opens, the most generally desirable and ~is shown because of imperfections of f!t How~
offset inner edge of the ends of the in Fig. 3 for a single~Oow tray and in ever, such imPerfections ot f:t art: or"
heavy side of the vn\ve are raised to, Fig. t~ for a doub!e~flow trny. Referring C:'lpillary magnitude and, as the :·irs·.
or above, the tangent point of the to Fig. 3, all trays utilize the exit weir traces of gas flow. starts and ~uf~:..·
rounded inside corner at the bottom of to seal the base of the entrance liquid cient pressure differential is est:lblisr.ec.
the sides of the brocket. This occurs downcomer on the opp~site side of the across the tray to initiate such flow, o::
as· the valve pivots aoout the heel of tmy nnd downcomer seal weirs are not wcepin~ in all areas of the tra·1 r.!:!as . ·s
.
the L·shaped section of the valve, which used adjacent to downcomers. The even \~(th normal liquid flo\V~ Eve:)
is offset and outside in respect to the tower may be fully loaded with either with mn:dmum liC]uid flow, the v::tlve-;
point of contact between the valve and liquid or gas flow initially without in- remain closed b~cause the weight. of
bracket. This construction permits free terruption or adjustment of either whtle liquid dow.:ward is always gre~:c~r ~har\
movement of the valve in its retaining turning on the other phase. the lateral force of liouid across r.h(:
brnckets through all oo.-dtions nnd h; tray against the uprigl;t web l.'f !h·:
economicnl to ffiunufncture. Opcr:ttion ••• t\«ume that the liquid valve.
 sorption-oil inlet piped interna\ly for
norm~! entry on the top tray. Lo.t~r, the
top 8 trnys were inst:\Jled and two tests
() were conducted on the full l6·trav unit.
This absorber is the left·hand to\ver in
the photograph. Duplicate s3mples were
carefully. taken and analyzed. The ob-
servations and evaluations of these tests
arc shown in Table 1.
Although these tests were made with
different numbers of trays and at "
wide sprend in 1oading1 t.hey show the.
tray performance to be remnr!<.ably the
same. Operating performance in plants
of three diffcrt.:>nt compnnies in which
all of the absorbers were changed over
to Ooat~valve trays show a reduced
fLxed gas retention for given product
recoveries. )n one of these plant~,
where a still, to be described l3ter 1 was
installed and the oil is comoletelv
· stripped, there is no pentanes ?lu·s cot{.
tent in the outlet gas.

Calculations ~ , , In development of the

summary results of T;.tble "1, the ::~.ctu:\1
per ce'nt recoveries were cntcu!nt<!d
from the moles -of entering gas deter~
mined from the analysis ;nd its m~·
tered volume divided into the Giff~r·
ence between same and the snme ev,ll~
U:ttion of the outlet gas strc::~.m. The

() Fig. 5--A '70'h·tn •• l,tl., 16-tray nbsorber used for test t~ on the left. O~ber two ,'\b~orbers
on right :~lw equipped with Oont-vnlve trays.
comparative per cents of theore~icnl re~
covery were based on an as'Sumed ~ 00
per cent tr"y efficiency by calculat!on,
Gn.< flow , , • The first small flow of gus of valves throughout the \ray start open- using Katz and Hackmuth equi:ibri~.~m
brings the light edges of the row of ing in unison through their range of constants for the entering volume nnd
valves adjacent to the exit weir into movement, and at ubout 70 per cent of quulity of gas at observed condi:ion:;.
operation. As rate of gas flow increases loading all valves are in fully open po· The analyses were somewhat errntic
the light edges of additional rows of sition. in the cut between iso and norm:ll b•.:-
valves, consecutively from their exit Beyond this point, full loading is·. tanes but, since they are both \.n '.h·:
weir towards the entrnnce downcomer, accomplished by increased pressure' key component range of extr~ction,
ru:e brought into operntlon. At approx- drop through the valves in full open they were grouped together for cO.'ll·
imately 20 per cent of design loading position. The stabuity o( the function- · parison between actual und theoret:c~!
of the gas pha$e, irrespective of liquid ing of the Jloat valve bas been observed recovery. Referr!ng to Tlble l, t.~e
loading, all the valves throughout the during many tests, both on dry valves actual recovery oi butanes !n the three:
tray nre brought into operation and full to develop capacity information and on tests is 98.25, 96.16, and 94.75 1)Cr c~n!,
efficiency of the \ray is realized. As a single commercial-size test tra)'· op~ respectively, or an aver~ge of 9·6.39 ?~r
rate of flow increases, additional load~ erating on absorption oil and natUral cent. A similar comparison of :~e tota:
ing occurs by increased differential gns. Their stability is also manifest by propane and heavier' products is 97.91
through the openings o( the light edges the perfect condition of trays in a com· per cent uvemge for the thrc~ !ests.
until at .about 40 per cent of design ffil!rcial tower observed on inspection !t appears, therefore, thar a tray -~::·>
loading the heavy edges start to rise. ar'ter many month.-; of operation. ciency in the ran;?",e of 75 per ccr.t :s
The g3s emitting from the light sides being attained.
of the valves urgt: the ends of the heavy Test and Oper::tting Perlomtnnce
Similar tests !n the same pl<lr.t on
edges of the valve snugly against the Test facilities for measurement of all Jive origin:\! absorbers befurc t.':v:
inside legs of the br1ckets and, nlso streams, sampling, and observation of changeover gave 14 to 20 per cent ·-:!·-
the upper edges of the light endr, against tempc:rntures and pressures have been ficiency on four of these :ow~:-<: D.:-'.1:.
the Jop of the inside of the t,rackcts. providr:d on imtallations in absorption, possibly 30 Der cent on one of !f~r..·
Thus, :my movement of the henvy edges distillation, and fractionation services. towets.' In t.his panicu!o.r ?lan~. ~~,·~
is dampened by these contacts. five o.bsorbers comisted of ~wo 7 -f~.·
Furthermore, nt the time all the light Absorbers • , • The first absorber, men· diamet~r 12-trav, one 7-ft.-di'J.r'.'.cto:r

0.., edges of the vnlves were brought into tioncd curlier, wns cq\lipped for testing 16-tray, one 6.ft:·diam·~~cr !6-tr~y (:e;:
operation, any previously existing liquid and has been tested three different tower), and one 5-it.-dinmeter t6.w:::·
.. · . grndien: ncross the troy disappeared. times. On tht:' first test,· only eight trays towers, before the change o·1er to f:o:-\~·
Thus, as loading progresses bey o n d were used and were installed in the valve travs. The latter three w<::-·~
about 40 per cent, all the heavy edges lower eight-tray positions with the nb .. equipped ~vith these tray-; and !h~ ('>W)
 TnlW DESY.Gl\1 \
7-(t.-diameter 12-trny towers were re-' "Since the first commercial installation of float-
moved from absorption servict!, Al·
though the original towers weN more. valve bubble trays in 1951, a total of 27 installations
than ample for gas capacity, they were ·- ....
handling about nil the oil they would ,
have been mode. Twenty of these ore in service and
take at an aggregate of 1,250,000 gal. the others will soon be operating."
daily. The present three towers appear
to be able to handle nbout 50 per cent
more gas loading than originally and T,\DLE 1-T~Sl' OllSF.RVA'l'IONS AND RESUY..TS ON A 6·FT.·DtAMETER ABSORBER
EQUIPPED WITH 3 AND !6 FLOAT VALVE TRAYS ON 24·L'i. SPAC!:>G
nrc handling 1,785,000 gal, of oil daily.
Opernting Test No. 1 Test No. 2 Test No. 3
First Installation , , , This test absorber conc!itiOn$- October22, 1951 Febru•ry22, 1952 April26, 1952
hns been loaded to full gas capacity No. of trays ... , , . , . , . g t6 16
Inlet gas, s.M.c.f./hr••. , t,soo.os 1,302.93 l,<Jol2.48
which was 51.3 million standard cubio 1 Outlet gas, s,,\Lc,f./hr •. 1,6S6.liS 1,639.19 979.66
feet per day, at which point the tower Pr(!SS. psia. base •. , , . , .. 242.13 23J,04 '200.8·~
did not (!ood but frothed excessively. Press. psi<\, top , ...... . 240.'13 230.44 199.1-~
Upon load reduction of less than· 1 per Lean oil gal. hr. at 60° F. 18,;0•1.40 21,117.62 14,773.4.3
cent, stnble operation resulted immcdi- .'
Tempemtur~s, °F.:
ntely. 1\ study on the effect of oil mte ' Gas .. . .. . .. . .. .. .. ln-92 Out-69 ln-81 Out--<10 ln-73 Out-71
on gas loading was made and it was Oil .. .. .. . .. .. .. .. . In-53 Out-9 t In-l2 Out-83 !n--&i Oul-S4
found the above gus load could be Specific gravities:
handled at nil oil rates used with iden- Gas , .......••• , • , . , !n....j),696 Ottt-M56 Jn....j).692 Out....j),643 Jn....j),695 Out....j),6l2
Lean oil ... , .. . .. .. . 0.8212 0.8100 0.8100
tical results. This being the first tow~r
to usc float-valve trays, the trays were· Mol. wt. of oil , : .. , , .. 216 . 2)6 ::!05
adapted to the existing downcomers.
Also, this original tray layout was not Per Cent Test Reconrlcs o( Product! as Comp::~~ed to Theoretical Recovery
Wltb As.~·utoed 100 Per Cent Tmy 1Wicteucy
· equipped with nearly so many float-
valve l.tnits as is now standard design.
Where this installation carried only 110
valves with ?V2 per cent downcomer
area, present design permits the use o(
about 150. Thus, ii excessive froth·
ing is the result of slot velocity, then
the prospect of appreciable increase in
capacity rating is very favorable, TAllLE 1.--..I)UMMARY OF TEST RESULTS ON A 61h·FI'.·DTAMETER STJT~L EQUIPPED
WlTH EIGHT DOUnLE·FLOW FLOAT VALVE Tll.AYS O!'{ 2·l·G'. SPAC!i':G .'-''\D
COMPAH..ATlV£ HESULTS 1:"\' PART FOR T\YO 6-FT.·DfA~lET£1~ STlLLS
Other installations , , , Operating re.' OPERATED 1N l'AHALLEL ?u'\D REPLACED BY :>EW FLOAT·
suits of the three plants, where ilont~ VALYE TRAY STILL
valve trays have replaced the former
Item- Ne.w still Old stills
trays, have in all instances reflected Date of test , ... , . , . , . , , , , , • , .• , , ..••..•.• , .. , , .. , . , , . , 9-23·53 No acmat test
product recoveries nt least ns good and Oil cir. gat./ d:>.y .... , •.••.•• , , •••••••..•••••• w ••••••••• 1,%9,601 l,SOO,OCO
apparently better than indicated by the Stripping ste::~m, lb. per day •... , .•......•. , , .. , . , ... , ... 303,120 552, ooo
above·dcscribed absorber tests. ln all Lb. stcatn/g':\1. oit ........... , , .. , .... , , • , .• , , , •..• , . : . . . 0.1539 0.3065
l.b.p. Jean oi.l, .. F ....... , , .. , • , . , • , .•• , • , . , , , •.. , ••.•. , , 44-l 36Q.J.Il
instances, the cross·scctional area of Per cent over at •100" F, .. , , , •• , , ..• , . , .. , ... , , ,', .. ,,,,., 0.3-'l.S
0
the towers left in service ore about 50 Per cent sat. by C01 meL~od , , . , .. , , . , .......... , . , ... , . 1 0.26 Not run
per cent of that originally used and, in Gasoline hydro.cnrhons passing still-moles per day:
most instances, the extra towers have (a) Liquid fet!d !o cle•:L1anizer .. , ..... , .... , ........... , 3,727 't'\o test di\t:'.
been utilized in other services. The (b) Vapor recycl.! w demethanize( , ..... , ... , , .. , . , . , , . , ·1,2"2"2. No t.:st J:>.t1
Retlux oil return from dt!ph., g.p.m: .. , ..... ,,.,., .... ,.,, 30.16 A}'prox. 60
present towers at each plant in aggre~ Pressure, psi:t, bas~ , . , , , , , . , , , , , , . , . , , . , , , , . , , , . , , , • , , , . .55.2 Nut t~sted
gate are capable of handling more ca~. Pr!!ssure, psia. top ...•.... , . , .. , . , ... , , •.. , ••... , , ••... , 54.15 · Ap:HQX, same
pncity of oil in combination with gas Oil ~ntering, °F. , , . , , .. , , , , , .. , . , , ... , . , , , .. , , . , , ... , . , 395 .:OJ
thun could be handled wit~l the origi~ Oil. te~ving, "F. ~.,.,,,,.,., ·.. ,., ... :,, .. ,,, .. , .. ,,,,.,, 33'1 400
Stnpp1ng steam, F.......•... , ..•...•. , ... , .•.......•.. 33'1 Approx. s.une
nat equipment. St~nrn superh~::at, op, , , .. , . , , , , , . , . , . , ... , . , , , , , , .... , , . 77 . A~.l?!OX. U.1lh'
Liquid levels in downcomers above top of entering tray decks.
Stills • , • The first installation of float- (Exit wdrs-1% in. hit:h):
vulvc trays in a still, steam absorption~ (a) Above Xo. 1 tray (bottom tray), in ..... , .. , . , ..... , 3 No Cat.'\
oil stripper in a nutural·gasoline plant, (b) Above Xo. '2 tr:>.y, in, ......... , .......... , , , ..... , 3'A ~o C;\;:t
(c) .r\bove >to. 5 tray (probably error), in ............... . 7:Y~ ){() c':H:!
was m.,de in July 1953. This is a 6·ft. {d) 1\bov~ ~o. '! tr:>.:, in .............. , , , .. , .... , .... . 3¥! ~o C:1;~
6·in. i:d. tower with eigin doublt~·flow !vto!. Wt, of oil in sy·n•:;n, :1.pprox. , ... , . , , .. , . , .•. , .... ,.. . 21G 21'.)
trays on 24·in. tray spacing and re· t\.P.J. of oil in systo:n1 ........••• , • , • , . , •.. , .• , , , , .• , .•. ·i0.7 .t LS
.' placed two 6·ft. i.d. towers operating !vfol. wt. o{ charge (new) oU ......•.. , • , ...•.. , •• , .....••.
t\.P.I. of ..:harg.e oit ....... , , . , , , .. , • , , . , . , • , , •• , •. , • , . , ,
130
42
1SO
·:1
in pttrulld with six singlc.flow trays ,\pprox. l.b.p. ch;uge oil, "F•. , , , , , , .. , .... , , , , .... , , ... , 415 1\5
each. Table '! is n summary of the test Approx, e.p. cho.r~~e oil, 9 F, .. , ... , , , ...... , , , , , , • , , , . , , , , l20 no
data and results showing aVailable com·
parutivc dutt\ on the two old stills. the landing 10 1,970,000. 1\ 100 per F. and SS psin. The results show :~·lis
Thi'i tower wns th.~siPnct.l to hand!~ ct:nt trav effickncv wns n'isumed in de· performance to be nccompEsh~.:d. The
2 1/.1 miltit.)n t!alh.'ns of \c~;tn oil dailv but sign rctiuiring 0. (55 lb. of stenm per il•vds of liquid st:\ckuo in !!w Gown~
i\V~Iilnblc oil:pump cap;.tcity hn'i li;nitcd g~t!lon of oil nt ~ tt.'mpcrature of 400° comers indlcnt·~ the to\'vcr co•J!O ~:lsi!y
 ~~-h,{ndl~ ih~ 2\1.• million gallons of cir- . Commercial Insin.lbtions
culation at 90 per cent of flood-point Since the first commercia! instalh·
loading. tion of float~valvc bubblt> tr:lys in an
i........ -· This still is handling twice. the load absorber in Seolembcr 1951, a totnl of
per square foot of <:ross·section and 27 installations have been madt!.
using half the stripping steam in com .. Twenty of these installations are !n
pnrison to the old stills. Most impor- service and the others will soon be
tant, it is throughly stripping the oil, operating. Of these instalbtions, ther·~
whc.reo.s the old stills were h:nving an are 10 absorbers, '~ rcabsorb~rs, nnd ~
undesirable ·per cent of saturation in c:~.ch of froctionators, stills, and rich-oil
the oil. dcmethanizcrs. These installations h2ve ·
been made to 11 companies, All ~o~s
' Fractionators ••• The first installation have fully met poriormance recom·
o( float-valve trays in a fractionntor wo.s mendations.
also made in July \953. This is a 4-ft. It is quite evident from the expe-
6-in, i.d. depropanizer with 30 trays on rience factor on these travs that ap·
. 13-in. tray spacing. The piping was in· preciable incre:lsed efficien"'cy and ca.
·- c: stalled so the tower could be operated pacity have been realized over ~onven·
:o:. with the full 30 trays, or could be re- tional trays now in general 'Jse. These
_,, fluxed and operar ~ on 14, 16, or IS trays are simple and durable in con·
.. trays (the regular •llernate feed entries) struction, are ensy to handlt! and !n~
:: o:with feed points lowered to al!ernate stall, are highly efficknt over wick
:.. .. :.points on trays 4, 6, and 8. loading ranges, and have many other
!OOl. ·!fwo tests using 30 trays in one in- merits, several of which nre summ:1·
... stance and lS trays in the other have rized as follows:
.::._.;..been conducted but becau<:>e of me· 1. Efficiency constnnt over wide op-
.... chanical trouble and poor measure· erating range due to uniform di:Hri~t~·
ment, the results are not presented. tion of gas and liquid forced by !o•.v
Time since has not permitted addi· valve prcssure·differential produced ':Jy
tional testing. Fir.. &--.This is 6\/J.ft•. dL·IUteler, e/ght·ir~y still float valves.
us~d (or test.
The tower operates very smoothly 2. Energy of gas passing Hf!o<:t
.... and abrupt changes in feed or reflux valves/' in narrow ribbons at \~av level
{ ·)-.,. · rates arc not reflected on the temper· i.d, 30-tray towers operated in parallel utilized for intimate contact. ·
\ ....... • :· ature or· pressure-recording controller in this same service previous to this 3. Time . contact at ma.:ximum be·
·::charts, The maximum loading. ever installation. cause of lateral h!gh vcloc:ty imping•:··
::.;placed' on' the tower wus during the 18-
ment at trny level.
tray test when 1,085 moles per hour Butane splitter • , • Of interest is a 3-ft.
4. Ivfist and entrainment to tr:~.y
··of combined reflux and overhead va- i.d., 46-tray butane splitter with 21-in.
above at minimum due to un:for::l
~ .!."pois, representing an 8.63 to 1 reflux. tray spacing put into operation in late
·- ratio, were handled in the fractiona- October 1953, producing isobutane with
blanket of froLoy liquid on tray.
tion section. At this time the tower was 1.8 per cent normal-butane contami- 5. Liquid gradient across tmy is at
-· operating ot 263 psia. and 127' F. top ... nation and normal butane with 3.2 per minimum because of no high obstruc-
· --teinperature. The overhead contained cent isobutane contamination. The re- tions on tray floor and, i~ fact, dis·
. - 2,3' per cent isobutane and the base flux ratio to overhead is 23.6 to I and appears above 20 per cent lo8.d:n,? .
-· product 0.90 per cent propane. There represents a vapor load in the frac- 6. Liquid cannot short-circuit tray
- wis no indication of bdilg close to n tionation section of tower of 505 moles at any operating rate ar1d exit wc:rs
· flooding loud although the liquid in per hour. The tower operates at 125 seal downcomers from opp'osit~ s\Ce o(
::-.'the gage· glasses on the downcomers psia. pressure with a top temperature tray.
·:\Vas standing 9 in. above tray deck of 134Q F. 7. Liquid capacity h'gh due to no
- level, or one·hnlf the dis~ancc between obstruction at tT3.)' cntr;.\nce and ucro:J~
P'trayS. . - . · Demethanizer stripper •.. A rich-oil tray.
""'.=-=··Th.is tower currently operates on 30 . demethanizer stripper was equipped 8. Interruptions of eithe:· g2s or 1iq-
~- 'tr~ys ilnd as an average with a feed of: -with float~v<llve bubble trays. This strip~ uid flow ma'v be restored \\~ithou~ :td-
•·~ 3;700, base product of '2,350, overhead per is an o!d ?·ft. 12~tray absorber ln jusrment of /tow of other mecEL::r..
~.:·product· of 1,350, and reflux of 7 ,2"25 which the downcomers, of single· flow 9. Fouling is m!nim:z:;d Cue ~o ::lC·
gal. per hour. The operating pressure type, were enlarged when the float- tive operation throughout tr;1y :lt :'l:i
is :?.63 psia. and tht! top tem~,erature valve trays were installed, This tower capacities.
127° F. The ref11,.1x noted is a raric is handling o rich-oil volume of ap- 10. Installation readily aotJ!k-ab!e to
... of 5.15 to I apinst the overhead pro- proximately 2,\00,000 gal. daily, or existing towers and dow~Com•!rs i!l
- pnne production. i'requent control about !,460 g. p.m. A volume of 1,600 either '""remov::~ble or fb:cd tny e~e!:>
;- !lflp\yses show the propane to contain M.c.f. daily o( vapor recycle off the ment.
.... -~, l.O per cent isobtWmc and the base raw product accumulator is introduced 11. Light weight yet extremely s~ur·
(_';, product to contnin-0.7 per eeoc pro· ·into the bottom of this tower as a strip- dy and price economical inc!ud::1?
... ,-pane a1 an average. These results re· ping agent. Although no tests have stainless-steel trim .
: i!.;:ct at k·nst 20 per cent more eficctivc been mnde on this system, operating 12. Design and engineering slmp~e
fractionntion per tray nod capJcity thnn results inc!icn!c it is more than meeting nnd positive, thus ris% or haza:--C o{
was being ~ccomp!ishcd with 'tWO 36-in. design requirements. improper design eliminated.
TIUtY m::smnr
( I 2 V2 IN.=:J END or
' SLOTTED

T
' SECTION
3 7/8 IN.
l'A" .... / ;

:;; i
:;;
TROUGH VAPOR

:. l ?
SE:CTION

'/ ' // 7 // ' / 7

"' sr.CTION
SLOTS/

..A"
.\-
_l
"\
"'
-
!Oi161
+13/16~
fr1I1/2TOIN.! lN .
f-3/16 IN.

Fly_.. 1-e~S" munb~r t~ fundamental unit of Socony.Vacuum Un!flux tray,

:by VJ. o. nowles

Q VER the past two decades the ex· sive hydraulic gradient and resultant un· vapor liquid contacting device, but to
4

perience of Socony·Vncuum Oil Co., equal slot submergence. develop a tray employing the same gen·
Inc., with fractionating trays has been em! principles as bubble-cap trays which
almost exclusively with trays of the Bubble cnps most common , , • \Vhile correct some of the faults in conwn·
bubble-cap type. During this period over the years there have been some tiona! bubble·cnp trny design and at
Socony has left the problems of rating notable innovations in fractionating tray the same time is simple, adaptable to
and designing trays almost entirely in design, no new or improved design has stnndardlzatipn, and inexpensive.
the hands of engineer·contractors and yet been offered that has received wide Socony TJniflux type fractionntin£.
has made little attempt to develop speci- generat acceptance. Bubble·cap trays trays perform in substantially the same
fications for, or standarize on bubble· arc still by far the most commonly em- manner and have many of the snm•:.-
cap trny designs. Consequently to pro· ployed fractionating trays today and characteristics as conve.ntionnl bubbl.::·
teet continuity of operation, it has been probably will be for at least some time cap trays. Since the design of these trays
necessary to warehouse a great variety to come. This is remarkable when one combines functional and mechanic:1l
of bubble caps and tray parts, with re· considers that there has been no such features, the tray is light in wci?ht and
su!tant high inventory cost for large significant improvement in bubble-cap therefore Jess expensiVe than cvompar·
stocks which are turned over very trays during the past 15 years, and that able bubble-cap trays. It is oi simple
slowly. this type of tray can be fabricated and design and can be rapidly inSialled in
It has been the experience of Socony installed more cheaply today than 15 fractionating towers and easily removed
that fractionntors have been genemlly years ago (disregnrd!ng the effect of in· if nccl.!ssarv for ma!ntcnnnce insoection
operated ultimately at rates substantially flation). It is not i~tended to deny that and cleani~g. ·
in excess of original design rate::; and there have been improvements, because
S Member
only rarely has it been considered ex- it is reco~nizcd that fabricators have
pedient to replace, or even significantly bent considerable effort to this end and The fundamental unit of the Unif1ux
alter any trays. Furthermore, the frnc 4
certainly with some success. On the tray is a:1 S member (sec Fig. 1). ThL'
tionating efficiency of bubb\e.cap trays other hand, it is oossiblc that m o r e vapor section of e:1ch S member is en·
appears to hold up remari;ably well, stringent rcquire~cnts and specitica· closed at each end by vaoor·closurc
even at rates of 40 to 70 per cent in tions of process designers have tended pieces. Intermediate st(ffcnefs are bs-
excess of original design rates. Addi· to nullify such gains. tened at specified intervals in the ~ro•.!g>.
tionally, experience indicates that the section, to insure a good s.;:ction mo·.::.
ultimate caprtcity of bubblt!·C:lp trays Uninux tray, .• Socony engineers have, ulus to the member, especial!v to t:-:l'
is . more often limited by downflow along with other operators ttnd design· free wall of the trough· sect:o;1. The':ie
rcstrictioR than any other cause. There ers of fractionating equipment, devoted vapor~closure pieces and tray stiffe:1crs
ls reason to believe that in some frac· consider~b!c attention to the develop· are slotted ::.0 that on ~1ssemblv ~hr S
tionritors where conditions of high mcnt and improvement of fraction.:.ting members are imbricated i!:> shown !n
downflow exist, fractionating efficiency tray design over the past several ye:lrs .. Fig, 2. Thus the basic S m('mbcr has
is not all it should be, because of execs· Th•: result has been the development of structural, as well us functionai (JrOper·
Uniflux fractionati.ng tril.y. The Socony ties. When fabric:~..tcd of 16-gag~ m-:l:c-
Author is with Socooy.Yncuum Oil Co.,
ln<: .. refinery enginc:~ring division, New Yor't approach to ~~is problem has been n0t ria\, it hns ample strcn~th to ~D~Hl a ~ 0-
City. to develop a rudicn!ly new ilnd differcr1t ft.·diametcr tOwer w!t'h ncg!(gible de~

3G
 ' flection under operating and mainte
''Egual slot submergence and elimination of areas of inter· nance conditions. In l a r ~ e r towers
4

ference will result in performance advantages."- · ~ '(up .to 20~ft. diameter) a ;ingle inter·
mediate support beam is used. :--:o other
support beams or purlins arc required.
(.. )
Assembly ••• These S members nrc
secured to con_ventional support rings
STirrr.NER by a single conventional tray cbmp at
TRAY - A each _end. The general assembly of a

~ ~--,:~"~;-~=l~~~:i;.""·1Y_;_._.;;p· single cross flow Uniflux tray rnny br:

.. -· 1
seen in Fig. 3. It may be seen that the
1( ·---11 f 't,.
~ I ,c·,-~ l ;t- I
r[.._ .3=- I ~._-; ,d,~ I S members are nssembled st:trting from

/~ .. :':~J! ~~ lF·:·
the trough at the downstre3.m Side oi

~..-~..~50 \J~u _______,/ the tray. The lost vapor section is fas-
tened to the vertical flang~ which is a
part of the segmental s~ction nt th\l
foot of the downcomcr.
For trays having more than seven
S members, a split S member is pro-
".v;ded on alternate trays so that v'er(ic:J.l
access is provided for mainter!ant;e,
cleaning, and inspection. Downcomers
are similar to those of conventional
bubble-<:ap trays, except that they are
VAPOf\•SECTION tNO CLOSURE ""'~'.·,s" M EM BEl\ ,-';'":. .~r ..-, ::·
standardized with Uniflux trays anC:
are so. designed that unrestricted down~
flow is insured. For trays of very h\r2e
Flf, 1-Craplllc view of typlc:tl Unlflux lrny sedions. downflow requirements the vertica~
flange of the segmental section :n t:te
inlet side. of the tray is extended up-

"
<"':;:-:;;~~~~AP P 1\T
ROY ED TRAY .CLAM P
END OF' EACH .. ward to provide a proper se::\! for tf1c
1
J TRAY SEGEMENT downcomer at the inlet to the tr2.y.
,-: :· . I Double crossflow trays employ the same
I :> 4·1 N. WIDE: basic _S member e!ements and :1.re of
\.._) I
I
.I :-l,
Q'l ~ "·BOLTING
BAR!
similar design to the single crosst!ow
design just described.
I
I
I
:I
II
. 21N.
Slot area ••.• The tray parts are so
·:.

I, proportioned that the available slot area

'I
I I is about 12 to 14 per cent of the
I I superficial tower ruea, or generally
I I'" equivalent to that provided in ccn~
I ·I vcntionfl.l bubble,-cap tray designs. Slnce

.•
,__i_l. the slotted sections are well seoar2.ted,
there is· no interference of the. va~ors
issuing from one slotted e!emt:nt ~.vith
those issuing from anothe: and the
entire len~th of the stotted sections :'i
fully effecl.ive. Furthermore, the V:lcors
issue forth in one direction :tnd in
sUch a manner as to boost ~he ~!cui(~
across the tray an C. .. compensat? ·for
hydraulic gradient under '\ll cond:tior~$
of flOw.· Therefore, nm::-.;ng slot -;ub~
··· mergeriC.e is uniform across the tny,

Riser area • , • This in general avem~es

2 to 4 times the riser o.re~ provtC~d
in bubble-c:lp tray designs. Li:·:?w:st\
revers~l area and equiva:i:-nt :1.~!\'..!::J.r
area are large. There!·ore, ~he U:1if!ux
tray .is <;harUcterized 'oy :1. lower fixcll

·-o
' . "~
.. ~r~ssure drop (exclusive of :h:~~ C•.:;;
to slot submerge·nc'e) thar. ?u0ble-c::>.J?
trays. Ffom data dev~:loped t~ J~tc w:~,,
Fig, J..-l\s.serobly phtn <~nil elevoulon for single aos:i' [low tr<~y,_ .the Uniflux tray it appe~Hs. evich.·r1t ~~~~H

37
 the combination of equal slot sub- Socony-Vacuum ha'i tnkcn the position petroleum and chemical industry on the
mergence (compensated hydraulic gra- that these trays are extraordinarily in- basis of a modest fee of S 1.00 per sq.
dient), low pressure drop and elimina- expensive and that on the bas}s of ft. of superficial tower area per tr:1.y.
lion of nrcns of interference will rcsutt merely equivalent performance very In actual practice fabric~tor.s .will be ·......·
in measurable performance ndva~tages. subsw.ntial savings in capital outlay licensed to construct Soconv l!!liflux
have been realized. fractionating trays and this fee wilt be
Fnhricntlon , , , Socony Uniflux 'trays The problem of warehousing spare included in the tray quotation by the
fabr.icatcd of 16-gagc material, com· parts for Socony Uniflux trays is quite fabricator. At the present time four
plcte with downcomers, weigh about simple and very inexpensive. For I!X- fabricators o( fractionating tr:\ys arc in
8 lb. per sq. ft. of superficial tower area. amplc, if Uniflux trays are being used a position to furnish quotations in the
Fabrication is very simple and can be in nn entire new refinery, it is merely United States and one fabricator in
accomplished with stand a r d press necessary to have on hand an appropfi .. France. The information necessary for
brrtkcs, shears, and die-stamping Ct1Uip- ate number of S members approxi- fabricators to build these trays will be
ment found in mnny light-metal shops. mately 10 ft. long nnd a S\titnblc supply made available to other firms on rc·
Thus the Uniflux tray is very inexpen- of vapor-closure pieces and tray stiff· quest. Rating of trays will be provided
sive. Experience tn the purchase of a eners. A suitable stock of 16-gn.ge flat by the refinery engineering division oi
large number of these trays indicates plate· can be provided for downcomer Socnny~ V~cuum, ::~.t the request of either
. costs in the range of 60 per cent of the ::~.prons, seal pans, and other miscel· engineer contractors or the ultimate
cost of conventional bubble-cap trays laneous tray parts. users. Such requests arc usually proc~
of similar alloy. esscd in a matter of a few d::tys, even
L!cettsing , , • Consistent with Socony- though they might include th.e entire
Uniflux In Service Vacuum policy the Uniflux fractionat- tray requirements for a complete pro<>
ing tray is being made available to the cssing unit, or an entire refinery.
Uniflux fractionating trays have been
in service on catalytic cracking faciJi ..
· ties at the refinery of Socony-Vacuum
Francaise in southern Frnnce. The ap·
plications at this refinery include the
t'eed·preparntion towers, the reactor cf ..
fluent fractionator and its strippers, the
amine absorption system, hydrocarbon
absorber, dcbutanizer, rtnd a side ..
stream stripper for recovering butane
cut from th~! dcbutanizer. This catalytic
cracking unit has a capacity of 'ap-
proximately 17,500 bbl. per day.
The fractionating equipment has per- by .n. c. Eld
formed in a completely satisfactory
manner from the start. While fraction- rpHE Gulf-designed perforated-plate- classified as low and high~trny-spacing~
ation is considered to have exceeded type tr~y has been in operation in type trays, which ·also are different in
expectations, it would be premature to Gulf Oil Corp. refineries for the past design and operation.
attach any quantitative efficiency or ca· 14 years. Fig. 1 illustatcs a low-tray~spacing.
pncity advantage on the basis of the It has been installed in towers, rang .. cap-type tray. This is L~e type oi ~:-ty
data that have been secured to date. ing in diameter from 2 ft.-S~in. to 10-ft. that most often is found in !~Xt booi:s.
The management in France is especially with tray spacing from !S to 30 in. A · A definite depth of liquid is established
impressed by the low cost of installa- tower of 13-ft. diameter with 36-in. tray and maintained on the tray, and the V~··
tion of these trays and by the very low spacing is about ready to be put into· por, passing through the fiser, revers:tll
maintenance rel!uirement since opera· operation. and the annular areas mert~!y bubbles
tions started. The tower pressures range from sub· through the slots, and the Eqdds on tr..e
stantially atmospheric to 650 psig. tray.
Developing specifications , , , In the Thirty-five towers equipped with Gulf Fig. 2 illustrates a high-tray~spaCing
United Stnte_s, design development has perforated-plate-type trays are now. cap-type tray. The vapor passe.$ thro•..1gh
been under wny in coopern.tion with operating, and an additional twenty~one the vo.rious riser and cap are:."\s. :\nd t:~e
tray fabricators, aimed largely at the tO\Yers are being built, most of which slots, with a consider::~S!y ~ig~~!" va?or
development of a uniform set of spcci· are scheduled to be opera.ting within a ve!ocity thn.n is the the cns.e for ~~1c
fications for the Socony Uniflux fmc .. year. 10\v~tray·spacing cap-type tray. Th·~ '1::>.·
tionating tray . .At present Uniflux frac· The Gulf perforated-plate-type tray is por velocily is sufficient tO push the Eq~
tionating trays arc being planned for a high-tray-spacing type tray (tray spac- uid up into the S!J:lCC b~twecn !he
more t~1an tOO fmctionating·tower in· ings above 1S in.) in contrast to the trays, so thnt the liquid th:J.t is c~a:-geC
stallationS in the United States and sieve-type truy, which is designed and to the tray through the do, . . n!Jow from
abroad, including the r''ltire fraction .. built by several other corporations as a the tray above, is immediate!y pkkcd
ating·trny rtquirt!ments [or two com .. Jow·tray-spacing~typc tray (~ray spacings up by the Y;l!)Or rising from the dlscn-
pletc reiin<ries, It is expected that less than 18 in.). . gaging areas.
soon a sufficil!nt ::~.mount of reliable The two types or g!'oups of trays are The liquid is held in suspcmion in
information will be avnilabte to more different in design ::~.nd in operation. the form of droplets above thl' t..ray :!s
qunntitativl'ir nsscss the anticipated P'!r~ llubble-cap·typc trays may also be it flow.~ from the in!t!t to the ot!tlet o;ic~
formancc advnntagcs of the Unifiux Atnhor is with Gulf Oil Corp., Pitts· of the tr::~.v. The slots :1re merelv l..!S·~
fractionating tray. ln the me:\ntime, bur~h. for d!striblttlon of the vapor. ThC'. ac~t·=:

3!l
 'fit Fl. Y DESIGN

.: .:SX!i;Y~§&,~IJ
,,. S',
. ·, I · ·· · ., I· · . • . .. • r j

Fig, 1-L()w•tray.spaclng cap·trve tray most ofttn

d10wn In te:q books.

·'
-----'vA~--1 .,., ......, . .•. . . . .... ·'. I
,Cd if;~~:'~·i::··;,:_:<:·{:>.·;..-;. . I

~- ''
Fig. 1-High-tra)'·spadng cap-type tray Fig,_ .$-Golf pedornted-p!ntc·t;vpe tr:ty,
· ha" hl~ln·r vapQ( 'Ttlocity. ......__ ,., hlv.h tray- s-pnciny,· for which weirs :trr:
uot required.

,,, '"' ,, ,,. "· .,,.,,,.

. i~:__ ,', •'I
''•'' ' I
•\:12. ~..!J.~;!J!..:..'U-•:•;.:1~' ~· • ~ .. ~~ SENGAGING AREA= 10,52 SO. IN,

l/£:~!~~~~~~t~
() AREJ\ ~ 14.2 SO. lN.

' Flg, 3--SieH·type tray1 !mV 'tray spacing, on wh!ch

more or ks.s violent Uolling nctlou occurs,

release of the vapor takes place in the

·disengaging area, and the cnp spacing
and shape should be designed accord-
ingly. Fig. 5 shows a typical Gulf cap
.. layout.·· · · ·- -
'Fig. 3 illustrates a !ow-tray-spacing ~''
perforated-plate-type tray. This tray de-·-··_
. ·sign is based on a· theory similar to that ··
· ~- usr:d for the low·trny-sp:idng cap-typC: · -. ·
--- ·tray·,· nnmcly, ttint a definite. volume o[- · · · -
liquid is kept on the tray; and the vapor · · · '-.L.:.:...:~.::c...:.....:.._______________-'-~-------'
•· ~meicl)' bubbles through the liquid, caws-· ::"'::.::.. ·- .. ~: FI~:-'5-Ti?k:tl b)'out of bubbk·e:lp assembly, CuH OU Corp.
. .. . .. .
•. TABLE 1-CAP-TYPE TRAYS- PERFORATF.D·PLATE-TYPE TRAYS_ ing more or less violent botling ~ct:on o:
Capao::illes of A-..niluble Vapor Areas tht: liquid, Quieting zones ::m.· pn."•!dl'd ·
. -TraY Spncing, in._,, ........ ·.. ~ ........; .•••... ; ........ ,;; ..... ~ 30 24 :· .. : JO- 24 ~ ?l inlet :md outlet v(_ the ..vapor are:.1 .
·- Densit•• Factor - DL - Dv - - .-· .. - •. -· .• - •.•• .. . : The ·holes in the p!a:~ ~\rC v~ ·in. or
· : : : : ; : . : · . . . . :.--:·::.·. . . :::;;~;.';';', 'lZO'
.;.,, • • • J •• - .•.•
180
' . Dv larger~ and the spacing is :1bout l :/J2
-·· ., ~p~fik~ in. The velocity of the vapor is fixe(
,.' , Capncities Vapor
cu. ft. per S<!cond/ cu. (t. pi!C second/ Area Required so that the resultirH-' oressure droo is ~-.t
C ':
. JY.-in. Gulf plnte·type cap ...... : • •..... S.)S
Sq. Ft.-%
100
Sq. Ft.-%
),'\9 100
%
100
%
tOO
least equal to the h~·ight of iiquid ;J~ :hc:-
tr:~.y. The height of iiquid equals th·:
·: ·-22.6 per cent free afen-petf. plato ...•... 6AS 120 4,1)7 l)) SJ 15
30 per cent free nrea-peri. plate , , . , , .... 8.75 163 6.34 · lS2 6[ 55 wc!r height 1 p!es the hdP,ht of 1iq~1:d ::s
J6 per cent free nreu-perf. plute , . , .. , •. 11.10 '2Q7 8.05 2J 1 48 ·!3 a functiOn Of liquid volume anJ we!:-
:: .•
39
 Tlll!."l DZ:S!Cl\1
far ·as the vapor is concerned. The flow
of liquid is also different. For the low·
tray-spacing-type trays, the liquid flows
over the weir into the downf!ow, anC
the length of ·the weir is of particular
importance. The size of the downfiow
area itself is, of course, also important.
For the high-tray-spacing-type trays,
the liquid does not, literally spca:,ing, .
•ot flow over the weir. It approaches th~
oJ
area above the downflow in!ct area,
where the preliminary separation of va-

r ••••••••• l.l--f--..!.r... --. --l--1- por and liquid takes place after the 'Ia·
per-liquid contact has taken place, o:-
rectly above the caps, or the perfonteC
plate. ·
Liquid and vapor mist falls into the
/" Nn AREA oF I downflow. The inlet area as well as any
restricted area in the downf!ow areJ
PERFORo\TED P~ATE ··12,45 SO. FT.
DOWNFLOW ARE~ ..... lo\.40 SQ. fT. may, in this case, be considered nn ori-
L,OS1' AR. EA····i····· ~SO. fT. ·fice, and it is the designer's problem to
TOTAL TOWER A EA 20.27 SQ. fT.
see that the height of va_?or-liquid mix-
r-\,~c·------1 ,.. . r---------1--~-,~.~--+i*ll ·'" ture Duid does not build uo excessively
above the various downf!Ow restrict~~~
areas-or orifices. Usually the inlet ar~a
FJg, 6-Typ!cal Gulf design, hv{n.flow perfornted·pl:tte type trny, is the imoortant area to c~nsider.
The b~ildup of liquid height in down-
TABLE 2-DETS.OPENTA~ER flows, in high-tray-spacing-(ype t ray')
length. Each of the holes in the plate
simulates n smal{ bubble cap, and opera~ Feed charge-J33 bbl./hr., or 8,000 bb!./dny due to pressure drop over the tray a~d
'I'my type-Env~lop<! tvpe resistnnce of f1\1id flowing through ~.:..:~
. tion of the truy is similar to the low· downtlows ·5¥.1-in. Gulf Gulf Perf, downflows, is usually not a serious
tray-spacing cap·type tray. Dubble Cnp J>Jate-36%
Tower diameter ...... lJ ft.-6 in.• 10 ft.-6 in. problem. That is often a problem lor
·... ..-
Tower pres~~1re-psig. 42 42 the low~tray-spacing-type trays, simply
Gulf design , , , Fig. 4 illustrates a Gulf Tray spacing . , , .. , . , 24 24 because there is not much height to giw
high·tra y·spacing perforated-plate· type Th~oretical number of away.
tray. The vapor velocity is high enough trays , . , ... , ... , .. , , 43 43
to push the liquid into the space be· Actual number of trays · 5S 58
Height of towi:r-ft. , , 116 126 Ynpor capaciiies , , . The vapor capaci-
tween the trays. No definite liquid level l04SS ties in cubic feet per SC!Uare feet of va~
Tray mnterial-typ" 304 SS
is maintained, and weirs are not re~ Estimated cost oi por area vary for .the various designs.
quircd. In fact, weirs do more harm travs-S .......... · 175,000 95,000 Obviously, the high-trny-spacir:g grou?s
than good, in that they would act as Estinlated cost of
tr<J.y installrttioJt-S • 33,000 19,000 of trays with high vapor velocities hJ'iC
baffles for this tray design, increasing Estinuted cost of a high vapor c:lpacity.
the height to which the liquid is pushed, tower shell-S . . . . . 36,000 26,000 The !ow-tray-spaccng tray design re·
which of course is not desirable. Estimu1ed cost of suits in short, wide towers, Jnd th~ hi:?h
There is no quieting ?.one. The total insulation-$ . , • , . , · 11,000 8,000
tray spacing tray deiign results in t:1!;,
available tower area is utili7.ed by the nnrrow towers.
Totnl-S ..•.....•. '!GQ,OOO 148,000
vapor area, and the liquid area, with the
exception of a small percentage taken TARLE 3--ECONOMIC COMPARISON OF TOWER DESICC{
up by supports and required material
thickness for weirs. T h c perforated Tower service Depropanizer Debut:!nizer
. ·•..•..•.
T OW(!r pressure-- lb s. ps1:1, • , •..•••••••••••......•••• · · · 35~• S5
holes are smaller, about 3/32 in., and
more closely spaced, 5/32 to 3/16 in. "X"' Cap Gut! "X" Cap Gut!
Tray type Type Perf. Plate Type Peri. P:::1te
Fig, 6 shows a typical Gulf design, Recti!ying Section
perforated plate tray layout. Liquid does Tow.:r diamt!ter, ft. , , r,,,,,,,,,.,,,,,,,,,,. 9Yl 3
not readily flow through small holes, 'fray spacing, in. , . , , . , , , , , , , . , , . , ..... , .• , 24
due to the surface-tension effect, and Nun1btr of uny'> .. , , , , •... , , , , , , ...... , • . . 1'2 '20
the adhesion of the liquid to the metal Stripping Section
surface, For this reason, high liquid
Tower di:lmeter, ft. .•..• , , , , , , . • • . • • . • • • • • • 9~ 7 s 5Yl
Tr:~oy spacing, in .. , , .. , . , , , , , , . , , .• , , , . , . . . 24 30 ~0
ratt> may be handled without necessarily
inr;reasing the vapor velocity to prevent
l'{urnber of trays .. , . , , , , , , . , .. , . . • • • . • . • . . 12
Height o{ tower, ft. . . . . • , , .. , , .. , ..•..•. , ...• , 55
t2
63
'20
19
'0
91
large volumes of \;quid 0 u m p i n g Tray nt:lterial ...• , •.•.. , ...........•.. , . . . . C. Stl. C.Stt. c.stt. .:. Stl.
through the perforated plate, rather Perf. plat~ material-alloy steel, type , ...... ·, ... . 116 )16
Shell thickness, in, , , . , . , ... , .... , ... , ......... , l ~ % & l·t/16 .y,
than flowing into the downflow ns it is Estimated cost o( trays (incl. install'n) ....... , , . S32,000 'SIJ •\00 SJS,OOO
supposed to do. Estimnled cost o{ sht:ll , , .. , .. , ... , , . , • . . • . • • . 47,000 ·n:1oo 16,'2QO
Opcrntiou difrercnccs • , • This indicates Totat ...... , .... , ...................... !79,000 SJ6,500 S6·1,'200 SJ5SC0
Difference in cost .............. , . , .......... . .S·:2,500 s:~:~.)C!J
tht.! difference in operation of Jow·!rny~ •For Tr:1ys of S.S. Type J16--Add S5,100
sp{1Cing and high·tmy~s.pacing tray.,, as tfor Tray.i o( S.S. Type 316-Add .$6,650
 Tll.ll"l DES!Gl\1
The··formcr ·group is purticulurly- .. here beirlg compared with an old, out· same. efficiency as bubble tr::tys, operat-
suited for towers thnt must be erected of·dnte cap-type tray design. This cap· ing results in many ranges hnve ind:--
inside buildings, culling for a limited type tray is fully up to date in design, cat~d higher efficiencies than would be
height. and is in competition with other C<\p· expected o( conventional bubble tr:1ys.
The latter group is considerably more type trays. Efficient use of the higher avaibblc
economical to build, but often is quite With further study and applications Benturi slot energv has resulted in
tall. Existing towers of this type arc 'S of these dnta, n cost estimate hn.'l been higher efficiencies i~ the ranges of bu-
hich as \40 ft .... worked up, a$ shown in Table 2, which tane and heavier· liquids. Incrc:1sed ef~
~\ c.ornpatison of capacities of up-to~ is a direct compndson of Gulf designs. ficiencics ar~ more pronounced in vi::~
date Gulf cap-type design towers, and The perforated tray is particularly fa- cous range such as absorption oils and
Gulf perfornted·plate-type design is vored in this instance because of the heavier. Butane separating efficiencies
shown in Table 1. In this table 'the high percentnge of vnpor area to gross well in excess of 100 per cent can be
trnys were designed for density factor tower area. predicted under many conditions. Ab~
DL Dv Table 3 shows nn economic compart .. sorber trny efficiencies in the rnnge of
== 180 son oi a depropnnizer and n 'ctebutani· 75 to 100 per cent have been -cxperi·
Dv zer, The given tower diameters, 9-ft. 6- enced, whereas bubble-tray efficiencic'l
which corresponds to an atmospheric in. find 8-ft. were recommended by in the range of 25 to 30 per cent would
tower, taking gasoline and (stripping) others, but these towers are now being be expected.
steam overhead. . built with the Gulf perforated-plate-typ~
The perforated-plate-type tray' is not tray, as indicated in Table 3. Slot velocity , • :Both the original Kas-
kade nnd the later Benturi have ver:y
similar slot velocity vs. efHciency
curves, the plateau of highest efiickn('y
beihg at higher operating ratc'l than
-------· .. ______ ___ . ---·-·----
0 " ,
usually allO\vab!e on conventional bub·
ble trays. Published ratings of Bent\!ri
tmyss nre purposely made con'5erv:>.tive
in many rnngt.!s and the mnnuf~cturer
should be consulted on final rating).
Flexitray
hy C. C. Th:d.H The demand for a trav of moder:\te
() THE Kaskade tray 1 2 3 4 which was Benturi tray . 'The evolution to the
capacity and extremely \vide opcrndn2:3
range has led to the developrn~nt ai
first used commercially in 1943, "Benturi" type, announced in 19525 the third design of KaskaCe known o.s
after extensive laboratory study, now after '2 years of commercial use, was a the FlexKray. ~This tray ?as ratings ap-
has approxim:'ll<:ly 350 installations in result of Cxtensive study of operating proaching those of a well-des\gned pt!r~
n.n extremely wide range of services results and design variables affecting forated tray yet retains an eiiicient .
from 7 mm. Hg absolute to 1,000 psig. plate efficiency 'and throughput. Th~ range exceeding that of conventional
pressure and as low as 7--in. tray simplicity and lower pressure drop per bubble trays.
spacing. 'theoretical plate h:1vc contributed great-
Author is with Koch Engineering Co., ly to its increased popularity. Despite Applications , •• Due to its simplicity,
Wichita, Kans. current design practice of using the
TADt.E 1 - PERFOR~!A:O:CE. OF LOW-
. PRESSURE ABSORUcR, TIF.:"\'TVR!
TAD[.E 1-CO)!l'ARtSON OF K,\SKADE WITH fiUfiU[.E TR,\ YS lN HIGH-PRESSURE
AfiSORJ•TION ... 1'RAY
rAbs.orption•~
Moles ab~orb~d*
Inlet Residue gns, mole · p<!r hom from -::- · " ··.-, ··: : .•: . ~· . ... ..... ·- . >(!oles/
100
gu.s, both per o.:ent gus Per cent absorbed Com· ..-Mole per cent~ moles Per cent
towers,
mole Bubble Xnskade. Bubble -Knsbde Bubble Knskade ponenr Rich go.s Lean gas rich ps on ft:o:d
.. _ · pe~ cent lf3.)'S tr_ays trays tr:1y.~ trays trays c, .. 79.37 s:J.1S o o
co, . .. .. .. ..... .. .. .. .. 0.60 0.20 0.40 .' . " C1 ...••. 6.45 1.13 Q •)
Ct .•.•.•. , • -·.... ... . . . . • ~9.60 93.69 94.09 120.00 124.00 13.4 12.4 CJ , .••.. 6.:'.0 4.11 2AS 40.0
C1 .. , , , • , • , , :::. :·. ·.,, ,·, . 5.60 • ·1.·1•\ 4.61 . 19,"21)· 19."20 J4A 30.6 iC-l 1.)5 0.33 l.O t 7-~.3
c>· .... : ................- 2 .•~) 1Al 0.7) l2.JO '20.40 50.6 75.0 uc, '2,)2 0.~3 ::.."20 9·;,,::
iC~- .....•.•... .-: . .. ,, .•. : o..t9 0."22 O.t1 3.10 ·1.00 6l.O .,. 1.5
nC~ ,.,.,-,,,,,,,,,.,,.,, 0,70 ...... 6.90 1.30 100.0 100.0 iCs 0.77 0,77 100.')
iCJ ••• , , , , , , , . , , .. , • , , , . 0.21 ~.:o 1.35 100.0 100,0 uc, 0.17 0.17 ~ 00.0
nCJ .~ ......••..•.....•. 0.2\ 2.10 '2.J5 100.0 100.0 C.11+ ' •• ' :un 2.S7 ~ 1)1.' .0
c. ..... ...... .. .. .. .. .. o.os 0.30 0.90 tOO.O 100.0
c,+ ............ ....... .o.os ...... o.ao 0.90 100.0 100.0 Total .. 100.00 .!00.00 9.50
Total .............. , . , too.oo 100.00 100.00 161,)0 131.90 .--:-nteven-trnv absorber---..
'fheore!i~.·nl
() Sp. gr .................. ,
Rich gas to tot'.·u, M.c.f.d ..
Lean oil to tower, g.p.m,.,,
0.6l9 0.594
!9.090
56.5
0.589
t10,'2"20.
56.5
- ...
, ...... _..
CJ --........ 2.
Per Cen!
tray5 (calo:;.) dfickncy
1S.2
P':r <:c:t!
'J.b5orC..:d
1.0
Prc::c<mrc, psig,. , . , .. , .. , , . 145 145 iC, ~:.-.-:.... 5· ·-· 45.5-
Lc:\n·oll temperature, '"F... 33 38 nc,. ' .. '''' 8 11.6

. •FrQm rkh oil analysis. tCnlculated (rom n::~.iduc ga1 m~tc:r rcadings.

4 l.
 Koch Benturl-type lr:tys nre
shown In these views. Column
llt left Is tqulppctl wllh the
F!e:dtro.y on the three upper
tr:trs. Au 111/2-ft.·diarneter
Hcnturl·type . K:15b<!c tray i~
shown lu photo ~\l1(1H, Typic"\
covp..-r ~on$tru~!lon of F!e:<i·
fr'ly for chemkt~l 1n{\ustry Is
-~hown at ri£hl.

TABLE 3-TYPICAL KASKADll·THROUGHPUT DATA

·,
Dlam,, ft.' , . , .. , , 6 {t, 6 in. 6ft 0 in. 6 ft. 0 in. 5 ft. 6 in. 5 ft. 0 in. 4 ft. 6 in. 2 {t, 6 in. 2 ft. 6 in.

Tower service , . , .Absorber De prop. Syn. Syn. Absorber Debut. De INt.• Deb~tt.
tower tower
Vapors:
Lb./hr. 117,000 148,000 32,430 74,700 121,000 83,000 )6,000 27,832·
pv, lb./~tl, ft. , , • O.Sl ).06 0.115 p.217 3.11) 2.34 2.0 l.S-~
lltid: ..
.p.m. . ....·... 1,070 860 )19 156 405 485 100 30
pL, lb./cu. fl. , , 50 29 1\5.1 4M 41,4 34.0 36 )6
Tray space, in. ". c4 24 )6 24 21 20 2·l 'l·'.

•Tests indicate greuter then 100 per cent troy effkiency; unu~unl !or such a small tower.

~ro-~TEMPERATURE
.,
~A" B(. NZOL 51RlPP(R Ci--tARG(.
•e" K;\ SKAO( DECK OVCRH(AO

•oo(·~~
SKAOE OECK BOTTOMS
l30L.!: OECK OV(Rtn:,\o
I I
300 I I I "C" BOTT~\<S
---r-=
J I
r-=-
Ji

20 JQ 40 50 60 70 90 100
P£R CE:-.tT OVER
Dlsllllntlon corns for benzol·cumene slrlppel'.
 '!'Rll.Y Dl!:SJ.Gl\T
low weight, and adaptability to mass ,,.r-TEMPERATURE Of ···---··- ...

uj
.... . -~--

production, this tray is rccommcnd~d

when the tray cost is inherently an ap-
,'
\_)
' preciable percentage of the total lower ,
installation costs, Typical installations ~ 00
I LaorLs

are .. essentially atmospheric or moder ...

.I ~I I I
ate pressure with carbon·stecl vessels, · I I I NO. e OV H 0. _;...........-

~~1==7
0
or above the feed zones in light·cnds
fractionators. Benturi trays are re I I
I
4

N0 • .5 OVHO,
quircd below the feed, due to the heavi- I ______,.. ! I I
I
er loading, but the lower loading above , 300
~NO.2 C~ARCt NO.4 OVHO,. - - l
the Iced docs not justify Benturi usc in .
this zone.
I I :,~
NO. J OVHO. .
'

I
I

~ :
The Flexitray, more completely de-
scribed in current Jiternture,e has lift·
able caps approximately ::! in. in diam·
eter, held down by plate stamped
"
t?. .... , ....
I I I I
I
..
'
I~
NO,
'

2 OVHO

·· I
:
l

I
i
I
f---1
10 0

spiders whose projections are clinched I I

on the underside of the deck, These
caps are spaced on n triangular pitch 0
I JO 40 !>O ~0 70
I
•o
' I
of 3 to 6 in. in a similar manner to
0 10

PER CENT OVER

"
bubble trays. The caps and holddowns A.S.T..~.'r!. cll.~tlllntloos of charge nnd proclud.'f on naphtha uoit.
are furnished in alloy even though the
remainder of the truv is of carbon~steel TABLE ~ll.EFJNERY DEPROPA.'\'JZER OPERATION BEFORE A.'ffi AFTER I:'ISTALL-
materials. • ING DENTUlU·TYPE KABKADE Tl<AYS
Depropanizer Depropanizer over· Stabtlized nnphthn
No hydraulic gradients • To insure charge"' hl!o.d gast disti!lnt¢¥
uniform vapor distribution at very low
1
Analysis, '!'olurne
per cl!nl~ Before Afler . Defore ,vter Before
loads on trays inadvertently installed N1 +co'] .......,....... ; 1.7
out of level alternate rows of caps are
1 0'1 •••.•.••...• ·, .• ,, ••••• O.l
of different weight, The absence of 0.5
hydraulic gradients and the fnct thnt
H'l ··················•·
CH" •.................•.
C'lH~ . ,, . , ............ .
..... ">:0;2,:-,
0.2
OA
the maximum slot velocity is directed
c.,rr.:. ................. . 7.l
() horizontally contribute appreciably to
the flexibility and high throughput of
c,H, ................ ..
c,H, ... , ............. . .... '
J.S
O.l
29.)
- ~-·-,_,.
63.3 ·.·.· .. \_•,
Flexitmys. : Total CJ cut ........••.. 2.1 1.9
0.7 15.3 lOA
The accompanying tables and charts iC• · · · · · · · · · · ·.-.· ·, ·, ·.-...
nCJ. .......•.. · · · ·•-::..·
· 19.)
o\A 2.5 2.4
indicate performance obtained with c .. paraffins ........ , . ; . 4.7 5.6
Knsknde trays. c" olefins .............• 0.1 0.1 .... 0.~
2.9 4.7 0.1 0.1 5.9 •\,0
The low pressure drop per theoret- iC• · · · · · · · · · · · · · · · · · · · · -· 0.6 - SA S.J
ical tray has made the Benturi tray ~:~ :::::::::::::::::;:. 84.1
6.1 5,) ~-

82.8 36.3 ~5.0

· attractive in vacuum flashers and many H<JS .. , ............... . o:2 30.9 "
12.9 o.o
heat-sensitive materials separations.
To1nt .... , ......... ·.. 100.0 100.0 100.0 100.0 100.0 !00.0
Besides the extensive usc of Kaskades
in the oil-refining, natmnl-gasoline, nod H1S, g./1.00 ml. , . , . . . . . .. 0.328 . 0.175 ',,, 0.0014 <0.0005
petrochemical industries, Knskades are · Tests:
be1ng incrcnsingly used in the. chem-. Grrwity, tA.P.J ........ 63.1 62.<1 62.2 GLG
ical industry. At this time a wide range . Vapor pressure , . : , : ::·. ··•. ;::.' .~ .. ~ .. .9.6 -· 1.>15 ' 1.548 6.4
. • (sp. gr.) (sp. gr.)
of materials arc being separated with Dist. (A.S.T.~;f.), oF,:
Kaskades. including: -· .. J.b.p. " " " "" .... 102, ____ 91 ____ ~·-- "" 110 100
Acetones 1 acetatCS 1 acetylene,.. ·alco-":·- --~5 ·per-cen~-..-.-.-134 136 .. , . . ... 1·13 13$
hols, amines, cy::mides, fntty acids to· per cent ..... , . . . 152, , IS4 16<.' \5·i
1
20 per c~nt .. , , , . , , . 1-:'S 178 172 l'.'G
formaldehyde, glycols, ketones, mercap- 30 per cent .. ~... c . . . . _ 1~1-- .... -193- t9·1 :%
tnnsl ph!!nols, polymers, resins 1 and tar~ 40 p<!r cent , . , . , , • , • 107 214 '2'.)6 'lJ-')

base oils. 50 per c<!nt .. .. .. .. . 220 ( 22S / 219 ~'?:6

. 60 per cent , ...•. , , , '131 --·-~~;__, ---~.:--.-.'...:-~. DO '.2·:0

Referen~::es ,-,.--~ _ _ _ _70. per ccn~---.--.---;-;·.-~·.--
.. -24-J _,~n '2·iD :~.5':.

1. H. Litwin, The Oil and , Gas J 1 SO per cent , ....... , '256 27':. 15·1- -::-:-u
..,,,n
ourna , 90 pa cent . , ...... , 17J 194 '270 -·J·'
~fnrch 22, 19-n.
95 per cent .. .. .. . .. '290. 310 _ ·..: ~_.__ _ _ _.,., ;s'2 )0-~
1. H. Litwin, World Pttroleu!11 1 • March E.~"--~--~~,·:=:·--:J36 __ _:::_ 3 36-· .:.--~~-~ J.,B "i-ll
19·18. ..:n.-cc:; vcrc.rnc--:-::~-=-~s~·~--~i::;.o ~-~.0 n.o
3. F. DunM Fuqua, Pc!ro!eum Processing,
November 1948, Yields, val. per cent .... , t21.16 127.46 §),2 §2.5 t20.50 t26. 79
4. Trnnsilctions, Institute of"Cheinlcnt ~n·
- ···---·--· lj%.3 ~9: .5
gineers (Dritish), Vol.)!, !')SJ, Riltes,, bbl. per hour ..... 375 2 t. 4 12 488 '2.1.511 J6J 476
M.c.f.h. M.c.f.h.
5. Thornton, Petroleum Processing, :\fay .. ____ -·-
1952. - - - - - - - - · - ·-- ·•tow tei-n'j)e'raturo !i-'aCtiOnat" clislillation. tMass sp~!ctrometer.' ~Oased on crude. ~n:o..~ed
6. Chemica! Engin<!cring, Nf:ly 1954. on dcpropaniur ch;Hg•:. Nt.\1C the incr..:-ascd capacity af~l!r Kasbde rcp]::~ccment of b·~Ob!c tr:-ty~.

(~ 3
 ' ,

... ~ct~2 ~.

THIS paper presents an accurate has been used widely throuohout the in the column depending upon the
method for calculntion of com· I
pe I ro.eum .mdus try to design
o ab~ value of the absorption factor for
ponent yields and temperature con~ sorbers. In the derivation of the · the component in question.
ditions ~n ordinary nonintercoo!t:d, method it wns assumed that the pres·
nonreboiled light • hydrocarbon ab· sure, temperature, and quantities of Edmister modification . , . More re-
oil and gas thrO\Ighout the column cently; Edmister· presented r~ modi-
1
~orb.ers.In the nnturnl·gas process~
tng industry, many cnlcu!ation meth· we~c con~t~nt. -These assumptions,
fication of the Horton and Frn.nk·
ods have been applied to operating whJie sufftciently valid for the theh· lin method. Edmister derived expres· .
ubso~bers, but the evaluations too prevailing low absorption pres-sures, sions for the effective absorption ·· ..../
i
frequently are inconsistent and il· led to serious errors in the desion ~actors in terms of the absorption
logical and leave doubt as to wheth· of higher~pressure absorbers, pr~rtfc­ !actors at the top and bottom of the
er the data, the equipment operation, u\arly .those treatino0 rich 0oases where column based on the assumption of
or the evaluation method itself is in a rc I attvely lnrre fraction of the rich a two-tray absorber.
gns is absorbed~ The above~mentioned absorber de·
error. The present method is sup·
sign methods consist primarily of
p~rted by a lar~e am?unt o( oper·
Effective Absorption Factor derivations of comoonent material
attng data covenng Wlde ranges of
ln 1940, Horton and Franklin' balance exoressions 'in terms of ab·
pressure, temperature, lean-oil mo-
derived an exact equation for the sorption faCtors. The evaluations of
lecular .weight, gas composition, and
fraction of component absorbed in the absorption factOrs, while ade-
absorption factor.
terms of individual absorption fac- quate for some situations, are based
Kremscr·Drown Method tors on each plate, In order to obtain on assumptions which arc not nee~
the factors on each plate they as· cssarUy consistent with conditions in
. Since the advent in the early twen- actual absorbers. To be of full orac-
tres of the oil·absorption process for sumed that pt:r cent absorption on
each plate throughout the absorber tical value, a method should in~lude
recovery of natural-gasoline compo- menns of evaluating absorotion fac·
nents from natural gas, the desi2.n wns constant nnd that the tempera~
tors in terms of actual absOrber con·
and evaluation· of absorbers hns re- turc change was proportional to the
gas contraction (shrinkage) across ditions.
ceived considerable attention. The Ho'v this may be done will be
first significant contribution to the the• tray. Althouoh
0
the .predicted
•
mn.~
.tenar balances and temperatures on shown in the installments to fo!low .
problem was in 1930. At this time The information to be presented w!ll
Kremser 1 introduced a method that each plate may tllfft·r considerably
f~om those aclUa!ly obtained in a
deal with an improved absorber de-
used the concept of "1\bsorption sign and evaluation procedure, in~
FaCtor" based on Raoult's l:'tw. In gtven absorbcr1 it is claimed that
the over-all component absorption clLJding: (1) empirical desi~n corre-
1932, Souders and Drown~ modi~ lations based on actual ~absorbe-r
fied ~remser's method by using ab~ agree.s closely with plate·to-plnte c~l~
culauons. data, and (2) the best nvnih:tbk cC'ui-
'?'PilOn factors ba!'d on equilib· librium and thermodynam!c data·.
num constants
Alternative method • , , ~Jorton ~~ld Reference-s
Rich ~nses, high pressures , , , The tive method Franklin also presented an a!terna~
that sim[)lified the use . I, Kremser, A., Nnt'l. Petroleum :-xews,
modified method is popularly known 22, No. 21 (t9)0).
as the Kremser~Brown method and · of individual absorption factors by 2. Souders, :\1. and Drown, c. 0., In9.
employing ,~e. concc:-pt of an effec- Eng. Chem. 24, .519 (19:'1"2.).
•California Research Corp., Richmond, tive ubsorpt10n factor for e3ch com· . J. Horton, 0, and Franklin, W. B.,
Calif, Paper pr~;sc:ntttl at C;l!i(ornia Natu· ponent. The effectiv~ absorption !ac~ !nd. Eng .. Chem., 32, tJS-~ (t9.;?)·
ral Oust.)lint! r\ss~iatit.)l\ mel!ting, Los An• . , 4, Ei.!mJs!er, W. C., Ind. Eng. Chern. J5
gelef, Octobtr ~.t), Iors are ~vn Iuat~c1 at different pmtcs SJ7~9 (1'J4J). · '
 ..-~·.""'",._.,..,.,, ...,.'"'"·'"'"'''-·--·-.-, • , A series reprinted from the Field
Processing pages of The Oil and Gas
Journal. Initial installment was pub-
···---· . ···- ~ ~ ..... ····.-~· ... ~·~ ...
lished November 9, 1953. Date of
final installment was March 15,
1954.

Y?rt.t~ 2
HE heal and material balance nnd he:tt5> methods will b<! dderred Until other A:::: area o( outside ·sudul·e of ,ab·
T ~ o( absorption will be: dealt with in this
lnstallm~nt o( the Ca!Hornia Re~earch
phnses of the design prrx:edure have been.
pres..:nted,
::;orb'er, sq. ft.
.\ H ::: totnl h"~t of absorption of o..b·
Corp. method for cak;ulating ubsorbers. ~orbed components, M. B.t.u./
Heat balance • , • Simplified absorber ·t.lay with refaence sub.:.cript~:
Material bnlance , , , The exact com- heat balances nre mnUe by equating the !0 for inteke gas
ponent materinl balance expressi0n de- products oi the flow rates, specific heal-;, DO {or discharge gu~
rived by Horton nnd Franklin for an :1nd temperature changes o( the process. LO for lean oil
:~bsorlxr of n theoretical plates· may ·be ·streamS· to the· heM evolved bY the -ab· ' RO for rich oil
sirnp!J'fied inh.l the following form: sorption o{ components from ihe intake S {or total shrink.r.ge
ps. Heat los~-:s or go.ins due 10 differ· AY for arithmetic average
cnces between column avernge tempera• AMB for ambient
= F = - - - ( 1 ) ture and ambiem temperature<; should also
b¢ included in. the heat balance. With Smoothed vnlues o( compon~nt h.:ats llf
intercooled ibsorbers, the intercooler dmy abs-orption are shown on fig. l. These
where: . is also included in the heat balance, The values were derived from th~ .. resptcli\'..:
ytn =mole fraCtion of componwt in in·· remainder of this study neglects the use componenl equilibrium charts by us.:: uf
o( intercoolers. Howe·1er, the methods
~~;~ e~:action
)'rm .;::;: of component in herein may 1x u~d in intercookr justi-
Equ:lli.on ~:
discharge gas fic-.ttion and localion studies. 1' 11 K 1 K 1 i)
ThtJ he:\t·batunc.: ~:-::pression ls as fo!-..,._:_~ ~.1?.;.~. .~:~:.l'?:t· '.-.T·!)· -~·
1 1
V w:::: mole~ o{ intake gas {.

Vrx. = mo!t:s of di~h-.uge gas lows: · ... wher~. ... .: ..... . • . :: ...
x.. ::: mol" fraction of light component· ..\!! ~heat 0( absorption, B.t.u ..'Jb. of
tn lean oil
component at' :w.:rag.e temp<!ra.
t.. :::: motes o( lean oil
0 F :::: (raction of coroponent absorbed
n:::: number o( .theoreticill trays + w~ Cp~ (Tan- T .;) ::..: ..\H.~ ·r ..
ture between Tt and T1
'1'1:::: absolute temperature, •.•R. Tern·
peratures used wete 540~ R.
· · ·- · · · (30~ F.) and sso~ R; t 120" F.l
The equation iS fn terms of an average - 0.024 UA_{T.H·- T.,Jlu) · ~·
eUcctive absorption -factor; Ar., and in- ·("")
- ;· ·X.'·:·
K.J .-=·equ.ilibriUm- constants corre·
c!ud.;~ a correction for the presence of where: · · "Spending to T1 and T1
solute in t.he oil fc::d 10 th~ ~bsorber in W:::: rate of flow, M. lb. per day R;:::: unlversa: ~:~.s constant, B.t.u.'
. lerm' of an ..absoq~tion f~ctor Ac. The:: C'p ::::: specific he;\t, O.(.u./lb.r F. · ' ·- - : .; .lb. mol..:/" F. · -·-
more d.:tailed expression~ of Edmi~tcr and :.. : i =temperature,, ~F. .. . . . M::::: molecular. weight of component,
Horton and Franklin fot the absorption U :::: hent·transf•:r C!X'ffiCient between lb./lb. mok
factors. Ar;. and Ac give a clut! to methods absorbt:r Outside surfa!.."¢ and the .
.~ of e\'alualing the::~ factor~ in actual ab .. a~mosphere, S.t.u./ sq. ft./ "F./hr.. : · • ."I:he. J~ :~al~es va_ry. st.:m\.ewha~ with tern· .
·~orber~; how~ver, disc~1ssi0n of th..:se .. Normally use. ).0. D.t.u./sq. ft./ perature, and if- the avemgt temperatur..:
Author~ with Cnlifornin Re~earch Corp., ·- o F./hr. of abSorption i:> ·outside the so~ ·I ~o~ F.
Richmond. Calif, Pllper presented at the range;-va!u..:s should..be calcullited dirccUy
C.N.G.t\, meeting, Los Ang..:k·s, Octo· [ SPEC!F'\C HE: AT OF' GAS from the equation ....
ber 8-9, . 8_._~~ L~./ 0 ;., C p Actuat. total heats. of absorpt!on b:'lsetl
on a lar~c numb..:r o( c:werimenta! ab-
•. - w. I IF OVE.R .2.0 VOL. Pf..R Cf.NI sorbd he'at: and materiat ·bll:lnce~ wr:rc
:~·: tt\HAT.100°f.,B,T.U./l8. ~... NONHYOFl~CARBON:~.~2.
or · found to· be gen.:rally in ·ve·ry g.ooU agro:e·
f r - USC Cp::0.2•1·rOR NON- / / \ - ment with ·computed total- heats of ;~b.
8 H'r'OROO.R30N .•
·-·.P.OHI!ON or..·.: ... :.
i
jl!l.
I V
X sorption based on compontnt he:"~ts ~'f
·.absorption read from Fig. L The !"Jr~t>t
GAS.-,c-~~~7f~1L-f~
deviations were with ·very !ow·pressur'"
I I abs.orber:J wh.ere thC "":eight rativ o( Jt'~\n
·. .-. oil-·to--lntnke· g:ts 'i."i '\'err:.h!gl .. In this
rang~. total.:_h,nts .of at>1orption com-
pute.ti (~om . t..:st ..dat:.J..'_arc..very sen.SI!I''t!
.in
\0_-inaccurac_i_..:s .ttmp~ra_ture. mea)UfO:·
· ntents.
SpecHic. heat values for nrious n:H·
ural g<1Se~ are shown On Fit. ~. Tho:~<
r· ~-Values .were derived from.~:-,·! enth".l:O\'·

N•C•e
30f---f-+- us: fOR c .. .,;\+)-"'<>1-'-.;:-i
I
r--+-+-l---f-n:
\ .. entroPy diagrams. of G. G. ~(own. 1 s?~­
cific heat Yfl.lues used for lean oil :~nd
~otr~l shrinkage are re-;re<.::ively O..'iO :wu'
r .. 0o\---'--;:,-';o"o_j~-,_,-1"'~~-'-'---,;';c,;_jl_-"z.~o"';;,~:.J '--'-,"'oo--'--,..,,o.Loo,-,--'-~.,.,•,.!.0-:0-'-:~-:,o"o-=0-' 0. tl5 B. t. u.! lo./ " F.
() -p

PRESSURE - PRESSURE- PSIG. 'Rtfet('nce

Flq. 1-Smooltled vuluu or C•lntpontnt ·r.t~. 2-Spt<:iflc 'hellt value{ for v:trlou~ 1. o·rown, G. G., N:G.S.M.A. F.nf_ineer·
la;.~U o( tU~orptlon, natur:.~l f.\ISl'S. ine Data Book, !951 edition, pp. S6·'7'2.
~·- .
._.,;~ :; :. . ···"·
 :
. ;
··~./

"
~!rntt~
by E. 3. Hun cmd E. Bnymond
r was pointed out earlier that this ble-poiill relation for bottoms-prod·
• nbsorber design procedure i11· uct tempernturc. Usc· of these rela·
lean oil from lean-oil temperature
.to nbsorber·top·section temper:\turc
eludes empirical correlations which tions is not applicable to abSorber (discharge·gns temperature) can be-'
are derived from actual absorber design where methane is a m~jor expected ·to correlate with the heat
performance. In order to be broadly . <:omponent. A large contribution to evolved in the top section. of the
applicable these correlations were the dew point is from small ~mounts , absorber pt•r unit weight of lean oil.
developed from the widest possible of lean oil of unknown equilibrium For reasons discussed later, the "top
range o( variables involved in ab~ characteristics and a large contribu~ section" may be thought .. of as the
sorber opcr<Hions, tion to the bubble point is from average of the top three. actual trays,
methane having equilibrium relation· or as the top theoretical tray. "'
Absorber correlations , • , Data were ships relative~y insensitive to tem- Fig. J shows the dischorge-gas
studied from 192 absorber test run$ perature changes. Thus, in order lo minus lenn·oil temperature relation-
which covered <\ wide range of pres~ establish the key temperatures for ship which was developed from test
sure, intake gas composition, lean· absorber design it is necessary cithci· dnta. 'The actual point devbtions
oil charncteristics, weight rntio o( , to resort to tedious lray·by~tray hc:1t from the correlation are surprisin~ly
lean oil to lnw.kc gas, and number, bal:::tnccs or to usc empirical corrl!- small, nbout ±3° F. m:\Ximum. The
design, and loading characteristics lations of, suit~tble nccu'racy that rc· average slope of'the line is 2,0° F.
of trays. Fig. 9, to be given Inter, 1.1uire little effort. \Vhen one of the temperature rise per B.t.u. heat re~
will· show the range of the investi~ key temperatures is estnblished, the lensed per pound of lean oil, obvi-
gation in terms of pressure and other re'adily is cakuhltcd by an ously because all" normal lean oils
weight ratio lean oil to intr~ke ·gas. over.all heat. balance using Equa· have specific heats near 0.5 B.t.u./
\Veight ratio rr~ther tht\n molal ratio tion 2. (Set! Part 2, issue of No· lb.!" F.
is used in cer1ain correlations herein vembcr \6,) Top-section heat of absorption
because the weight ratio is more im~ ~ At the usual commercial extrnc· can be calculated from the composi-
portant when considering absorber tion percentages for propane and tion and quantity of liquid in the
temperature rise, butane, absorbers in the 40-90U· top section of the absorber. :'Yfo!e
psig. range all have weight ratios of fractions of components absorbed in
Pressure·· rnnges , , • Treatment of lean oil to intake gas from slightly. the top secrion liquid are determined
the absorber data was divided into below 1.0 to 5.0 or above. The. from Equation 4 which is derived
operation in the 40-900-psig. range methods and correlations given here from the basic material~ba!ancc
and operation in the I ,400-2, 100· are not suitnblc for ~w~900·psig. ab·. equation and the equilibrium rela 4

psig. range. No datn were available sorbers operating nt weight ratios tion Yno = Kn 1: x 1, where YDG is the
in the 900-l ,.\00-psig. range which less than about 0.3. mole fraction of the component ln
is understandable because it is a scl· question in the. discharge gns, X on .
dom used' range between maximum Lenn~oil temperature • , , Because is. the equilibrium constant for. the
gas-transmission line pressures and lean oil is the principal he~\t·ba\ance component in questlon at the col-
minimum recy~;:liogTplant pressure~. quantity in the top section ·of the umn pressure and discharge·gas krn··
Certain corrdalions developed from column, the temperature rise of the perature (rop~scction temperature),
the data are applicable only to the and x\ is the mole fraction of t~e
specific pressure range in question f(ToG ~TJ..o), ot component in question ~n top-sec-
while others arc applicable to both 20 tion liquid.
pressure ranges. /

1. Temtina.l Temperatures • / X1 = Yw (VwiV 00 )

(1- F)
----

The key to the over-all absorbor

heat balance is the establishment of 0 I /1
either the discharge-gas or the rich~
A
i

oil temperoture. In light-hydrocar-

I
'
+ (4)
bon. ft.lctionator design the corre· • AG-~OG P~!G,--j

sponding problem can be handled

easily by application of the familiar
dew-point relations for the column·
0/
0 2
I
...,
I I
.j 1!1 10
where light fractions in lean oil ar~J
negligible, the second term of Equa-
tio;"l. '~ becomes zero. Trial und error
overhead, temperature and the .bub- TOP S£.CTION t\H, B.T.V./LB,
L£.AN OIL arc invclved in evaluating the nbovt:-
Authors ar<:' with California Rc'SJ!'arch expression because Kno is depend·
Corp., L>\ Habra, Cnlif. P:'!per prcsc-01cd Fl){. :'r-Disch:~r:.:~ S:ll-~· mlnU.! lenn..oU !em·
at C.N'.G.A. meeting, Los Angdes, Oo;;. p-<r.:thtre, us dcYdop~·d !rom lt.\i dat:A ovo:r cnt upon the dischar_g~·gas tt'mpcnt-
tobc( a.9, pro::.:.\Ut<: tllrli:C from ·~ to 9-00 pslg.. t1Jrc. However, this i'i not n tc<Jiou'\
 procedure because the equilibrium from which the contribution to top- are based, lnsofar as possible, on
\
constants of the components present section heat of absorption can be data for systems similar to norm:.~!
in the top section liquid arc not very obtained by converting from the.. absorption oil systems. Olefin hy~
temperature sensitive. . mole basis to :t weight basis and droc.:~rbon values may be obt~int?d
The total quantity of liquid in the multiplying· by the component llH by plotting log K vs. norm<~! boil-
i \ top ~ction is determined from the value obtained from Fig. 1. of ing p<~int for several paraffins :~t
'--.../ expression: Part 2. the desired pressure. Olefin value-;
L, Equilibrium Constnnts
arc then obtained by interpolation
L,=--- (5) at their own normal boiling point:-.
..... >:;;r=
"An eqcrilibrium-constant chart is
1
presented here for nitrogen. Other . References
·where !x 1 is the sum of mole frac~ charts wilt be given for carbon di~ 1. .Jacoby, R. H., and Rzasa, ~f. J.,
lions o( components absorbed in the oxide,' hydrogen sulfide, 1 and the · J. Pet. T<eh., 195, 99-t 10 (1951).
top-section liquid. The quantity of paraffinic light hydrocarbons, meth .. 1. Poenman, ~. H., and Katz, D. L.,
each componr:ont absorbed is L 1 x 1 ane through n·hexane. These charts Ind. Eng. Ch,m. 33, 530 (19•<6),

'r )

'i:

., .
·' .

.., "i .·

0 ::.

: ·'·. ,.

.,

"

· ....
..
; ·~

~:&~····.·.~·····;;:;·
I ...
.
~? ... .
. ..
. ·.·• •. •. 7;~.·
'
~

:: ... :-...: :

·-
... ··. .;.;;.•

·~·-~~3~·;··---
... ~41~'2
.. .
·~ ..
. ... .. .'·!
..
' .
..
' '
.,
,.
' .. ··.
'
. .'ic.' ,
..: ..... ,....,... :.. · ;··:--·,;_;•
..
..
;-.;--.-.•-. r !~ . l . .
!.'
'·
:: I· ··• ::: : •: :··I ... ,;., ,...· . :r·-) "·~
()
.
I' I
I'
,, .. 'I<
.::~1. !,' ,·.' I•' :'• !"'
. ~~·
•.· ~·.
' " I '
' ·' " ~
I
w
.' I: :j, ''.
lo) ,,~·
..~
•oJ
; . ,, ~

·~
';'·I I" ' :. I~ :'

"
Equll!bdum con.\f;tnf.~ for nlfro',!c.>n, Rtfer('nce: J, P~:t, Ttch. 19S, 99 (1952),

-·· ., ... -.·• ... - . -· ... : ' .....

·:~·

47
 poR absorbers operated in the . nary estimate of the average tray this plot should be within 10' F,
1,400,2, tOO-psig. range, the tempernture in terms of known of the correct average temperature.
weight ratto of lean oil to intake quantities in order to speed the cal- A more precise nvemge temper:\·
gas may be much lower than LO. culation procedure. Broadly, al- ture is requir~d for subsequ~nt cal-
,\t such low weight ratios the lenn though many other factors enter in, culations. Fig. 6 shows the average
oil is not the pri~cipal heat-balance the weight ratio ot lean oil to intake temperature rela~lon as a function
quantity, and, as ·might be expected, gns may be used to obtain reason- of top section hl)at of absorption.
no satisfactory correlation could he ably average temperature estimates. This plot' should be used in oll cal·
developed. for tho temperature rise Fig. 5 shows an average temperature culations made r\fter the first trial.
in the top section o( the column. relation which is suitable for first · Considering Figs. 3 (lnstaUment 3)
However, on the basis o( a limited trial cakt1lations. EstimatCs from and 6, the following may be devel-
amount of data, a satisfactory corre-
lation was developed for the tem- ... ' ...
perature difference between rich oil
and intake g<"~s.
Fig. 4 shows the
relation as for t,400·
devdopcd
2,100-psig. absorbers.
2. Column Tcmpernture Gradient
Extractions of light hydrocarbons, SUBTRACT fROM
particularly that of the key com· r (TAy-~~),•f,
ponent, are known to be tempera-
ture sensitive. From a study of a
large amount of actual absorber
•
6
T1G ·TLOAPPROACH ,
C0R~EC110H
I I I
l
·
temperature gradient data, it was
found that the arithmetic average
tray temperature_ was required for
proper evaluation of component ex~
tractions. For this renson, correlu-
tions of average tray temperature
o,p==""
were developed, RATIO LB. l[M.' OIL
, U. HHM[C.AS
I '
'
Avemge tray temperature It is I 1 ) "' ~ ~ 7 ~ I 1->
desirable to hnve a good prelimi~ TOP SE;CT!ON 6H, BTU./LB l.EAN OIL
Authors with C:l!ifornia Rcse~n:h Corp .. Fl~. 6-Average tentpemture relntion ps function Df top sec•
La H:1br;1, Cnlif. Po.po::r presented at !ion hettt of absorption, Insert ~iYe'i :.tpproxima(e corrr:clion
C.N.Od\. meetin~, Los An~cks-, Octo-- for hcnt~xchnn~<! dfect due to diff('rcnces between i11l.:l ~p~
ber 3·9. ~ • and oH h:mp~rJlun.•s.

__~(T~o~R~·-T~,o~)~,·-~~----,------,-­
,_-
"r

"r--t-----+--- ---tY'----t--
\400 • 2-)00 PStG.

J I T,o .:co)'r.-r+-
T""7 I o
I i'
•" -~ • ~2
VOL. PER C~Ni
4 10

c2c•:
... 1- IN INTAKE· GAS

o•=~l
.

,) \) ~ I 0 I ~
.
l 0 ~ I ~~ ) I l ' .,
RATIO LB. LEAN OtL/t.B. INTAKE. CAS RATIO LB. L(AN OlL / L8. lNTAXf: GAS
Fl~. 4--R~hlllol\ for temperature diffnence between rich oil nnd In· FIJ.'!. 5-1\\'crn:!e l~mpernlure relation .stli!..1blt for (trH
tnkc ~a~. lrl.al cnlcul:ttion.
 oped: (a) For example, from Fig, 3, tntys. In effect, Fig. 6 serves to tire column. The temperature in-
a 4.0 B.t.u, per pound leun oil, top cvnluntc the obc;crved ''bulge'' in a,b· crcnsc· is large at the bot(om of the
.section or c.lbchurge gas temperature sorbcr~temjXraturc gmdicnts. column, nnd fades out ~s _the top
is found tO be T 1,., + S' F. (b) If a
() straight-line tcmpcruture-gn\dicnt rc· Correction , , , The insert on Fig. fi
of the column is approached. Thl.'
net effect is a considerably hight:r
lationship existed' between the top gives on appcoximnte correction ... fOr average column temperr\tore. r\t
and bottom of the column, T A" ;::: l\ hent·exchnngc dfcct due to dif·. higher lean oil to into.ke gas weight
Cl\o + 8 + T".,)/2 or T,,v''- fercnces bctwtcn inlet oil and i<1let ratios a relatively steep temperat'..lrt.:
{TL0 .+ 1'u0 )/2 =: 4' F. (c) How- gns temperatures. 1\t lean oil to in- gradient is -produced near the bot-
'· ever, from Fig. 6, T,w - take ga~ weight ratios near 1.0, l.l tom of the absorber with relntivdy
(TLo + T" 0 )/2 =: 6' to 12' F. de· relatively hot ·intake gns incrcnses \ittle effect on the over-an aver~tg·:
pending on the number of actual the temperatures throughout the en- column tcm~ramre.

:~: :~~~~·.:-d/
:W:lm, -5~&~~~~u~~~r·: ::: ~'~ ·. ·. ·.·; ··.:.~.'~:
.·.· .: ,:,.:.., "'· ·,' . :,· •, • :..:-.~.,.·· :.....·,·
~82

! :
,.
,_.;'

':
....
...
:. ., . . !·:g'·~:~::.:·
.:_· -·-~·':. ~~~~!-
:·. ·; · _ . : : .;. :: ... : . 1..
:,.;!:
.;;:·
J

:'
: ..

'I
I,,
I.
. :; .

, , ,I

. ': ·. :'
·.:

C.>

I':'.:.;.:~. ;X . : ..,-;,. :,-;;j

: /.~'( . xo·
1 ....
. ::::;:.~~:
z·· '·· ·.·
~ ... I . "c:\.
. 0'?--";.,;:·:: ..,...
. .: ..
·~:
' ..i: .. . .::.:p: : ... . .....
··:
·, i' :'
' .. ·, "''.,, .

':
'
Equltlbr!um con.\!ant1 for cnr'oon dlo:l:!de.
 )
·..... /

A CORRELATION of average with points nlready located in (2). as a function of a range of absorp·
tray temperature suttn.ble for the The shifts mlJst always be mnde lo tion, fnctors. In order to compute
design of absorbers in the 1,400· meet the known T.\v for the entire absorption factors on a given tray,
2,100-psig. range is shown on Fig. column. ·Gradi'cnts made in this Horton and Franklin made the Js-
1. On the basis of a limited amount manner were found to' be within sumption that the per cent absorp·
of data, it was found that the pro· about 2' F. of actual operating tion on each tray is constant and
pan~·and·heavier content of the in· gradients at any location in the col· that the· temperature· chan3e from
take gas satisfactorily expressed the umn. tray to tray is proportional to in-
"bulge" in the temperature gradient. 3, Effective Absorption Factor take gas shrinkage across the tray.
The insert plot on Fig. 7 is an esti· ·These assumptions may be expressed
mated correction (or pressure .that In the discussion of the basic ma- · as follows:
,is based on the approximate ratio of terial balance Equation 1, it was
component heats. of absorption at pointed out that the detailed ex- (6)
the low and high ends of the range. pressions of Edmister and Horton
For all absorbers, it the actual and Franklin for the absorption fac· nod
entire column tempernl\.1re grndicnt tors Ar. and A 0 give a clue to meth-
is desired in orde-r to make inter.- ods of evaluating these factors in y HI +
1 - y nl· T 01 -Tm-1
coolcr studies the following steps operating nbsorb~rs. Their e~pres- - - - - - - =: - - - - ( 7 )
·~ are necessary: (1) a plot is made sions for A F. show th!\t, !or light .. Yw-Voo TRo -Tt.o
having temperature as one axis and components with low absorption fnc-
theoretical trays (number of actual . tors, the effective absorption factor where:
··~···
trays divided by three) as the other should correspond to conditions at Vn~· V10 are a~ given previOusly
axis; (2) T 00 is located at the top the. bottom of the column, and for Y01 ::::::: gas volume leaving tray m
theoretical tray and Tno is located components with high absorption
factors, the effective factor should
RO =
rich oil
at the bottom theoretical tray; (3) LO =lean oil
knowing T.vv for the entire column, correspond to aver~ge condition<> in m ::::: any tray in question
the average temperaturt of the in" the column. The expression for Ac · n = total theoretic~! trays
termedin.te theoretical trays is com .. shows the same relation for light
putcd by simple arithmetic; (4) the. components with low absorption Edmister used Horton and Frank-
intermediate points from (3) arc all factors; however, for components lin's metho,d of computing condi-
1ocated at constant temp.::mture, then with high nbsorption factors, the tions on a given tray, but derived
shifted to meet a smooth gradient factor 1\ 0 corresponds to conditions the following analytical expressions
at the top of the absorber. for the iactors Ar. and A 0 by ::\ssum-
Author1 are with California Research Horton and Franklin presented a ing that the absorber consisted of
Corp., L..\ Habra, Calif. Paper presented
l.t C.N.G.A. meeting,. 'Los Angl!les, Oc· table that shows the tray correspond- only two trays:
Iober S·9, ing to the effective absorption factor
A, =: (A, (A 1 + l).+ 0.25)'"
. T -+T )
MULTIPLY (rxv- R02. 0<'\
-0.50 (3)
VALUE: BY THIS fAC1'0'R and
A, (A 1 + I)
(9)
A,+ 1
where:
A 1 =absorption factor .on top
theor~tical tray
A 11 ::::: absorption factor on bottom
tray (tray n)

''The methods of H o r to n and

Franklin and of Edmlstcr are prob-
0 0~ 1,0 ably as good as co.n be developed
RATIO l.O. LCAN OIL/LB. from theory and still hav~ practic:tl
Fit- 1-ComllltJon o( .ucrv\."t tr.ty l~ntptrnturt I>UIInblc for l:lhsorbu clulill In 1,400-o s!mpliciry: however, the mcthoc!s
2~l00·P<I'IC.
l'!lll:l;t. Jmo:rt 0Ht t~·uumttd corrtctlc;~n (or prt.s.nrr.;. have the following limitations: ( 1)
 nctu<1l internal conditions muy vury und Fn1nklin''i table for locating the mister's assumption of a two·tray
considerably from lho'c predicle<i effective :lbsorption factor is not a absorber is of .necessity an over~
hy Equalions 6 and 7: {2) Horton continuous function, and (J) Ed~ simpl.i fication.

()

.~·--~ .F.qulllbrlun~ <.'On~!unl In ab~orpt!on oU £or hydro~en sulfide (Reference: J. Pel. Te-ch, 195,· 99,
1951). 'fhe refe-rence {or t11e C:lrhon dlo:tlJo: dt!lr( pulltbhed hl th.: precetlin-g ht~!:.~:lmenl oi
thl1 '><:'rle.1 W:l'i: lntl. ~11\Z. Cht-m. JS, SJU (l'>'..U).

s!
A METHOD of evaluating the fac.
tors Ae and Ac, which over·
comes the limitations of previously
discussed methods, is shown on Fig.
8. The effective absorption factor
A~ is presented in terms of the frac·
tlon:\l distance between the bottom
and average conditions in the col·
umn, und the (actor Ao is presented
in terms of the fractional distance
between the bottom and top condi·
tions in the column. The values of
Tno, Toe• TAY• (LIVJ.""' (L/Yho•••
and (LIYJ.w to be used in Fig. S
arc developed as follows:
THat Toe, TA,.-rich oil, dis·
charg~ gas, and :uithmetic avemge
tray temperatures are obtained bv
methods explained in previous SeC.
tions,
(L!Y)uT>!- a preliminary value
for first trial calculation purposes h ~~ ,5 .G .1 :Q 1.0 1.5 2 3•\561810 15 20
taken as (L, + S)/Yw; where L, LB. LEAN OIL
and Y IG are ns defined- previously, LB. IN1AXE G1\S
and S is total shrinkage. This value
Fig. 9-lL'lnge o[ investigation.
shoUld be correc1cd in, subsequent
calculatio.ns· to obtain (LI Y)nT>I in
per theoretical tray in the column. 4. Over-all Tray Efficiencies
terms o( the vapor leaving the bot~
1om tray. An approximate multiplier Values of this multiplier vary from Actual component yields were de-
1.005 to 1.035.
to obtain the corrected (LI Y)nT>I is termined for a!\ absorber runs h:w·
equal to 1.000 plus the fractional (L/Yhor-Equations 4 and 5 are
ing requisite analytical data. lnCi-
shrinkage o( propane and heavier used to obtain the quantity of liquid
in the top section o( the absorber. cated over·all tray efficiencies were
Authors are with Ca\iforni:J. Research Then, (L/Y)TOP is simply, L,IY oo• then Qbtained for componcn~s ·,ye~l
Corp., L:o. Habrn, Calif. P:lp.:r presented away from the infinite plate line by
at C,N.G.A. meeting, Los Angeles, Oc· (L/Y)xv--the 11rithme1ic average
tober 3·9, 1953. of (LIY)nT>I and (L/YhoP· relating the calculated :1bsorption
factors to the actual vietds and usi:1g
the Kremser~Brown a·~sorpt!on chan,
I
Fig. l 0 (next page). For ent::-·~ ;xes·
I sure r::lnge from 100 to :'.,lCO ps!~ .•
'
I vv-!
A--.
'
I
' average over·all tray efficiencies ob·
tained were 37 per cent for ?ropan~.
l/j and 33 per cent for !he butanes.

v
/ I I I Tray e(ficit::ncies for propane ranged
from JO to 38 per tent, :\nd th•.21$.:?
/I !' for butanes rnn~ed from 23 to Jt)
per cent. Consid~ri1ig the wide r:,n;~
1/
I I I
of number, design, and lo11ding ch:lf·
acterit':cs o( !favs covered in the

l7
.., ,,
"' O.f
Ac'·' on
1.0
Ac
I
I I
abson.. et ~est data: the individual ao.
sorber efficiencies were rem::".r'<ably
consistent. No column showet! 1.!!1·
usually high efficiency. ::".nd ihe one
V!SC ORDINATt AT Lr;FT FOR Ac usr. OROlNAT!. AT RIGHT rOO Ae column which showed nbnormu:;y
Fig, 8-Er!tdlve nhrorption f:<ctors. low efficiency was rcpbced.
 ricld Processing (Continued)
. ' -. '·.

•,

;N
I
Y--'----?'"--1 ..,
'"
I

.~-1-f~-1-f---1~1~~~·C-~~L_----.:~.

ri-'-'r.-·t-c,..:.:..;._,-, --~----:r'--+f-+-8-c:::,.--i-~,..L-=.,..:--:___;'.-"
.
I '
+.!c-+.-'4+--!o----;,',.-,- - - - - ! ( ( ' 1' =~
"• : 0 ~...;: .;... ~...;
(
.. ~
\..._ /
: <)..· ~--.

1
'001 X ( ---1...!:..!.'.':'.\.N6t:l::)Vlll.X3 .tNJJ llJd
. v-1+\JvJ

::: : ;- ;' ' .

 ·,
)

5, Lean-Oll Molecular Weigbl~ A.S.T.M. distillation 50 per cent curacy. On the other hand, errors in
temperature is shown on Fig. \ I. ln the usual rnndom experimentally de·
THE interrelation of lean~ oil . absorber evn\tlution worX., it is tcrmtt\ed molcctllar weights may be
molecular weight, gravity, and preferable to use V~\\Ues from this as high as l 0 per cent.
All!hors are wi1h California Rese:~.rch chart instead of sing:e labonHory Methane . , . Katz and Hachmuth
Corp., iUchmond, Calif. Paper presented
at C~\ifornia N:.t\ural Gasoline 1\ssocia· values. In our experience, the chart values from Lnd. Eng. Chern. 29,
u'on mce\ing, Los t\ngdes, October 8-9, always has given rcosonab\t! ac- 1072-1077 (1937) arc used with a,

,. .'

'I '· "

!.

,, ' [:C:'.: .,,

:r:.J '". :::'. /

h~i· l:c: · 'i:o'::t'~' h~"'

·'
 correction for the characterization fects of system boiling range and terns, hence, corrected methan~ X
factor of the liquid phase. The cot· density, and )s b~sed on data for values should be applicable to a wide
rcction factor thus include's the ef. methnne in n wide variety of sys~ range of. absorption systems.

10 ~
,. IM .,
~ I~
"" "" ""'
. ,•
!
J-..''-·V<2
".;'7' .
.. '""rri-'r
100

" " '

I ,
,.
!' I I •'
''' '· ·I:·: ''pit ·•·F~•;c ';1'··1 ' '.
I';·! 'I I .. : '! .. '
., •I, I '·'•. . , . :· <O '
' ·~· ...
.· ... ' : ~~~ ~}@:'"
. ''" . ; . . . . •: . . Y.l

10

fE2'.:;
&0 • :ffi . ' ..
. .ii·'. ·.•. .. ;
·,•c•::

~~411R1 M
.... ';::
, ..-... ' . ,_
............. '. ;· ·-· ··=-
....... : -- .. : , ..:

lr-:
! •.

,.
' '.
. \.: i
:''1
'::".'•::.' :-..:e
;, ., • . . ,., j .•

.~~-:~~~·.·.···· .::.,·:.c ... ":~-·i'

,,,,.. - "'.."'
;;; .
1

'
•
""'
. . . . . .. •:. ·~'~···· '
. •:S':c:._,,,_.,;:;.'£· '~":
'
'"·'··
• C'""~;o·' "';
'
• n~c_;:~;c·· '"' !:::'"
:• •
'
' . .' .
~-·~·.
·.I ..

() '

,·..
 E:umplt dtt1~n enrcnlu!lons , , • Althout:h D. ln!a't.e gM: Q, Comput.O tlrtt tn!.l :'!lnnXage using
abwrber design method outlined in preced· thc.orotict! tn.)"l :::::: tenl:l.t trays X 0.3) ::::
ing installments is c:~puble of e:<ceHent .-.c 4
~---V,o-----. 6.6, S<>':l ctlcultcion t9.btllut.ioru on Table
M,c.t,/ Lb/ M. lb,/
~uracy, it hns the admitt~d app~ar:mce of to b<l ~ in :1. h.tu installment, Caku ·
being cumbersome to U$.1!, AcmaUy, only Vol.% day M.~Lc.f. day !:t.t«< 1-llhl!.S" ).s cloa-e enough to proceeJ
)~ hours' time is rtquired for a complete co, . '.' 0.50 135 116.2 22 with moro Cel.lil.e-:1 c..-J.cub.tlo!l~
solution. This time comp ... rcs quite favor• c1 .... . 3J.73 30,998 •IM 1,31·1
11 bly with other less p rccise methods, con~
c') ..... . 2,901 79.4 2)0
H. Now, Vw = )1,000 - S = ·)7,000
sldering that a complete he::tt and mate· C:t .•.. ·.' ·1,99 1,346 116.4 215
iC"' ..•.. 0,68 252 153.5 39
rinl balance automatkallv derives fror:n
oC"' ...... \,1·1 422 153.5 65 - 3,00{ = 33;)96 M.c.f, per day
this mrthod but not from' the other sim--
plified methods, In order to illu.mate the C,+ 1.01 )96 •Jt o.o 33
CC~atpttt.t .eo.o~ fnctioru of components
absorber desi~n pro..::edure a sample ct.l· shwrW.o::! in tO}' ll!Vtion, MJ from the~c
culation is presented bdow, The examp\o 100.00 J1,ooo· 1,963
mole fru.;ticru, coroouk too ~ction tOt:ll
il!ustr:Hes the procedure for absorb~rl iu hett o( J..btorption, · '
or-Estimated.
th~ 40·900·psig. range. Calcu!:ltion f.ttp1
for absorber~ in the l,•100·2,100·psi~.. r:a.nze M, M.
are outlincd following the cxnmpk. lb./ B,t.u,} D.t.u./
r--Yto X (V1o/Vo-o) X (1- F)/K:.-4---. :t1 Lc"<r C:ty lb, day
!, DI:.SIC:-1 PROCEDURE IN 4~900· co, ' " " (0,0050) (11,000/)),996) (I- o,OGJ)/2,)5 = 0.0011 1 1 . 69 69
PStG, Rt\1'\G"£ c. (0,3J'/~) (1.0884) (1- 0.0291)/6.3-'. ::;:: O.D% ?OG JO 65 1,95 1.'
c, (0,{)7S•~) (1,083•\) (1-0.1..(0)/LlJ. = 0,0556 ~21 2::' l·t.~ 3,163
llASIC DE.._'HCN INFORMATION c, (0.0•\99) (1.0334) . (l-0.3t.G)/O.~ == O.tXD.~ 3.St 41 DS 5,65S
iC.-. (0.006:l). {t.0384) .(1 ~0.76.:5)/0.132 == (!,0075 JS 6 lJ:'.. 192
A. Intake gas: nc. , ..... , (0.0114) (LOSS4) (1- 0.900)/0.171 = 0.0012 JG 6 lJ:l 192 ·...-·
l. RIHe-)7,000 M.c.f, per day {Vro)
2. T.:mp<-rnture-15~ F. tTrn} 1..1:1 == 0.2311 t2,.~:9

), Compo~ition-votume per cent 3,6)5

C. Weight ratio lean, oit per intake g:u L.
L, = = = 5,056
Carbon dioxide .. , . , . 0.50
Methane , , ... , , , , , , , 83.18
= 2,016/1,968 = t.02 1-).::u 0,71S9
' ' Eth;~ne .. ., ......... 7.84 First trial average column temperature
Propane , , , . , .. , . , , . 4.99 1. Top w:tion heal of absorption pcr-
!~om Fig. 5:
!sobutnno , . , . , . , ... , 0.6S pount! oL 1;:..u1 ,a ::: 1'1,4'29/'2,016 : : : : 6.:!
n·Outano ...• , . , , . , . 1.14 T.u·-TLo::: 30 4 F.;it,O:::: so• F.;,', Tu From Pig. 3, Tw - 'fw ::::: 1·~o F., 1'1 ••1
Pentane, plu! , . , , , . . 1.01 .= 110 4 F.
:=:: ~o· r~ .. ·r:H : : ; 94" F.
J. Cornru~ 1'1 0 b'J over all he:>.t anJ
4

Total , , , , . , . , . 100.00 D. M.c.C. txr day lean oil materi1J bfh~:

M lb./day X !73.6 2,016 X 378.6 M, lb./day X cu, (t./lb, mole
B. Lean oil:
t. Rate-200 g.p.m.
=
Mol. wt.
=
210
(M.W,)w = M.c.{./dny
'2. T.emperamre-30 4 F. {Tt.o)
). Gr:wity-J6.SQ r\.P.I. = J,6Jl
4. Molecular weight-210
(1,96! - m) J7S.6
5. Vapor pressure (pre~enco of light E, Try intake gas shrinkage, S = 3,200 =--------= 18.)
J7,0C.0 - ),00·~
hydrocarbons)--neglec t M.c.f. pu day

C. Column pressure-620 psig. F. First trbl (L/V)rlTl.t ::: (Lo + S)/V1o

Spt"...cilic he:tt o{ tli.o;charze gas = 0.590
front Fig. 2..
D. Number of actual tr:lys-20. J,63S + ~.:.oo
As3umo 6-(t.~e!er column w!th 'Z·ft.
E. Ambient temperatur.c-10QF, (I'.uu) = ------:::;; 0.135. tn.y tp:tcint, 0tcn :tpproximate ·coh:m!1
)7,000 s.tc.!. = 3.1~ X 6 X 20 X ·2. ~ 75~ sq. ft.
CALCULATION STf.PS
WLO Cp1.o (f:o~o- Tw) +, Wot Cpw (!\.~- 'f1<~) + W, Cpr (f~~..o- 'f1o)
A. M. lb. pel' day lean oil
::: L\Hs- 0.02-<UA (1'J.T-1'.un)
g.p.m. X lb, per gal. X min. per da.y
= (2,016) (0,50) <T>o- SO) + (l,<"ll) (0.590) (9', -15) + (125) (0,65) (f,o -75)
1,000 = '.0,426 - (0.01~) (l.O) (75<) (1 \0 -10)
200 X 7,001 X 1;440 t,OOS (Tll.,_ SO) + 18,411 + 211 (T1.o- 7~) :=. 40,.(7.6 - 2,160
:::; = 2,016
1,000 !,COS T11o- 80,640 + 1¢,418 + 211 TJ..O- lS,rl~ =:::: 1..0,416.- 2,16lJ
•Calif<Hnia P.esenrch Corp .• L.t Ht.bra, !,219 T,o = tt6,lt3
Calif. P:\pcr prescrHed at Califorub. Ntt·
ur:.l Gasoline 1\ssodntion mectinr., Los T~o =: 93• F.
Angek~. October 8·9, 19.53: (Continued on ne'-1 page)
 ·TUJ+T,,, 30+95
Field K. = =sa· F.
what by pha~ compo~1~10ns and, !or thti
ro:ason, olculated ethane yields m>ty b<
' I 2 2 ~lightly in error fo.t some sy.stcms. How.
\..j Processing F.thnne Katz Md H.ilChmuth (fnd.
ever, 1his is not considered to be too irn•
portant in that ordinarily ethnne l~ not
Ene- Chcm. 29, p. 1072, 19J7) Yalues are now :.\ commercially important k.cy com·
(Con-tinued) uscJ, Ethane X. nlucs are affected som<:~ ponenr.

()

()

57
 '
i

t.h~ (L/Vho~> + (L/V)ano~ only minor changes·, therefore, no addi·

T• HE4U.900·pslg,
S3mple design calculation for
Hu\ge is conc\\ldc:d in P, (L/Y).v =
,),
tiona! c:dculations are no:ccss.nry nnd the
heat bnln.nce valtlcs. already calculo.tc:U
thls irutallm.:nt,
I... From Fig. 6, nt 6.2 B,t,u. per pound may be: taken ns is.
0.149 + 0.180
lean oil and 10 trays: = -.---
2
= 0.164. 2. Dl'.SICN CALCULATION lN 1,400·
1,100-PSlG. R.A.:--I"CE
TU> + T)lo The s.teps arc similar to those in the
T.v- - - - - = 18' F, Q, Compute sec on d trial shrinkage 40·900-p-;iz. ranee except:
2 · Y:aling temperatt.Jrl:!:i from T11.o toT ... " and ,\, Rich-oit tcmper:\Cure, T :ao, is oO.
U~ing T11o ,from Step J: {L/V) ratiO$ from (L/Y)BTlol: to (L/V)H talned directly from Fig. 4,
:H.:cordtng to Fig. 3. Usc 37 per cent effi· B, T .n• is taken equnl to T no and Ci\1·,
To\., = 8S + 18 = \06' F. cier\cy for prop;U1c ('JA theoretical trays)
and 31 per cent dfickncy for butttne:s (6.6
culations are made to a ch~ck on assumed
shrinkage. Care should be ~a~cn to de-·
M. (L/Yho' = L,;y.,. = l,Ol6/ll,996 t.heoretka.t trny:;). t:ornpm(l rich-oil mo· vdop predse K values for melh~U\e, and
= 0,1•1?, kcutu weight nad A.P.I. gnwiry to oir to· run all c:tlculations to esS<!ntially p<r·
N. 0 b t a In mul!lplier for corrected tain correction fao.;tor for methane equiUb· feet checks on each assumed !!hrinkagc.
(L/V)anb Mu!Uplier is c:qul\1 lO 1.000 rium constant: C. Heat nnd materi<:~t babnce is m;lde
plus the fractional shdnl:.nge of propane 2,)t, 1 X 178.6 and Too is calculated.
and hc:'lvicr per thc:orc:tic<tl tray. (M,W.)>o = = Ill D. T.n• is obtained from T:ta and T o<l
6,619 by use of J:Oig:. 7,
1,681 E. 111e calculation is repeated if neces--
Multiplier = 1.000 + 7,141 X 1,000 with L~e final T.1.v.
3::tf)'
37,000 X 6.6 (lb./gal.),o =·----- = 6.376 As stated beiore, calculation tline for
the complete procedure in dthcr prcs.sur~
= 1.000 + 0.007 = 1.007 367,130
range takes about 3-4 hours, Sucee~sive
0. For stcond trial assume: shrinkage
or 53.3~ A.P.I. calculations on ~imilar compo~itions, prt3· .. _)
~urcs, oit rates, e,tc., take t-: hours or k$~.
= J,OOO M.c.f. per day, then: Correction factor :;::: 0.96
Re!erenees
(L/Y)nw = 1.001 The calcub.ted shrinkage (sec Table 1) J. Webber, C. E., A.I.M.E., Tech. Pub.
l,6lS + l,OOO is within 1 ptr cent of the as.sumed value. No. 1252 {1940).
X = 0,180 Comparison of th.: nc'i'i rich-oil cornposi· '2. Kellogg Equilibrium Charts, Chem.
17,000 tion with the first tri~ composition shows Eng, Progrcs.s 46, 1\o, 3, '20 (1950).

TAllLE 1-8A~!PLE DESIGN CALCULATION

First shrinkage trial ,--Material quantities based on first shrinkage t r i a l - - .

Ab-1. inc· Pu cent (Y•o- S)
v,. K,at tor A:::; rec, s s M. ~ Voo v.. L,o L,
M.c.f. 110' F. and (L/Y)nw 6.6 thco. M.c.f. M.lb, B.. t.u. B.t.u, M.c.f, M.!b, :'>1, lb. ;r, gal.
p<r day . 635 psia. X 1/K trays p-:r drLy p<r day p<r lb. per day per day p<r day p<r day p-:r oJ;\y
co, " ...... ' ISS 1.95 0.063 6.3 12 I 69 69 11J 21 1 '2l6
c, ........... l0,99S 6.40 X 0,99 0.029 :.9 905 38 65 2,470 30,093 1,276 3S 15,3~j
c, ........... 2,90\ 1.32 0,140 14.0 .ft)(j J2 1•\4 4,608 2,495 198 72 10,}5J
c, ... "'.' ... I,R-16 0.-13 0.)86 :t'3.6 '112 33 llS 11.-15·1 1,13·1 tJ2 33 19,530
iC4 ... , ... ., .. 252 0.232 0.7?3 ·:6.5 19) 30 13'2 3,960 59 9 30 6,31 I
nc. ........... ·122 0.1"11 1.082 90.0 no 5S 132 1,C56 -12 7 53 11,9i0
c,+ ......... 396 3% 83 123 10,20<) 33 15,365
L.O. . .. " ..... 2,016 2S8,!..itl0

37,000 :3,C()4 J25 40,426 Jl,996 1,6-13 2,34l J67,BO

SECOl\'D Sllll.!NKAGE TRrAL-AFTER HEAT DALANCE

Y1o Temp. oF. K Per cr.nt S :\Lc.f.
A,
M.cJ.
1SS
f-<!! day forK
96
615 psitt.
2.74
(L/Y)
0.179 0.065
rccov.:ry
6.5
.
''"X:! (bV
12
30,993 95 6.30 X 0.96 0.17? 0,0296 2.96 9\ ~
2,90 t 97 1.21 o.m 0.1-16 1·1.6 ·!14
t,S46 101 OA·L o.tn 0.391 39.1 T2'2
:!52 10.~ 0.1.15 0.166 Q,T/2 7).1 ! J·l
Authors aro with California Re~arch ·~2'2 106 0.163 0.1&1 1.0% S7.0 J(.?
Corp., t .t Habra, C:ttif. Pupcr pre~;;nted 396
;t.tCali(orniJ. NL\tuut Gl\So!ine A~ocia·
tion. mC"Ciing, Los t\ngde1, October S-9, J1,GC<"J J,O~J
 Pt~pntto tquiUbrlun\ conslanh , , , This tho chart gives n\ue~ suitable !or most terns, propane equilibrium values can best
eh:~.rtis bns.<:d on the dnta of Webbu,1 synems, b~ taken (rorn th~ Kcl\ogg 1 fugacity
Katz ;~.nd Haehmuth, and our own e:< .. Thtl values cMnot be used with ac· charts for light hydrocarbons. It :;hould
curacy for ~ystems defived from an in·
\__) ptriroental ~esu\ts. Although ·propane K. ta1:.e gas composed, for example, of 50
be noted th~\l the Kdlog,g charts arc
not suitable for normal absorption ·oil
values art:~ affected c.bove 200 ·psi g. by IX' cent meth:tne, 20 p-:r cent ethane, systenv; bec:tuse the liquid phn'>e is not
phase compositions within the range o{ :!0 per cent pi"op:m~.:, and 10 per cent light enough to Q..:have as :.. "light" hy·
·compositions encountered ln this work, butanes and he;,wicr. Foe such rich sys· drocarbon mixture.

. '• ' '.

.i
.,
'· I·

." ~ ·'. ,..

.
··rr·~
. ..
-
. .
·. .~.
;,•

''1'
''•

o· '•"

•. ·.~ ~ ..
I'

··'), ~' s' ·':

7' '• ~ .
·'"""'' . '

':-:?': .·
'I

>'
·-! ·1~,· •: !>; .-,,,. ·: .. ... ,'
.- ..
· ·.!:
'· ..

y;
::::;rc c.~,.

'. .: ·'
: r
:,":;.~:.::.;: ..
I \+~··.
·l
'lib '
'I
i.
···:
'' , \I ' ; ! ,, ' .. ,.
.I
() ··-...:.. .. ., .,, . !., •
.. Ti1'' I
·l .'

'1
I
I',,
'
•
·,•;
' ' "",.., "" ''"' '" " ,..,,.
.. l ):'
" '" "'
{:;qullihrlurv con:o:lunt'i fnr prop•ute. Rderenc<-: C. E. Wcbht•r, Pt'f. lech, Nov. 1940: K:lt1 mttl H•1chmuth, lnd,
EHR• Cheltl., ~9, lU72 ~l9J7),
 ·,.

THE evaluation of performance ,,f In the 1,400-2,100-psig. range, ysis. This error makes the vields of
actual absorbers is a subject of the same procedure wns U$ed except <;ither component }Jo.ry 25 'per cent
considerable importance ·to plant that intake·gn') rates were used· as or ·more. Considering an.:tlys~s from
operators. \he proper evaluation of !\ basis bec;use discharge-gas rates many different laboratories, manv
absorber performance is not a simple were \!SUally not available. Calcula- pairs of indicated low methane an~·
. matter, for it require$ the high~st tions were carried to a check on as~ hi!:!h ethnne recoveries •.vcrc found
degree of accuracy ln metering, tern .. sumcd shrinkage, and yields were fo~ runS having no rich-oil n~alysis. ·
perature mcnsurcment, and in the calculated by difference between 2. Ethane vie\ds based on rich-o:!
\!Sc of fractional analysis and other quantities o( component in intake ·analysis tertd to b~ high when a !ar~e
analytical tools. Because of the im. and discharge g::ts. amount of carbon dioxide is present
portance of this subject, many of The ::tbovc methods of computing in the intake gns. In many co.sc.:;,
!he common errors encountered in component yidds are recommended apparent ethane yidds based on
determining component yields ~ud for all absorber evaluations even if rich-oil analvsis \\re identical to the
over-all tray efficiencies from nb· both intake and discharge·gas rates combined ctl!culnted yields of ethnn~
sorber test data will be discussed, are available. Using metered dif· nnd carbon dioxide which indicates
Component yields , •• In our work, ferenccs, ns such, of intake nnd dis- that carbon dioxide is inc\I.H:i~d ln
the computation of actual campo .. charge·gas rates to compute gross the ethane cut in rich-oil ~n:1!yses.
shrinkage or to compute individual Such rich oils should be scrubb<!d ·,j
ncnt yields for. all absorber runs
compo~ent babnccs almost always with caustic before low-temperntere
having requisite. analytical data was
systematized. In the 40-900-psig. leads to incorre-:t and inconclusive nnfllysis. The c:1rbon dioxide content
range the runs were computed using yield datu. may be determin~d by topping rt
the intake-gas nna\y'lis to check on In our wor:<, there was generally ~ep:1rate samole for ffi:1ss-soectrom·
the available discharge-gas rate and good a~reement bctwee.n calculated eter ano.lysis.' '
assumed shrinkage. If rich-oil anal- and acwal component yields in both 3. Where pen lanes(+) in dis-
f' pressure ranges. Significant devia~
yses were not available, yields of charge gas are high and butanes are
all components were compured using tions found arc probably the same low, analyses tend to be low on
both intake and discharge-gas an~l­ as wilt be found in other evaluations. butaneS. Discharge-gas fractionation
vses as (amount of component to These deviations are as follows and for trace butanes is difficult at b·~s:,
mtake gas minus same component nre ~dl based on the a~sumption th:1t and this case is probably the worst
in discharge gas)/(amount of com- \oW·tcmpcraturc fractional analysis from the standpoint of accuracy.
ponent in intake gas). \Vhcn rich-oil is used. This situation occurs at high bu ..
analyses were available, yields of 1. \Vhere rich-oil analyses are not tanes absorption factor in low-pres-
components having extractions of available and yields are based only sure absorbers when usin£! nve:-age
about 25 per cent or less were taken on inta!<e and discharge-g::\s rates or poorly .stripped lean oil: •
from rich-oil analyses; those from and analyses, methane and ethane
yields are often unreliable. For ex~ 4. \Vhen an improper lenn-o:l
:!5-85 per cent wcr"! taken as ~he molecul:.lr weight is used there m~w
arithmetic average cf the cxtrnctlOn amo!c, at methane absorption of
fro~ 1 to 2 per cent, the methane be abnormally._ poor checks betw~c:,
indicated by the rich oil and that cnlcu\ated and actu:1l yields for botn
indico.ted by int:1ke and discharge- yields arc based on small differences
of large numbNs and app:1rent propane and but~nes: all being of:"
gas analyses; those from 85-lOO per in the same direction. Le:'ln·oll mo-
cent \\'ere taken from mtak.e and yields fluctuute widely. Our labora~
tory wor!< hns shown that a corumon lecular wei(1hts are undoubced:v t::.?
dischnrg.e·gns nnalyses. Thi·r cla~si­ most unrefi:-~b!e single facto; !'or
fication is belicv~d to minimize \to- error in di~ch:trgeMgas analysis is
the loss of 1· to S rer cent .of the proper comparison o( c:1lcula~ed anc!
certainties of yie;d, and is based on actu.al yir!d data. This ['Dint c:anr.ot
Xnowled,ge of the relative analytical ethane cut into thC methaf'~ cut.
This is caused by too-high methane he overemphnsized.
difficulties of various strenms.
takeoff rates being used. The nat~· Tray efficiencies wilt be discussed
Author1 1\re with Cl\\i(ornia Resel\r<:h ural tendency i~ to usc high rates in the nc:xt installment and an ab-
Corp., L:~. Habra, C::~lif. Paper pr~s_cn~ed
at Cltifornia Nntur:~.l Gasoline AssOCt;l\ton to save tiaH~. on the large samples sorber performance ev:1lu::ttion •:.\·
mee!in~, Lo! Angeles, 1953, that must be used in dry·gns anal- nmp!e ,Vi\1 be pres•.:nted in the rl!-

(!0
 m•ining installments· of the series, affected by phase composition, no butanes and heavier in rtU absorption
reliable vnluc:i arc available for rela~ systems. The isobutane chart is pre·
Butanes and henvicr , , , ' These tively rich systems. Until more suit~ sented here. Other charts through
charts arc based on the data o( able information is available, the · n-hexane will be presented in later
·~) Webber. Although the K values arc \Vcbbcr data arc recommended for installments.

'f., .. ·· •: '"'' "'.. '"~·. .

-' 9- .· 'N nR " • . c#~ ~~·" : · .;:\!c"=.:f.f'
~f~~-t '
•
}!:1/~:.::!. :!-f 1: .: • : ••• : • i' . . ·tB~ ~ '
• • .. • '' 1 • • • ' • • • • • " t •

Jtrtli'' ·1·,··· '•:c::'·''·.:. :·

0 'IT;iiif~:'' i,,·i-.:·l!:· :' : ;; ' ' ; '
......,,. \.. ·11"'.·:
:': ''' :' ... i ,·! ,.,. ·: c
. . - '- ' .• • : ..• · .. . •. 1- ... :.' . :· .... . :. . l ~ ' ;• : .

I::· (.'... . ' .. : ' ... ·. ' ·.. 1:-::: .. ,' ' ~
;_
.• •. 1;.·,· ': : ," ;·.• :.:•.. '.; :
''
. . . .
: '::::'; ''
'
i ' i :·:' ' ..
· . .7c, :"if
' . '1 ••• • • .i!iE . 'I'"·:'.'· •:, , .. ; '' .. <'.-•:'#. ;:;:;; :':'(,)! . ;. 'fj~l·
'·;/ ':
, ..
., '' . . . . . .. I. .
' ., c.,_L· -~?.' ·. r;;.r-.c."' r.::,...>?:;z/7 · · : .. :; ., ·
' ·'· ''
. '. ~- . ··;:;;i. ' . ' I_? .. >·~i-.r(
. .;;:
• f

·:f+r , , . ;.or
.e-r2 ~~r:. ·. ' '"·'"''
""' ..~;;: i'./· . :: '' ·...}-- ·~·ii ;·· t.o
. .,
(
'.
... ·.·• ,,•,1' >.::'•"".'' '"'· ' . ' '' ·'
r.

..
.1

.'

) :
·'
i.
, I

•• , ' l
·!-·'

Equll!brlum COOSIAOIS for l~ol,ulfl!J~. lhftrtnct: c. F.. Wtblxr, ret. 'Ttch., NoT. 1N.O.

'6 1
 '
•'
.
(}

THE method of obtaining com 4

charge gas r'::ttes and analyses; (c) '2. T DO - Tr.o = 7S - 65 = ! 3° F.
divide the number of theoretical From Fig. J, obtain estimated top s.ec!ion
ponent over-a!! tray efficiencies _heal of absorption :::::: S.3 B.t,u./lb. k:~.n oil.
and the range of values to be used trays found in (b) by the number
l. CT<.o + T ,o)/2 = (65 + 3·1)1' =
for design h';we been discussed pre· of actual trays for the combined effi· 74.5° F. From Fig . .6 at .5.3 B.t.u./1b. o.nd
viously. · ciency. . ~4 trays.
2, Some runs showed low or high
Over-all tray efficiencies , , , With recoveries for propane and corre· Tw+T~o

few exceptions, propnnc and butanes spo.ndingly low or high efficiencies

T\,. ----= 13.5
2
tray efficiencies were in good ngree- for butanes~ probably because of an
ment wi!h the recommended design impro~r assignment of molecular T., = 74.5 +18.5 = 93' F.
values. ·significant deviations were as weight ·to the Jean oil used in the 4. Assume shrin.k:1ge, S =: 900 \Ld.
follows:· cn.Jculntions. If the butanes absorp· per day.
I. Many of the deviations in indi· lion factors were multiplied by the S. First trial (L/V)sT~ = (L., 7 S)/Y1r.
cated tray e-fficiencies were due to ratio of._ nctua! to calculated propane 2,406 + 900
unalytical errors mentioned above, Jt yields, the butanes efficiencies usual· = = 0.509.
is obvious that errors in analytical ly came out ncnr 33 per cent. 6,.500
resu!t.s will give improper component 3. In the noncommercial range of
40-65 per cent butanes recovery, cal· TAnLE 2-ABSORBER T£ST DATA
recoveries and in turn improper in-
dicated over-all tray efficiencies. , culated butanes yields tended to be Inta~e Discharge L.:an
Other analytical errors that may high and indicated efficiencies were g:u g:ts oil
have significant effects on eHiciency verv low. It is believed that the dis· Pres~>Ure, psia. ;As
dcterrhinations are the reporting of crciJ:lncies in this range are due to Temp., GF... , , , 66 73 65
trace amounts of butanes in the dis- :;;hortcomings o( the absorption fac~ Rate, M.c.E./day . 6,500
charge gas as "0.01" per cent, and tor calculacion method itself, prob· Ratl!', "gt~t./day .. , 1.51,100
neglecting traces of ·butanes present ably became of the assignment of Gravity, 4 t\.P ..f. , ' ")9.0
in poorly stripped lean oil. AI very improper effective absorption tern· Mol, wt. •170.1)
high butane absorption factors these pcraturcs. (Rich-oil iemp.·.S4°, ....
·F.)·, · ...
. .
errors may result in low indicated Example Absorber Performan<e Composition: ,--Vo.p. %-.. t.iq. '7~
tray efficiencies. Evaluation Air ......... 6.\S 6.66 . .. "
Unavoidable errors in isobutane· ;\ ~ample ev:1lu:ui6n is presented in or· :\-iethnne 54.)6 61."25 0.011
n~butnne separations in ve.ry low bu~ tier to illustrah! procedures used in check· Eth:'l.ne " " .. 1-6.65 :'_'i,OS 0.491
tane content discharge gas analyses ing absorber pt·r(ormance. The example is
ba~ed on actual :tbsorb~r test data and is
Propane ..... l:".S 1 1.01 l ,9)·!
may tx minimized by calculating n cho~n to il!ustrate th<i m~!hod o( cor-
Butanes plus ...
combined butanes efficiency. The recting cornponr:nt recoveries wht:n the Normal lean
combined efficiency is obtn!ned as oH .... 91.55?.
follows: (a) calculate the individual
butanes recoveries for various num~
!e::m oH cont.tins !ow boiling solutes. The
lest d:11n are pr.:s-:nted on T:l.ble 2.
I. C:dcula!e mole fracuons and quan·
-- --
100.00 100.00 J('l),GOO
litic:s of compont•n!s in kan oil:
bers of theoretical trays at the rc·
spective ca/culatcc.l absorption fnc. Liq.% Gal./ day M.c.f./day M. !b./d:J.)'
c, '·
tors, and plot the totnl 'butanes re·
c, .. '. .' .. '. .' ......, ' 0.011 26
15 I 19
I 0.0804
covery versus the number of theoreti·
cal trays: (b) obtain from this plot
(', ....
"Normal lean oil"
0."7
1.9J4
9'/,552
2,924
147,499
lOG
2,270
').01""20
O.l"-1-tl
0.$'·'.35 1.0! ':1
j' '
the number of theoretical tr;ws re·
q•Jired to give the actual tot:i.l bu· 100.000 ·151,"200 :~,w.s 1.0<.'<:0 1,03J
lanes recovery i:JS indicated by the
difference between intake and dis·
"Normal lean oil" rille
Authors are wilh California Research
1·17,499 X 6.910 X O.J1S6 Note: Column-24. lrays, :o.in, tt:ly
Corp., Richmond, Calif. Paper prcstn!(d
at Ct~Hfornia Nnturnt Gasoline t\5SO¢ia· ' " - - -110- - - - !.po.cing, 43·in.·diamcter o::o!umn.
tion meeting, Los Angeles, OctQ:ber 3·9,
19$3, :::: 2.~10 Nl.c.f./day •Propertks o( normal kan oil.
 field Processing (Conthmetl)

.j ··-+-··-· .. ~---~···-~···- 1......... r...,.. ., ....... ../ .. ........ /=-·-'· -=/.-..r,.... ::z .... _,. . ....... ,_.c.::,..
..::::~...:J ;;::'";:;:t:::···:!~!::~~;-~:,.;o:::-_:·;T_~::::::_:::-;z;;~~··:r~::., ...-t..:::;::::~,::::7-~:::~.:-f-7..::~::;;.~~•.:;::::;:~~
.,,,_1 ... ,\.-+···r-r\•-·-· 'r"'"''' • .,,....,, )~r-··r--•~·-.. t··"p.-r-;..-•~il.,.,.,,.,..(-:,i:.::..,_.,...f ... -y: .... ,, .... ~...•__.. •.. ,, .. -,--....,• ,, .... ,.~,.. 1 ' ·•·H
-~r·· -~r-·,_.~,,.~ ....,......--~.~--·r·-.n•·pp···" t··z1 --1 · ,, ... ,. ·.....-r"'r-,...:;d~~t·~~;.:.!,-"''''-r,_.r.;·<r,....-. .. ··,::r~-- ·-··~~
1
·~
-:r;::.:,:::;:::~--::!:;:-,..:..:;;--:R-;:·;r.r::.:...~~:. ::-.:!-;:-/_(,-;:::~·:~~: :=-c;l__:.:?..--:..,.;.:r;:':?--f.';=;,:-t::· ···;:~.-::0:1!::=:...,

~
~--r:-·-~-~cOLr-·~17' "T"'7''<-•o•"' '""':!=;;c::·"~;"'"·rorr;7C""--;><"-~?C·~m··'"] t
TF'rr.::~.:<.:J~:.r-~·~·~·-;;;::::;:cz::z;;z::, -i{-~;-!:':zf:::~·;:;r.::PJ.?c-£Z'~~-;;~rL:;.j
~~p .. ..-- ·.2r-·r-;-~::~~-r0;k!.f;:~-~J:.~?-(~~T~~~rf.f~~~:~7-2C~:2:::~
~ ·~- ' ~' ;.c;..< • C:::;:;.:.-;.Cc~~~·
,L..-i;?';C: ~ ',;- ~A·-'---"··-·L;.... -~~1
' rl",rlr·-~:~~-
If I 1 ct.o ' r-r-10\1
, -:r.~~-~~'(..l~l.l(f_l'
'[;;;(, --~yl'-~---·,'"-..:::~~.;-~~,....,.
I,C '.o ../.'-'.:/.~ ?' !·!·VI' I
~·---7~f:4-,.~·l
,,
1,::~~-~~~-~~--~
I r,?'f'"' '<"•;,.:j
~1 T.T·Fr.- 1 • ·n fi'TI'...Y' ~-;- ;li'(r::--~-~: ~it _,_;':" ... 'l;';'prr..-;-,-,,--~r~r:;•r
A" '' • / ,, •
,·;--;r.r•tT".--..-r~,·~ ,-~

I"r::~;m~~~~[~~~~~~~::;~~~~~;~~I.O
.7

·'
.<

('"'·
\ ..... / ·'
Gx.... J....,i.'Cx:;r-.. -,-{Z;:,:'·-~··22.,_$ ..c-·.z·--,,·-··/c..v:?'i.:c--···'·,>··--·"-....~,·-~·...-..

tr~£f_;;~~t£~-c~i_:~&-~s~:~~~~r$ffii~:_~~=g:~.~~-t_#~~-?:~~2~.:~
• r·~--~- . ~(--':7'~.-. ,A.=::z.;.;:;-z-;_Zr-y";-'--r'-J"~~-~- . . ...--............-·r~,.....,..~ .t
1
~_:.ir~~;/r~':e:__:p~ 1
'ZI'2(.c;(."2;';c~)'d'c:'c_S;:;:£:;"¢f,':~"~~;;;~,8';,_:.~:,~ .·:c~T'"·,:!~
·;:·\: . ~ ·., :~~)~~ -.:.,..,Lr·: .. ·:~~Z.I7.24=:;~~;~-:~jj:8~1:;:~~t:+:~ffP~i~!-~~~
• ~ • 1/ I
.,_).'-·T~/,
~"'01 ,.
I
·~
...~.
~
-(';- 6
:rx··'-c·~· ,,~
•·.: "'\-_,~·-~• _,. ·- ~-~-£:-~·
;
,;;-r··_:.,..
</."..f --':•>-' ''
/('-;:;r·r ,_....-'1:-:;; _.,-;;lr',,, :..
~ p;f_.~
1:
_ :- . .)----:~,;_:,
, ... .. ,_,. ~-~~-
"·IL
.•-!'
.. .,...,r."""
1
· ·,
"I
'.-"~
, 1•1• • I
OJ:-.,;,rt I L .. :
1_., ,,
t t,-, . 1''>•!• • •"
','!_ ,.1 ·I :<T'-;-_.,----..,,,
-~"""fiT" .. ,.,~ 11~~~1'G... .,.,..,.....
t •I
··.tJ.--~.:.J:~.. ~Pr+J;{' '·····~;__,.:_~
H • L1 ;.
.,
o!-1,
• •, ,. ··" '• :'I

''·'I .. ~ . r: · 2> ,._ .n ·. 1,. ·. •·l' .,. :! : •. J!.:1,.;1') _,J~.T ··-k-t,l' ,.. ,·.,. :·1' -~.~-~~ ,,,,,.,1 ,, "'i>~
· . I'.--
!··\ !:~::·1~~- ·... 3~ :p-, 1
1 __ ·' -i~0<:--'lk'f 'GJ:.:~t7r~n;.r;-tl.~.!-~,\Cf~T-fi.T0T~u~~
:r--r-~4. '11(: ·I : 1 · _._;0,-:r'~: tr'.T~i-~[· =;: =d0ffl1~ f:Tollt:t;Tr;T: · ,·· /, ;d'; 1!-:' lt: 1::!,"l ;,:i;:!d .r

:-!

"'~~+:-.~·: r: r::-~--;~k~f.:: k2f!:--;;..·;·~::-;;:~rh%;:!;.·:-r,· :: ·_:~. [l":j;;·~si-\;rr. :l r.ttf.~~r~.IT'.·. '.· · ~- . rf·~.·.~n-rr:n_ :r·. nr-rr·~~. g"'
:-rr ;··!j:!Mr\,.:1 1 ·H~~.;.:;:y;:,_};n;•,.,;r-::::'· i: ··:!~~:;~~r.~·--~:,i':.'8f,Hf1'f:IIT:~rTI."-~1[1l:·~'7f:'"':·~;:-----::··:~~-.
· ·.
, ·f-· .. 1 .n.-·:):1.Jfr!rJ····~·-- 1 r.pr<:;::,.--<!r.rr .-.. 1"l:',...,,.TT.. .,.r-rr·rr-rr'T"t·l"r-• .... rr--m· rm--:-~r-:-.,.·rrrn~ 1 ·rr-•r-+-r!rf-l<l:---f1"":':-rr·,.T·..---,--.--.-
·~ .;1; ·~r.~.J,jf~nP.?:·n~-'-i'!!,t}t.'f-')');~ft'.!t::....,~_;~<~;.:_ :,·,·~~;~;.;-·::'.~;I:~~~~: !I,: ],;<~!.,,'!~' ·'··~~.;.;.:,rf~~,--\..:...~;.J.~~:,.~~~;~._:'..!.
lf" i!PI :1 IL~ (.! ;,~f,.,,r:·~JI''..-r.~H · t.~<).''J;/1 '•''· !fS'f.l;
t-. t· frf'C/1T0f1.c~ ,1,' diTJiTfdJ.T'lf·l·' 1''1. : :,. '' 1•i· :•' -!·: · •:. ·, :,: · , \1
-----nJ--~.:J r~u;--~to·~-c.v~-To·~~.~iTU_._ ....._TI.'Ci--(fl) :~·--rn;r-rfl'.);)~"'o--t'C~J

f.qull1brhun tou*u;ts for n-but:otn~. Rdl·ren<:('! <:. E. \Y('bbtr, J>d. Tt('h,_, Kov. \9-W.

G3
 ·..... ...

Ym X (V1n/VDtJ) X (1-F)/Koo +
CONTINUING with the example of 3b·
5orber performance c>\lculation which
(0.0613) (6500/5611) (1-0.007)
Wi\$ lxgun in fns!nllrnent II:
6, Compute fiw trial shrinkage u'lng
,\ir ..:... , . , , , . , , , , , . ; , , . , . , ::::: 0.0009
"•
+· ..... = O.C009
79.0
eight th~oretical trays and sc;;~.ling tem·
perature1 from T.n to T.w. See cnlcu!ntion (0.5436) (1.153) ( \ - 0.033) (0.033) (I)
tabulation on Table 3. Calculated shrink· "Co
14.4
+ -------
(14.4) (5,611) (0,033)
=0.042) + . ". . =0.042l
age is do~ enough to proceed with more
detaikd calc.ulations. (0.2665) (1.153) (1- 0.216) (0"216) (29)
7. Now, VM " 6,500- S =; 6,500-
c, - - - - - - " 0.1100 + 0.0023
+ -(2.20} "0.1123
2.20 (5,611) (0.216)
S89::::: 5,61 t M.c.f. per day. Compute mok
fr:Jctions 9f components in top section (0.1231) (1.!53) (1-0.636) (0.63 6) (I 06)
liquid. Us.e actual discharge gns tempera· c, - - - - - - = 0.0794 + 0.0274
+ -(O.oS) = 0.1068
ture, 73 4 F., to obtain component equilib· O.oS (5,61" I) (0.6'-1)
rium constants.
Authors are with California Research where:·
1 Corp., 1_\ichmond, Calif. Paper presented xr' is mOle fruction of component ab~ component in lean oil.
at Calil:ornia Natural Gasoline Associa· sorbed in top section. xr is total mole fraction o{ cOmponent
' tion meeting, Los ,.\ngc!es. xr" is mole fraction due to presence of in top section.

Table 3-Sample Absorber Performance Evaluation Calculation

FIR.~T SHRL"\'I<AGE '!'RIAl...
yroVuJ K A= (L/V)a"T}.( F ' {yro1'rr. s•
M.c.€./day 245 psia, X 1/K S theo. trays L,x./A - L..x../ A) M.c.f./day
Air • , •• , .•• '"·' , , , , , , , • , .. , 1 •• , . , • • 402 loi.O 0.001 0.001 402 3
Co .................... , ........... 3,533 15.0 X 1.025 0.033 0.033 JO 3,503 116
c~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . t,7J2 2.35 0.21 6 0.216 iJ4 . 1.59:::: )4S
CJ •••• , • ~· •••• , ••• , .• , ••.••.• ,.,. 833 0.79 0.6-14 " 0.636 164 669 .fl5

6,500 S39
SECOND SHRINKAGE TRIAL
y1oYro K K (yruYur- $·~
M.c.f./day T, 'f.· :245 psia. (L/V} A, F T, op, 245 psia, (L/V) Ac L....'(~/ Ao Lx .. / Ac) ~Lc.i./ 0.
Air 402 34 ?oi.O 0.520 0.007 0.007 84 74.0 0.520 0.007 401 J
c. 3,D3 S4 15.0 X I.Q2S 0.520 0.034 O.OJ4 34 15.0 X I.OlS 0.521 O.OJ.I 29 J,50·~ 119
c, 1,732 8'/ 2.35 0.525 0.27.3 0.123 82 2.25 0.531 o:.D6 123 1,609 3S';}
c, 833 91 0.79 0.533 0.675 0.665 so 0.70 0.548 0.714 \JS 69:-S •\(~~

6,500 9·15
FINAL CALCULATION
YtoYrn K K (ywYtr.- St
M.c,f./day T, ~F. "245 psfa. (L/V) A, F T, ~F. 245 psin, (L/V} Ac L..x .. / Ar. L,."~: .. /Ae) M.c.UD.
Air 402 84 74.0 0.517 0.001 0.001 84 14.0 O.Sll 0.007 402 )
c, J,SJJ 24 15.0 X 1.025 0.517 0.034 0J)J4 84 15.(1 X 1.025 0.518 0.034 29 3,504 119
c, 1,732 88 1.37 0.52) 0.2"21 0.2'21 S2 2.25 0.57.9 0.2:\S 123 1,60') 3.56
c, SJJ 94 0.8 IS O.lJ \ 0.651 0.&-l·~ so 0.70 0.548 0.7:3) lll (,')H 450

6,500 (}'l"~

COMPONENt' RECOVERIES DASED ON INTAKE AND DJSCHARG£ GAS ANALYSES

c.:nl rcco\·crv
Actual percent (from !irn:'
Shrinkage recover'! C:l!C'.l!atio;~J
..---lnta!<.e·gas-, r-Discharge gas-----., M.c.f./d;ty_ytnVto:~y~>.;Vt><< S
Yap. % M.c,f./Cay Vaj).% :\·1.c.f./day ytoYtr. - - - - -
Jill X 100 yroYtn Jl.!t.. X 100, 'jO<JVno - ynr:V!llt JtnYto VhtYtol
Air , .•.. , . , ... , , •.•. , .. , .•.....• , , . 6.18 <102 6.66 171 " 31 7.7 "!0.7
c. .. .. . . .. .. . . .. .. .. .. .. .. .. . .. .. 54.16 J,5JJ 61.25 3,4\J I "20 J.4 J ••~
C1 . , , .. , ....... , • .. .. .. • . .. .. .. 26.65 1,73'2 25.03 1,397 )35 19.) 10,6
c, .............................. , 12.81 3JJ 7.01 391 4-12 5),1 ~ol.U

100.00 100.00 5,512

•s =: F(yroV,o-L .. x.,/A). tS::::: F(ytoYuJ-L ..."<Q/At·). !S:::: F(yruYm-L .. x.. /Ad.

iSc<J St~p 14 in t-:~1, in later in:'ilal!ment,

64
 by n . .J. HuH
and Il'.. naymond.

·.)

=
T HIS installment continues the e:o:ample
for an absorber tt~t
evaluation. A
14. Calculate component recoveries
bnsed on intake a1td dh:charge gas anal-
l\H:./W~ 1'2,369/35 :;:::::;· B.t.u. per !b.
shrinkage (calculated). .
· flnnl inst::~llmcnt, to appcu In n later issue, yses (St:e 'fable )) Componl!nt ro:covuks This result is a good check with the
will present the hut chart to appear with based on compone!lt differences in intnke experimental result obtained by the o~·er-a\1
the S¢rit:·s, namely an equilibrium-constant nnd discharge ga~s compare fairly wdl heat balance,
chart 'for n·he;.;ane. with computed r~coveries. It is apparent The results o( the above sample ab-
Total mole fractions o{ components in that the ana!ySt:s for air in either the in- sorber test .:valuation are fairly represent:\·
the top section may 3lso be olHnined bv take or discharge g:'ls, (or both), are in tive of results obtained for many absorb-:r
th~ relation yno =: Koo xr, 1,1sing the actua'l error: howe·tcr, this does not Mfect the ev<\luations. fn most evaluntions, the qu:tn·
di~harge ,gus analysis to obtain yoll (or rccoveri1:s of other components striously. tities of light solute in the kolr\ oil ur1'
each component, It is interesting to com 4
An over-a!! tray efficiency for propane so low that their effect$ m•lY be negkcted
pare the results of the two methods, would not be reliable ar a propane absorp· and the evaluation~ are corre5pondin~t:'
tion factor of only 0.651. simo!er.
y,., Koo · · 15. If.so desired, the heat of absorption The sample <C:Vnluation is an exce\!ent
Air ...... " ' 0.0666 79.0 "
0.0008 per pound of tot;;~! shrinb.ge may b~ com- illuHration of the importance of ind•Jd·
c,. ··········
c, ..........
0.6125
1),2508
14.4
1.10
0.0415
0.11·!0
puted by 010 over-all heat and material ing the avtrage absorber ttmpt:rawre in
balance .
c, ' ........ ' 0.0701 0.6S O,l0l1
..--Intake ga'i------->. ,--Shrinkage------, r-- Discharge gas~
0.2604 M.c.f./day M.lb./dny M.c.f./day ~1. !b./dny M.c.f.!dny ,\].!b./day
8. Compute quantity of liquid in the top 1\ir • , , • , .. : . . a J n H9 J2
$t!ctlon, In computing the quantity of Jiq· Ct .••..• , ,, J,5J3 150 119 5 :'1,4!4 1·~5
u/d in the top srctio·n, it must bo re· C'l •.•.... ·•·· 1,7J2 13-7 3% 2S l,J76 tl)9
membered th!!..t the terrn 1 - ~XI repre- Co 8ll 97 450 l2 J3) 45
5ents the mote fraction of normal le:tn
oil, thcrdorc: 6,500 416 85 5,571 Jl I
Lo = 2,270/0.7J77 = 3,077 Molecular w~ight of hydrOcarbon por·
9. As.sume shrlnknge, S :::: 950 M.c.f, (Jlt -l2) X )78.6 the calculation of comoof\ent recovl!rie~.
per d;ty, th.;n: tion o( discha~ge g:t1 ::::: - - - - - - - The effectin temperatu'res of absoro!ioo
3,017 5,572-J99 for lhe various eomponen!S wae obt:,;ncd
(L/V)roo• =- = 0.554 ::::: 22.0; tlir content 6.66 per cent volume by scaling between the rich-oil lemyen-
5,550 or 9.73 r-er cent wdght; Cpnn O.SJ. = ture o! 34° F. and an n•1ern~e :J.bsori)er
Neglect hent·lr:'lnsfer effects into or out temperature of 91~ F. In coo;parison, ef.
425 2,407 + 950 of column. fective tempcriltures obtain!!d by •.Jse o(
(L/V),,. = (1 + ---) ( ) eith~t the Edmister or Horton and Fr:J.nk-
6,500 X 8 6,500 WLO Cpt¢ '(1'"o - T1.n) !in ntethods would be between the lean-
= 1.008 X 0.5\6 =0.520 +
+Woo Cpoa (Too- T1o)
Wa Cps (T"o- Tw) = dH~
oil temperature of 65° F. and tho: rich·oil
temperature of 84 4 F. As a re~ull, c:d·
(L/V),, = (0,554 + 0.520)/2 = 0.5)1 (I,Oll) (0.50) (34 - 65)
culnted recoveries of ethane and ?fO?::lne
would be about 10 pe~ o:ent higher :h:m
10, Compute second trial Shrinkage. Scale + (331) (0.53) (78 - 66) shown in the example and agre.:rnent be-
temp<:ri\tures nnd (L/V) ratios accorQing + = ,m, ·
(S5) (0.65) (34 - 66) tw~en actual and co.!cu:atcd w;overies
to Fig. 8. s~e Table 3. Computed shrink· would. be poor.
nge checks assumed shrinkage, ther<!fore, 6H, = 9,314 + 2,105 + 99·i = 12,913
M. B.t.u, per day, nererences
no additional shrink:tge trials are n<!CI!~snry.
1 t. Compute mole fractions of com- .:.\l-h/Ws ::::: 12,9\:l/SS :;::: 152 R.t.u. per I. Brown, G. G., Xatural GnsO!ine Sup·
ponents absorbed in top section liguid, and lb. shrini:o.go: (!lctu:.1l). ply ~f<n's Association Engi:-H:cring D1ta
from these compute top S(Ction he;lt of 15. Compute heat ot absorption ~r flook-1951 Edition, 36-n. \
absorption. 2. California N:ttura! Gasoline As~ocia­
tion, "Physical Constant~ of the Com·
M.lb. D.t.u. M. B.t.u. ponents of Natural Gus ;~.nd Ga'loline,"
y1o X (VIll/Ynn) X (1- F)/Ku., Bulletin 'fS----!01 {1953).
t\lr (0.0618) (6,500!5,555) (1 - 0.007)/79.0
-· :";!' LlX1
1
p~r day per lb.. per day
3. Dodge, D. F., "Chemical En1?ineerirw
Co (0.54)6) (1.170)
= 0.0009 3
Thamodynamics," }.kGraw,Hill n~x•l: Co.',
(I - O.Oll)/14.4
c, (0.2665) '(1.170) (I - 0.223)12.20 = 0.0427 [J t 5 77 385
Inc., New York, !944.
= 0.1101 )")9 27 158 4,266
Co (0.1281) (!.170) (1 - 0.665)/0.63 = 0.07)S ::!~7 26 \55 4,0)0 4, Edmister, W. C., Ind. Eng. Chern.,
)5, 837-9 (194)),
3,681 5, Horton, G., :tnJ Franklin, W. B., Ind.
Eng. Ch~m., 3'2, lJS·t (!9•10).
12. Top section heat of absorption per 6. Jac.oby, R. H., and Rza~a. M. !.,
pound of lean ot.l :;: : :; S,6S 1/1 ,OJJ =: ::;.4. 1. Pet. Tech., .195, 99·110 (1952).
From Fig. 3, Too - Tt,J) =: 19~ F.: since pound of shrinkage from individual com• I. Katz, D. L., :tnd H:tchmuth, K. H.,
Tv, = 65, Too (computed) ::::: 3•!' F. ponent shrinbgc:s: Ind. Eng. Chr:m., 2'J. 101~·10"!'7 (!93.,').
13. Using lhe above nlue of SA D.t.u. S. Kellogg Equilibrium Ch:tns, Chem.
p<:r lb. of lean oil, the corrected. T.~.,. :::::: Shd~kage Eng. Progrt!%,' ~6. ~o. J, '20 ~t950).
4
97 F. as read from Fig, 1), Assume shrink- M. ~b. ll.t.u. M, D.t.u. 9. Kremser, A., ~all, P~trc!tum Nc•..,.:;,
age, S :::::; 925 M.c.{, per d:'ly, and pra<:e'!d p~r"day JXr lb. pl!r day 22, No. 21 (19JO).
with {inal _calculation. Co 5 77 385 10. Poettman, F. H., and K:ttz, D. !....,
Author$ are wtth Ca!Hornia Rtsearch
c, 2S 158 4,·S24
Ind. Eng. Chem., 38, '530 (l9·16).
11. Sot.!ders, :-.t, :tnd Brown, C. G,, !nd,
Corp., Richmond, Calif, Paper. presented Co 52 155 8,060 ErH~. Chcm., ?.<1, 51') {19)2),
at California Natura! Oaso!i!le Asro.:iation 12. Webber, C. E., A.I.>.LE., Tee!"\. P'Jb,
meeting, Los Ange!e.1. 85 12,869 No. 12.52 (19·!0).

66
 Field Processing (Continued}

·"
I
'. . .'!'""''
,1 , .•
. :'
···7·
;' :;. .,~,+. ~'
'; I!. ',!
',\:l:l!A:': ...
I ' !.i({l I! ' '. ! I
.. f~

! .7

1 • :

()
,._;, 7':ce'·C'·'··/· ..
t·i
. '•'. YY..::~0:0 :d ,,, ·
'' . ·.':bt
''
 - - - . , / ......... L . C _ , •-

)
Ir.. Uttymond

)1

: amifi)
I

Author~ ar~ with C.1!i·

fornia R~s.cnrch Co r p.,
Richmond, C:l\i(, Pap<!r
pre~ntecJ nt California
Naturnl Ga:ioline r\ssod·
Btion mtt:!ing, Lo~ An·
go:lc:s.
Equl!lhdunt comiUnU for
n·ho:'(an~. u~iw:uce: c.
t::.
,ytbllu, I'd. Tt'.:h.: No..
vtrnL('r t9..ft),