You are on page 1of 10

486 PETROLEUM REFINERY ENGINEERING

in Fig. 16-8 and in Fig. 16-9 which shows a newly developed cascade-
type tray . 14 , 31 The cascade tray ( Kaskade, Benturi , and Flexitray
designs) apparently exhibits a higher efficiency and greater capacity
(about 35 per cent ) than conventional ’ trays, but it is more complicated .
The single enlarged Kaskade unit indicated in Fig . 16-9 is too small for
most commercial operations , and hence several of them are lined up
across the tower with each adjacent pair of units dumping into a com¬
mon downcomer as indicated around the enlargement of Fig. 16-9. The
exaggerated crossflow of vapor indicated in the bubble-cap tray of Fig.
16-8 is not troublesome in small towers ; but in large towers , Sd particu¬
larly if I -beam tray supports extend into the path of vapor crossflow , the
operation of the trays may be completely upset .
Davies 32 has derived Eq ( 16-4) for the liquid gradient across trays of
.

caps spaced in an equilaterial triangle arrangement .

.{
S/


where a = U / h
AT A ,. ,

^ )
- 1.4 4- 3)K + «(a - 1)]

Cd liquid gradient factor ( Fig. 16-10)


J =
2

AM = liquid gradient (difference in clear liquid levels between inlet


and outlet sides of toy.sec.tion.LJor u
,
^^ (16-1)

o
A* = liquid gradient , for any condition , inches liquid
do = total flowing clear liquid depth adjacent to overflow weir, in .
l\ = total free space between caps normal to liquid flow (average
of various rows) , in .
12 = total free space between risers normal to liquid flow ( average
of various rows) , in.
Q = clear liquid flow, gal per min
p0 ~ vapor density , lb per cu ft
r = number of rows of caps perpendicular to liquid flow
8 = cap skirt clearance (distance of bottom edge of cap above
tray floor) , inches
u = superficial tower vapor velocity (using total tower area) , ft
per sec
In solving the equation, Cd obtained from Fig . 16-10 is inserted in Eq~ .
( 16-4) along with other variables to solve for AM . The value of u y/ pa I
is computed , and then by the use of Fig. 16-10, the actual A * can be 1
determined .
Tray Performance. The over-all performance of a bubble tray can be
outlined or defined as in Fig. 16-1157 or 16-12. *3 At minimum vapor
11Harrington et al .# No Peace for Fractionators, OH Go* J ., Nov. 24, 1945, p. 135.
" Davies, J . A., Ind . Eng. Chem ., 3ft, 774 ( 1947 ). See Rodriguea, F. , Chem. Eng . ,
November, 1956, p. 230, for a direct nomographic solution .
** Munk, P. , New Approach to Tray Design , Pei. Refiner , July, 1955, p . 104 .
FRACTIONATION AND TOWERS 487
CORRECTION FACTOR
120 FOR ADJUSTING LIQUID
GRADIENTS FOR VALUES
OFUVp OTHER THANU
no
too
§
90

80
-J2 -
70
03

60

500 20 40
ppm (hoi )
60
ft ( overage overoll flow width )
2.0


Holddown bars below

ID
- '
Ho /d,down bars tTbcve
03
03 7
OJ
06
// /
r
06 1
/ 1
0.4 // HYDRAULIC GRADIENT EQUATION
03 T1 CONSTANT FOR uVfi = /. /

0.2
gpm ( hot)
It (average overall flow width )
Fio. 16-10. Liquid-gradient-factor correlation (at u -\ PB — 1 - 1) &nd correction factor
/
/
for other values of u y pa . (Jnd . Eng. Chem. )

rates the caps pass vapor only intermittently, and this is called pulsation ,
and between burps liquid tends to overflow through the vapor risers. At
the other extreme of vapor velocity, excessive entrainment occurs by
spouting or jetting of liquid to the tray above, and in the extreme the
cap slots may be so overloaded as to cause excessive pressure drop,
spouting, or coning . At extremes in liquid flow , liquid or froth backs
up in the downpipe ( downspout buildup ) into the tray above and may
m PETKOLEFH REFFNEftT ENOtttKJSR1 NO

finally appear at the top of the column by what is called jfoodmp * Like¬
wise. at the highest liquid loads , liquid may dump or dram to the next -
lower tray through raps that poos no vapor ( Fig. ld ft j , but it in com¬ -
mon practice to design for such a liquid gradient and vapor velocity that
all of the cape are working, arid thi « rcuulta in what in called tUibiLiiy. A.
scrica of charts similar to Fig, 16-12 for vorirma cap arrangements, spruv -
ings, etc . ,, are presented by P, Munk ." The number of rows of cap*

600

••

i
,
-.
U
i

i i-
m
srisncTORr
ceffiAiONi
ft
!ii &
/

CO
f -
i
/
7
200 400 GOO
/ SCO iQOQ tpoe
Lijimf tw( gpm
-
Fra Jd ll Qualitative effect vf liquid
formance, ( /*«1- TroiaMtufl. )
ujd vapor loads DO -
buhble^cap tray per*

vitaily affects the stability of the tray , and from Fig, 10 12 it appe& ra -
that the use rtf several trey lavela ( cascading) should ho practiced more
frequently than in the past -
Pressuro Drop and Downspout Backup, Pressure drop through the
caps can bu a little larger than the hydraulic gradient computed by the
< -
Davies equation [Fq . 16 4';] for tray stability.w At the same time, preo -
sure drop cause* liquid to bock up into the downcomer , Pressure drops,
by the study or T. C, Dauphinc11 for three atea of cape, based on the
M J taotorsi dissertation , MMB&nhuuette Jaatihita ot TocluiDlogy , 1939.
*
FRACTIONATION AND TOWERS 489
Cop diom. 4* Cop spacing 2* Cops per sq. ft 4.6
200 Rows of caps
B\6

ISO
DOwnspout
rs

120
* /

3?
3
V 80

Stability

40
V df* Vopar density, lb per cv ft
d Liquid density, lb per cv ft
^-
L mtr length, ft
Minimum
vapor rote
— Q^Uqutd rate, Qpm
w= Vapar rote, to per hr per cap '

'
80 120 /60
OIL
FIG. 16-12. Operating limits of bubble trays (4-in. cap diameter, 2-in. cap spacing,
and 4.6 caps per sq ft ) . ( Pet . Refiner )
.

assumption11 that the annular area , riser area , slot area , and reversal
area are all equal, and equal to one-half of the inside area of the cap, are :
For 3-in . caps :
rL
u(d,) ° - s
He = 1.939 ( d , -
2
r
~

( 16-5)
For 4-in. caps:

[_
Ha = 2.234 ( dL - d*)'o »«] ‘
«( « T
(1 M)
.
For 6-in . caps :
Hc = 2.768 (
L
r - <t) “ “ J
T" ( 16-7)
*
where Ht = wet cap pressure drop , inches at flow conditions
u = slot velocity , ft per sec
d, and dL = vapor and liquid densities, lb per cu ft
i
490 PETROLEUM REFINERY ENGINEERING

The total buildup in the downcomer in terms of clear liquid heads above
-
the lowest point (the overflow weir) is (1) wet cap pressure drop, (2)
crest above overflow weir, (3) half of hydraulic gradient across plate, (4)
height to distribution weir and its crest if such a weir is used , and (5)
pressure drops for flow or contraction through the downspout [see (Eq .
16-3)]. For low liquid flows the backed-up height may approach two-
thirds 34 of the tray spacing, but at higher rates, lower backups are neces¬
sary because of the need of time for the disengaging of vapor from the
froth. Muhk3S suggests decreasing the allowable backup by 1 inch for
every 10 gpm per ft of weir length above a liquid rate of 65 gpm .
Pressure drop in vacuum towers is vitally important, as well as the
pressure drop through any condensing equipment and the vapor line.
The drop through pressed-steel caps installed in vacuum towers operate
ing at about 40-mm pressure and with a slot submergence of in-
ranges from about 0.7 mm per plate at a superficial velocity of 7 ft per
sec to about 1.2 mm at a velocity of 13 ft per sec .
Plate Spacing and Entrainment The obtainment of equilibrium on a
bubble plate is governed by two opposing factors. One of these, the
intimacy of vapor-liquid contact, tends to produce equilibrium ; but if the
contacting or bubbling is too violent, liquid particles will be carried from
one plate to the next by the vapor and will tend to destroy the separation
that has been obtained. One kind of entrainment is a distinct splashing,
.
spraying, or spouting of liquid particles The particles are thrown
upward by the velocity they attain at the slot of the cap, and they will
fall back into the liquid if they expend their energy content before they
reach the next plate. The liquid particles are relatively large, and a
relatively high velocity is necessary to cause them to be thrown to a
height of 18 in . This kind of entrainment can be eliminated almost
entirely by placing the trays far apart .
Another kind of entrainment might be referred to as “ carrying.” Very
small particles of liquid do not fall through the vapor as fast as the vapor
rises, and they are carried by the vapor stream to the plate above. At a
given velocity, all particles smaller than a certain size will be carried by
the vapor to the plate above and the larger particles will fall slowly back
to the parent plate. At ordinary vapor velocities, the size of the particle
that can be carried from plate to plate by the vapor appears to be between
0.1 and 0.2 mm in diameter.
Chillas and Weir 85 report that entrainment is negligible at a velocity of
2 ft per see but that it amounts to 7 and 20.5 per cent at 4 and 5.5 ft
per sec. In commercial columns, having trays spaced at 22 in , the .
entrainment 36 is approximately as shown in Table 16-4.
» Ind . Eng . Cftem. , 22, 206 ( 1930 ) .
” Analyses of commercial plates by the author.
FRACTIONATION AND TOWERS 491

TABLE 16-4

Linear velocity , Per cent


Material Pressure lb per sec entrainment

Gasoline Atmospheric 1.4 0.25


Gasoline Atmospheric 3.0 0.8
Gas oil 20 mm 10.0 1.8
Air-water Atmospheric 0.3 0.04
Air-water Atmospheric 2.2 0.7

-
Atkins22 suggests the entrainments of Table 16 5. The a / A designa¬
tion refers to the ratio of the required cap-covered area to satisfy the
widely used Brown and Souders37 vapor-velocity formula [Eq. (16-8) ,
curve 4], to the actual area provided. Inasmuch as the Brown and
Souders formula is conservative and is based on the total cross sec¬

tional area, a ratio of a / A 1.0 is thought to be 22 a good basis for design .

TABLE 16-5. ENTRAINMENT AS A FUNCTION or VAPOR LOADING ®

Vapor loading , Gal per min


a/ A per sq ft

Dry vacuum , low-pressure drop 0.4 0.03


Typical wet-vacuum distil) ation 0.6 0.1
Flash zone of crude tower 0.8 0.3
Basis for design capacity 1.0 1
Refluxed plates, atmospheric tower 1.1 2
Refluxed plates, butane splitter 1.2 5
Stripping section , butane splitter 1.25 10
Ultimate capacity 1.3 25

“ Atkina, G . T., Chem . Eng . Progr 50, 116 ( 1954).

Spraying or spouting of liquid seldom extends to higher than 18 in.


above the tray, and accordingly this type of entrainment is not involved
to any extent in the commercial operation of petroleum towers with spac-
ings of 18 in. or more, nor is such entrainment contemplated in any of
the figures given heretofore except possibly at the very highest load ¬
ings. Thus, the entrainment figures apply to the liquid droplets that
are entrained at a theoretical suspending velocity 37 [see Eq. (16-8) ].
The effect of tray spacing for this type of entrainment is indicated in
Fig. 16-13.
17

January , 1934, p . 32.



Fractionating Columns Entrainment and Capacity, Ref . Nat . Gaso. Mfr*
492 PETROLEUM REFINERY ENGINEERING

In actuality, trays are spaced primarily by mechanical consideration*.


Manways must be provided , and this requires a spacing of at least 22 in .
Although manholes may not be used on every tray, it is nevertheless nec¬
essary for a man to crawl across the tray for inspection and repair, and
-
this also requires a 20 to 24-in. spacing. The maximum practical num¬
ber of trays between manholes has beeh set at 20 by one designer, 27 but
more manholes are usually employed. Means are provided for descend ¬
ing or ascending through trays that are not served by manholes.
Although baffles, louver-like constructions, and wire mesh have long 38,19
been employed in efforts to alleviate entrainment, not until recently has
the use of wire^mesh entrainment blankets become widespread .46,11, 45
The allowable41 velocity u (ft per sec) ip gas separators equipped with
-
wire mesh mist extractors is:
» = K [ ( PL -
in which K ranges from 0.12 to 0.43 and pL and p0 are densities (lb per
cu ft) of the liquid and the vapor. Under many conditions the amount
of entrained liquid that is removed is 90 to 99 per cent. In vacuum
tower or jug operations 45 it is possible to attain the vapor capacities of
-
Fig. 16 13 and curve 4, without 'contaminating the gas oil product with
the metal contaminants that are so harmful to catalytic cracking catar-
-
lysts, and in many instances it has been possible to produce penetration
asphalt as the bottom product while at the same time producing a satis¬
factory catalytic feedstock . The mesh blankets (4 to 8 in . thick ) must
be continuously washed with a small amount of liquid to prevent the
accumulation of coke or debris. Blankets are being used by some
refiners below each draw plate and at the top of the column, but the
point of major usefulness is between the feed plate and the lowest side-
draw plate.
Many years ago, a device called the Centrifix38 was found to be effec¬
tive in removing colored material (Table 16-6) from a lubricating-oil
stock that was withdrawn above the vaporizer and in reducing the
amount of acid that was required to treat the product. A 9-in . layer of
steel wool *8 was able to reduce the color to about one-twentieth of the
color when no wool was used. Chillas and Weir 39 have also studied the
effect of a special baffle, similar to Venetian blinds, on entrainment.
“ Anon., Prevention of Entrainment . . . , Ref . Nat . Gaso. Mfr . , February , 1934,
p. 70.
M
Chilian and Weir , Ind . Eng . Chem., 22, 206 ( 1030 ) .
40 Reynolds, 8 C., Oil Gas J ., Aug. 10, 1953, p. 117 ; and Anon., Pet . Processing ,
,

February, 1954, p. 227.


41 Campbell , J . M. , Knitted-wire- mist Extractors . . . , Oil Gas J . , Mar . 5. 1956,

p. 115.
41 Nelson, W. L., Capacity of Vacuum Towers, Oil Gas J . , Apr. 9, 1956, p. 131.
FRACTIONATION AND TOWEBS 493
Diameter of Bubble Towers. It is evident from the foregoing that no
simple method of establishing the diameter of a fractionator is possible.
Perhaps the maximum in simplification is the preparation of charts simi¬
lar to Fig. 16-11 or 16-12 for the situation at hand , or for a number of
common situations. Nevertheless, some rapid means of approximation
is useful, so the familiar Brown and Souders equation (16 8) will be -
used . This equation was originally based on entrainment, but some
doubt has arisen17 regarding its relationship to entrainment. The equa¬
tion was also related to surface tension of the liquid, and this concept has
also been discredited .11 Finally, everyone agrees that the K constants

-
TABLB 16 0. REMOVAL OE ENTHAINMJSNT FROM VAPOH

Hi vis product

Vis¬
Crude Per cosity O . D. color O. D. color Color reduced to
® ®

cent of without with these per cents


at
crude Centrifix Centrifix of original color
210°F

Reagan 6 140 5 , 800 1 , 550 27.0


East Texas 6 140 11 , 900 1 , 650 13.8
Little Panhandle 5 125 13 , 400 4, 000 30.0
Barbers Hill 8 145 3 , 600 1 , 500 42.0

“The optical density (O . D.) color scale of the Atlantic Refining Company is directlj
proportional to the depth of color. See page 389 .

used in the original equations were conservative. Accordingly, empirical


K values based on general evidence in the literature will be presented
here (Fig . 16-13) .
The allowable vapor rate is a function of the densities p, and pL (Tb per
eu ft) at tower conditions and of a constant K which is related to service
and to tray spacing ( Fig. 16-13) ,
.
w / a = K \/ P ( PL - Pi ) . (16-8)
tP K- { pi, Pcf (16'9)
=

The symbol w refers to pounds per hour , and a to the area of the entire
tower in square feet . The constant K is dependent primarily on the tray
spacing (Fig. 16-13) but also on the type of tray and the kind of service.
The various general types of services are:
f The "Chemical Engineer ’s Handbook ," 3d ed . McGraw -Hill Book Company, Inc.,
f

New York , 1950, shows the equivalent of Eq. ( lfi-9) .


494 PETROLEUM REFINERY ENGINEERING

1. Maximum rates with perforated, sieve- type, Turhogrid, and Kaskade trays.
2. Maximum for perfectly designed bubble trays operating at moat favorable liquid
_
loads.**'11 Also , normal performance of Kaskade, Turbogrid , perforated, and
similar plate constructions.1Ma 4 # Also, liquid-washed aide-to-side pans.
1300
S"

- pounds per hr -sq ft


A , & A = dens( ttes, /
too — ——
fb per cu ft Si
/ A s’

/Jt
§ 900 '
/& /
/
* / / ze
s 700
_ u/ f
/
l .

^
Sbie 2 PSion
rfog

I2 OfidJ*ijder*
/

12
£ !
in0
jfi
I
500 /
/
i 50
^affS SSSi
f
/
5.

^ ^
too
to 20/5 25 30
Tray spacing, in.
PIG. 16-13. Effect of tray spacing and type of service on allowable vapor load of
fractionators [see Eq. (16-8)].
3. Normal performance of bubble plates through normal range of liquid loads” ’ *3, 48
at atmospheric and higher pressures. The lowest values of K apply to high liquid
loads and wide cap spacinga.
11 . ,
Thronton, D. P., Pei Processing May, 1952, p. 263; Kellogg, R. G., Oil Gas J
.
Apr. 18, 1955, p 128; Duren and Buck , Oil Gas JFeb. 28, 1955, p. 122; and Mayfield
et al „ Ind . Eng . Chem.., 44, 2238 (1952).
44
Fractionator Plate Symposium , Pel Engr., May, 1954, pp. C 17 to C 31. . - -
45 Barnes, K . B., Oil Gas J ., Sept. 15, 1952, p. 72; Shell Develop. Co., Chem Eng . .
Progr., February , 1954, p. 57; and Pel. Refiner November , 1952, p. 105 , .
44
Strang, L. C., J . InsL PeL Tech , 22, 166 (1936). .
FRACTIONATION AND TOWERS 495
4. Original Brown and Souders17 values for K dating from 1934 when tower design
was not fully understood . Also, vacuum- jug or dry-vacuum service using washed
-.
wire mesh mist extractors.4* Also, crude flash section of atmospheric topping
towers **
5. Stripping stills in absorption plants, and wet-vacuum refinery services.
6. Absorption towers, *3 47
. -
7 Foaming due to excessively high temperatures (vacuum decomposition ) in vacuum
service,12 or very high-viscosity liquids in vacuum service, 22 or high-boiling aro¬
matic fractions used as absorption oils in absorbers.48 Mist extractors are not of
much help for these services.
All of the allowable vapor velocity rates [Fig. 16-13 and Eq . (16-8 ) ] pro¬
duce about the same degree of fractionation . Note also that the rates
are based on the total free cross-sectional area of the tower and that
actual velocities in the immediate region of the part of the plate covered
.

with caps is therefore higher. Linear velocity is altered so greatly by


various conditions of temperature and pressure that it is not of much use
for purposes of comparison . However , for estimating purposes Table
16-7 is helpful .
TABLE 16 7. SUPERFICIAL VELOCITIES
- IN TOWERS

Pressure, Tray Superficial


Operation psia or spacing, tower velocity,
mm in . ft per sec

Topping 17 Jb 22 2.6-33
Cracking 401b 22 1.5-2.2
Pressure diet , rerun 20 lb 22 2.8-3.7“
Solution rerun 251b 22 2.8-35
Pressed dist. rerun 25 lb 22 2.8-3.9“
Pressed dist. rerun 60 mm 24 6.6-9.0
Vacuum 30 mm 30 9.0-12.0
Vaccum 90 mm 24 5.0-8.0
Stabilizer 160 lb 18 2.2-2.8
Nat. gaso. absorber 501b 14 1.0-1.3
Nat. gaso. absorber 4001b 18 0.5-0.8

e Greatly dependent on quantity of steam .


-. .
Example 16 6 Diameter of Topping Tower See Examples 16 1, 16 3, and 16 4. - - -
The quantities and conditions will be taken from these examples. Density of vapor
at top of column ( the reflux in the column is always hot reflux ):
Lb gasoline (Example 16-1) 3 , 415
Lb internal or hot reflux (Example 16-3) 15 , 500
Lb steam (Example 16 4)- 567
Total lb per hr 19 , 482
-
Total moles (Example 16 3) 203.5
47 .
Hutchinson , A. J , Pet. Refiner , April, 1951, p. 119.
. .
** Scliutt, H. C , Pet Refiner, July, 1945, p. 93.

You might also like