Designing piping for gravity flow

Gas entrained in liquid flowing by means of gravity from

a vessel can reduce the outlet pipe's capacity and cause
flow to surge cyclically. These problems Gin be avoided by
carefully designing for either full-liquid or two-phase flow.

P. D. Hills, Imperial Chemical hulllsiries PLC

o Entrainment curtails l iq uid g rav ity Bow from vessels I. For full !low, with the outlet piping size based on
hy raising the pressure drop (abovc that for single-phasc single-phase criteria.
flow) through the outlet piping, and reducing the static 2. For self�velltillg, with the liquid velocity in the outlet
head available for overcoming li It. pressure drop. A simi­ pipe kept 101\' enough to allow gas to now counter­
lar problem can arise wher. a liquid is near its boiling currently to the liquid.
point, or contains dissolved gas, especially if the absolute 3. For gas entrainment, but with the system designed
pressure at any point in the piping ralls below atmos­ to accommodate it.
pheric pressure, as occurs in a syphon. In gene ral. the first approach can be expected to result

J
l:onsider the Glse of liquid flow ing from the bottom of in the smallest pi pe diameter and should be given prefer­
an absorption column through a pipe that has been si zed ence. However, in many instances, it is not possible to
,. for full liquid flow (Fig. I). ensure full pipe now-in which case, the alternatives
ii When the liquid level in the column is low enough, the may have 1.0 hc adopted.
i-
liquid cntrains gas (Fig. la). The resulting i ncre ase in
pressure drop and reduction of head restrict the !1ow­ Designing for flooded flow
rate. and the liq uid level rises (Fig. Ib). Eventually, the To avoid gas entrainment in the full-pipe-flow desig n ,
level rises high enough to stop entrainment (Fig. \ c). the liquid level in the \"esscl must alwa ys be high enough
Howcver, gas s tilI in the oULlet p ipe causes I hc Icvel to to kee p I he pipe inlet floodcd. To achicvc this. some
C<JlIt.inlle to rise lllllil the g as is all swept out (Fig. Id). form of COllt rol will he necessarv , such as via a cOlltrol
Now, t he outlet pipe is ru nning fllllllow (as was ass ul I le d valve (Fig. 2a) 01- a vert icalloop
i� the piping (Fig. 2b). If
s
in the design), but the static head, becoming higher t h a n the latter is used, a syphon break will be necessary
\\'as assumcd, c reates excessive i 1o w. which causes the (shown in Fig. 2b), and the piping c1ownstl'eam of the
). level to (i111 u n til cntrainmelll ou:urs again and the cycle sypholl break cannot be assumed to run f l oo ded because
is repeated (Fig. I e). gas is likely to JlC
enlrained at the syp h on break. or
Such oscillations can be severe. d epending on syslem
,IT
course. either arrangement will increase the system pres­
I!;CI>ll1etry. III one case, the peak !low from a t ank ex­ sure drop and reduce somewhat the benelits of the
ceeded the ca;)acity of the vacuum breaker suniciently 1.0 lIoodecl-f1ow design.
Single-phase criteria Gll1 be applied to designing sec­
I .....
colla pse it.
P')- tions of outlet piping in which flow can be expected to be
General expression for liquid flow rate nood ed. I I' pip i ng that is certain to be Hooded is pre·
'·R.
In this anide, liquid lIowrates are generally expressed ceded by a self-venting s ection , the self-venting section's
ill terms or a di m e nsio n less superficial v olum etr ic flux. min im ulIl Icngth should be n.5 lll. to provide for gas dis­
}to which is defined by: engagement, beforc the piping is reduced for single­
phasc f low (Fig. 2c).
Jt = 4Q,/7Tlt2(!5d)1/2 ( I)
,I. rite criteria for Ilooded outlets are Eg. (2) for outlets
.\. a lIere. 0. is the volumetric !lowrate; d is the pipe I. D.; frolll the base of vessels. and Eg. (3) for outlets from the
linn
,I ,lIId g is the gravitational acceleration. Eq. (\) is si m i la l· to s ide of vesscl.�:
IlIe Froude number. It is used in prefe r cnce 10 the
,I
If.' 01 I'mude number, because the lalter's dclinilion I'arics.
.It < I.()(hld)�:h > 0_892(0.f/gdJo.25 (2)
circulllstances. All eq ll atio lls in I h is articlc Here, It is Ihe liq u id depth in the vessel away from the
Ic\\'
II"lIl
depending Oil

arc in consisten t units. t'egion of the outlet.

Designing for gravity flow . J r < (2hld)t/2:h > 0.81 I (0)2Igd" (3)
Th ree ap proaches to the design of gravity tl ra i na ge Here. It is the liquid height above the top of the outlet
<\sle ms arc pos sib le : away from the regi on of th e ou tle t .

1 (:III-·,�IH:"t. t-:N(:IN'·.F.I�IN(: SF.I'TE!\tHER �I. IllIn 111

 CRAVITY·U.OW PIP INC

Oesigning unflooded (self-venting) piping Near-lwriZlmtal piping-If such a pipeline will run on y
.
�
Side-outlet piping-Coming off li'om the side of a vessel, partially full, it must be inclined to provide the statrc
.
piping should be sized such that:
Jt. < O.3:d > (4Qt./(O.31Tgl12}'fJ.4 o·'t (4)
t:::Yih. head to overcome friction losses. A minimum slope qi
1:40 is recolllmended.

d.: ::: �1G. To avoid having the liquid carrying gas forward, ade­
This ensures that the line will run less than half full at its quate free area mllst be left in the pipe to al\?w �as �o
pass backward. For pipes up to 200 mm dla., hqUlU
I
entrance. The level in the vessel away from the outlet will .

be less than 0.8 d above the base of the line. The capacity depths should not be more than half the pipe diameter.
of such an overflow line can be found from Curve 1 in For larger pipes, depths up to three-fourths of the diam­
Fig. 3. eter may be possible.

Gas

l a. Gas is entrained in outlet,

restricting flow. Liquid flow in
exceeds flow out, and level rises.

Gas

J d. Liquid level now stops gas flow

o in.All gas has been swept from the
q outlet piping, and a syphon is
o formed. Single-phase flow in out·
let piping is helped by higher head.
.
o
. Now, outlet flow exceeds inlet
and fall s
o
flow, level rapidly.

��o�o�.�o��o�.---�

Gas

b. Higher liquid level reduces

entrainment, but flow out is still
restricted. Level continues to
rise.

Gas

� e. Entrainment starts again. Liquid

flow is still high, helped by syphon.
Built·up gas pressure in inlet pipe
is released in a surge. System is
Gas

J
returning to first stage.

__--L"""':
c. liquid level is now higb n o u",h"-- _
=��
to prevent further entrainment.
However, some bubbles are still
trapped in outlet, restricting flow.
Level continues to rise. Gas inlet
flow is beginning to be choked.

Entrainment causes surging by increasing pressure drop in piping and lowering head in vessel Fig. 1

C!HEMICAL ENGlNEF.RINC SEPTEMBER 5. /98:1

 � I-----I ---------------...,.
I
I
T II
r
(
I
I
I
I

I
I

Unflooded.'
I section \
\
\

[ >0.5
a. Control valve maintains level in vessel
m

\,

--Syphon break

i
_

c. Self·venting section precedes flooded piping

r.=====.:::::: ....- Un flooded

...; - -7 section
I
I
I

I
I
j
I
)

1/ i
1,000

' �
. !t:: ::i stabl;shed
,

r-- Curve 3IA-3I4.full ,o/ugh p.pel�"'" - +-flow --,-

500
b. :
il�'/ � �,<J'-:S
",' 'S'
Hydraulic loop ensures minimum level
I / 0° • V
"-r- 0"� .§-.,.,�� �,Q ",
� II V
I
",.01' /.
/....\�� .§:o,\0�o;
'(\\.�'b�.
�. '
200
/7'
�..,
Designs ensure flooded flow in outlet piping Fig. 2 ':'-1' /e.,q, I I

�
,\,,;� '\

'"
�'/(l
f)\'fI'- 4,(,
"v'?-
",.
/
r I
,
100
,
::..� ..}("
.
, G
J
.
When flow in a partially filled pipe is uniform (i.e.,
.
L
.'- /
<:::
constant depth), the energy lost through friction is bal·
, \
E 1// / / J
in�na­ 50
I

anced by the potential-energy ch ange clue to the

2]- '4 "f 'J / I !
tion of the pipe. In sllch a case. the mean vel{)cit�·, FL, is �
�I/ h
3: / \ �
.-
.9
I
related to the inclination and the de pth of !lo wing liquid
I
by Eq. (5) [1]:
L.L.

20 If / I, j

� V I

I
VL (32gmi)1I2Iog{[e/I4.8mJ + [0.2211Im(g"mi)I/�]} (5) I I

I
I

I
=

Here, m is the hydraulic mean depth (flow area/wetted

perimeter); i is the inclination of the pipe from the hori­
10 11 L !
:f,
zontal; is the pipe roughness; and is the kinematic ft .
,
€ v . !
v iscosity. II II I
5 I ;

Fig. 3 gives the volumetric capacity for established 110w I !

,
in half-full and three-quarters-full rough and smooth / I :
pipes. The curves were calculated via Eq. (5) for pipes of ! i
and a l1uid havin g a kinematic viscosity of i
,
slope 1:40 2 i I

I I I
/0-0 JJ12/S (e.g., water at 20°C). The absolute roughness I II
used for the rough pipes was 0.25 mm (moderately rusty
mild steel). The results are n ot very sensitive to liquid
viscosity. The capacity of a rough pipe is in creased by
II
50 100
l 150
I
200 250 300
!
350
i
I
400
I
45'

about 1 % ('0[, a totally inviscid liquid, and is only reduced Pipe dia., mm

by about 10% {'o[' a liquid having a kinematic viscosit}' 0('

Capacities for established
10-5 m�/s. Thus, the Fig. 3 water curves can be sale!}'
flows in unflooded pipelines Fig. 3
used l'or most liquids.

CHEMICAl. F.NGI�EERING SF.ITn'MIlF.R >. I"R:. II:.!

!EE PAZ.
 The initial velocity in an o tltlet line dcsign ed to Hill chosen lur self- vcnting" 1I0w (as in a prior section OIl de­
half full is less than the equilibrium velocity in a pipe signing unlloocled piping) , entrainment and su rging
having a slope of 1 :40. As the liquid accelerates down the may still occur due to the effects of the bends. The de­
pipe. the liquid depth diminishes with distance to that of sign recolllmendations now given are. therefore . offered
I'
the depth correspo nding to the established Ilow at a only tentatively.
I
i given flowrate. To maintain a constant relative depth. a Bends in the horizontal (or nearly horizontal) plane
I
tapered pipe would be necessary. As this is impractical. willnot necessarily cause problems if the 1: 40 slope is
reducing the pipe diameter stepwise is recommended. continued with the bend and the bend is gentle (prefera­
Tapered reducers should be installed to avoid sudden bly , the radius equaling five diamet e rs) .
disturbances in the !low. [n the vertical plane. the number of bends should be
For long lengths of pipes. the following design ap­ limited as mllch as possible. Gently sloping piping is
proach is suggested: preferabl e to vertical runs. The radius of bends should
1. Size the outlet line on the side of a vessel for Jt = be at least live diameters.
0.3 (Curve 1 of Fig. 3). If the resulting pipe size is not Bends from. or to. venical sections should be sized as
standard. choose the standard size higher thall the calcu­ fi)r vertical pipillg. Inclined piping follow ing a vertical
lated size. Continue the size so choscn for at least tell section (<In he sized ()!. half-full established !low \'ia the
I' pipe diameters. criteria ('or near-horizontal piping in the previous discll5-
2. Determine the pipe diameter corresponding to sion on designing unlloodcd (self-venting) piping.

Ii:
i'
I,
'1
half-full established flow for the required Ilowrate (using
Curve 2A or 2B of Fig. 3). Again. select the nearest
Changes in diameter should be made by means of ;\s\'m­
metric tapered reduccrs whose lengths are equal to twice
. .
I'
. ;
s tandard pipe size higher than the calculated size. the larger diameter. and which are installed so that the
1:: 3. Reduce the pipe diameter from the outlet size to thc bottom of the reducer has a slope equal to that or the
/\
established-flow size. using an eccentric reducer that will piping at either end.
not change the slope of the bottom of the pipe. Prefera­
bly. the reducer's minimum length should be twice that If entrainment is acceptable
of the upstream pipe diameter. There are lIIany occasions when it is not ncccssar\' to

If the foregoing procedure is followed for pipes of preve nt entrainment. Sometimes. moderate surging will
I: 40 slope , the liquid depth after the reducer will n ot not present a problem. In such cases. piping can be sized
exceed 75% of the pipe diameter. lor smaller diameters at considerable savings.
For long. large-diameter (>200 mm) inclined pipes, it Sometimes, sUl'ging caused by gas entrainmellt can be
may be worth considering a second reduction down to reduced hy providing a means for the gas to escape :u a
the size corresponding to an establis hed- Oow relative point downstream in the out let pipe . such as via some,
depth of 75%. This reduction can be made after 50 pipe type of gas - liquid separator. If t his is practical. the pi p ing
diameters (see Curve 3A or :m of Fig. 3). can be of smal ler r!iameter. However, because it is !lot
For short pipe runs, the additional cost of tapered re­ possible to predict the extent of entrainment-and.
ducers-especially if of a gentle angle , as recolllll1emled hence. calculate t.he pft.�ssure drop wit h certaint\,-'lI1�'
(which may lIot be standard) . or of lined pipe - may ex­ such approach should be auopteu cautiously.
ceed the savings in goin g (() smaller-diameter pi pin g. In
such cases. the entire length of the pipe should be of the Acknowledgment
large size. The author wishes to thank Imperial Chemical lndus­
tries ('or its permiss ioll to publish this article.
Self-venting flow in vertical pipes
Liqliid llowing vertically clown does so as an annular J. Matley. !:tI,/iII
!11m. In such cases. low superficial \ clocities are n ec e ssar y
to avoid gas being sucked down with the liqu id. Silll,;­
son's suggestion or basing pipe outlet diameters 1)1) a lim­ References
itin g Froude number o/" 0.3 is recommend ed [2J:
I. ,\ckCl"s. P. . r;IIJI(�s for 111(' h\"llriluli,: d('!ii�n nf "(flfm draill.... "('\n�IS illUI piPl"
Jt.< 0.3 «l) Iii 1(.'5. lin :\lajl'S(\",.. Sralillllcn' ()Ilkc. IlJf;�J.
2. Simpo;on, l.. 1.... SIzing piping ("or pr{)(css plillllS. (.'I"m. 1-:"1(.. 111111' Ii. 1(lIi�
Eg. «(j) being the same as Eg. (4). p ipe diameters can be
d etemlined rrom Curve 1 in Fig. :1.
This approach should be adopted when gas ent rain ­ The author
ment is to be avoided. as when a vertical pipe extends r. D. Hills is all ('II1,{im'cl ill� fkp:U11111'1t1

into a vessel to below the liquid surface. or ",hen the

hC.'ilHI :1IIo;I(,'r ami fluid·llow '1)('( iali�1 \\"j111
IlIIpe'nal Cht'micill 1lIIllIlOlri('<iI PI.C ,P.O,

dowllStream pip ing must be designed li)r flooded !lO\\!. Bw( �I). i, "'iIlJlillg-IfIIl. :"\lllIh\\-il"l1.
(:hc,'"hiu' (:\\,H 1 1>1. EI1�l;tnd\. Pf't..'\ iOllsly.
Smaller pipe than that dictated bv Eq. (fi) can be ex­ !'(' II:uIIH"t'II;' r"(HC'c;.o; ('II�iIU'('r ill\I,I ••."I'd

pected to cause surging. �\ illl 1,,1.1( t·...... il'n'/"llIlH'lJl. 111;1111

Itrlllhll'"hoOliIW; ;11111 (lnu (·....s Ik�il{ll. '-k

hllld� a B.Selbll,!:.) awl;1 I'h.D. 11"11111

Self-venting flow in complex systems 11II1'{'ri:d (:ull('gl.', l.ondlln.

Little infi)rtnation is available on lInl1oo<ied (low in

svstems that include hcnds. especially for !low changes
from vertica l to nearlv horizontal. and \'ice versa. Lim­
ited evidence suggests t. ha t e\'en if the pipe diameter is

114