You are on page 1of 0

AI R- ACT UAT E D PUMPI NG T E C H NO L O GY IN URBAN DRAI NAGE

b y
D . Aa r on Bohnen
B . A S c , The Uni ver si t y o f Br i t i s h Col umb i a, 1996
A TH E S I S S U B M I T T E D I N P A R T I A L F U L F I L M E NT OF
T H E R E Q U I R E M E NT S F OR T H E D E G R E E OF
M A S T E R OF A P P L I E D S C I E NC E
i n
' TH E F A C UL TY OF G R A D U A TE S TUD I E S
C I V I L E NG I NE E R I NG
We accept this thesis as c onfor mi ng
to the required standard
T H E U NI V E R S I T Y OF BR I TI S H C O L U M B I A
Apr i l 2004
D . Aa r on Bohnen, 2004
Library Authorization
In presenting this thesis in partial fulfillment of the requirements for an advanced
degree at the University of British Columbia, I agree that the Library shall make it
freely available for reference and study. I further agree that permission for
extensive copying of this thesis for scholarly purposes may be granted by the
head of my department or by his or her representatives. It is understood that
copying or publication of this thesis for financial gain shall not be allowed without
my written permission.
Name of Author (please print) Date ~(dd/mm/yyyy)
Title of Thesis: 4'/g^/}^fU/tT&h PUH4P'AJ6r TS^A/rJiFLO fry
Degree: Year:
Department of CWlL- &KT&? I K I S . ^ : fZ ?r4 ^
The University of British Columbia
Vancouver, BC Canada
A B S T R A C T
Ai r l i f t P ump t ec hnol ogy is bri efl y summar i zed and the potential appl i c at i on of airlift
t ec hnol ogy to l ow- l i ft , l ow- submergenc e, hi gh- fl ow applic ations suc h as open c hannel
fl ow i n urban st orm drainage, is expl ored. F our experi ment al setups are desc ri bed,
i nc l udi ng one prototype urban storm drainage instal l ation. Three desc ri pt i ve model s for
airlift pump operation are devel oped and one adopted for appl i c at i on i n l ow- l i ft , l ow-
submergenc e, hi gh- fl ow appl ic ations. The model al l ows a si mpl e desi gn proc edure for
airlift pumps i n this regi me. A si mpl e design hand-c al c ul ations proc edure is devel oped,
and t wo personal c omputer-based impl ementations are desc ribed. A si mpl e desi gn
exampl e is presented and rec ommendations for further research and devel opment
di rec t i ons are made.
TABLE OF CONTENTS
Ab st r ac t i i
Tab l e o f Contents i i i
L i s t o f Fi gures v
L i s t o f Tabl es v i
Ac knowl edgement s v i i
C H A P T E R 1 Over vi ew & S ummar y
1.1 I nt roduc t i on to Ai r l i f t P umps 1
1.2 Desc r i pt i on o f an Ai r l i f t P ump 6
1.3 Appl i c at i ons o f Ai r l i f t P umps 9
1.4 Projec t Sc ope and Rat i onal e 11
1.5 Two- P hase F l o w Regi mes 14
1.6 Operat i onal Effi c i enc y o f Ai r l i f t P umps 17
1.7 S ummar y 20
C H A P T E R 2 Literature R evi ew
2.1 I nt roduc t i on 22
2. 2 Devel opment o f Ai r l i f t P ump Theor y , 1797 to Present 22
2. 3 S ummar y 30
C H A P T E R 3 Exper i ment al Pr ogr am
3.1 Over vi ew o f the Exper i ment al Pr ogr am 32
3. 2 The Exper i ment al Setups 36
3. 2 Resul t s o f the Exper i ment al Pr ogr am 56
i i i
TABLE OF CONTE NTS (cont'd)
C H A P T E R 4 Three Ai r l i f t P ump Theoret i c al Model s
4. 1 Ai r l i f t P ump Mo de l for F i xed Bub b l e Vel oc i t i es 57
4. 2 Ai r l i f t P ump Mo de l for Var i ab l e Bub b l e S l i p Vel oc i t i es 65
4. 3 Ai r l i f t P ump Mo de l for Turbul ent M i x i n g 72
4. 4 S ummar i zi ng the Three Model s 79
C H A P T E R 5 P r el i mi nar y Desi gn o f Ai r l i f t P umps
5.1 P r el i mi nar y Desi gn Proc edure Cal c ul at i ons 81
5.2 Desi gn Cal c ul at i ons for Personal Comput er 86
5.3 Pr ac t i c al Consi derat i ons for P r el i mi nar y Ai r l i f t P ump Desi gn. 91
C H A P T E R 6 Conc l usi ons and Researc h Rec ommendat i ons
6.1 Conc l usi ons 94
6. 2 Researc h Rec ommendat i ons 95
R E F E R E NC E S 96
A P P E ND L X 1 99
LI ST OF FI GURE S
F i gur e 1 - Sc hemat i c Ai r l i f t P ump Lay out and Ter mi nol ogy 7 (also 57)
F i gur e 2 - Two- P hase Ai r - Wat er F l o w Regi mes 15
Fi gur e 3 - Two- P hase F l o w Regi mes characterized by Gas F l ux and Mi xt ur e 16
V o i d Fr ac t i on
F i gur e 4 - Fi r st Lab or at or y Ai r l i f t P umpi ng Sy st em 37
F i gur e 5 - Fi rst Lab or at or y Ai r l i f t P umpi ng S y st em Resul t s 38
F i gur e 6 - Prot ot ype Sc al e Lab or at or y Ai r l i f t P umpi ng S y st em 39
F i gur e 7 - Gi l b er t R oa d Prototype Ai r l i f t S y st em Lay out 42
F i gur e 8 - Gi l b er t R oa d Prototype Ai r l i f t S y st em Component s 43
F i gur e 9 - Gi l b er t R oa d Compr essed Ai r S uppl y Subsyst em 45
F i gur e 10 - Gi l b er t R oa d Prototype Ai r l i f t Sy st em i n Operat i on 46
F i gur e 11 - Sl ot - c onfi gur ed Ai r l i f t P ump Sy st em i n Operat i on 51
F i gur e 12 - R i c hmond P ub l i c Wor ks Exper i ment al Setup 54
F i gur e 13 - Bub b l e Ri s e Vel oc i t i es i n S t i l l Wat er 61
F i gur e 14 - Mi xt ur e and Gas F l ux Rates 65
F i gur e 15 - Compar i son of Exper i ment al and Cal c ul at ed Performanc e 78
v
L I S T OF F I G UR E S (cont' d)
F i gur e 16 - S i mpl e Desi gn E xampl e Lay out 82
F i gur e 17 - Ai r l i f t P ump Chur n F l o w Wor ksheet 87
F i gur e 18 - V B A Code for Chur n F l o w Ai r l i f t P ump Desi gn 89
F i gur e 19 - V B A Code for Chur n F l o w Ai r l i f t P ump Desi gn 90
L I S T OF TA BL E S
Tab l e 1 - Ai r l i f t P ump Nomenc l at ur e and Var i ab l es 8
Tab l e 2 - Prot ot ype Sc al e Laborat ory Ai r l i f t P umpi ng S y st em Resul t s 40
Tab l e 3 - Gi l b er t R oa d Prot ot ype Ai r l i f t Sy st em Sampl e Exper i ment al Dat a 48
Tab l e 4 - Gi l b er t R oa d Prototype Ai r l i f t S y st em Sampl e Exper i ment al 48
Resul t s
Tab l e 5 - Gi l b er t R oa d Prototype Ai r l i f t Sy st em Leakage Tests 49
Tab l e 6 - Gi l b er t R oa d Prot ot ype Ai r l i f t Sy st em Sampl e Exper i ment al 50
Resul t s 2
Tab l e 7 - R i c hmond P ub l i c Wor ks Sampl e Exper i ment al Dat a 55
vi
A C K N O W L E D G E M E N T S
I woul d l i ke to gratefully ac knowl edge my research supervisor, Deni s S. O. Russel l ,
Professor Emer i t us o f the Uni ver si t y o f Br i t i s h Col umb i a Depart ment o f Ci v i l
Engi neer i ng, for his c ontinuous support throughout this project. Thanks also to Professor
J i m At wat er and Professor Al a n Russel l , also o f the Uni ver si t y o f Br i t i s h Col umb i a
Depart ment o f Ci v i l Engi neer i ng. Thei r assistance and encouragement was par t i c ul ar l y
val uabl e. Thanks also to the Engi neer i ng Depart ment at the Ci t y o f R i c hmond, and M r .
Ar t hur L oui e whose interest i n the appl i c at i on o f airlift pumps to urban drainage made
this study possi bl e. F i nal l y , thanks also to the Nat i onal Sc i enc e and Resear c h Counc i l o f
Canada for their c ont ri but i on duri ng the earl y phases o f this project.
C H A P T E R 1
1.1 - I nt r oduc t i on t o Ai r l i f t P u mps
Ai r l i f t pumps are c ommonl y c onsi dered to be part of a uni que class of "alternate"
pumpi ng tec hnol ogies. These alternate pumpi ng technologies are requi red when c ommon
r ot ody nami c pumps are unsuitable for a gi ven project or appl i c at i on. S ome appl i c at i ons
that c ommonl y benefit fr om alternate pumpi ng technologies i nvol ve fl ui d/ sol i d mi xt ures,
very vi sc ous fl ui ds, hazardous fl ui ds, l i ve organisms suspended i n fl ui ds, l ow- head or
l ow- submergenc e situations, scenarios wi t h variabl e inlet water surface l evel s, etc. Ai r l i f t
and other alternate pumpi ng technologies pr ovi de a means for engineers to approac h
these scenarios.
Despi t e the success of the airlift pump i n several other areas, the airlift pump has not
gai ned acceptance i n c i vi l engineering appl ic ations. S pec i fi c al l y it has not been used for
management of st orm waters, pumpi ng fluids i n open channels, nor i n any other hi gh
disc harge, l ow l ift, l ow head appl ic ations despite the fact that i n some cases it may
pr omi se some advantages i n these settings. I n fact, an extensive literature search on airlift
pump research and devel opment found no references at al l to airlift pumps used i n hi gh-
discharge, l ow- head, l ow- l i ft capacities i n c i vi l engineering appl ic ations or otherwise.
Nevert hel ess, there are potential applic ations for l ow head, hi gh c apac i t y pumpi ng of
water i n open channels, and spec i fi c al l y of storm runoff i n drainage c ondui t s. Thi s
research pr ogr am is foc used on investigating those possi bi l i t i es.
1
The mai n feature of an airlift pump is a vert i c al tube wi t h the l ower end submerged i n
water and a suppl y of c ompressed air pr ovi ded to the l ower end. As the c ompressed air
fl ows into the l ower end of this tube bubbles are formed and a mi xt ure of water and air
bubbl es results. S i nc e this mi xt ure of air and water is less dense and thus l ighter than
water, the l evel of the air and water mi xt ure i n the vertic al tube rises above that of the
surroundi ng water surface. Wi t h a suitable phy si c al arrangement, this results i n
c ont i nuous "l i ft i ng" of the water to a higher l evel than the or i gi nal water surface - i n
effect c reat i ng an "ai rl i ft pump".
Ca r l Loesc her , a Ger man mi ni ng engineer, reportedly devel oped the or i gi nal airlift pump
c onc ept i n 1797 and the t ec hnol ogy began to gai n wi despread acceptance i n the mi ddl e
1800' s (War d 1924). By the early 1900' s several patents had been issued for vari ous
arrangements of airlift pumps and they were wi del y used for pumpi ng water, often fr om
quite deep wel l s, unt i l bei ng superceded by rel iabl e el ec t r i c al l y - dr i ven submersi bl e
r ot ody nami c pumps. There was a c onsiderabl e amount of early research into the airlift
concept, but broad interest on the t opi c waned as airlift pumps were superceded i n the
early 1900' s. Despi t e havi ng been repl ac ed i n c ommon use, airlift pumps have c ont i nued
to be used i n several spec i al i zed appl ic ations, whi c h are desc ri bed i n mor e detail later i n
this chapter.
2
The c i t y of R i c hmond i n Br i t i s h Col umb i a, Canada is situated i n the mout h of the Fraser
R i ver and experiences an average 1100 mm of rai nfal l per year
1
. The average el evat i on of
the c i t y is appr oxi mat el y one metre above mean sea l evel . Bec ause of this ver y l ow
el evat i on, muc h of the c i t y woul d be submerged under tidal or ri ver water dur i ng vari ous
parts of the year were it not for the extensive system of levees protec ting R i c hmo n d fr om
the Fraser R i ver and the ocean waters of Geor gi a Strait. Rec ent initiatives to further
i mpr ove the c i t y ' s prot ec t i on fr om ri ver and sea fl ood waters have been to pl an for the
i nst al l at i on of a so- c al l ed mi d- i sl and di ke to help prevent Fraser R i ver waters fr om
i nundat i ng central R i c hmond i n the event of a levee breach i n the East ern r egi on of the
muni c i pal i t y .
The average ground slope i n R i c hmond is zero and thus the muni c i pal stormwater
management syst em is constrained to very l ow slopes i n its' mai n c ondui t s. The pr ob l em
of l ow slopes is c ompounded b y the necessary l evee system used to protect the c i t y . The
levees create a need to pump stormwater out of R i c hmond when tides are unfavorabl e,
part i c ul arl y i n the wi nt er when the tides are rel at i vel y hi gh and constant. The stormwater
management syst em i n R i c hmond relies on l ow tides to al l ow the outfal l flapgates to
open. I n the wi nt er months the tides tend to be hi gh and constant, wi t h the dai l y sec ond
l ow tide st i l l very hi gh. Thi s is unfortunate t i mi ng since the wi nt er months are often ver y
rai ny i n the L o wer Ma i nl a nd. These factors result i n a real and ongoi ng danger of wi nt er
fl oodi ng i n the c i t y of R i c hmond.
1
City of Richmond Engineering staff graciously provided the background information on their stormwater
drainage system as presented in this brief section during various site visits, meetings and conversations that
took place from 1997 to 1999 throughout the development of this research project.
3
D ur i ng heavy rains, pumpi ng stations at the perimeter outfalls of the syst em c an pul l the
l oc al water l evel s down to shutoff but there can st i l l be fl oodi ng i n central R i c hmond
because of the i nab i l i t y to move st orm water qui c kl y enough to the outfalls. Rec ent
experi enc e has shown that R i c hmond experiences unacceptable st orm water l evel s and
fl oodi ng i n some areas as frequently as once every t wo or three years.
Bec ause of these concerns and i nc reasi ng high-density devel opment i n the ur b ani zed c ore
of R i c hmond, the c i t y has been c onsi deri ng options for i mpr ovi ng the c apac i t y of their
stormwater management system. Conc ent rat i on times are short so either faster r emoval of
runoff or detention and storage is required. Det ent i on and storage is probl emat i c gi ven
the hi gh water table i n R i c hmond, so the approach has been to c onsi der options foc used
on i nc reasi ng the rate of runoff r emoval .
The first opt i on presented was to introduce more and larger c onduits. Unfort unat el y this
strategy woul d be ext remel y expensi ve and very probl emat i c i n pub l i c i nc onveni enc e
since many of the mai n stormwater conduits are i nst al l ed i n bui l t - up areas and under
mai n c i t y roads. Addi t i onal l y , instal l ation of large concrete b ox culverts has b ec ome very
unattractive since Br i t i s h Col umb i a wor ker ' s protec tion l egi sl at i on c onc er ni ng the
c ondi t i ons requi red for their maintenance is so strict that it makes the upkeep of suc h
c ondui t s i mpr ac t i c al and very expensive.
The sec ond alternative was to investigate means of i nc reasi ng the effective sl ope of the
syst em b y i nc reasi ng the water surface grade wi t hi n the c onduits b y pumpi ng. Thi s
4
sec ond approac h woul d accelerate the mean flow vel oc i t i es and thus r emove stormwater
fr om the c i t y c ore at an increased rate.
A need for hi gh c apac ity, l ow head pumps that c oul d be i nst al l ed i n st orm drainage
c ondui t s to lift st orm water between 1 and three feet (0. 3 to 1.0 m) had devel oped. S uc h
pumps woul d increase the effective slope and hence the discharge c apac i t y of the exi st i ng
st orm drainage infrastructure. These pumps woul d onl y be required for short durations
under the c omb i nat i on of heavy rainfalls and hi gh tides.
Co mmo n r ot ody nami c pumps do not c onfor m to this hi gh- fl ow, l ow- head requirement
and i f pump units c oul d be found to satisfy these requirements they woul d st i l l be
expensi ve to i nst al l and house i n the R i c hmond system. Thi s is because of their need for
mi n i mum submergence l evel s at their inlets, necessitating substantial exc avat i on and
pl ac ement of infrastructure i n an area wi t h sandy soi l s and a hi gh watertable.
The di ffi c ul t i es and i mprac t i c al i t i es i n both of the proposed sol ut i on strategies have
effec t i vel y stopped Ri c hmond' s progress towards an i mpr oved stormwater management
system. Despi t e the i mpasse however, the danger of fl oodi ng i n central R i c hmond is real
and i nc reasi ng as urban devel opment continues.
Real i zi ng the need for a way forward, ah alternative pumpi ng t ec hnol ogy was sought and
this requirement spurred R i c hmond into sponsori ng the appl i ed research pr ogr am that is
desc ri bed i n this thesis.
5
1.2 - Description of an Airlift Pump
A n airlift pump i t sel f is c ompr i sed of fi ve major c omponents, namel y the air suppl y
apparatus, the air i njec t i on or aeration system, the water intake, the riser pi pe and the
pump outlet. F i gur e 1 shows the mai n elements of an airlift pump. Nomenc l at ur e used i n
that figure and other variabl es of interest are defined i n Tab l e 1.
A n airlift pump may also feature a so- c al l ed "foot pi ec e", a lengthened sec t i on of the
mai n riser pi pe l oc at ed b el ow the aeration poi nt and i n whi c h onl y single-phase water
fl ows. A foot pi ec e al l ows an airlift pump to entrain water fr om a depth greater than its'
aeration depth. Thi s al l ows a means for pump units wi t h l ow head ai r- suppl y apparatus to
suc c essful l y pump l i qui d fr om muc h deeper l evel s than they woul d otherwise be c apabl e
of. S i nc e foot pieces are required onl y i n scenarios i n whi c h the water to be pumped is to
rise fr om a great depth not al l airlift pumps feature foot pieces. In fact, most short airlift
pumps suc h as those i n this study, do not use foot pieces.
6
F I G U R E 1 - S c hemat i c Ai r l i f t P u mp L a y o u t
T A B L E 1 - Ai r l i f t P u mp Nomenc l a t ur e a n d V a r i a b l es
Ar ea = c ross-sec tional area of airlift pump tube
b = t uni ng parameter i n air phase vel oc i t y / mi xt ur e vel oc i t y rel at i onshi p
A
a
ir = area of fl ow mi xt ure cross section oc c upi ed b y air
A
w
= area of fl ow mi xt ure cross section oc c upi ed b y water
c = t uni ng parameter i n air phase vel oc i t y / mi xt ur e vel oc i t y rel at i onshi p
d = t uni ng parameter i n head loss equation
D i a m = diameter of airlift pump tube
Dens = relative density of the air-water mi xt ure i n the airlift pump tube
e = t uni ng parameter i n head loss equation
g
= ac c eleration due to gravity
Hdrive
= dr i vi ng head appl i ed to airlift pump
Hlift = l ift height of air-water mi xt ure i n airlift pump tube
Hioss
= head loss i n airlift pump tube
Htota
= height of pump lift above aeration poi nt
Hfoot
= height of pump tube footpiece b el ow aeration poi nt
H
s u
b
= height of standing water surface above aeration poi nt
Kent
= pump tube entrance loss factor
Kexit
= pump tube exit loss factor
Kpipe = pump tube pi pe loss factor
Ktotal
= total pump loss factor
Qair
= vol ume fl ow rate of air i n airlift pump tube
Qmix
= vol ume fl ow rate of the air-water mi xt ure i n airlift pump tube
Qwater
= vol ume fl ow rate of water i n the airlift pump tube
V
a i r
= vel oc i t y of the air fraction i n the air-water mi xt ure i n airlift pump tube
V
v
mix
= vel oc i t y of the air-water mi xt ure i n airlift pump tube
V
r e
l = relative vel oc i t y of the air phase to the water phase i n the airlift pump tube
Vwater
= vel oc i t y of the water fraction i n the air-water mi xt ur e i n airlift pump tube
Tlsystem
= airlift pump system effi c i enc y
Tlairdelivery
= airlift pump air del i very subsystem effi c i enc y
Tlriser
= airlift pump riser tube subsystem effi c i enc y
Pair
= density of gas phase
Pwater
= density of l i qui d phase
8
As ment i oned pr evi ousl y , i n this study onl y airlift pumps wi t h zero-l ength footpieces are
c onsi dered, so Hf
0 0
t = 0 and the total length of the pump riser tube is equal to the sum of
the submerged and unsubmerged lengths, H
S U
b and Hijf
t
.
1. 3 - Applications of Airlift Pumps
Despi t e havi ng been superceded b y submersibl e rot odynami c pumps i n most c ommon
appl i c at i ons, airlift pumps are st i l l used i n several spec i al i zed settings. Ty pi c a l moder n
appl i c at i ons of exi st i ng airlift pump t ec hnol ogy i nc l ude use of these pumps i n deep water
wel l s, where a related syst em known as a "geyser pump" is also b ec omi ng i nc r easi ngl y
c ommon where smal l - di amet er pump tubes are feasible. Ai r l i f t pumps also st i l l
frequently serve deep shaft and wel l dr i l l i ng appl ic ations.
Ai r l i f t pumps are also used i n modern wi ndmi l l - dr i ven pneumat i c al l y- operat ed water-
wel l pumpi ng appl i c at i ons, suc h as those avai l abl e as turnkey systems fr om Ai r l i f t
Tec hnol ogi es of Redl ands, C A .
Despi t e the fact that mi ni ng t ec hnol ogy has devel oped dramat i c al l y, airlift pumps are st i l l
a staple i n mi ne dewat eri ng, and modern exampl es are remarkabl y si mi l ar to the or i gi nal
syst em devel oped b y Loesc her i n 1797. Ai r l i f t pumps are also often used i n process
appl i c at i ons i n whi c h c orrosi ve or vi sc ous l i qui ds suc h as sand-water slurries, salt
sol ut i on, oi l s and various other waste products make traditional r ot ody nami c pumps less
suitable. (Gi ot , 1982) The oi l industry uses airlift pumps i n ret ri evi ng c rude oi l fr om dead
wel l s. The nuc l ear industry uses c areful l y c al ibrated smal l diameter airlift units to pump
fl ui ds i n nuc l ear fuel retreatment (Cl ar k & Dab ol t 1986).
9
Wast ewat er treatment plants are c urrently the most c ommon appl i c at i on for airlift pumps,
where exc el l ent aeration and subsequent oxygenat i on of the pumped mi xt ur e that is
der i ved fr om the injected air is a strong benefit. The Sanitaire c ompany of Br o wn Deer ,
W I b ui l ds stainless steel airlift pumps for this appl i c at i on.
Ai r l i f t pumps are often used i n aquaculture and fi sh farmi ng operations where their l ac k
of movi ng parts provi des necessary safety for fi sh and the air i nt roduc ed into the water
c ol umn i mpr oves oxygenat i on (Wurt s, M c Ne i l l & Overhul t s, 1994). The Aquac ar e
c ompany of Bel l i ngham, W A manufactures airlift pumps for fish far mi ng appl i c at i ons.
The c ompet i ng tec hnol ogies used i n fish farmi ng, namel y geyser pumps and propel l er
pumps, have the respective disadvantages of noise and possi bl e damage to fi sh safety i n
aquaculture appl ic ations.
Offshore mi ner al exc avat i on and di amond mi ni ng is an emergi ng appl i c at i on for airlift
pumps, where the l ac k of movi ng parts and abi l i t y to handle particulates make t hem
part i c ul arl y suitable. Ai r l i f t pumps are also sometimes used i n a si mi l ar manner for
underwater rec overy and salvage operations, where an airlift tube may be r i gged and
power ed fr om the surface, al l owi ng divers to pl ac e smal l items at the intake of the pump
and have the items c arri ed to the surface. Ai r l i ft pumps for use i n deepwater salvage often
feature tapered riser pipes, presumabl y so that as air bubbles increase i n size dur i ng their
rise fr om the aeration poi nt towards the surface the voi d ratio of the mi xt ure i n the pump
tube does not increase too muc h and reduce effi c i enc y. The airlift pump is very wel l
suited to underwater rec overy purposes since c ompressed air is a staple aboard salvage
10
vessels and the turbulent nature of the fl ow i n the airlift pump tube as wel l as the
upwar ds- openi ng shape of the c ommonl y - used airlift pump barrels i n this appl i c at i on are
doubtless hel pful i n avoi di ng any potential j ammi ng i rregul arl y shaped items may
experi enc e i n the pump risers.
The fi nal c ommon appl i c at i on of airlift pumps is i n l ake turnover, where these pumps are
used to counter the effects of lake stratification (Parker & Suttle 1987). I n l ake
destratification appl i c at i ons airlift pumps often float on smal l buoys wi t h their outlets at
the water surface and c ompressed air del i vered b y fl oat i ng suppl y l ines (Wurt s, M c Ne i l l
& Over hul t s 1994).
1.4 - Project Scope and Rationale
Thi s research project aims to investigate the sui t abi l i t y and behavi our of airl ift pumps i n a
new class of appl ic ations - namel y l ow- l i ft , hi gh- fl ow, l ow- submergenc e scenarios suc h
as pumpi ng i n open channels and management of urban storm drainage. Despi t e the
unort hodox concept, airlift pumps promi se many advantages i n suc h appl i c at i ons.
I nstal l ed costs are l ow since the pumps are si mpl e, c omposed pr i mar i l y of c ommonl y
avai l abl e P V C or steel pi pe fittings. Ai r l i f t pumps are very robust and nearl y
maintenance-free since they have no underwater movi ng parts (de Cac har d & Del hay e
1996). Addi t i onal l y , their air suppl y systems c an be l oc ated c onveni ent l y above gr ound to
mi ni mi ze i nst al l at i on costs and facilitate i nspec t i on and maintenance.
11
The following discussion of airlift pump efficiency suggests that the low-head, low-
submergence, high-flow, necessarily large diameter airlift pumps that would be required
in open-channel and urban storm drainage applications would be energy inefficient units.
Despite this inefficiency, the author believes that airlift pumps may offer enough other
cost and service advantages to offset the operational inefficiency of the airlift pumps that
would be applied in these settings.
Some of the advantages airlift pumps may offer to urban drainage applications include
low installation cost and maintenance cost, very low supporting infrastructure cost, and a
possibly huge placement benefit in the potential option for portable pump units and/or
portable power units, thus potentially eliminating entirely the need for a pump house or
similar infrastructure.
The aeration of storm runoff may also be a reason to consider the application of airlift
pumps to urban drainage applications. Urban storm runoff often contain high levels of
heavy metals, petrocarbons, chemicals from spills, and other roadwash pollutants and
tend to create potentially significant environmental impacts to the bodies of water into
which they discharge. (Turer, Maynard & Sansalone, 1996). Airlift pumps are used
routinely in aquaculture and wastewater treatment because of their significant benefit in
aerating the pumped liquid. Using airlift pumps for urban drainage could provide the
additional benefit of aerating the storm runoff, thereby mimicking the aeration process
used in many municipal mixed-sewage treatment plants, potentially accelerating
12
oxi dat i on of the roadwash and other stormwater pollutants, and al l owi ng for a decrease i n
resul t i ng envi ronment al impac ts.
Thi s c ompel l i ng array of advantages, part i c ul arl y the very l ow cost of i nst al l at i on and
maintenanc e of airlift pumps, the l ac k of a need for permanently i nst al l ed power and
c ont rol systems wi t h their attendant housi ng infrastructure, and the potential benefit of
aerating the runoff waters make the i nvest i gat i on of airlift pumps for these appl i c at i ons
very attractive.
13
1.5 - Two-Phase Flow Regimes
Ai r l i f t pumps are two-phase fl ui d fl ow devi c es. Gas and l i qui d (i n most cases air and
water) fl ow upwards together i n a vertic al pipe. Thi s two-phase fl ui d mi xt ure c an take
several different forms, and the various fl ow patterns of the t wo phases i n these forms
have si gni fi c ant l y di fferi ng hy dr aul i c behaviours. Thi s is signific ant to the sc ienc e and
desi gn of airlift pumpi ng systems since any of these forms of air-water mi xt ures are
possi bl e, and the for m found i n the system of interest is a very important vari abl e since
phy si c al relationships and deri ved mathematic al relationships are uni que for eac h for m.
The exact descriptions of the fl ow patterns vary somewhat b y author, t er mi nol ogy is not
al way s c ommon, and some fl ows are desc ribed as c ombi nat i ons of patterns (Shel t on &
Stewart, 2002).
A summar y of the fi ve basic forms observed i n the two-phase fl ow of water and air i n
vert i c al pipes, al ong wi t h their most c ommon names are shown i n Fi gur e 2. (modi fi ed
fr om Tai t el , Bor nea & Buc kl er , 1980). Fi gur e 3 shows the same fl ow regi mes
c harac t eri zed b y gas fl ux and mi xt ure vel oc i t y .
14
FI GURE 2 - Two-Phase Air-Water Flow Regimes
Bubbly Churn Slug Dispersed Annular
Flow "Froth" Flow Annular "Film"
Flow "Ripple" Flow
Flow
15
F I GURE 3 - Two Phase Flow Regimes characterized by Gas Flux and Mixture Void
Fraction, adapted from Wallis (1969)
PERFORATED PLATES
3 0
U o

x
NO. OF ORIFICES' DIAMETER(cm) SQUARE ARRAY SPACINGfcm^
I
4 9
100
2 8 9
4 . 0 6 X 10
4 . 0 6 X 10
1. 52 X 10
0. 41 X 10
- l
- l
6. 25 X
9 . 5 0 X
6 . 2 5 X
10
10
10
-I
I
-I
2 5
E
o
o
2 0
X
ID
CO
<
CD
15
10
CHURN
TURBULENT
REGIME
X
X
X
X
z
UJ
_ l
QC
Z>
X
o
XX
X X X
XX X
x
x
X X X *
X
X
X x
X
X
X
X
x A
XXX X
X XX
X X
XX
IDEAL BUBBLING
REGIME
0. 1 0. 2 0. 3
VOID FRACTION - oc
. 4
UJ
Z
o
u
cc
z
o
z
<t
cc
16
1.6 - Operat i onal Efficiency of Ai r l i ft Pumps
Ai r l i f t pump effi c i enc y c an be defined as the ratio of energy del i vered to the pump uni t to
the unit energy output i n the for m of vel oc i t y and head of the pumped l i qui d. The over al l
syst em effi c i enc y c an be c onsi dered a product of the air del i ver y and airlift riser
subsyst em effi c i enc i es. The effi c i enc y of the air del i very subsystem depends on the type
and c onfi gurat i on of the air suppl y equipment, pi pi ng, c onduits and c ont rol s. Effi c i ent
del i ver y of air through i nst al l ed conduits at desired pressures and fl ow rates is a wel l -
expl or ed and mature branc h of mec hani c al engineering.
Co mmo n wi s dom i n the design and use of airlift pumps suggests the effi c i enc y of an
airlift pump riser subsyst em is maxi mi zed when deep submergence is avai l abl e, the lift
height is l ow, and l i qui d and air fl ow rates are l ow. D e Cac har d & Del hay e (1995) and
D e Cac har d (1989) also suggest a very strong c ont ri but i ng effect i n the length-to-
diameter ratio, namel y that slender pumps wi t h hi gh length-to-diameter ratios are greatly
mor e efficient than their l ow length-to-diameter ratio counterparts.
The most efficient airlift pumps feature a situation i n whi c h the air and water phases have
very si mi l ar vel oc i t i es, air bubbles are either spheric al and very smal l or are large, dart-
shaped Tay l or bubbl es wi t h a cross section near the entire pi pe diameter. I n bot h of these
ma xi ma l l y efficient cases, the sl i p vel oc i t y between the air bubbles and water is
mi ni mi zed.
17
Ai r l i f t pump effi c i enc y is further enhanced b y use of the smallest possi bl e stable voi d
fraction - thus pumpi ng the ma xi mum amount of water per amount of air injected.
Aer at i on effi c i enc y is also an important factor i n det ermi ni ng the effi c i enc y of short
airlift pumps al t hough it matters less i n l ong pumps (Wal l i s 1968). Thi s phenomenon
appears to oc c ur because the l ong airlift pumps tend to operate i n sl ug fl ow. S l ug fl ow
oc c urs i n pumps l ong enough that smal l bubbles can accrete together to for m
homogeneousl y spaced large Tay l or bubbles c l ose i n cross section to the pi pe diameter
(Tai t el & al . 1980). In this fl ow regi me the fl ui d fl ows c ont i nuousl y i n contact wi t h the
pi pe wal l s c reating losses di rec t l y dependent on the fl ui d vel oc i t y .
At the entrance of l ong pumps (and i n shorter airlift pumps i n whi c h smal l bubbl es do not
have the opport uni t y to accrete into Tay l or bubbles before exi t i ng the pump riser) the air
and water mi xt ur e fl ow is turbulent and rec irc ul atory. Tai t el & al (1980) c harac terize this
fl ow regi me as "frot h" or "c hur n" fl ow and identify it b y the osc i l l at or y nature of the
l i qui d' s upwar d and downwar d mot i on between and around bubbles. A n aerator assembl y
that diffuses many smal l evenl y distributed bubbles into the fl ow fi el d helps reduce this
turbulence and rec i rc ul at i on, reduc i ng losses and i nc reasi ng effi c i enc y. Mor r i s on & al .
(1987) suggest that this is also true for the b ub b l y fl ow regi me where "mul t i por t i njec t i on
is more efficient".
Despi t e early and c ont radi c t ory observations suc h as those b y Wa r d (1924) and Bauer &
P ol l ar d (1945) on large diameter airlift pump systems, riser diameter also pl ays a rol e i n
airlift pump effi c i enc y for a gi ven lift height since larger diameter airlift pumps tend to be
18
less efficient than their smal l er counterparts. Thi s is because the larger diameter pumps
must be very l ong before the efficient Tay l or bubbl e- i nduc ed sl ug fl ow regi me c an
st abi l i ze (De Cac har d & Del hay e 1995). In fact, as the pi pe diameter increases the cross
sec tional area increases even more r api dl y , thus di mi ni shi ng the abi l i t y of surface tension
forces to hol d large bubbles intact against the i nfl uenc e of a c ompl ex turbulent shear fi el d
i n the air/water mi xt ure c ol umn. D e Cac har d & Del hay e (1995) also suggest that surface
tension forces i n bubbl es reduce sl i p vel oc i t i es between the air water phases. In that case
since bubbl e surface tension forces are increased i n smal l diameter pipes, the reduc ed
effi c i enc y of large diameter pumps may be due to greater sl i p vel oc i t i es, themselves due
to the reduc ed relative effect of surface tension forces.
Ob ser vat i on suggests that as the pi pe diameter increases above a ma xi mum feasible
bubbl e diameter the thickness of the f i l m i n the annular regi on surroundi ng the Tay l or
bubbl es i n the sl ug fl ow mi xt ure may begi n to t hi c ken rapi dl y. Thi s rapi dl y t hi c keni ng
f i l m c oul d then pr ovi de a dramat i c al l y increased fl ow path area for l i qui d fr om the regi on
ahead of any Tay l or bubbl e to sl i p downwards through the annular-shaped regi on, past
the Tay l or bubbl e and into the regi on behi nd the bubbl e. As the fl ow rate of the
downwar d- t r avel i ng fl ui d i n the annulus regions increases, the over al l fri c t i onal shear on
the pi pe tube may bec ome downwar d (Wal l i s 1968). In suc h cases the over al l lift
effi c i enc y falls r api dl y . Thus, i nc reasi ng pi pe diameter above the stable bubbl e diameter
for a gi ven fl ow fi el d may reduce efficiencies for l ong airlift pumps operating i n the sl ug
fl ow regi me.
19
1.7 - Summary
The mot i vat i on for this wor k is "Can appl y airlift pump t ec hnol ogy be pr ac t i c al l y appl i ed
to c i vi l engi neeri ng wor ks suc h as open c hannel drainage of urban st ormwat er?"
A n airlift pump is a dec ept i vel y si mpl e two-phase fl ow devi c e than can operate i n several
fl ow regi mes, dependi ng on several geometric and fl ow parameters. Ai r l i f t pumps have
been the subject of a smal l amount of research since their i nvent i on i n 1797 by Ca r l
Loesc her . S i nc e then they have been appl i ed ext ensi vel y i n a smal l number of spec i al i zed
appl i c at i ons but not to hi gh- fl ow, l ow- l i ft , l ow- submergenc e c i vi l engi neeri ng
appl i c at i ons suc h as open-c hannel drainage and st orm water management. Ai r l i f t pump
effi c i enc y is maxi mi zed i n scenarios i n whi c h submergence is hi gh, gas and l i qui d fl ow
rates are l ow and aeration effi c i enc y is hi gh. Despi t e the fact that l ow- head hi gh- fl ow
appl i c at i ons do not promi se very efficient operation of airlift pumps there are si gni fi c ant
reasons suc h as l ow i nst al l ed cost, ease of maintenance, reduc tion of envi ronment al
i mpac t of runoff water, etc. to investigate t hem for these uses.
Thi s thesis c onsiders the appl i c at i on of airlift pumps to these c i vi l engi neeri ng
appl i c at i ons and outlines a four-stage experi ment al program undertaken at the Uni ver si t y
of Br i t i s h Col umb i a and the Ci t y of Ri c hmond, Br i t i sh Col umb i a . Thi s study had several
objectives. These were namel y: to first evaluate the potential for airlift pumps i n urban
drainage and other l ow- sl ope open-c hannel appl ic ations, to create a mat hemat i c al
desc ri pt i ve model of l ow- head, hi gh discharge airlift pump systems, to devel op a
prac t i c al desi gn met hod for such pumps usi ng the mathematic al model above, and to
2 0
illustrate the use of that method. To these ends, small-scale and full-scale models were
built and tested. Water and air flows and levels were recorded. A broad literature study
was undertaken. From this theoretical background and experimental observations, three
mathematical models for predicting airlift pump behaviour in these settings are
developed. One is suggested as representative. A simple hand-calculator design
procedure is explained and two personal computer-based solutions are suggested. A
practical design example is presented, and conclusions and recommendations for further
development are made.
21
CHAPT E R 2
2.1 - Introduction - Literature Review
Thi s chapter describes a b r i ef history of the open literature on airlift pumps, pr ovi di ng an
over vi ew of the devel opment and theory behi nd their operation as wel l as an over vi ew of
airlift theory devel opment to the present day. Thi s literature search first investigated the
hi st ori c al use of airlift pumps i n c i vi l engineering appl ic ations. No ment i on was found.
Br oadeni ng the scope of the search revealed a ni c he body of literature c onc er ni ng airlift
pumps i n the process engi neeri ng fi el d, doc ument ed mai nl y i n the di sc i pl i nes of
Aquac ul t ur e and Chemi c a l Engi neer i ng.
2.2 - Development of Airlift Pump Theory, 1797 to Present
As not ed i n Chapt er 1, Ca r l Loesc her, a Ger man mi ni ng engineer, is thought to have been
first i nvent ed the airlift pump i n 1797 (Gi ot 1982). Loesc her ' s i nvent i on was an attempt
to si mpl i fy the pumpi ng tasks i n deep mi nes. Submersi bl e rot omac hi nery was not
avai l abl e i n the late 1700' s and the benefits of a pneumat i c al l y- operat ed syst em are
i mmedi at el y evident i n that context.
Ai r l i f t pumps bec ame popul ar several decades after Loesc her ' s first model s, dur i ng the
mi ddl e 1800' s (War d 1924). At this t i me direct pneumatic power was wi del y avai l abl e i n
the for m of b oi l er steam whi c h was easi l y generated at hi gh pressure. Pneumat i c power
was also avai l abl e fr om steam-powered c ompressors. Far aday ' s di sc over y of
el ec tromagnetic i nduc t i on i n 1831 l ed the way to the i nvent i on of the el ec tric mot or. Thi s
22
and the appearance of the internal c omb ust i on engine pi oneered b y R udol f Di esel and
others at the end of the same century made c ompressed air a vi ab l e source of power .
S haw (1920) first suggested a vol ume ratio for the gas and l i qui d phases i n a l ong airlift
pump riser tube operating at 100% effi c i enc y:
Volume
air =
Q
WMer
g-H,
\7~1 f rt
Volume
waler
discharge
In
(
P
aerationdepth
(1)
aerati
P
discharge J
S haw' s is the first attempt found i n the open literature to quantify airlift pump b ehavi our
on a phy si c al basis. Evi dent l y his relationship was suc c essful l y used i n desi gn wi t h an
effi c i enc y mul t i pl i er added, on the order of 50% (Z enz 1993).
Wa r d at the Uni ver si t y of Wi s c ons i n di d the first serious experi ment al study of airl ift
pumps found i n 1924. Thi s study focused on the behavi our of l ong airlift pumps and
attempted to create func t i onal relationships between the air and water phase fl ow rates,
effi c i enc y and pump riser geometry suc h as pump length and diameter. Wa r d devel oped
an elaborate c urve- fi t t i ng al gori t hm for use i n design but was onl y moderat el y satisfied
wi t h the results and qual i fi ed the tec hnique' s appl i c at i on to the l ong pump risers i n his
study.
23
Wa r d presented sixteen summary c onc l usi ons i n his study. Ma n y of Wa r d' s results and
suggestions for m the ongoi ng c ommon basis for subsequent use and underst andi ng of
airlift pumpi ng systems. Here is a summar y of Wa r d' s eight most salient c onc l usi ons:
1. The effi c i enc y of l ong airlift pumps depends pr i mar i l y on fl ow c ondi t i ons i n the riser
pi pe, and thus great refinement i n aeration and foot piec e design b ey ond ensuri ng
mi n i mum fl ow restriction at the entrance are not necessary i n most cases.
2. There is a ma xi mum effi c i enc y for every c omb i nat i on of pump geometry and
submergenc e that depends on water fl ow rate.
3. M a xi mu m effi c i enc y occurs at submergence ratios of greater than 70% i n most cases,
(ie: when over 70% of the total length of the riser tube is submerged) al t hough very
smal l diameter pumps can operate wi t h rel at i vel y hi gh effi c i enc y at l ower
submergenc e ratios.
4. H i gh effi c i enc y is possi bl e at l ower submergence ratios i f the aeration depth is deep.
5. The c omb i ned fri c t i on and sl i p losses due to the fl ow i n airlift riser pipes fol l ow a
different l aw than those that govern the fl ow of water or air i n a pi pe.
6. There is a rel at i vel y si mpl e relation between fri c t i onal losses and vel oc i t y of fl ow i n
an airlift riser pi pe for any partic ul ar mi xt ure of air and water.
7. S moot h joi nt s i n airlift riser pipes are necessary to avoi d unnecessary losses. Sudden
expansi on or c ontrac tion is very detrimental to efficient operation.
8. Ai r lift pumps of less than forty feet i n length are l i kel y to gi ve results muc h different
to those encountered i n l ong pumps. Losses that are rel at i vel y i nsi gni fi c ant i n large
pumps bec ome important i n short airlift pumps.
24
Ei ght years later i n 1932, Pi c ker t publ i shed "The Theor y of the Ai r l i f t P ump" i n an
attempt to elaborate on the mec hani c s of the fl ow i n these units. Hi s study di d not present
results greatly c ont ri but ory to the behavi our of the large diameter, l ow lift, l ow
submergenc e hi gh fl ow pumps of interest i n this study.
Mo r e than 25 years passed unt i l Govi er , Radfor d & Dunn' s "The Upwar ds Ver t i c al F l o w
of Ai r - Wa t er Mi xt ur es " appeared i n 1957. Thei r experi ment al study was based on a
1. 025" diameter pump riser 30' l ong (ie: length-to-diameter ratio approxi mat el y 350: 1).
They were able to accurately predict fl ow pattern, head loss and sl i p vel oc i t y but
restricted the appl i c at i on of their results to the behavi our of pump units of si mi l ar riser
tube diameters when pumpi ng mi xt ures of si mi l ar gas and l i qui d properties.
D J Ni c kl i n ' s "The Ai r l i f t P ump: Theor y and Opt i mi zat i on" of 1963 presented the first
satisfactory expl anat i on of the behavi our of smal l -diameter airlift pumps i n the b ub b l y
and mor e i mport ant l y, the sl ug- fl ow regimes. Ni c kl i n ' s moment um balance, 2-phase drift
fl ux model based on mass fl ow forms the basis of the b ul k of subsequent research into
airlift pumps and sl ug fl ow theory and behaviour. The most broadl y used of Ni c kl i n ' s
c onc l usi ons (recast here i n consistent t ermi nol ogy for this study) is used to characterize
the vel oc i t y of Tay l or bubbles i n the sl ug fl ow regi me i n st i l l water:
(2)
V
taylorbuhble
~ 0. 35 g Diam
25
Ni c kl i n also observed (l i ke War d) that although many aerators have been desi gned to
mi ni mi ze bubbl e size and maxi mi ze bubbl e di st ri but i on, none were successful i n l ong
airlift pumps. H e also first c l ari fi ed the one-to-one relationship between the submergenc e
ratio and the average pressure gradient i n the pump riser tube.
Mul t i phase fl ow was st i l l a nascent fi el d i n the 1960' s and devel opments i n this area were
happeni ng r api dl y . I n 1964, Duc kl er , Wi c ks & Cl evel and pub l i shed a two-part study
"F r i c t i onal pressure drop i n two-phase fl ow". Thei r results are i l l ust rat i ve of the st i l l -
devel opi ng nature of two-phase fl ow theory at that time. They found the exi st i ng
c orrel ations for pressure loss i n two-phase pi pe fl ow to be inadequate and asserted that
"There is not even a phenomenol ogi c al understanding of this type of f l ow. "
As two-phase fl ow theory was further devel oped, and due possi b l y to the expl i c i t sol ut i on
for sl ug fl ow operation as suggested b y Ni c kl i n , the study of airlift pumps c ont i nued to
focus i nc reasi ngl y on the mec hani c s of Tay l or bubbles i n the sl ug fl ow regi me, and to an
i nc r easi ngl y lesser degree on the b ub b l y fl ow, c hur ni ng fl ow and annular fl ow regi mes.
Wa l l i s ' defi ni t i ve wor k, One Dimensional Two-Phase Flow, appeared i n 1969. Wa l l i s '
text is st i l l one of the best sources for a broadl y- foc used c ol l ec t i on of most of the open
theory of one- di mensi onal two-phase fl ow. Wa l l i s ' wor k exposes the tremendous
c ompl exi t y i n mul t i phase fl ow behavi our and provi des muc h of the foundat i on for t wo-
phase fl ow as used today. Ma n y fri c t i onal and vel oc i t y relationships devel oped b y Wa l l i s
are st i l l state of the art i n modern two-phase fl ow theory.
26
Todor os ki , Sato and Honda fol l owed Ni c kl i n ten years later i n 1973 wi t h "Performanc e
of Ai r l i f t P umps", whi c h elaborated sl i ght l y on Ni c kl i n ' s approac h to the sl ug- fl ow
regi me fl ow of these devi c es. Todor oski , Sato and Honda modi fi ed Ni c kl i n ' s
experi ment al basis for determination of the sl i p vel oc i t i es i n sl ug fl ow.
The interpretation of the various regimes of vertic al two-phase fl ow was mai nl y
desc ri pt i ve i n nature unt i l 1980 when Tai t el , Bar nea & Duc kl er pub l i shed "Mo del i n g
F l o w Pattern Transi t i ons for Steady Upwa r d G a s - L i qui d F l o w i n Ver t i c al Tub es". Thei r
study undertook mat hemat i c al l y predi c t i ng the transitions between these patterns. They
were able to predic t whi c h pattern or regi me of two-phase fl ow woul d oc c ur under a
gi ven set of c ondi t i ons, and their approach is st i l l used today. Tai t el , Bar nea & D uc kl er
also pr ovi de the best of the vi sual descriptions of two-phase fl ow regimes (whi c h
suppl i ed the basis for Fi gur e 2 i n chapter 1). Thei r most useful fi ndi ng for this current
study suggests that (i n cases i n whi c h sl ug fl ow c an develop) the l ength of the turbulent
entrance or transition zone regi on fr om the aeration poi nt to the poi nt where sl ug fl ow
c an devel op depends on the mi xt ure vel oc i t y and pi pe diameter:
France = 40.6 Diam
yjg Diam
+ 0. 22 (3)
J
In 1982, Mar kat os & Si nghal produc ed a numeri c al analysis process for two-phase fl ow.
Thi s study was foc used on b ub b l y and sl ug fl ow, muc h the same as those that had
prec eded it. It appears that i n the b ub b l y and part i c ul arl y the sl ug fl ow regi mes the
mat hemat i c al formul at i on for the fri c t i on and loss terms is easier to ac c ompl i sh since the
27
rel at i vel y fi xed geometry of the round or Tay l or bubbles al l ow a sol ut i on that requires
less experi ment al data for c orrel at i on. Mar kat os & S i nghal ' s technique was devel oped for
use i n deep water wel l s and depends on the b r eakdown of l ong vert i c al risers into smal l er
c ont i guous segments, i n effect creating a "gradual l y var y i ng f l ow" for mul at i on. It is
suitable onl y to l ong riser pipes.
L o n g, smal l diameter airlift pumps are wi del y used i n nuclear fuel reproc essing. V e r y
accurate estimates of fl ow rates are required i n those settings. In 1986 Cl ar k & Dab ol t
devel oped a general set of design equations for airlift pumps i n sl ug fl ow for use i n the
nuc lear industry. They foc used pr i mar i l y on accurately predi c t i ng the fl ow rate b ehavi our
i n their appl i c at i ons. Despi t e their admitted i nab i l i t y to accurately c al c ul ate the over al l
fri c t i onal losses i n pipes of 38 mm diameter they di d provi de an accurate desi gn model
for suc h pumps i n the sl ug fl ow regime. I nterestingly they also attempted to appl y
Ni c kl i n ' s model to a short pump and found that Ni c kl i n ' s model overpredi c t ed the pump
effi c i enc y, unl i ke its' better agreement when appl i ed to l onger units. Cl ar k & Dab ol t ' s
general desi gn equation for l ong, smal l -diameter pumps does not address pump effi c i enc y
i n great detail but does pr ovi de an accurate and prac t i c al means of desi gn for very l ong
slender airlift pumps.
In 1993 Z enz produc ed "E xpl or e Potential of Ai r l i f t Pumps and Mul t i phase Sy st ems"
pr i mar i l y expl or i ng airlift pumps i n three-phase scenarios. Z enz' study was c onc erned
mai nl y wi t h sl ug fl ow and particulate entrainment, again i n l ong pipes. Ai r l i f t pump riser
pipes are general l y c onsi dered l ong when length to diameter ratios are 50: 1 or more.
28
Wur t s, M c Ne i l l and Overhul t s (1994) pr ovi ded a si mpl e c urve- fi t t i ng approac h to airlift
pump performanc e for near- 100% submergence i n aquaculture and destratification
appl i c at i ons.
I n 1995 Tr amb a, Topal i dou, Kast r i naki s, Ny c as , Franc oi s & Sc ri vener c ompl et ed their
"V i s ua l St udy of an Ai r l i f t P ump Operat i ng at L o w Submergenc e Rat i os" whi c h is not
part i c ul arl y hel pful for this present study since it is mai nl y c onc erned wi t h bubbl e
format i on at a jet inlet and inc l udes no performance data for non- sl ug fl ows.
F ol l owi ng Cl ar k & Dab ol t i n the nuc lear fuel reproc essing industry, D e Cac har d &
Cal hay e created a steady-state model for very smal l diameter, l ong lift pumps i n 1995. D e
Cac har d & Cal hay e' s is c ertainl y the most extensive study found. It is c onc erned
pr i mar i l y wi t h c reating an accurate model for gravitational and fri c t i onal c omponent s of
the airlift pump riser pressure gradient. L i ke Cl ar k & Dab ol t ' s wor k, it is foc used on very
l ong "sl ender" airlift pumps i n the sl ug fl ow regime. D e Cac har d & Del hay e are not
c onc erned wi t h opt i mi zat i on of energy effi c i enc y since energy inputs are very smal l i n
their cases of interest. D e Cac har d & Del hay e observed c hur ni ng fl ow i n the l ower
sections of their study pump units and c onc ur wi t h previ ous researchers that c hurn fl ow is
a devel opment phase for the sl ug fl ow pattern. However , they also found that c hurn fl ow
c oul d exi st as a stable fl ow pattern at hi gh gas fl ow rates. D e Cac har d & Del hay e
devel oped the most detail ed and accurate analysis framework avai l abl e for airlift pumps
of under 40 mm diameter and wi t h length-to-diameter ratios above 250: 1.
29
Mo s t rec ently Nenes, Assi mac opul os, Mar kat os, & Mi t s oul i s c ompl et ed "S i mul at i on of
Ai r l i f t P umps for Deep Wat er We l l s " i n 1998. Thei r anal yt i c al framework i nvol ves an
interspersed c ont i nua model and solves a system of differential equations per Mar kat os
(1982). Resul t s fr om their system are very accurate but unfortunately suitable onl y for
very t al l pi pe units i n whi c h the pump riser tube may be broken into tens of i nt ernal l y
c ont i guous discrete elements.
2. 3 - S u mma r y of L i t er a t ur e R evi ew
Ai r l i f t pumps have been a ni c he interest area i n process, c hemi c al and mec hani c al
engi neeri ng as wel l as aquaculture. Pub l i c at i on on the topic has been sparse and the
literature has tended towards attempts to expl ai n the behavi our of these devic es i n t wo
distinc t fl ow regimes. Ni c kl i n ' s model (1963) continues as the base for al most al l
theoretical devel opment whereas the numer i c al techniques of Mar kat os, Nenes et al .
(1992) pr omi se a power ful toolset for eval uat i ng the behavi our of l ong airlift units.
There has been l ittl e reason to evaluate the hi gh- fl ow, l ow- head, l ow- sub mer genc e airlift
systems and subsequently those appl ic ations are st i l l unexpl ored fr om theoretical and
desi gn standpoints. Thi s has not stopped suc h pumps fr om bei ng used spor adi c al l y and
often uni nt ent i onal l y since i n a prac t i c al sense, si mpl i c i t y i n fi el d use has tended towards
i nst al l i ng a pi pe at an appropriate depth, addi ng air i n an appropriate vol ume and at an
appropriate depth to produc e the desired results i f possi bl e. In a research sense, the
i nab i l i t y to accurately measure the relative vel oc i t i es of the air and water phases exc ept i n
the b ub b l y and sl ug fl ow regimes have tended towards t uni ng a theory foc used on those
30
regimes alone. H i gh submergence, smal l diameter, short lifts have been t radi t i onal l y
investigated since l ow- submergenc e, high- l ift units pr ovi de dec reasing effi c i enc i es.
As Wa r d (1924) c onc l uded, there remai n inherent gaps i n our understanding of theory
that mai nl y arise fr om the fact that the sizes, speeds and di st ri but i on of the air bubbles are
not known, yet the si ze and the rate of ascent of the bubbles through the air-water mi xt ur e
are c r i t i c al variabl es.
Rec ent research wor k on air entrainment i n fast fl owi ng water usi ng laser opt i c al probe
t ec hnol ogy wi t h the abi l i t y to measure the behavi our of the air fraction i n a two-phase
air-water fl ow mi xt ure as distinct from the over al l air-water fl ui d mi xt ur e has rec ently
b ec ome avai l abl e. Thi s technique was empl oy ed first by Cart el l i er (1992) and later
refined b y S er dul a & L oewen (1998), whose techniques mi ght seem to pr omi se a more
rigorous approac h to air lift pump design b y the direct measurement of gas and l i qui d
phase vel oc i t i es and voi d ratio. Unfort unat el y, investigation of the experi ment al
equi pment and personal conversations wi t h L oewen suggest that since the laser syst em
empl oy ed is a' point measurement system it is not suitable for air-water mi xt ures wi t h
rec i rc ul at ory movement . It cannot resolve differences between upwar d- movi ng and
rec i rc ul at i ng air bubbl es and does not function wel l wi t hout distinct boundaries between
the air and water phases at the bubbl e boundaries suc h as are found i n b ub b l y and sl ug
fl ow. Unfort unat el y this means that at least c urrently the laser measurement t ec hnol ogy is
i nappl i c ab l e to the study of airlift pumps i n the c hurn fl ow regime.
31
C HA PT E R 3
3. 1 - Overview of the experimental program
F or airlift pumps to be used effec tivel y i n short lift, hi gh fl ow appl i c at i ons suc h as st orm
drainage c ondui t s, the airlift pump must be able to "l ift" large vol umes of water through a
height of about 0. 3 to 0. 6 m. F or exampl e, suc h an increase i n head over a run of 5000
feet i n rel at i vel y flat terrain suc h as exists i n Ri c hmond, B. C i n the 5' b y 9' concrete b ox
c ul vert l eadi ng fr om Gr anvi l l e Street to the Gi l b er t R oa d outfall c oul d easi l y result i n a
doub l ed water surface slope and subsequent dramatic i mprovement s i n water fl ow
vel oc i t y , and subsequent reduc tion of l oc al fl oodi ng duri ng extreme rai nfal l events.
A pump syst em for oc c asi onal use i n emergenc y situations suc h as mi ght be experi enc ed
b y R i c hmond dur i ng the 10-year desi gn fl ow shoul d i deal l y be i nexpensi ve and
maintenanc e- friendl y. S i nc e the pump units woul d run onl y dur i ng extreme c ondi t i ons,
and sinc e operating costs i n extreme c ondi t i ons are often ac c ounted for differentl y than
ongoi ng costs, effi c i enc y is onl y important i n so muc h as it affects the first cost of the
i nst al l at i on. Runni ng costs are less important as the pumps are onl y used dur i ng extreme
st orm events - i n the order of once every t wo or three years. However , i nst al l ed cost is
important - of course less expensi ve is preferred. Ai r l i f t pumps pr omi se a very attractive
mat c h to these c riteria. The c ompressed air suppl y can be dry and out of the way , i n fact
there is no need for the suppl y apparatus to be permanentl y l oc ated since air c an be
suppl i ed through a fl exi b l e pipe. Thi s means that a permanent pumphouse need not
nec essaril y be used wi t h an airlift system. The pumps themselves c an be b ui l t quite
32
i nexpensi vel y as they onl y consist of sets of tubes and aerators wi t h an air suppl y.
Bec ause airlift pumps are not ori ousl y inefficient c ompared wi t h their r ot ody nami c
counterparts and air compressors are expensi ve, it is very desirable to have the ratio of
water l ifted to c ompressed air used be as hi gh as possi bl e.
These desi gn c onsiderations and potential benefits were investigated, mot i vat i ng the
current study. P r el i mi nar y c al c ul at i ons were made and t wo series of pr el i mi nar y
qual itative experiments were run at the Uni ver si t y of Br i t i s h Col umb i a Ci v i l Engi neer i ng
Hy dr aul i c s Lab or at or y to c hec k the concept. Cl assi c al airlift pump c omponent s and
layouts were c onsi dered and adapted for use i n a l ow- l i ft situation. S i nc e the desire was
to determine whether a vi ab l e airlift pump c oul d be devel oped for these appl i c at i ons it
was reasonable to start wi t h a system that was c onfi gured si mi l ar l y to what a wor ki ng
unit mi ght be.
Fi r st a smal l - sc al e airlift unit was bui l t usi ng the ful l wi dt h of a 6 i nc h wi de
undergraduate student hydraul i c s lab fl ume, expl or i ng the c onc eptual l ayout for a ful l -
wi dt h b ox c ul vert instal l ation.
Subsequent l y a si mi l ar larger-scale airlift unit was bui l t usi ng the ful l 20-inch wi dt h of a
large hy dr aul i c s l ab fl ume to c hec k i f the system woul d be func tional on a larger scale.
33
The laboratory experiments pr ovi ded some val uabl e insights into the nature of airlift
performanc e at l ow lifts, l ow submergence and l ow voi d ratios. They also showed that
airlift pumps of this type c oul d be prac t i c al .
Next a sequence of ful l -sc al e prototype experiments was c arri ed out at the Gi l b er t R oa d
st orm drainage out fal l i n Ri c hmond. Several ful l -sc al e experi ment al prototype l ayouts
and systems were pl anned and tried. Var i ous c ombi nat i ons of upstream and downst r eam
water l evel s were investigated wi t h var y i ng rates of air injec tion. Ri ser pi pe diameter was
investigated, as was aerator geometry.
S i nc e eval uat i ng and ma xi mi zi ng water fl ow for these pump systems was the end goal of
this study, i n al l cases it was attempted to determine the water fl owrate possi b l e for a
gi ven upst ream and downst ream depth, rate of air injec tion, pump riser diameter and
aerator geometry.
There were many prac t i c al diffic ul ties suc h as unusual l y l ow fl ows and water l evel s i n
the drainage c ondui t that was used as the site. However , these experiments pr oduc ed
several ver y useful results. They demonstrated that l ow- head, hi gh- fl ow, l ow-
submergenc e airlift systems di d pr ovi de a vi abl e and prac t i c al alternative i n an i nst al l ed
st orm drainage system. However , they also showed that there was c onsi derabl e
c i r c ul at i on i n the pump tubes and as a result, the water was b ei ng l ifted several times wi t h
resul t i ng l ow over al l effi c i enc y. They also showed that there was never a steady state
situation suc h as is usual l y assumed i n der i vi ng theory and formul ae. The air-water
34
mi xt ur e was very turbulent and the air bubbles were i nc reasi ng and dec reasing i n si ze b y
shearing and c oalesc enc e as they rose, i n effect al ways i n a transient c ondi t i on.
H a vi ng i dent i fi ed the major pr ob l em of c i r c ul at i on and r e- c y c l i ng, a fi nal set of
experiments were set up at the R i c hmond P ub l i c Wor ks Y a r d, usi ng banks of tubes 75 to
200 mm i n diameter, as opposed to the 250 and 300 mm tubes used i n the Gi l b er t R oa d
prototype experiments. Theor y suggested that smal l er diameter pipes woul d al l ow less
c i r c ul at i on due to a more evenl y distributed air phase. Al s o , the riser pi pe wal l s woul d
have a muc h greater i nfl uenc e over a larger cross-sectional fl ow area. Tr i al s were also
made wi t h the tubes i nc l i ned instead of vert i c al , as i nc l i ned tubes woul d be easier to
i nst al l and c oul d be pot ent i al l y less c ost l y due to a reduc ed number of fittings requi red
for c onst ruc t i on.
The results fr om the fourth experi ment al setup were successful and showed l ittl e
c i r c ul at i on, al t hough there was st i l l considerable uncertainty as a result of the transient
nature of the under l y i ng phenomenon of the air bubbles ri si ng, expandi ng, shearing and
c oal esc i ng. S i nc e the experi ment al "pumps" were si mi l ar to the proposed fi nal desi gn the
results were c onsi dered acceptably accurate. The design concept was thus c onsi dered
pr oven and the relationships devel oped sufficient for design of a prac t i c al operating airlift
pump.
35
3. 2 - The E xpe r i me n t a l Set ups
The first experi ment al laboratory setup is shown i n Fi gur e 4. The purpose of this first
setup was to b ui l d a vi sual model of the airlift pump as a ful l - wi dt h element i n a st orm
drai n sc enario. Bec ause of known l i mi t at i ons i n wi dt h, discharge rate, fl ow rate
measurement equipment and theoretical knowl edge, this first setup was intended to serve
as a base for experi ment at i on to ai d i n understanding airlift pump behavi our, rather than
as an instrumented data c ol l ec t i on experiment. Thi s system was desi gned to si mul at e on a
very smal l scale the or i gi nal proposed l ayout of an i n- c ul vert airlift pumpi ng syst em.
Upst r eam water fl owed into the system under a baffle. A n aerator i nst al l ed on the base of
the c hannel suppl i ed bubbles to the water c ol umn. The air-water mi xt ure then fl owed
between the upstream baffle and a downst ream baffle, exi t i ng the pump unit at the hi gher
downst r eam l evel .
The smal l - sc al e syst em was i nst al l ed i n a si x- i nc h wi de student hydraul i c s lab fl ume to
test the c onc ept of a ful l - wi dt h airlift system i n a rectangular c hannel . Var i ous
c omb i nat i ons of upstream and downst ream water l evel s and air vol ume inputs were t ri ed
i n order to maxi mi ze the water flowrate gi ven any c omb i nat i on of upstream and
downst ream l evel s. Several geometries were tried since al l of the vari ous syst em
elements were modul ar. The most effective l ayout found is shown i n Fi gur e 4.
36
F I G U R E 4 - F i r s t L a b o r a t o r y Ai r l i f t P u mpi n g S y s t em
Preliminary Lab Airlift Pumping System
1
n
,
o
to
all dimensions in mm
Several aerator designs were also tried, wi t h little i mprovement i n effi c i enc y. Ci r c ul at i on
was evident as a very important (and pr evi ousl y muc h under-estimated) effect i n these
hi gh- fl ow, l ow- submergenc e scenarios.
The best per for mi ng setup i n the pr el i mi nar y experiment was measured for performanc e
and pr ovi ded the first clues to the shapes of pump discharge curves for systems of this
nature. F i gur e 5 shows the sample data set and resulting discharge c urve for this syst em.
The sec ond experi ment al setup was a si mpl e test intended to determine the possi b i l i t y of
a larger-scale syst em based on the same c onc eptual l ayout as the first smal l - sc al e system.
The larger scale syst em bui l t i n a large 20- i nc h wi de hydraul i c s fl ume at the Uni ver si t y of
Br i t i s h Col umb i a Depart ment of Ci v i l Engi neer i ng. Thi s system was desi gned to
determine the feasi bi l i t y of airlift pump t ec hnol ogy at prototype scale for very shal l ow
37
F I GURE 5 - First Laboratory Airlift Pumping System Results
Preliminary Airlift Pump Curve
0.800
0.700
0.600
0.500
$
5 0.400
s
x
0.300
0.200
0.100
0.000 -
0.000

4
y -
O.O003x
J
+ 0.O< 24x+0.719-\
10.000 20.000 30.000
Flow Rate, gpm
40.000 50.000
Q, gpm
Poly- (Q. gpm)
These operating characteristics are for the model pump which has a width of 156mm and hence art
operational lift chamber plan area of 29640 mm
2
.
Lift, mm Q, 1/s Lift, feet Q, gpm
226.000 0.333 0.753 5.283
205.000 0.714 0.683 11.321
182.000 1.176 0.607 " 18.647
162.000 1.667 0.540 26.417
116.000 2.222 0.387 35.222
139.000 2.353 0.463 37.294
67.000 2.500 0.223 39.625
89.000 2.500 0.297 39.625
39.000 2.857 0.130 45.286
7.000 3.077 0.023 48.769
i nsert i on depths. Several tests were run and results were pr omi si ng. Ci r c ul at i on was very
st rongl y evident i n the large 18" x 20" lift chamber. Ac c ur at e instruments for measuri ng
the ai r fl ow i n the system were not avai l abl e and therefore direct numer i c al results for
38
ai r fl ow were not c ol l ec t ed. Thi s was not c onsi dered a drawbac k because the larger scale
l aboratory airlift unit di d prove the concept at the prototype scale despite serious
submergenc e l i mi t at i ons and l i mi t at i ons i n the air fl ow rate possi bl e fr om the i nst al l ed
sc rew c ompressor- based air del i very system. Fi gur e 6 shows the sec ond experi ment al
setup and Tab l e 2 shows numer i c al results of this phase of the study.
F I GURE 6 - Prototype Scale Laboratory Airlift Pumping System
lift chamber length 18"
V
Qwater i ^ > H u/s
Hd/ s
If Qair
S ec ond Phase Ai r l i f t P ump Test Setup
U B C C i v i l E ngi neer i ng Hy dr aul i c s Lab or at or y
flume width 20"
three-tined 3/4" brass aerator, 3 x 30 ea. orifices 1 mm dia.
lift chamber plan view with aerator
T
18"
_L
20"
39
T A B L E 2 - Prototype Scale Laboratory Airlift Pumping System Results
Airlift Flow Test 2
Aaron Bohnen
UBC Civil Engineering Hydraulics Lab
Flow baffle height Hds
Weir crest height Pweir
Weir crest width Wweir
Coefficient of Discharge = 0.63 for this
test.
Q=Cd*(2/3)*L*H
A
(3/2)*SQRT(2*g)
25.5 in
23 in
19.5 in
0.65 m
0.59 m
0.50 m
Head d/s Head u/s
(Hds, in) (Hus, in)
Weir head
(Hweir, in)
lift
(delta
H, in)
Flow
(m
A
3/s)
Flow
(l/s)
Flow
(cfs)
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
25.5
26.0
24.5
24.0
23.0
22.5
22.0
21.5
20.3
18.5
17.0
16.0
15.0
14.5
14.0
5.5
5.3
5.0
4.5
4.3
4.3
4.0
3.5
2.8
2.0
1.5
1.0
0.5
0.3
-0.5
1.0
I. 5
2.5
3.0
3.5
4.0
5.3
7.0
8.5
9.5
10.5
II. 0
11.5
0.050
0.046
0.043
0.037
0.034
0.034
0.031
0.025
0.018
0.011
0.007
0.004
0.001
0.000
49.5
46.2
42.9
36.6
33.6
33.6
30.7
25.1
17.5
10.9
7.1
3.8
1.4
0.5
Notes:
Pressure at 100 psi delivered, approx. 4 psi at
aerator
20" wide in 39" tall flume
four-tined 3/4" diam aerator -1 mm holes
1.75
1.63
1.52
1.29
1.19
1.19
1.09
0.89
0.62
0.38
0.25
0.14
0.05
0.02
The l i mi t ed ai r fl ow avai l abl e fr om the screw-based c ompressor sl owed devel opment unt i l
a four horsepower gasol ine engi ne- powered centrifugal b l ower was obtained fr om the
uni versi t y equi pment salvage program. The b l ower was overhaul ed, performanc e was
eval uated and fittings were designed to adapt it to the experi ment al setup. The c ent ri fugal
40
b l ower was not i nst al l ed i n the sec ond laboratory experi ment al setup sinc e at this t i me a
full-scale prototype l oc at i on was selected i n the Ci t y o f R i c hmond and the experi ment al
pr ogr am shifted focus to that l oc at i on.
The t hi rd set of prototype experi ment al equipment was devel oped, assembl ed and
i nst al l ed just upstream o f one o f the fl ood boxes at the outlet o f a drainage c ondui t at the
Gi l b er t R oa d storm water outfal l i n Ri c hmond. The experi ment al setup was i nst al l ed over
the wi nt er o f 1997 to 1998. It or i gi nal l y c onsisted o f three different pump designs ai med
at i nvest i gat i ng key features o f the proposed pumps. The or i gi nal equi pment was
c ompr i sed o f t wo, t en- i nc h diameter pump units and one, t wel ve- i nc h diameter unit. One
o f the t en- i nc h diameter pump units was constructed fr om clear ac r y l i c pi pe, al l owi ng
vi sual i nspec t i on o f the mi xt ur e fl ow regi me wi t hi n the pump riser pi pe.
Wat er was i nt roduc ed fr om an upstream c hamber over a V - not c h wei r and pumped b y the
experi ment al airlift units into a downst ream chamber. Anot her V - not c h wei r at the output
o f the downst ream c hamber enabl ed the water pumpi ng fl owrate to be measured. Wat er
l evel s were read fr om staff gauges.
Tests were r un unt i l the syst em had st abi l i zed, at whi c h poi nt measurements o f al l the
water l evel s and ai r fl ow rates were taken. The water l evel s were used to find the
flowrates b y c onvent i onal V- not c h wei r analysis. Fi gures 7 and 8 show the or i gi nal
prototype l ayout at the Gi l b er t R oa d l oc at i on. Addi t i onal large-scale syst em and site
drawi ngs c an be found i n Appendi x 1.
41
F I G UR E 7 - Gi l b er t Road Prototype Ai r l i ft System Lay out
F I GURE 8 - Gilbert Road Prototype Airlift System Components
Compr essed air was suppl i ed to diffusers l oc ated beneath eac h pump uni t b y a 28
horsepower Comai r - Rot r on posi t i ve displ ac ement b l ower uni t and pi pe di st ri but i on
system. Thi s unit was i nst al l ed i n a soundproofed shed and equi pped wi t h intake and
disc harge filters and silencers. Si nc e posi t i ve displ ac ement b l ower s pr ovi de a constant
rate o f ai r fl ow, i ndi vi dual val ves were i nst al l ed at eac h pump uni t and a bypass added. I n
this wa y eac h pump unit c oul d be tested i ndi vi dual l y . Vent ur i - st y l e air vel oc i t y meters
were also i nst al l ed to al l ow air fl ow to eac h pump unit to be i ndi vi dual l y moni t or ed.
F i gur e 9 shows the air suppl y system at the Gi l b er t R oa d site. Fi gur e 10 shows the
prototype syst em i n operation. Large- sc al e site drawi ngs i n Appendi x 2 show the
i nst al l at i on o f the air suppl y subsystem at the Gi l b er t R oa d l oc at i on.
44
F I GURE 9 - Gilbert Road Compressed Air Supply Subsystem
2"> LONG RADIUS
90" ELBOW (TYP.)
3"0 AIR FLOW METER
AIR FLOW
3"0 AIR PIPE
3 0 3-WAY WYE
4"x3" REDUCER
4 LONG RADIUS
90' ELBOW
STEEL LADDER
2"x3" REDUCER
2 0 CONTROL VALVE
4 0 AIR SUPPLY
DETAI L Y
45
F I G U R E 10 - G i l b e r t R o a d P r ot ot y pe Ai r l i f t S y st em i n Oper a t i on
The or i gi nal system as tested was moderatel y successful at pumpi ng water and pr ovi ded
tremendous insight into the operation o f the pump units, part i c ul arl y b y the ab i l i t y to
observe the mi xt ure flow pattern i n the ac r y l i c t en- i nc h diameter unit (the centre uni t
shown i n Fi gures 9 and 10). The or i gi nal prototype setup i dent i fi ed several weaknesses i n
the phy si c al l ayout and c onst ruc t i on o f the layout at the prototype site. Leakage bet ween
upstream and downst ream chambers was found to be part i c ul arl y probl emat i c .
Addi t i onal l y , the ab nor mal l y l ow l evel s o f water i n the drainage c ondui t l eadi ng to the
site over the wi nt er o f 1997/ 1998 made testing at onl y one l evel o f upstream fl ow
possi bl e. A portable pump was i nt roduc ed to increase the l evel i n the upstream c ondui t
but this had onl y l i mi t ed success.
46
These di ffi c ul t i es l ed to del ay i n this phase o f the pr ogr am as several revi si ons to the site
syst em l ayout and air di st ri but i on l ayout were devel oped, drafted and i mpl ement ed. Thi s
wor k progressed over muc h o f the S pr i ng o f 1998.
The results fr om the tests o f these prototype units were more scattered than expec t ed but
c l ear l y showed the importanc e o f desi gn details. Al t hough reduc ed, leakage bet ween
c hambers c ont i nued to be probl emat i c despite the system revi si ons, and a b r eakdown o f
the c ompressed air del i ver y system created further di ffi c ul t i es.
S ome o f the data and c al c ul at i ons fr om the t hi rd experi ment al setup are shown i n Tab l e 3.
Tab l e 4 shows sampl e vel oc i t y and loss c al c ul at i ons for the data shown i n Tab l e 3. Tab l e
5 shows the leakage test data c ol l ec t ed at this site. Tab l e 6 shows a sampl e o f later data
and also c al c ul at i ons for the loss coefficients o f the three or i gi nal prototype pump
systems at the Gi l b er t R oa d site.
Whe n anal y zed, the results o f this phase o f testing i ndi c at ed syst emi c pr ob l ems wi t h the
alternatives b ei ng investigated. The large diameters pipes had ver y l o w pac ki ng densi t y
i n the space-constrained storm drai n scenario. Al s o no aerator geometry was found to be
hi ghl y successful i n sharpl y ma xi mi zi ng effic ienc y. Thi s was l i kel y due to several
factors, pr i mar i l y the prac t i c al issue o f leakage and the prevalent rec i rc ul at i ng f l ow
patterns i n the pump risers.
47
T A BL E 3 - Gilbert Road Prototype Airlift System Sample Experimental Data
Results march 9,1998
Outside wl = 33 Vee notch = 34 inches
Pump U/wl d/swl aircfm Flow pipe dia area qmix vmix headless Mix density vwat vair vrel
2.5 8
35.000 40.25 250 0 676762 0 833 0 54498 4.843429 8.887344 35.7942 0.391054 0.485614 12.55535 12.06974
34.500 39.75 250 0 578271 0 833 0 54498 4.744937 8.70662 35.313 0.395151 0.419289 12.64>t 12.22111
34.500 39 250 0 450973 0 833 0 54498 4.617639 8.473037 33.44365 0.411212 0.340279 12.98521 12.64493
39.000 41.5 . 250 0 974318 0 833 0 54498 5.140985 9.433339 41.45398 0.413587 0.739413 13.0378 12.29839
38.750 41.25 200 0 908704 0 833 0 54498 4.242037 7.783834 28.22426 0.500209 0.834052 12.23798 11.40393
35.250 40.75 200 0 786746 0 833 0 54498 4.120079 7.56005 26.6247 0.45895 0.66255 11.30473 10.64218
35.000 40.5 200 0 730287 0 833 0 54498 4.06362 7.456452 25.9 0.462329 0.619532 11.37578 10.75624
35.000 40.5 150 0 730287 0 833 0 54498 3.230287 5.927345 16.3665 0.552526 0.740398 10.25159 9.511189
35.000 40.75 150 0 786746 0 833 0 54498 3.286746 6.030943 16.94361 0.543329 0.784363 10.04514 9.26078
37.750 41.62 150 1 006934 0 833 0 54498 3.506934 6.434972 19.28984 0.562277 1.038892 10.47997 9.441076
38.000 41.5 100 0 974318 0 833 0 54498 2.640985 4.846018 10.9397 0.675072 1.206897 9.411976 6.205079
34.500 40 100 0 626111 0 833 0 54498 2.292777 4.207082 8.245128 0.658465 0.756489 8.954315 8.197826
40.250 41 250 0 846199 0 833 0 54498 5.012866 9.198249 39.41356 0.406102 0.63056 12.87348 12.24292
33.500 38 250 0 316525 0 833 0 54498 4.483191 8.226335 31.5245 0.356521 0.207068 11.88156 11.6745
35.000 38 200 0 316525 0 833 0 54498 3.649858 6.697228 20.89419 0.443392 0.257522 10.98875 10.73123
39.750 41 200 0 846199 0 833 0 54498 4.179532 7.669142 27.39864 0.471078 0.731449 11.56395 10.8325
37.000 40 150 0 626111 0 833 0 54498 3.126111 5.736189 15.32789 0.524148 0.602177 9.640225 9.038048
35.500 39 150 0 450973 0 833 0 54498 2.950973 5.414823 13.65853 0.5125 0.424095 9.40989 8.985795
T A BL E 4 - Gilbert Road Prototype Airlift System Sample Experimental Results
Results March 9, 1998
1 2 3 4 5 6 7 8 9 10 11 12 13
Pump Us/ wl d/s wl Air Water Vmix Dens Vw Head loss t Wm
A
2 Vm
A
2*Den UV w
A
2 -exit/vw
A
2
ins. ins. cfs cfs ft/sec ft/sec feet
2 35.00 57. 75 4.17 0.68 8.96 0.23 5.53 1.44 1.16 5.12 3.03 2.44
34. 50 57. 75 4. 17 0.58 8.78 0.21 5.05 1.46 1.22 5.78 3.69 3.05
34. 50 57. 75 4.17 0.45 8.54 0.19 4.34 1.55 1.36 7.10 5.29 4. 55
39. 00 57.75 4. 17 0.97 9.51 0.27 6.76 1.60 1.14 4.28 2.26 1.73
1.00
2 38.75 57.75 3.33 0.91 7.83 0.30 5.54 1.42 1.49 4.92 2.98 2.38
35. 25 57. 75 3.33 0.79 7.61 0.29 5.10 1.21 1.34 4.71 2.99 2. 36
35. 00 57.75 3.33 0.73 7.51 0.28
i nn.
4.89 1.23 1.40 5.07 3.31 2.65
2 35.00 57.75 2.5 0.73 5.97
1 .UU
0.34 4.00 0.96 1.74 5.16 3.87 3.12
35.00 57. 75 2.5 0.79 6.08 0.35 4.19 0.92 1.61 4.63 3.37 2.65
37. 75 57.75 2.5 1.01 6.48 0.38 4.87 1.00 1.53 3.99 2.71 2.03
1.00
2 38. 00 57.75 1.67 0.97 4.89 0.48 3.79 0.62 1.66 3.49 2.76 1.97
34.50 57.75 1.67 0.63 4.24 0.41 2.81 0.60 2.15 5.22 4. 89 3.95
1.00
1 40. 25 54. 5 4.17 0.85 9.27 0.25 6.27 1.85 1.38 5.54 3.03 2.48
33. 50 54.5 4.17 0.32 8.29 0.17 3.43 1.60 1.50 8.80 8.77 7.77
35. 00 54.5 3.33 0.32 6.74 0.20 2.86 1.59 2.25 11.00 12.54 11.40
39. 75 54. 5 3.33 0.85 7.72 0.29 5.32 1.62 1.76 5.97 3.70 3.08
37. 00 54.5 2.5 0.63 5.78 0.32 3.64 1.30 2.50 7.86 6.32 5.51
35.50 54.5 2.5 0.45 5.45 0.28 2.93 1.31 2.84 9.98 9.80 8.82
Outside wl = 33 Vee notch = 34 Average #2 1.48 4.96 3.43 2.74
Av. #1 2.04 8.19 7.36 6.51
Sdev #2 0.28 0.91 0.89 0.82
Sdev #1 0.58 2.17 3.69 3.46
CV#2 0.19 0.18 0.26 0.30
CV#1 0.29 0.26 0.50 0.53
48
T A B L E 5 - Gi l b er t Road Prototype Ai r l i ft System Leakage Tests
Pump Data from May 1 tests
No 3 pump-12 "
Air U/ sW,L D/ SW. L Weir Weir flow Leaks Total
250 46.5 50.5 41.75 1.127673 0.33775 1.465423
250 44.5 50 41.75 0.974561 0.334608 1.309169
250 37.5 48 41.75 0.489537 0.321734 0.811271
250 38 48.75 41.25 0.769405 0.326621 1.096026
250 46 50 41.25 1.127673 0.334608 1.462281
250 42.5 49.5 41,25 0.974561 0.331436 1.305998
200 45 50 41.75 0.974561 0.334608 1.309169
200 38.25 48 41.75 0.489537 0.321734 0.811271
150 47 50 41.75 0.974561 0.334608 1.309169
150 44.5 50 41.75 0.974561 0.334608 1.309169
150 38 48 41.75 0.489537 0.321734 0.811271
100 45.5 49 41.75 0.707362 0.328234 1.035596
No 2 pump
250 48 49 41.75 0.707362 0.328234 1.035596
250 45 48 41.75 0.489537 0.321734 0.811271
250 37.5 47 41.75 0.317686 0.315099 0.632785
220 39 48 41.25 0.592484 0.321734 0.914218
200 45.5 48.5 41.75 0.592484 0.325 0.917484
150 48.5 49 41.75 0.707362 0.328234 1.035596
150 45.5 48.5 41.75 0.592484 0.325 0.917484
150 39 47 41.75 0.317686 0.315099 0.632785
100 45.5 48 41.75 0.489537 0.321734 0.811271
No 1 pump
250 48 49 41.75 0.707362 0.328234 1.035596
250 45 48 41.75 0.489537 0.321734 0.811271
250 38.5 47 41.75 0.317686 0.315099 0.632785
200 46 48.5 41.75 0.592484 0.325 0.917484
200 38 47 41.75 0.317686 0.315099 0.632785
150 49 49 41.75 0.707362 0.328234 1.035596
150 46 49 41.75 0.707362 0.328234 1.035596
150 39.5 47 41.75 0.317686 0.315099 0.632785
100 46.5 48.5 41.75 0.592484 0.325 0.917484
49
TABLE 6 - Gilbert Road Prototype Airlift System Sample Experimental Results 2
Tests May 1, 1998
1 2 3 4 5 6 7 8 9 10 11 12 13
Pump U/Swl d/swl Air Water Pipe area V mix Dens Vw Head loss UVm*2 Ud"vm
A
2 UVw2
ins. ins. cfs cfs sq. ft. ft/sec ft/sec feet
3 46.50 53.50 4.17 1,80 0.79 7.60 0.38 5.99 1.86 2.07 5.42 3.34
3 44.50 53.50 4.17 1,31 0.79 6.97 0.33 4.98 1.88 2.50 7.46 4.89
3 37.50 53.50 4.17 0.81 0.79 6.34 0.28 3.73 1.53 2.45 8.84 7.08
3 38.00 53.50 4.17 1.10 0.79 6.70 0.31 4.49 1.44 2.06 6.62 4.59
3 46.00 53.50 4.17 1.46 0.79 7.17 0,35 5.31 1.95 2.44 6.96 4.44
3 42.50 53.50 4.17 1.31 0.79 6.97 0.33 4.98 1.72 2.28 6.82 4.47
3 45.00 53.50 3.33 1.70 0.79 6.41 0.43 5.07 1.56 2.44 5.72 3.91
3 38.25 53.50 3.33 1.20 0.79 5.77 0.37 . 4.09 1.21 2.34 6.27 4.65
3 47.00 53.50 2.50 1.31 0.79 4.85 0.46 3.66 1.61 4.41 9.68 7.77
3 44.50 53.50 2.50 1.31 . 0.79 4.85. 0.46 3.66 1.40 3.84 8.43 6.76
3 38.00 53.50 2.50 0.81 0.79 4.22 0.39- .2.65 1.12 4.07 10.45 10.30
3 45.50 53.50 1.67 1.04 0.79 3.44 0.52 2.53 1.22 6.66 12.76 12.37
2 48.00 57.75 4.17 1.04 0.55 9.55 0.28 6:91- 2.31 1.64 5.95 3.12
2 45.00 57.75 4.17 0.81 0.55 9.13 0.2S 6.06 2.19 1.69 6.89 3.84
2 37.50 57.75 4.17 0.63 0.55 8.81 0.22 S.27, 1.67 1.39 6.31 3,89
2 39.00 57.75 3.67 0.91 0.55 8.41 0.28 5.89 1.52 1.39 4.87 2.82
2 45.50 57.75 3.33 0.92 0.55 7.80 0.30 5.52 1.98 2.09 6.86 4.18
2 48.50 57.75 2.50 1.04 0.55 6.49 0.39 4.91 1.87 2.86 7.39 5.01
2 45.50 57.75 2.50 0.92 0.55 6.27 0.37 4.56 1.70 2.78 7.54 5.26
2 39.00 57.75 2.50 0.83 0.55 5.75 0,32 3.63 1.37 2.67 8.33 6.71
2 45.50 57.75 1.67 0.81 0.55 4.55 0.45 3.32 1.36 4.23 9.42 7.94
!
48.00 54.50 4,17' 1.04 0.55 9.55 0.28 6.91 2.39 1.69 6.14 3.22
1 45.00 54.50 4.17 0.81 0.55 9.13 0.25 6.06 2.26 1.74 7.10 3.96
1 38.50 54.50 4.17 0.63 0.55 8.81 0.22 5.27 1.82 1.51 6.85 4.22
1 46.00 54.50 3.33 0.92 0.55 7.80 0.30 5.52 2.10 2.22 7.29 4.44
1 38.00 , 54.50 3.33 0.63 0.55 7.28 0.26 4.45 1.61 1.96 7.51 5.25
1 49.00 54.50 2.50 1.04 0.55 6.49 0.39 4.91 2.02 3.09 7.97 5.40
1 46.00 54.50 2.50 1.04 0.55
1
6.49 0.39 4.91 1.77 2.71 6.98 4.73
1 39.50 54.50 2.50 0.63 0.55 5.75 0.32 3.63 1.50 2.92 9.11 7.34
1
46.50 54.50 1,67 0.92 0.55 4.74 0.47 3.60 1,49 4.26 9.11 7.38
Average #3 3.13 7.95 6,21
Stdsv #3 1.38 2.17 2.77
CV#3 0.44 0.27 0.45
: Average #2 2.30 7.06 4.75
Std#2 0.93 1.33 1.68
CV#2 0.40 0:19 0.35
Average #1 2.45 7.56 5.10
Std #1 0.88 1.01 1.44
CV#1 0.36 0.13 0.28
A second prototype concept was then designed and buil t at the Gil bert Road l ocation.
This system was conceived i n an attempt to maximize the use of the available pl an area i n
the constrained space of a drainage conduit. The concept was based on the original lab
model that used the ful l width of a smal l flume. A ful l -width buil t- in pump assembly was
designed. It formed a continuous side-to-side element i n the base of the drainage conduit.
50
S uc h a unit woul d have superior aeration density and make maxi mal l y - effi c i ent use o f the
l i mi t ed pl an area i n the base o f the conduit. The result was a lateral slot-based pump and
aerator. Thi s unit was bui l t i n the for m o f a slot four feet l ong b y one foot wi de for the
air-water mi xt ure. A hor i zont al l y oriented c y l i ndr i c al aerator was designed and i nst al l ed
at the base o f the unit. I f this pr oved successful the intention was to adopt the desi gn
c onc ept and b ui l d airlift pump units that c oul d span across the entire wi dt h o f rectangular
b ox c ul vert. Large- sc al e drawi ngs o f the sl ot-c onfigured airlift pump system c an be found
i n Appendi x 2. Fi gur e 11 shows the sl ot-c onfigured airlift pump i n operation.
Fi gure 11 - Slot-Configured Ai r l i ft P ump i n Operation
51
This system was tested under adverse conditions with very little upstream depth
available. Nevertheless, it identified an unanticipated and major problem with the "slot"
design. Water tended to "slosh" from side to side in the pump unit body, but very little
was effectively lifted. It was observed that when the water was high at one end of the slot
it gave a large enough back pressure to the orifices there to create an increased flow of air
from the orifices at the other end. In this way the air tended to escape from the system.
When the water sloshed back to the other end of the system the air escaped from the first
end. Thus, although a considerable amount of spray was created very little water was
pumped. The immediate failure of this design showed convincingly that large capacity
airlift pump systems must be designed to prevent this sloshing behaviour, effectively a
one-dimensional recirculation effect analogous to that which had been observed in the
cylindrical units. This test confirmed the findings of the first test, namely that circulation
could easily develop within the pump riser pipes, greatly reducing pump efficiency.
The third experimental phase was successful in pointing the way forward. The design
concept was revised to minimize the two most severe problems encountered at the Gilbert
Road site, namely circulation within the riser pipes and leakage between the test
chambers. The large ten and twelve-inch diameter pump barrels were replaced with eight,
six, four, and three-inch units. This was considered a useful means of reducing circulation
and turned out to be very effective. It was also decided to test the effect of inclining the
pump tubes. Wallis (1969) asserts that inclination up to approximately 40 degrees from
the vertical does not adversely affect bubble velocity, and inclining the pump riser tubes
in this way promised reduced construction costs by requiring fewer pipe fittings.
52
This fourth phase of the experimental program was carried out over the course of several
months at the Ri chmond Publ ic Works Yard. Plastic pipes were set up i n a tank with a
metered water supply and metered compressed air supplied from a single jet at the bottom
of each pump' s riser tube. Since aerator configuration had little discernible effect i n
previous phases of the project, aerator designs were not tested i n this phase. The pipes
were incl ined from zero to thirty degrees from the vertical. The fl ow of air and water
were set and when the water l evel stabilized i n the tank, the stable depth of water i n the
tank was read. Figure 12 shows the experimental setup for this phase of the project.
Various combinations of air and water fl ow and pipe diameters were tested i n an
"evol utionary" manner - starting with one pipe at one upstream water l evel and varying
downstream water levels for a given air fl ow rate. The tests showed that for a given fl ow
of air the incl ination away from the vertical of the pump riser pipes wi thi n the range of
zero to thirty degrees did not seem to affect the water fl ow rates. The overal l results stil l
evidenced some scatter but this was at least partly due to unavoidable variations i n the
position of the air jets i n the bottom of each pipe riser tube and other minor factors such
as the resolution of the meters used, etc.
Data from this experimental phase was far more consistent than that from the third
experimental setup since upstream and downstream water levels coul d be precisely
control l ed and there was no leakage from the sealed tanks. Tabl e 7 shows sample
experimental data from this series of experimental tests.
53
FI GURE 12 - Richmond Public Works Experimental Setup
f
m
+->
CO
(D
H
OH > H
H CO
a
o
3
4H
<
CO
P H
O
P H
T 3
O
I
P H
o
a
IT)
CN
CN
CN
X
CN
m
GO
S-H
*->
"5b
PH
1
*S ' 5
54
T A B L E 7 - R i c h mo n d P u b l i c W o r k s S a mpl e E xper i men t a l D a t a
o
1
3
.a
s
I
I 5?
S i
PO ^ ?5 ^
s s s s
d o o o
n n w w P J
d o d d d d
CD CD CD O O)
S N N N N
<r *r v
d 6 o' o* d
o o o o o o
CD <D (O <>
V V f ^ V
d d d d d
w u> w w i n trj i n i n vn i n jp jn ip jp i n i SP IP IP
N N N K K f*. K p l* N . K J *. |s. fx. f N- r N - h-
flCtcooqp co co <o co" co co iri i ri i n i ri tn s s N s N N
N N N W
t r t IA n tf) i o i n h~ f
I O in i d in ifl i n in in ui tn in m w to w in j w w
13 <3 13 pi S$ <8 i8 !3 5$ $5 S
V "t V ^
<o u> m to <o <x> m i n m N w n r UJ u> >q
J: e
2 to!
CQ m o cn p s m M i n ;* ro
rs.^r*. ^- > o P - r*
3 8 8 8 3 8 8 8 8 8 3 8 8 8 S 8 3 8 8 8 S 8 8 8 8 8 ? S ? 88
1 l d o* d o o d d d o d d d d d d d d d d ddd d d o d o d o
i i
2 Q J
pj n n n v n ^ < v 10 n o o o <o co to 03 to co
1
At this poi nt the head loss relationships desc ribed i n equation s (35) and (36) i n chapter 3
were created and used to characterize the system behaviour. A fi nal set o f tests were
made wi t h a bundl e o f ni ne, four- i nc h diameter pl ast i c pipes at an i nc l i nat i on o f 0 to 40
degrees from the vert i c al . Compr essed air was suppl i ed t hrough a mani fol d o f pi pes wi t h
a one i nc h jet at the centre o f eac h pump riser pi pe.
Al t hough some prac t i c al probl ems remai ned, the performance o f this experi ment al setup
c onfi r med that the rel ationships devel oped i n the model for turbulent mi xi ng for med a
reasonable basis for airl ift pump desi gn i n the c hurn turbulent regi me. Thi s suc c essful
phase o f the pr ogr am resulted i n a rel iabl e data set, pr ovi di ng the basis for c al i b r at i on o f
the theoretic al model and poi nt ed the way towards a vi ab l e airlift pump desi gn for the
situation i n R i c hmond.
3.3 - Results of the Experi ment al Pr ogr am
The lessons fr om the first t wo laboratory-based phases o f the experi ment al pr ogr am
i ndi c at ed the vi ab i l i t y o f the concept o f l ow- l i ft hi gh- fl ow l ow- submergenc e airlift pumps
for urban st orm drainage. The Gi l b er t R oa d prototype syst em suggested several prac t i c al
c onsiderations for ful l -sc al e appl ic ations and l ed the way to the fi nal experi ment al phase.
The fi nal phase pr oduc ed the rel iabl e data set used to calibrate the head loss rel ationships
devel oped i n the t hi rd theoretical model . Thi s experi ment al pr ogr am also l ed to a vi ab l e
prac t i c al desi gn. The results were also used to ver i fy the theoretical model devel oped to
expl ai n l ow- l i ft , l ow- submergenc e, hi gh- fl ow airlift pump behavi our i n these scenarios,
whi c h i n t urn l ed to a vi ab l e prac t i c al engineering desi gn procedure.
56
CHAPTE R 4
4.1 - Airlift Pump Model for Fixed Bubble Slip Velocities
Refer again to Figure 1, reproduced here for convenience:
' j j Qwat er
In static conditions, the pressures inside and outside of the airl ift pump tube are equal at
the point of aeration.
H
sub =
H
t e al '
D e n S (4)
Under dynamic conditions air bubbles are rising through the water col umn within the
airlift pump tube and a driving head must be added to the system as described i n (4) to
maintain the pumping action:
57
H
sub =
H
total +Hdrive
H
drive =H
xllh
-H
lolal
-Dens
which can be rearranged to form
(5)
For equil ibrium, the driving head must be equal to the losses i n the system.
H
drive ~
H
hss (6)
Fl ui d fl ow losses are commonl y expressed i n the form of:
V
2
headloss = K (7)
2g
so for the case of entrance, pipe and exit losses i n the airlift pump system, and assuming
that the entrance, pipe and exit losses due to viscosity and fl uid friction due to air wi l l be
much less than those due to water:
v
2
V
2
V
2
TT T/' water . js water , jy- water / o \
^ loss ~ ^entrance '
+
^ pipe 2g ' 2g
or for the case i n which loss factors for various entrance, pipe and exit geometries are not
expl icitl y considered separately, a combined loss factor can be used:
Kate'
H
loss =
K
,o,al
-
(
9
)
2g
58
Comb i ni ng (3) and (2) and rearranging, get
H
suh
-H
lolar
Dens = K,
vi
water
total
2g
(10)
The goal i s to determine the c omb i ned loss factors K
tota
i for representative geometries so
that airl ift pump performance c an be model ed si mpl y b y (7). We want to determine the
loss factor, so rearrange:
Equat i on (11) provi des the total loss factor for the airlift pump, dependent on the
submergenc e and total pump length, as wel l as the density o f the air-water mi xt ur e and
the vel oc i t y o f the water phase.
To sol ve (11) for the total loss factor we need the density o f the air-water mi xt ur e i n the
airlift pump tube and the vel oc i t y o f the water phase i n the airlift pump tube. Consi der i ng
a representative c ross-sec tion o f the air-water mi xt ure fl owi ng i n the airlift pump tube,
the relative density o f the air-water mi xt ure i n the airlift pump tube is gi ven by:
K

a
, =
2
8
H
suh ~
H

vi
water
tolal
Dens
(11)
Dens =
water
(12)
Area
59
To sol ve (11), the area o f the pump cross sec tion oc c upi ed b y the water phase is al so
needed. To obt ai n the area oc c upi ed b y the water phase, the vel oc i t y o f the water phase
and the vel oc i t y and area oc c upi ed b y the air phase are required.
To get the vel oc i t y o f the water fraction o f the mi xt ure, c onsi der c ont i nui t y o f the vol ume
flow rates o f the mi xt ur e and o f eac h phase i n the airlift pump tube:
Qmix ^
r
mix ^mix (13)
Qair ^air ^air 0 4)
Q water ^water ^water 0^)
also the vol ume flow rate o f the mi xt ure is c omposed o f the sum o f the vol ume f l ow rates
o f the air and water phases:
Qmix ^mix ^mix water ^water 0^)
and the total c ross-sec tional area i n the airlift pump tube is si mpl y c omposed o f the sum
o f the areas oc c upi ed b y the air and water phases:
A A + A
mix water air V /
Ni c k l i n (1962) suggests that i n the sl ug flow regi me for st i l l water and where Tay l or
bubbl e diameter and pi pe diameter are ver y si mi l ar,
60
(2)
If Taylor bubbles were to be found in the 3 to 12-inch diameter airlift pump riser tubes in
this study, Nicklin (1962)'s equation (2) would suggest their rise velocities to be within
the 1.1 to 2 foot per second range. Classical observations of bubble rise speeds outside
the slug flow regime (i.e.: smaller bubbles not constrained directly by pipe boundaries)
suggest the terminal velocity of a single bubble is relatively constant between 25 to 45
cm/s over a broad range of bubble diameters, as shown in Figure 12, here reproduced
from Wallis (1969):
FI GURE 13 - Bubble Rise Velocities in Still Water, from Wallis (1969)
/ I 1 1 I I I I I I I I !- I ' I L l l 'I l l I I I I I I I I I I I I I I
O.OI 0.02 0.04 0.04,0.01,0.1 O.Z O.H OA 0.&/.O 2.0 4.0
Equivalent Radius in centimeters Ry
61
Tai t el & al. (1980) suggest that above a c r i t i c al diameter (approxi mat el y 1.5 mm) air
bubbl es tend to deform and adopt an erratic path. Thus the sl i ght l y sl ower effective
b ub b l e rise speeds o f the unc onstrained non- Tay l or bubbles may be expl ai ned b y the
i rregul ari t y o f the smal l er bubbl es' rise trajectories c ompared wi t h the c onstrained-
ver t i c al rise trajectories o f the Tay l or bubbles i n the sl ug fl ow regi me.
Us i n g the c onc ept o f a constant t ermi nal rise vel oc i t y for bubbl es i n st i l l water, we
introduc e the relative vel oc i t y o f the air phase to the water phase i n the airlift pump tube,
water rel
(18)
substituting (18) into (14) results i n:
Qair = (Kwater
(19)
substituting (17) into (19) results i n:
Qair = 'water +Kel X
Are
" 4 water )
(20)
rearranging (15) and substituting into (20):
(21)
62
Equat i on (21) provi des a functional rel ationship between the measured fl ow rates o f the
air and water phases Q
air
and Q
waler
, the known c ross-sec tional area o f the airl ift pump
tube Area, the known relative vel oc i t y o f the air phase to the water phase V
re
i, and the
unknown vel oc i t y V
waler
o f the water phase. Therefore, under these assumptions the
vel oc i t y o f the water phase i n the airlift pump tube c an be c al c ul at ed for any c omb i nat i on
o f the measured val ues. Thi s water phase vel oc i t y c an then be used to sol ve (11) for the
desi red over al l loss factor K
tota
i.
However , to sol ve (11) we also need the density o f the air-water mi xt ur e i n the ai rl i ft
pump tube. Rear r angi ng (15) and substituting into (12):
Dens = (22)
V
water '
A r e a
substituting (22) into (11) and rearranging, get:
v
water
(
H 0 ^
_ *
x
total XL. water
V *
v
water'Ar;
(23)
Equat i on (23) gi ves the pump loss factor as a func t i on o f the water phase vel oc i t y ,
diameter, total l engt h and submergence o f the pump tube, vol ume flow rate o f water and
vel oc i t y o f the water phase i n the airlift pump tube. The vel oc i t y o f the water phase c an
be det ermi ned fr om (21) and thus the pump loss factor determined for a vari et y o f flow
and submergenc e c ondi t i ons.
63
I n this wa y it was hoped the characteristic behavi our o f the airlift pump syst em c oul d be
det ermi ned fr om the pump loss coefficient, enabl i ng a clear understanding o f the pump
syst em operat i on and as wel l creating a si mpl e desi gn procedure.
Whe n the experi ment al data was c ompared wi t h this model it bec ame apparent that losses
found i n the pump units i n this study were not accurately predic ted. The summar y val ues
i n c ol umn 13 o f Tab l e 6 correlate the measured system losses wi t h the water phase
vel oc i t y . The wi de spread i n the deri ved loss factors as wel l as the large c oeffic ients o f
var i at i on indic ate c l ear l y that this model is not appl i c abl e to the pumps i n this study.
Furt her observations o f the experi ment al units i n operation and mor e research suggested
that the assumpt i on o f a t ermi nal bubbl e vel oc i t y as expl ai ned b y Fi gur e 13 was not
appl i c abl e i n this case.
The assumpt i on o f a t ermi nal bubbl e speed relative to st i l l water makes this model
possi b l y mor e suitable for l ow voi d -ratio fl ows and l ow mi xt ure fl ow vel oc i t i es, suc h as
mi ght be encountered i n a l ong, large diameter riser suc h as used i n l ake aeration or
destratification or harbour de- i c i ng. Al t hough this model for airlift pump performanc e is
not hel pful i n the case o f l ow- l i ft , l ow- head hi gh fl ow pumps suc h as are c onsi dered here
it does have pr omi se and may prove useful i n anal ysis o f cases suc h as those ment i oned
above.
64
4.2 - Airlift Pump Model for Variable Bubble Slip Velocities
Wa l l i s (1968)' s Fi gur e (9. 5) i n the sec tion c onc er ni ng c hur ni ng fl ow presents the mi xt ur e
and gas phase mass fl ux rates i n terms o f the mi xt ure mass fl ux rate and a gas phase "dri ft
flux" rate relative to the mixture flux. Wa l l i s ' figure is reproduc ed as Fi gur e 13 here.
F I GURE 14 - Mixture and Gas Flux Rates, Wallis (1969
OVERALL FLUX Jt m/sec
65
It indic ates the rel ationship between the mass fl ow rate o f the gas and the mass f l ow rate
o f the mi xt ur e, expressed as a mass fl ux rate per unit area for the mi xt ur e and a rel ative,
or "dr i ft " mass fl ux rate per unit area for the gas phase. I nspec tion o f Fi gur e 13 suggests
that the gas phase average vel oc i t y is dependent on the flux rate o f the mi xt ur e.
Consi der a case suc h as ours i n whi c h the densities o f the gas and l i qui d phases are
known fi xed quantities, the pump geometry is known and the density o f the gas phase is
negl i gi b l e c ompar ed to that o f bot h the l i qui d phase and that o f the mi xt ur e. I n suc h a
case Fi gur e 9 suggests that the drift fl ux rate and mi xt ure flux rate are dependent on the
mi x ratio and c omponent phase vel oc i t i es onl y . So for any gi ven mi x ratio the straight-
l i ne rel at i onshi p i n the ratio o f the gas drift flux and mi xt ure fl ux rate shoul d be equal l y
representative o f the gas and mi xt ure fraction vel oc i t i es. I n that case for known fixed
densities o f l i qui d (water) and gas (air) phases, and for a known voi d frac tion, the
vel oc i t y o f the gas phase o f the mi xt ure i n a churn-turbulent two-phase flow depends not
on the vel oc i t y o f the water phase as suggested b y Fi gur e 8 and as found i n l ow voi d-
frac tion st i l l water and b ub b l y fl ow, but rather depends on the vel oc i t y o f the mi xt ur e
instead.
Wa l l i s ' equat i on 9. 36 suggests a different for m o f this rel ationship. That is expec ted since
hi s flow anal ysi s was moment um- based and di d not require the relative vel oc i t i es o f the
c omponent mi xt ur e phases.
66
Nevert hel ess, the c onc l usi on is powerful - namel y that i n cases where the gas densi t y is
negl i gi b l y l o w i n c ompar i son wi t h the mi xt ure density the gas phase vel oc i t y is greater
than, and rises l i near l y wi t h the mi xt ure vel oc i t y . Thi s provi des a val uabl e c omponent
mi ssi ng so far i n the anal ysi s o f these short-lift systems. It is reassuring to note that D e
Cac har d & Del hay e (1995) also found a si mi l ar result for mi xt ure and gas phase
vel oc i t i es up to approxi mat el y 6 m/s i n smal l diameter, l ong lift pump risers.
S o, expressi ng the air phase vel oc i t y as a linear func t i on o f the mi xt ur e vel oc i t y :
Equat i on (25) model s Fi gur e 13 to r emar kab l y good agreement i n units o f feet per
sec ond. Thi s equat i on fit corresponds wi t h the bubbl e t ermi nal vel oc i t y o f appr oxi mat el y
30 c m/ s, whi c h is approxi mat el y equal to 1 foot per sec ond as i n Fi gur e 12 and suggested
b y Wa l l i s for st i l l water and used i n the first model . The \ 2V
mix
t erm is also fami l i ar
since it represents the ratio o f the centreline vel oc i t y to the average vel oc i t y i n the ful l y
devel oped turbulent flow field wi t hi n a c l osed pi pe.
(24)
and fitting the l inear rel ationship i n (25) to Wa l l i s ' data i n Fi gur e 13 suggests:
(25)
67
The for m and val ues o f (25) as interpreted here fr om Wa l l i s ' data are ver y si mi l ar to
those suggested b y Ni c k l i n (1962) before hi s wor k on airlift pumps. Ni c k l i n suggested for
the vel oc i t y o f a sl ug bubbl e r i si ng i n a two-phase mi xt ure at Rey nol d' s numbers under
I n consistent units for a representative pump riser tube o f si x- i nc h internal diameter,
equat i on (26) bec omes
Ni c k l i n (1962) qual ifies (28) above as b ei ng accurate for Rey nol d' s numbers b el ow 8000
and approxi mat e for Rey nol d' s numbers over 8000.
Furt hermore, Fernandes, Semi at & Duc kl er (1983) independentl y suggest that the Tay l or
bubbl e rise vel oc i t y i n larger diameter pipes than those studied b y Tai t el et al . (1980) is
gi ven b y
Equat i ons (27) and (28) are ver y si mi l ar to one another and suggest val ues for the air
phase vel oc i t y for sl ug fl ow just sl i ght l y greater than suggested b y Wa l l i s ' experiments
8000:
V
air
=\.2V
mix
+0.35jg-Diam (26)
(27)
V
air
= \.2W
mix
+0. 35 Jg-Diam (28)
68
for b ub b l y f l ow as gi ven i n (25). These findings i nspi re c onfidenc e i n this study that the
, for m and val ues i n equation (25) are rel iabl e.
Therefore, substituting (25) into (14):
a,>=(i.o+i.2F;,
fc
R (29)
Subst i t ut i ng (13) into (25):
1.0 + 1 . 2 ^
^mix J
(30)
Rear r angi ng (30) to sol ve for the cross sec tional area oc c upi ed b y the air phase,
A - Qair
1.0 + 1 . 2 ^
V ^mix J
(31)
substituting
:
(13) into (28):
(
V,, =1. 0 + 1.2
'mix
Area.
(32)
69
also substituting (32) into (14):
f
1.0 + 1.2
0,,
Area
(33)
and substituting (33) into (15) and usi ng (16):
V , =Q
water w
Area - -
1.0 + 1.2
Q water Qair
Area
(34)
J)
Equat i on (34) wi l l then gi ve the vel oc i t y o f the water phase i n the airlift pump tube as a
func t i on o f the measured fl ow rates o f air and water, and the cross sec t i onal area o f the
airl ift pump tube for churn-turbulent fl ows, assumi ng the air and water phase fl ow
vel oc i t i es are ac c uratel y represented b y equation (22) whi c h was der i ved fr om a fi xed-
densities and mi x ratios anal ysis o f Wa l l i s (1968) data i n Fi gur e 14 and bol stered b y
Ni c k l i n (1962).
H a vi ng det ermi ned the vel oc i t y o f the water phase i n the airlift pump tube fr om equat i on
(34), and knowi ng the water phase vol ume fl ow rate and pump geometry, the val ues c an
be used to fi nd the val ue for the pump loss coefficient as det ermi ned b y equat i on (23):
K
2g
total
H
suh ~
H
water \
total water
V
water '
A r e a
J
(23)
70
Thi s approac h hol ds more promi se than the first for i mpr oved and more rel i abl e results.
The experi ment al data was reanal yzed. However , even wi t h this more rel i abl e approac h
for c al c ul at i ng the vel oc i t y o f the water phase and despite good evi denc e to support
equat i on (29), the head losses were st i l l found not to be proport i onal to the square o f the
water phase vel oc i t y .
Despi t e the fact that this model cannot be used to expl ai n the behavi our o f the l ow-
submergenc e, l ow- l i ft , hi gh- fl ow pump units i n this study it does hol d pr omi se for use i n
mi d- vel oc i t y b ub b l y fl ow pump units. I n suc h units head losses are pr i mar i l y due to pi pe
fr i c t i on as suggested b y Wa r d (1924) and this model may hel p pr ovi de a si mpl e anal ysi s
t ool for that class o f airl ift pump systems.
71
4.3 - Airlift Pump Model for Turbulent Mixing
G i ve n the i nab i l i t y o f the sec ond model to accurately predict the head losses usi ng the
i mpr oved met hod for c al c ul at i ng the water phase vel oc i t y , a new approac h is c l ear l y
necessary. E vi dent l y the assumpt i on that the losses are pr i mar i l y due to pi pe fri c t i on,
entrance and exi t losses and are dependent on the vel oc i t y o f the water phase must be
reexami ned.
The f l ow o f the mi xt ur e i n the airlift pump tube is ver y turbulent wi t h si gni fi c ant vi sual
evi denc e o f c hurni ng. and rec i rc ul at i on, so the assumpt i on that the i nfl uenc e o f the air
phase is negl i gi b l e may be suspect. Wa l l i s (1968) suggests i n passi ng that the majori t y o f
energy dissipated i n the pi pe fl ow o f c hur ni ng two-phase mi xt ures results fr om internal
losses rather than pipe-fric tion-related causes. Cl ar k & Dab ol t (1986) also argue that the
fri c t i onal head losses are a second-order effect wi t hi n prac t i c al lengths for non- sl ug fl ows
al t hough they do not quantify what the fri c t i onal head losses are.
Further research and passi ng suggestions i n several other references pr ovi de some c l ues
to the mec hani sm o f these losses. Wa l l i s (1968) mentions that i n c hur ni ng f l ow the
c haot i c movement o f water i n the fl ow mi xt ure causes the most energy l oss, and
furthermore, that i n the majority o f prac t i c al cases b ub b l y fl ow never bec omes ful l y
devel oped and entrance effects dominate the regi on before sl ug fl ow devel ops. Wa r d
(1924) ment i ons that short pumps have losses not important i n l ong pumps. Mor r i s on &
al . (1987) suggest that c hur ni ng fl ow is i n fact a transition regi me usual l y exi st i ng fr om
15 to 35 pi pe diameters away fr om the aeration point, before si gni fi c ant enough bubbl e
72
ac c ret i on c an oc c ur to create sl ug fl ow. DeCac har d & Del hay e (1995) suggest that the
l engt h effects fr om the devel opment al r egi on o f c hurn flow l eadi ng to st abi l i zed sl ug
flow may create hi gher than predic ted head losses up to lengths several hundred t i mes the
pi pe diameter away fr om the entrance. Thus a ful l y st abi l i zed sl ug flow regi me may not
devel op wi t hi n a l ength up to even t wo hundred times the pi pe diameter. Tai t el & al
(1980) quant i fi ed a mi n i mum l ength for the turbulent entrance transition zone as
Lauras =
4 0
-
6
-
D i a m
f
v. ^
+ 0. 22
yjg Diam
(35)
It oc c urs that the short pump losses ment i oned b y Wa r d (1924) must be due to the
entrance and t ransi t i on zone turbulence. The pumps i n this study are c onc l usi vel y "short "
- substantial l y shorter than 15 to 35 to several hundred diameters l ong, and subst ant i al l y
shorter than the entrance lengths suggested b y Tai t al & al . (1980) b y equat i on (35) above.
Therefore it is reasonable to assume that the mi xt ure i n the entire pump riser tubes is
exc l usi vel y exper i enc i ng the turbulent transition zone fl ow regi me. I n that case, the losses
i n short airlift pumps suc h as those i n this study must be pr i mar i l y turbulent i n nature -
and not pi pe fr i c t i on losses dependent on the water phase vel oc i t y as was assumed i n the
first t wo model s, and as c ommonl y assumed i n the pr i mar i l y sl ug- fl ow model s devel oped,
to date.
I n this case the c hal l enge then bec omes how to quantify the mi xt ur e turbulence and relate
the pump head losses to that turbulence.
73
It was observed i n experi ment al trials that the air-water mi xt ure bec ame i nc r easi ngl y
turbulent wi t h i nc reasi ng mi xt ure vel oc i t i es, and that ver y hi gh gas phase vel oc i t i es at
hi gh voi d fractions resulted i n large losses and ver y little l i qui d fl ow. These observations
suggest that mi xt ur e vel oc i t y i n the c hur ni ng regi me is a good i ndi c at or o f mi xt ur e
turbulence and hence o f losses i n these short pump units.
Furt her research di sc over ed I shi i and Z ub er ' s (1979) c l ai m that i n turbulent fl ow regi mes
the bubbl es i nfl uenc e the surroundi ng fl ui d and also other bubbl es, and that thus bubbl es
c an be entrained i n eac h others' wakes, and therefore the losses i n suc h fl ows shoul d be
c onsi dered relative to the mi xt ure vel oc i t y rather than that o f the l i qui d phase. Wa l l i s
(1968) also suggests a si mi l ar general for m.
We therefore propose a func tional for m for the turbulent head losses i n these short airl ift
pumps, dependent on the mi xt ure density and turbulence, and represented b y the densi t y
and vel oc i t y o f the entire mi xt ure rather than the vel oc i t y o f the water phase alone:
H
lms
=d-Dens-V;
iix
(36)
The t uni ng parameters d and e wi l l be experi ment al l y determined fr om the research
pr ogr am data.
The for m o f equat i on (36) wi l l effec tivel y parameterize the head losses but does not
pr omi se a great advanc ement i n terms o f the details o f the head loss mec hani sm. Thi s i s
74
not ent i rel y surpri si ng since D e Cac har d & Cal hay e (1995) expl ai n that a model for wa l l
fr i c t i on i n c hur ni ng flow is not yet avai l abl e, and that the c haotic mot i on i n c hur ni ng flow
makes empi r i c al c onsiderations for wa l l fri c t i on losses a necessity. G o van et al (1991)
agree, suggesting that creating a realistic model for c hurn flow mec hani c s is "par t i c ul ar l y
c hal l engi ng".
D e Cac har d & Cal hay e (1995) r ec ommend a for mul at i on si mi l ar to equat i on (36) for l ong
slender pumps, based on the l i qui d phase vel oc i t y . F ol l owi ng their suggest i on and usi ng
their equat i on [48] and Bl as i us ' for mul a for fri c t i onal losses i n the boundary l ayer as
gi ven i n their equat i on [11] their sol ut i on proposes a fri c t i on loss t erm for c hur n flow as:
- 0. 316- Dens
l i q u i d
_.
2S 2
Hhss
=
Hlolal IDiam liquid ^liquid (37)
for the Rey nol d' s Numb er R e based on the vel oc i t y o f the l i qui d phase Vn
qui
d. The for m
o f equat i on (37) is reassuri ngl y si mi l ar to the for m o f equation (36), arri ved at
independentl y. The pr i mar y difference between D e Cac har d & Del hay e' s for m and that
suggested i n this study is that their expressi on is calibrated for smal l diameter t al l risers
and uses the l i qui d phase vel oc i t y as was suggested i n the sec ond model above, whereas
equat i on (36) rel ies on the mi xt ure vel oc i t y as an i ndi c at or o f turbulenc e.
To use (36), we substitute (6) into (5):
H
loss
=H
suh
-H
lolar
Dens (38)
75
and substitute (36) into (38):
d Dens V
e
mix
= H
suh
- H
lolal
Dens (39)
now substituting (13) into (38):
d
(^water ^ l y e
V Area j
' mix
n
sub
n
total
(A \
\Area j
(40)
Thi s t hi r d model for airlift pumps continues to make use o f the relative vel oc i t y o f the air
phase to the vel oc i t y o f the mi xt ure as gi ven b y the experi ment al l y det ermi ned equat i on
(26) and suggested fr om Wa l l i s (1968)' s data and reflected i n Fi gur e 13.
S o, usi ng (29) for the water phase cross sec tional area, and substituting into (40):
Area-
Qa
1 + 1.2
Qm
Area J
V = M M
' mix
n
sub
n
total
Area-
Qa,
1 + 1.2
Qm
Area
(41)
F i nal l y , b y substituting (11) into (41), get:
Area
a,
1 + 1.2
Qm
Area
Qair Qwater
Area
Y
= H,
h
-H,
sub
1 1
total
Area - -
Qm
1 + 1.2
V Area
(42)
76
B y usi ng the measured water and air phase fl ow rates and pump geometry, equat i on (42)
al l ows the t uni ng parameters d and e to be determined fr om the experi ment al results.
The summari es o f c ol umns 10 and 11 i n Tab l e 4 and c ol umns 11 and 12 i n Tab l e 6 show
good c orrel at i on between the head losses i n the pumps and the mi xt ur e vel oc i t y as a
measure o f turbulence. Thi s c orrel at i on was found throughout the experi ment al results,
albeit mor e c onvi nc i ngl y fr om the last phase o f the pr ogr am i n whi c h results were mor e
rel i abl e than those previ ous due to factors already discussed.
Fi gur e 15 shows a summar y o f the experi ment al data and model predi c t i ons. The l i ne o f
best fit for the experi ment al data leads to the fol l owi ng rel ationship for the head loss as a
func t i on o f the mi xt ur e fl ow vel oc i t y :
c onst ruc t i ng a sl i ght l y more c onservative c urve fit fr om the data l ead to the fol l owi ng
rel at i onshi p for the head loss as a func t i on o f the mi xt ure fl ow vel oc i t y ,
Equat i on (44) c oul d be mor e suitable for desi gn since it predicts a sl i ght l y hi gher head
loss than the l i ne o f best fit and woul d thus be a c onservative estimate for pump c apac i t y.
(43)
H
loxs
=0.62-dens-V
n
(44)
77
Figure 15 - Comparison of Experimental and Calculated Performance: Log(head
loss/density) vs. Log(Mixture velocity)
Thi s model for airlift pump performance performs wel l i n predi c t i ng the performanc e o f
the l ow- submergenc e, l ow- l i ft , hi gh- fl ow units investigated i n this project. It also
pr ovi des the basis for a si mpl e approac h to eval uat i ng the behavi our o f these units and
leads to a reasonabl y direct and prac t i c al desi gn approach.
4.4 - Summarizing the three models
The first model desc ri bed i n 4.1 - Airlift Pump Model for Fixed Bubble Slip Velocities
relies on the assumpt i on o f a rel at i vel y constant bubbl e rise speed, and fri c t i onal pump
losses governed b y the vel oc i t y o f the water phase i n the air-water mi xt ur e. I nvestigations
o f the experi ment al results and further research indic ate that this assumpt i on is not va l i d
i n the c hur n fl ow regi me experi enc ed b y the pump units i n this study. Thi s model does
have pr omi se i n appl i c at i ons where the constant bubbl e rise speed is supported, and ma y
find use i n large diameter systems suc h as are used for lake destratification and harbour
de- i c i ng.
The sec ond model desc ri bed i n 4.2 -Airlift Pump Model for Variable Bubble Slip
Velocities also assumes losses governed b y the vel oc i t y o f the water phase i n the air-
water mi xt ur e. It features a refined estimate for the mean bubbl e vel oc i t y as a func t i on o f
the mi xt ur e vel oc i t y , a refinement based on the experi ment al wor k o f several previ ous
researchers. Thi s model does not accurately describe the behavi our o f the l ow- l i ft , hi gh-
fl ow, l ow- submergenc e pumps i n this study but does hol d promi se for use i n mor e
energetic b ub b l y fl ow regi me pumps suc h as those used i n aquaculture and wastewater
treatment appl i c at i ons.
79
The t hi r d model as desc ri bed i n 4.3- Airlift Pump Model for Turbulent Mixing is
spec i fi c to the c hur n flow regi me. The refinement introduc ed i n the sec ond model is
retained, but a new for mul at i on for the nature o f the head losses is ut i l i zed. Head losses i n
the t hi r d model are assumed to be proport i onal to the turbulent mot i on o f the air and
water phases i n the mi xt ure. Mi xt ur e vel oc i t y is found to be a good i ndi c at or for mi xt ur e
turbulence and is thus used as a basis for c al c ul at i ng the turbulent head losses. Thi s
model predicts the behavi our o f the pumps i n this study wi t h good ac c urac y and forms
the basis o f the si mpl e desi gn procedure presented i n Chapter 5.
None o f the procedures suggested i n the open literature are prac t i c al for the engineer
wi s hi ng to desi gn a l ow- submergenc e, l ow- l i ft , hi gh- fl ow airlift pump syst em. Wa r d' s
(1924) approac h to desi gn o f ver y l ong airlift pumps b y c urve mat c hi ng i nc l udes no data
for short l engt h pumps and hi gh voi d fractions. Ni c kl i n ' s (1963) technique for desi gn o f
airlift pumps i n sl ug flow, and al l o f the suggested refinements to Ni c kl i n ' s wor k
suggested b y subsequent researchers do not describe turbulent losses i n a c hur ni ng
system. Tr amb a' s (1982) and Nenes & al ' s (1995) mul t i - c el l ed si mul at i on- based
numer i c al approaches for deep- wel l airlift pump anal ysis relies on di vi di ng the pump
pi pe riser b ody into di fferi ng c ont i guous sections, each wi t h i ndi vi dual f l ow
characteristics, a process not feasible for the short pumps desc ribed here.
However , 4.3 - Airlift Pump Model for Turbulent Mixing desc ri bed i n the previ ous
chapter, pr ovi des the mi ssi ng basis for a si mpl e and prac t i c al desi gn proc edure a l ow-
submergenc e, l ow- l i ft hi gh- fl ow airlift pump system.
80
C HA PT E R 5
5.1 - A Preliminary Design Procedure for Low-lift, Low-Submergence Airlift Pumps
in the Churn Flow Regime
Thi s proc edure al l ows a designer to qui c kl y c ompl et e the pr el i mi nar y c al c ul at i ons for an
airl ift pump i n a l ow- head, hi gh- fl ow, l ow- submergenc e appl i c at i on for prac t i c al pump
diameters i n the approxi mat el y 3 i nc h to 12 i nc h range. Si nc e airlift pumps are
i nexpensi ve to construct, a prototype unit may then be bui l t and the performanc e ver i fi ed.
Bec ause this desi gn approac h is si mpl e and based on the fri c t i on and vel oc i t y c orrel ations
devel oped i n this research pr ogr am it shoul d be used wi t h care i n cases o f muc h hi gher
lift and muc h deeper submergence. I n those cases the pi pe- fl ui d fri c t i on losses wi l l b egi n
to pl ay a larger part i n over al l system behavi our as the bubbles i n the c hurn fl ow b egi n to
c oal esc e into Tay l or bubbl es and arrange themselves into a sl ug f l ow pattern. The desi gn
proc edure o f Cl ar k & Dab ol t (1986) is r ec ommended for use i n suc h cases.
Thi s desi gn proc edure is used to predict the vol ume fl ow rate o f water expec ted fr om a
l ow- head, hi gh- fl ow, l ow- submergenc e airlift pump operating i n the c hur n f l ow regi me.
The desi gn parameters required are:
Qair = the intended vol ume fl ow rate o f air, i n c ub i c feet per sec ond
Qwater = the intended water fl ow rate, i n c ub i c feet per sec ond
81
Upst r = the upstream water l evel , i n feet
Dnst rdes = the desired downst ream water l evel , i n feet
The desi gn proc edure wi l l be illustrated b y a si mpl e exampl e. I n this exampl e an engineer
desires to aerate and pump 10 c ub i c feet per sec ond of water over a 1.5 foot lift usi ng a
si ngl e or mul t i pl e- pi pe airlift pump system wi t h 8 i nc h diameter riser tubes. Thi s
pumpi ng syst em is set i n a smal l concrete drainage c hannel 8 feet wi de b y 5 feet deep.
F i gur e 16 shows the proposed l ayout o f the system.
FI GURE 16 - Simple Design Example Layout
A Simple 10-Step Design Process:
1. Thi s desi gn appears to require several pump units to ac c ompl i sh the requi red flow
rate and lift. I n suc h a case, assume an air flow rate and water flow rate for a
Qwater
82
single pipe. Wi t h no other i nformat i on avai l abl e, 50% o f the air fl ow rate i s found
to be a reasonable starting estimate o f the water fl ow rate. As s umi ng 2. 5 cfs o f air,
and 1.25 cfs o f water:
a, > =2. 5c fs, e
w
^ = 1. 25c f s ( D I )
2. F or l ow i nsert i on depths air c an be c onsi dered i nc ompressi bl e, so c alc ulate the
vol ume f l ow rate o f the mi xt ure b y addi ng the vol ume fl ow rates o f the air and
water:
& * - Qair + QWa,er = (2. 5) + (l . 25) = 3. 75 cfs (D2)
3. Det er mi ne the mi xt ure vel oc i t y b y di vi di ng the mi xt ure fl ow rate b y the c ross-
sec t i onal area o f the riser pi pe:
Qmix (3. 75 cfs)
0. 785 (0. 75 ft )
:
= 10.7fps (D3)
4. Det er mi ne the vel oc i t y o f the air phase i n the pump riser pi pe fr om equat i on (25):
V
air
= 1 + 1.2-V
mix
= 1 +1. 2 (10. 7) = 13. 9 fps (D4)
83
5. Det er mi ne the sec tional area oc c upi ed b y the air phase i n the pump riser pi pe:
V
air
(13.9 fps)
)
6. Det er mi ne the relative density o f the air-water mi xt ure i n the pump riser pi pe:
Dens = 1
\Area) (0. 35 sfj
7. Det er mi ne the system head loss from equation (44):
- 0 . 5 6 - = 0 . 4 8 - ( 1 0 . 7 f p s )
2
=1. 18ft (D7)
8. Cal c ul at e the expec ted downst ream water l evel fr om equations (4) and (5):
fc
,
ra
,
c =
fc^)
=
MzM)
= 4
. 78f t (D8)
Dens 0. 48
9. Compar e the c al c ul at ed downst ream water depth fr om equation (D8) i n Step 8 to
the desired downst ream water depth. I f the c al c ul at ed water l evel fr om Step 8 is
b el ow the desired l evel the pump unit cannot pr ovi de the desired fl ow rate at the
desired lift and gi ven air fl ow rate. I n suc h a case the water fl ow rate must be
84
decreased and/or the air fl ow rate must be increased. The reverse is true i f the
c al c ul at ed downst ream l evel is above the desired height.
10. I n this exampl e we select a l ower water fl ow rate, leave the ai r fl ow rate as i s and
re-enter the process at step 1, now decreasing our assumpt i on o f the water f l ow
rate to Q
wate
r = 1.15 cfs for this pump unit at this ai rfl ow. Car r y out the steps agai n
starting at step 1 and c hec k the new result for the downst ream water l evel . Onc e
reasonable agreement has been reached the pr el i mi nar y desi gn i s c ompl et e. I n this
case the sec ond t ri al for the water fl ow rate was al most exact and thus a
pr el i mi nar y estimate o f the pump unit performance has been made.
I n this exampl e a prac t i c al operating poi nt o f 1.15 cfs o f water and 2. 5 cfs o f air i n a
si ngl e 8" diameter pump wi t h a lift o f 1.5 feet has been established. Ot her operat i ng
points ma y be expl or ed usi ng the same technique unt i l a satisfactory operating poi nt is
selected. As s umi ng the pr el i mi nar y desi gn gi ven above is satisfactory, and gi ven the
desi gn requirement for 10 cfs o f water, a reasonable suggestion woul d be to i nst al l 9 o f
the pumps as desc ribed, for a total water flowrate o f approxi mat el y 10. 5 cfs o f water
r equi r i ng appr oxi mat el y 23 cfs o f air.
G i ve n that step 7 as shown uses the fit values for the parameters i n equat i on (43) rather
than the c onservative val ues o f equation (44) it is reasonable to b ui l d a prototype uni t
based on these spec ific ations to c hec k performance. A more c onservat i ve approac h woul d
be to use the "envel ope c ur ve" parameters from equation (44) i n step 7 instead. D o i n g so
85
results i n a predi c t ed water fl ow rate for the exampl e pump o f 1.05 cfs, i ndi c at i ng that 10
rather than 9 units c oul d be required.
Al t ernat e pi pe diameters c an be investigated easil y, as c an the i nfl uenc e o f a greater
aeration depth, possi b l y devel oped t hrough exc avat i ng a sump at the site, etc. F or
exampl e, the model suggests that the same pump c oul d produc e a water fl owrate o f 2. 14
cfs i f the aerator were pl ac ed i n a three-foot deep sump. However , this increase i n water
fl owrate at the same air flowrate is not entirel y free since the air must be del i ver ed at a
c onsequent l y hi gher pressure, and subsequently a possi b l y hi gher cost.
Thi s desi gn approac h is c l ear l y suitable for hand c al c ul at i on and c an easi l y be automated
b y pr ogr ammi ng into a poc ket c al c ul at or suc h as the Hewl et t - Pac kar d 48 series or others
o f si mi l ar c apabi l i t y.
5.2 - Design Calculations for Personal Comput er
The desi gn proc edure out l i ned above is also suitable for i mpl ement at i on i n a c ommon
spreadsheet software suc h as Mi c r osoft E xc el . Fi gur e 17 shows a si mpl e formatted
spreadsheet i mpl ement at i on o f this desi gn technique. The user enters desi gn val ues i n the
b oxed c el l s mar ked as "I nput" and the spreadsheet automates the subsequent c al c ul at i on
steps desc ri bed above. S i mpl e changes to the system characteristics c an be made and
effects investigated. F i t parameters for the air phase vel oc i t y to mi xt ur e vel oc i t y
rel at i onshi p c an also be adjusted i f desired, as c an the fit parameters for the head loss to
mi xt ur e vel oc i t y rel at i onshi p. I n this wa y the performance predi c t i ons resul t i ng fr om the
86
direct fit and c onservative parameters c an be investigated. The spreadsheet sol ut i on also
al l ows for automated fast iteration to accurate sol utions b y means o f the Mi c r osoft
E xc e l "S ol ver ", " G o a l Seek", or equival ent user- i mpl ement ed system.
Figure 17 - Airlift Pump Churn Flow Worksheet
Sample Airlift Pump Churn Flow Worksheet
AB December 2003
Dens
Hloss
Qair = volume flow rate of air = 2.50 cfs Input
Qw = volume flow rate of water = 1.16 cfs Input
Diani = diameter of airlift pump tube = 0.67 ft Input
Upstr = upstream water level above aerator = 3.50 ft Input
Dnstrdes = desired downstream water level above aerator = 5.00 ft Input
grav = acceleration due to gravity = 31.90 fpss Parameter low high fit
a = curve fitting parameter in Vair=a+b*Vmix = 1.00 fps Parameter 1 3 1
b = curve fitting parameter in Vair=a+b*Vmix = 1.20 n/a Parameter 1.2 1.29 1.2
d = curve fitting parameter in Hloss = d*Dens*Vmix
A
e = 0.56 n/a Parameter 0.56 0.62 0.56
e = curve fitting parameter in Hloss = d*Dens*Vmix
A
e = 0.62 n/a Parameter 0.62 0.64 0.62
theoretical value for a above given slug flow
athy = 0.35*sqrt(grav*Diam)
-
1.61 fps Calculated
volume flow rate of the mixture
Qmix = Qair+Qw = 3.660 cfs Calculated
cross sectional area of the airlift pump tube
Area = PI()/4*Diam
A
2 = 0.349 sf Calculated
velocity of the mixture
Vmix = Qmix/Area = 10.486 fps Calculated
velocity of the air phase in the airlift pump tube
Vair = a+b*Vmix = 13.583 fps Calculated
cross sectional area occupied by the air phase
Aair = Qair/Vair = 0.184 sf Calculated
relative density of the mixture in the airlift pump tube
l-(Aair/Area) = 0.473 n/a Calculated
head loss from curve fitting experimental data
d*Dens*Vmix
A
e = 1.136 ft Calculated
calculated downstream water level
Dnstrcalc = (Upstr-Hloss)/Dens =
difference in calculated and desired downstream water
Dnstrdiff = Dnstrcalc-Dnstrdes =
5.000 ft
levels
Calculated
0.000 ft Calculated
87
The user selects val ues for the system inputs and parameters and c an expl ore vari ous
aspects o f the pump uni t ' s predic ted performance. I teration is si mpl e as the user adjusts
the air and/or water fl ow rate unt i l the desired downst ream and c al c ul at ed downst r eam
depths are equal . The difference i n these depths is c al c ul at ed at the bot t om o f the
worksheet to facilitate the process. A goal - seeki ng al gor i t hm or syst em may also be used.
The desi gn proc edure c an also be c oded into a func t i onal for m for i nc l usi on i n other
spreadsheets. Thi s approac h makes the c al c ul at i on o f airlift pump b ehavi our i mmedi at e.
Thi s approac h i s also wel l suited for tabulating predic ted airlift pump b ehavi our and
generating predi c t ed performance val ues for vari ous c ombi nat i ons o f desi gn vari abl es.
The desi gn proc edure was c oded into a set o f Mi c r osoft V i s ua l Bas i c for Appl i c a t i o n s
func tions for use wi t h Mi c r osoft E xc el spreadsheets. .
Fi gur es 18 and 19 show the Mi c r osoft V i s ua l Bas i c for Appl i c at i ons functions.
88
F i gu r e 18 - V BA Co de for Ch u r n F l o w Ai r l i f t P u mp D es i gn
F unc t i on Dnst r c al c ( By V a l Qai r As Si ngl e, By V a l Q w As Si ngl e, By V a l D i a m As S i ngl e,
By V a l Upst r As Si ngl e) As Si ngl e
' Thi s func t i on c omputes the downst ream water l evel gi ven the fl ow o f water (i n cfs),
' f l ow o f air (i n cfs), the pi pe riser diameter, and the upstream water l evel (both i n feet)
D i m Q mi x A s Si ngl e, V mi x As Si ngl e, V a i r As Si ngl e, Dens As Si ngl e
D i m Hl os s As Si ngl e, Ar ea As Si ngl e, Ai r As Si ngl e
Q mi x = Qai r + Qwat er: Ar ea = 0. 785 * (Di am)
A
2
Aa i r = Qai r / V a i r : Dens = 1 - (Aai r / Ar ea) : Hl oss = 0. 56 * Dens * ( V mi x
A
0. 62)
Dnst r c al c = (Upst r - Hl os s ) / Dens
E n d F unc t i on
89
F i gu r e 19 - V BA Co de for Ch u r n F l o w Ai r l i f t P u mp D es i gn
F unc t i on Q wa t er ( By V a l Qai r As Si ngl e, By V a l D i a m As Si ngl e, By V a l Upst r As Si ngl e,
By V a l Dnst r As Si ngl e) As Si ngl e
' Thi s func t i on c omputes the fl ow o f water gi ven the fl ow o f air (i n cfs), the pump riser
' diameter, the upstream water l evel and the downst ream water l evel (al l i n feet). It sets
' the water f l ow rate to zero and raises it i n smal l steps usi ng Dnst r c al c unt i l the c al c ul at ed
' and desi red downst r eam l evel s are equal .
D i m Q w l As Si ngl e, D ns t r l As Si ngl e
Q w l = 0: D ns t r l = Dnst r c al c (Qai r, Q w, D i a m, Upst r )
I f D ns t r l <= Dnst r Then Qwat er = 0
D o Un t i l D ns t r l <= Dnst r
Q w l = Q w l + 0. 01: D ns t r l = Dnst r c al c (Qai r , Q w, D i a m, Upst r )
L o o p
Qwat er = Q w l
E n d F unc t i on
The disadvantage to the func t i onal for m desc ribed here is that it isolates the user fr om the
intermediate val ues o f mi xt ure density, air, water, and mi xt ure vel oc i t y , etc. There is a
greater opport uni t y for the user to trust possi b l y questionable results because o f this
disc onnec t.
90
5.3 - Practical Considerations for Preliminary Airlift Pump Design.
Mixture Density:
There are several prac t i c al considerations when usi ng this approac h. It wi l l b ec ome
evident b y usi ng this desi gn technique that the l ow- l i ft , l ow- submergenc e c hur n fl ow
airlift pump system is sensitive to the mi xt ure relative density Dens. Whe n the mi xt ur e
relative densi t y falls muc h b el ow 0. 5, di mi ni shi ng returns set i n qui c kl y i n terms o f
inc reased water fl ow rate wi t h increased ai r fl ow rate. Onc e the mi xt ur e rel ative densi t y
has fal l en muc h b el ow 0. 45, i nc reasi ng the ai r fl ow rate even dramat i c al l y wi l l produc e
ver y l ittl e increase i n fl ow o f water. I n practice, i nc reasi ng ai r fl ow past this l evel wi l l
event ual l y reduce the f l ow o f water since the air is di spl ac i ng water i n the pi pe riser tube.
The model presented here does not capture this behavi our at ver y hi gh air f l ow rates.
However , that is not c onsi dered a fai l i ng because the phenomenon oc c urs far outside the
prac t i c al range o f desi gn. I f a designer finds hi m or hersel f attempting to b ui l d an airl ift
syst em to operate i n suc h a scenario, prototype testing wi l l be requi red sinc e the pump
uni t wi l l l i kel y be operating i n the annular or mi st fl ow regimes, whi c h exi st i ng airl ift
pump theory cannot quantify.
Air Pressure Required:
The air pressure requi red for an airlift pump system is theoretic al l y equal to the static
water pressure at the aeration depth and an al l owanc e for losses i n the air di st ri but i on
system. I n prac tic e, i f usi ng a mul t i - port aerator the aerator ports shoul d c ontribute a
reasonable head loss themselves. P r ovi di ng a notable pressure drop across the ports hel ps
91
ensure that al l ports pr ovi de equal ai r fl ow, thus ma xi mi zi ng the aeration effi c i enc y o f the
mul t i - port aerator. Thus the system designer shoul d be prepared to pr ovi de ai r fl ow at
appr oxi mat el y 0. 5 to 1 psi greater than predic ted b y the aerator submergenc e and air
syst em di st ri but i on losses.
Compressor Types
Compr essed air at l ow pressures and hi gh vol ume fl ow rates suc h as is requi red b y an
airlift pump syst em o f this type c an be obtained b y several means. Ener gy effi c i enc y o f
these systems i s l ow since muc h is lost i n turbulence and mi xi ng. Bec ause o f this energy
i neffi c i enc y , requirements for power are reasonably hi gh. (Fortunately, portabl e gas-
power ed sources are a ver y vi ab l e alternative and c an be used onl y when necessary).
Cent ri fugal b l ower s are the most ec onomi c al means o f suppl y i ng c ompressed air to an
airlift pump system, pr oduc i ng hi gh rates o f fl ow at l ow heads, t y pi c al l y b el ow 3 to 4 psi .
The Vor t r on Z 40, for exampl e c an easi l y generate 1000 sc fm at 3 psi wi t h a 40 hp mot or.
The c entrifugal units operate at ver y hi gh rotational rates, on the order o f 25 000 r pm, and
must be muffl ed appropriatel y to avoi d exc essi ve noise output. Regenerat i ve b l ower s are
somewhat more expensi ve than centrifugal bl owers but have the potential for a mul tistage
desi gn. I n suc h systems operating pressures o f up to 9 psi i n the 200 to 250 sc fm range
c an be reached. The F P Z S CL - 115 - D H , for exampl e, c an generate 475 sc fm at 9 psi wi t h
a 40 hp mot or. The last type o f air suppl y mac hi ner y suitable for use i n airl ift pumpi ng
systems is the posi t i ve displ ac ement b l ower . Bec ause o f their desi gn these units del i ver a
rel at i vel y constant suppl y o f air governed b y displ ac ement o f their internal lobes and the
92
rotational speed o f their i mpel l ers. Del i ver y pressures up to 15 psi are possi bl e wi t h
single-stage units. F or exampl e, the Sutorbil t 8 D H c an generate 300 sc fm at 15 psi wi t h a
36 hp mot or. A n airlift pump system requi ri ng an air suppl y wi t h del i ver y pressure above
15 psi woul d feature an aerator submergence muc h greater than those treated i n this
study. I n suc h a case air suppl y woul d l i kel y be suppl i ed b y a rotary sc r ew c ompressor
(suc h as that used i n the sec ond experi ment al phase o f this project). I n suc h a case the
desi gn proc edure o f Cl ar k & Dab ol t (1986) woul d be rec ommended.
93
C H A P T E R 6
6.1 - Conc l us i ons .
Interest in low-head, high-flow, low-submergence airlift pump units has historically been
low since such pumps are not particularly energy efficient and have been superceded by
submersible electric rotomachinery for many decades.
Despite having been replaced with more modern technology, airlift pumps are still used
in several niche applications and offer some promising potential benefits in the field of
urban stormwater management and other open-channel civil-engineering applications.
Existing theory was evaluated and found inadequate to describe the behaviour of the low-
head, high-flow, low-submergence airlift pumps. A four-stage experimental program was
developed and implemented, including a full-scale prototype application in an urban
storm drainage application in the city of Richmond, British Columbia. Performance data
was collected.
Three theoretical models were developed, with one satisfactorily fitting the experimental
data. The model was translated into a practical procedure that an engineer may easily use
to develop preliminary designs for airlift pumps operating in the churn flow regime. The
design procedure was implemented in two personal-computer-based applications and thus
can be quickly and easily completed. Some practical considerations for design of airlift
pumps operating in the churn flow regime are given.
94
6. 2 - R es ea r c h R ec ommenda t i ons
Now that the behaviour of airlift pumps in the churn flow regime has been modeled, the
potential for uses of these systems in other low-lift, high-flow, low-submergence
applications than those mentioned in this project should be explored. For example, airlift
pumps may be useful in irrigation and other pumping in open channels. If so, methods of
optimizing their performance in those scenarios must be developed. Additionally, the
aquaculture potential of airlift pumping in shrimp and other invertebrate farming drainage
applications should be investigated - the range of lift and flow rates are similar to those
in urban drainage and aeration of the water may provide additional productivity benefits
and cost savings through reducing the need for aeration equipment.
The airlift pump seems to offer many advantages in the urban drainage setting, and the
details of those advantages deserve to be investigated. Portable airlift pump units for local
flood control may be practical, as might portable or "emergency only" trailer-mounted
gasoline-powered air supply subsystems for permanently-installed units. The possibility
of reduced environmental impacts in urban drainage subject to aeration as a side effect of
airlift pumps should be investigated and subsequent benefits quantified.
More work is also needed to better understand the underlying phenomena of the two-
phase churn flow regime. Details such as the turbulent fluid behaviour at high void
fractions, the manner in which bubbles accrete at high void fractions and the influence of
aeration efficiency on regime stability are all unexplored. The development of high-speed
3-dimensional laser imaging technology may provide the necessary tools.
95
Refer enc es
Cart el l i er, A . "Si mul t aneous V o i d Fr ac t i on Measurement , Bub b l e Vel oc i t y , and S i ze
Est i mat e Us i ng a Si ngl e Opt i c al Probe i n Ga s - L i qui d Two- P hase F l ows ", R evi ew
o f Sc i ent i fi c I nstruments 63(11), (1992)
Cl ar k, N. N. & R . J . Dab ol t , " A general desi gn equation for air lift pumps operating i n sl ug
fl ow", Amer i c a n Institute o f Chemi c al Engi neers J ournal 32(1), (1986)
D e Cac har d, F . & J . M . Del hay e, " A sl ug- c hurn model for smal l - di amet er airl ift pumps",
I nternational J ournal o f Mul t i phase F l o w 22(4), (1996)
Duc kl er A . E . , M . Wi c ks & R . G . Cl evel and, "F r i c t i onal pressure drop i n two-phase fl ow:
A . A c ompar i son o f exi st i ng correlations for pressure loss and hol dup", Amer i c a n
Institute o f Chemi c al Engi neers J ournal (1964)
Duc kl er A . E . , M . Wi c ks & R . G . Cl evel and, "F r i c t i onal pressure drop i n two-phase fl ow:
B. A n approac h t hrough si mi l ar i t y anal ysi s", Amer i c a n Institute o f Chemi c a l
Engi neers J ournal (1964)
E l - K a ma s h, M . K . , M . R . L oewen & N Rajaratnam, "Measurement s o f bubbl e properties
and voi d fraction on a stepped chute", Proc eedi ngs o f the 15
t h
Canadi an
Hy dr ot ec hni c al Conver enc e of the Canadi an Soc i et y for Ci v i l Engi neer i ng (2001)
Fernandes, R . C. , R . Semi at & A . E . Duc kl er "Hy dr ody nami c model for gas- l i qui d sl ug
f l ow i n ver t i c al tubes", Amer i c a n Institute o f Chemi c al Engi neers J ournal 29,
(1983)
Gi ot , M . , "Handb ook o f Mul t i phase Systems", G . Het sr oni , ed. , Hemi s pher e- Mc Gr a w
H i l l , Washi ngt on, D C (1982).
96
Gova n, A . H . , G . F . Hewi t t , H . J . Ri c ht er & A . Scott, "F l oodi ng and c hur n flow i n ver t i c al
pi pes", I nternational J ournal o f Mul t i phase F l o w 17, 27- 44 (1991).
Govi er , G . W. , B. A . Radfor d & J . S . C. Dunn, "The upwards vert i c al flow o f air-water
mi xt ur es", Canadi an J ournal o f Chemi c al Engi neer i ng (1957).
I shi i , M . & N . Z ub er , "D r a g coefficient and relative vel oc i t y i n b ub b l y , dropl et or
particulate flows", Amer i c a n Institute o f Chemi c al Engi neers J ournal 25, (1979)
J ayant i , S. & G . F . Hewi t t , "Pr edi c t i on o f the sl ug- to- c hurn fl ow t ransi t i on i n ver t i c al t wo-
phase flow", I nternational J ournal o f Mul t i phase F l o w 18(6), (1992)
L a wni c za k, F . , P . Fr anc oi s, O. Sc ri vener, E . G . Kast r i naki s & S . G. Ny c has , "The
Effi c i enc y o f Short Ai r l i f t P umps Operat i ng at L o w Submergenc e Rat i os",
Canadi an J ournal o f Chemi c al Engi neer i ng 77 (1999)
Mar kat os, N. C . & A . K . S i nghal , "Numer i c al Anal y s i s o f One- Di mensi onal , Two- P hase
F l o w i n a Ver t i c al Cy l i ndr i c a l Passage", Advanc ed Engi neer i ng Software 4, 99-
106(1982).
Mor r i s on G . L , T. I . Z ei neddi ne, M . Henr i ksen & G . B. Tatterson, "Exper i ment al Anal y s i s
o f the Mec hani c s o f Reverse Ci r c ul at i on i n Ai r l i f t P umps", I ndustrial Engi neer i ng
Chemi c a l Resear c h 26 (1987
Nenes, A . , Assi mac opoul os, D . , Mar kat os, N . & Mi t s oul i s , E . , "S i mul at i on o f Ai r l i f t
P umps for Deep Wat er Wel l s ", Canadi an J ournal o f Chemi c al Engi neer i ng 74,
448- 456 ( D ATE )
Ni c kl i n , D . J . , "The air-lift pump: theory and opt i mi zat i on", Transac t i ons o f the Institute
o f Chemi c al Engi neers 41 (1963)
97
Ni c kl i n , D . J . , M . A . Wi l kes & M . A . Davi dson, "Two- phase fl ow i n ver t i c al tubes",
Transac t i ons o f the Institute o f Chemi c al Engi neers 40 (1962)
Pi c ker t , F . , "The theory o f the air-lift pump", Engi neer i ng, 34 (1932)
Serdul a, C D . & M . R . L oewen, "Exper i ment s I nvestigating The Us e o f F i b er - Opt i c
Probes F or Measur i ng Bub b l e S i ze Di st ri b ut i ons", J ournal o f Oc eani c
Engi neer i ng 23, (1998)
S haw, "Unwat er i ng the Ti r o Gener al Mi n e b y Ai r l i f t " Transac tions o f the A I M E 421,
(1920)
Shel t on, S. , & S . W. Stewart, "Bub b l e P ump Desi gn for Si ngl e Pressure Ab s or pt i on
Refri gerat i on Cy c l es ", A S H R A E Transac tions 108 (2002)
Tai t el , Y . , D . Bor nea & A . E . Buc kl er , "Model l i ng F l o w Pattern Transi t i ons for Steady
Upwa r d G a s - L i qui d F l o w i n Ver t i c al Tubes", Amer i c a n Institute o f Chemi c a l
Engi neers J ournal 26, 345- 354 (1980).
Turer, D . , B. J . Ma y na r d & J . J . Sansalone, "Hea vy metal c ont ami nat i on i n soi l s o f urban
hi ghway s: Compar i son between r unoff and soi l c onc entrations", J ournal o f
Wat er, Ai r and S oi l P ol l ut i on, 132(3) (2001).
Wa l l i s , G . B. , "One- Di mensi onal Two- Phase F l o w", M c G r a w- H i l l , Ne w Y o r k (1969).
Wa r d, D . "Exper i ment al St udy o f Ai r L i f t P umps " Bul l et i n o f the Uni ver si t y o f
Wi s c ons i n Engi neer i ng Servi c es 3, 4 (1924)
Wur t s, A . W . , S . G. M c Ne i l l & D . G . Overhul t s, "Performanc e and desi gn characteristics
o f airl ift pumps for fi el d appl i c at i ons", Wo r l d Aquac ul t ur e 25(4) (1994)
Z enz, F . A. , "E xpl or e the Pot ent i al o f Ai r - L i f t P umps and Mul t i phase Syst ems", Chemi c a l
Engi neer i ng Progress, 89(8), August (1993)
98
Si
a
1111111111111
mJIM
n n
nul l ill""
HI ill' ilF"|||i"
IIIIIIIIIIII
iiiinni
1
illiiiiiill
1
can
lllllllllljl
III ill'
llll"llll
IIIIIINllllllll
,,1111111111
"llllllllllllll
o
>
C: /AIRPUMP/AIRPUMP- 3.DWG
-a
o
!'!<i
- 7
< 5< 5<
o<5<
<<<<<
<5<5<
<5<5<
<5<5<
<<<<<
<<<<<
<<<<<
<<<<<
FLOW
TYPE 4
12"x4' PVC
I
k
<<<<
<<<<
<<<<
<<<<'
.I
>
O
"O
n
XI
>
zi
O
z
o
>
o
X)
>