HAL Id: hal-00454440
https://hal.archives-ouvertes.fr/hal-00454440
Submitted on 8 Feb 2010
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Discussion of ”Investigation of Flow Upstream of
Orifices” by D. B. Bryant, A. A. Khan and N. M. Aziz,
Journal of Hydraulic Engineering, January 1, 2008, Vol.
134, No. 1, pp. 98-104
Gilles Belaud, X. Litrico
To cite this version:
Gilles Belaud, X. Litrico. Discussion of ”Investigation of Flow Upstream of Orifices” by D. B. Bryant,
A. A. Khan and N. M. Aziz, Journal of Hydraulic Engineering, January 1, 2008, Vol. 134, No. 1, pp.
98-104. Journal of Hydraulic Engineering, American Society of Civil Engineers, 2009, 135 (2), p. 155
- p. 157. �10.1061/(ASCE)0733-9429(2009)135:2(155.2)�. �hal-00454440�
Khan and N. M. Aziz
Journal of Hydrauli Engineering, January1,2008, Vol. 134, No. 1,pp. 98-104.
DOI: 10.1061/(ASCE)0733-9429(2008)134:1( 98)
GillesBelaud 1
and XavierLitri o 2
Intheirpaper(Bryantetal.,2008)theauthorsusepotentialowtheorytostudyowupstream
of ori es, and ompare theoreti al results to a large set of experimental data. As pointed
out previously by dierent authors among whi h Shammaa et al. (2005) and Belaud and
Litri o (2007,2008), potential ow assumptions give a qui k and e ient method to estimate
thevelo ityeldgeneratedbyanori eandmaybesu ientformanyengineeringappli ations
that do not require to take all the real uid ee ts into a ount, su h asvorti es. Thispaper
thereforeprovidesanother interesting appli ation of the potential ow theoryto real ases.
However, there seems to be some onfusion about the notions used in the paper, espe ially
the radial velo ityandthe velo itymagnitude,and thisdis ussionaimsat bringing some
om-plements to larify these points. We also larify the way one should apply the prin iple of
superposition, and nally provide some theoreti al ba kground for the study of the velo ity
distribution.
1
Resear her, UMR Gestion del'Eau, A teurs, Usages, IRD,Maison des S ien es de l'Eau, 300 av. Emile
Jeanbrau34095MontpellierCedex5,Fran e,belaudmsem.univ-montp2.fr 2
Resear her, UMR Gestion de l'Eau, A teurs, Usages, Cemagref, 361, rue JF Breton, B.P. 5095, 34196
Thepaperusesthepotential methoddes ribed inShammaa etal.(2005). A ording to Bryant
et al. (2008), the appli ation of this method givesa good des ription of the ow far from the
ori e but a rather bad des ription of the owpattern nearthe ori e. Theauthors therefore
propose a so- alled new model that better reprodu es the ow pattern. This dis repan y
between the two modelsis surprising, sin e the method usedbyBryant et al. (2008) to derive
the potential fun tion is exa tly the same as the one originally proposed by Shammaa et al.
(2005). In fa t, the model proposed by the authors is obtained from the original model of
Shammaa et al. (2005) via a simple hange of oordinates, whi h doesnot justifyin our view
the denomination of new model.
Moreover, the dis repan y appears to be due to a misunderstanding of the original model of
Shammaa et al.(2005) whi h orre tly representsthe velo itypattern (seeFig. 3 in Shammaa
et al. (2005)).
First of all, let us re all that the ow potential
Φ
, in the planex − z
(the same notation as Bryant etal. isused),isafun tionofr
andθ
, ontrarilytothenotation usedbythe authorsinEq. (1). The radial and transverse velo ities are then obtained asfollows (see e.g., Bat helor,
1967,pp.100, 600):
V
r
=
∂Φ
∂r
V
θ
=
∂Φ
r∂θ
and~
V = V
r
e
~
r
+ V
θ
e
~
θ
. Were allinFigure1thedenitionofthe radialandtransversevelo itiesV
r
andV
θ
and the unitve torse
~
r
ande
~
θ
.Inthe verti alplane
x − z
, the velo itymagnitudeV
an be omputed asfollows:Figure1: Denitionof velo ity omponents in averti al plane(
x − z
)Assuming that the potential dependsonly on
r
leads to assume thatV
θ
iszero, whi h in turngives
V = |V
r
|
. NotethatV
r
should benegativein the ase ofa sink.In fa t, it is not lear whi h quantity is displayed in the gures of the paper. The legend
states
V
r
, i.e. the radial velo ity, but the dis ussers think that the plot orresponding to thenew solution andthe experimental resultsin fa trepresent the velo itymagnitude, whilethe
original solution orrespondsto the magnitudeof the radial velo ity.
Therefore, we think that the apparent dis repan y between the original and new solutions is
due toa onfusion between radial velo ityand velo itymagnitudes. Provided the originof the
polar oordinatesisintheori e,thesetwoquantitiesare losetoea hotherfarfromtheori e
aspointed out bythe authors, butlargely deviate aswe approa h the ori e.
We now usean analyti al modelderived by Belaud and Litri o (2007) to illustrate these
on- epts.
Radial Velo ity and Velo ity Magnitude
Let us onsider a square ori e of side
2c
entered in 0, and use potential ow theory toexpress the velo ity omponents in Cartesian oordinates. As shown by Belaud and Litri o
(2007) for any re tangular ori e, the potential ow solution an be expressed in losed-form
omponents in
x − z
plane. Dueto symmetry, the omponentV
y
is null. We introdu eM
= −
Q
8πc
,
(2)r
1
=
pc
2
+ x
2
+ (c − z)
2
,
(3)r
2
=
pc
2
+ x
2
+ (c + z)
2
,
(4)λ
1
=
px
2
+ (c − z)
2
+ (c − z)
x
,
(5)λ
2
=
px
2
+ (c + z)
2
− (c + z)
x
,
(6)X
1
=
c
r
1
+
px
2
+ (c − z)
2
,
(7)X
2
=
c
r
2
+
px
2
+ (c + z)
2
,
(8)in whi h
Q
isthe dis hargethrough the ori e.Usingtheresultsdevelopedin BelaudandLitri o(2007),
V
x
andV
z
anbe omputedusingthefollowing analyti alexpressions:
V
x
= 2M
arctan(λ
1
X
1
) − arctan
X
1
λ
1
− arctan(λ
2
X
2
) + arctan
X
2
λ
2
(9)V
z
= M log
(r
1
+ c)(r
2
− c)
(r
1
− c)(r
2
+ c)
.
(10)From this set of equations, we an ompute the radial velo ity
V
r
and the velo itymagnitudeV
asfollows:V
r
=
~
V .
−→
OP|
−→
OP|
=
xV
√
x
+ zV
z
x
2
+ z
2
(11)V
=
pV
2
x
+ V
2
z
.
(12)Thesevelo itiesareplottedatdierentdistan esfromthe ori einFigure2. Thedistan esare
normalized by the equivalent diameter
d = 2p4/πc
whi h gives the same area for the square ori e asfora ir ularori eofdiameterd
. Theplotsareverysimilarto thoseofBryant etal.x/d = 0.1
0.2
0.3
0.5
x/d = 0.1
0.2
0.3
0.5
x/d = 0.1
0.2
0.3
0.5
x/d = 0.1
0.2
0.3
0.5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
−1.5
−1.0
−0.5
0.0
0.5
1.0
1.5
Normalized velocity |Vr|/Uo or V/Uo
z/d
Radial velocity Vr/Uo
Velocity magnitude V/Uo
Radial velocity Vr/Uo
Velocity magnitude V/Uo
Radial velocity Vr/Uo
Velocity magnitude V/Uo
Radial velocity |Vr|/Uo
Velocity magnitude V/Uo
Figure2: Velo ity valuesalong
z
oordinate at dierent distan esx
fromthe ori e plane.et al. (2005), while the solution presented as the new solution appears to be the velo ity
magnitude. Far from the ori e, the transverse omponent of the velo ity be omes small and
bothquantitiesbe ome very lose. Figure3shows the iso-velo itylinesin the plane
x − z
and an be ompared to Figure3of Bryant etal. (2008): the dottedlinesdepi tthe radial velo ityV
r
, while the plain lines depi t the velo itymagnitude. We an also point out that the ori eshape haslittle inuen ein thedomain of study.
Asa sideremark,we notethat usingEq. (1),wehave:
V
2
r
≤ V
2
.
(13)Thisinequalityexplains why, in Figures3and 4ofBryantet al.(2008), the originalsolutionis
always lower than the so- alled newsolution, in whi h the velo itymagnitude
V
is al ulatedfromEq. (12)of thisdis ussion andEqs. (5)to (7)of Bryant et al. (2008).
Prin iple of Superposition
A ordingto theprin ipleofsuperposition, thepotential fun tion,the streamfun tionandthe
0.2
0.5
0.4
0.3
0.2
0.5
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
−1.0
−0.8
−0.6
−0.4
−0.2
0.0
0.2
0.4
0.6
0.8
1.0
x/d
z/d
Radial velocity |Vr|/Uo
Velocity magnitude V/Uo
Figure3: Iso-velo itylinesin
x − z
planean ertainlynot beadded.
Let us illustrate this point in the ase of the superposition of two velo ityelds. We want to
nd the resulting velo ityelds bysuperposition ofthe elds
~
V
1
and~
V
2
. The resulting eldisgivenby:
~
V = ~
V
1
+ ~
V
2
.
This anbeexpressedinany oordinatesystembyaddingthe omponentsofthevelo ityve tor
alongthe oordinate basis. InCartesian oordinates, we get:
V
x
= V
1
x
+ V
2
x
V
z
= V
1
z
+ V
2
z
.
Inpolar oordinates, weget:
V
r
= V
1
r
+ V
2
r
the transverse omponent
V
θ
is needed in order to ompute the velo ity ve tor. This is valid onlyifthesame oordinatesystemisusedfor allthesuperposedsinks. Inthepolar oordinatesase, itis important to he kthatthe two systems aredes ribed usingthe same origin.
Theexpressionsprovided byShammaaet al.(2005)inEqs. (1011)are orre tandleadtothe
same results as in Cartesian oordinates. The dis repan y between the original solution and
the newsolutionin Figures10,12,13and 14ofBryantet al.(2008)isagainprobablydue toa
onfusionbetweenradialvelo itiesandvelo itymagnitudes. A ording tothedis ussers'
al u-lations, theseplots (originalsolution)mayresultfromtheaddition ofthe velo itymagnitudes
or fromaddition ofradial velo ities al ulated with a dierent origin for both sour es.
Ee t of the Velo ity Distribution
The authors used an empiri al method to determine the limit when the velo ity distribution
should be onsidered. Thepresent dis ussion brings some theoreti alelements onsistent with
theexperimentalresults. Tosimplify, we onsiderasquareori eofheight
2c
. Theee toftheori e size an be analyzed by onsidering that an ori e is omposedof two parts, an upper
partof height
c
, entered in+c/2
, and alower part ofheightc
, entered in−c/2
.Themeanheadabove theupperori e is
h
0
− c/2
, whilethe meanhead above thelower ori e ish
0
+ c/2
. Sin etheori estrengthisproportionaltoitsdis hargeandthereforeto thesquarerootofthe head,themeanerror
ǫ
onthe strengthbetween thelowerand theupperori es anbeestimatedby:
ǫ =
ph
0
+ c/2 −
√
ph
0
− c/2
h
0
(14) whi h givesǫ ≃
c
2h
0
(15)If
d = 2c
is the height of the ori e and ifwe takeǫ = 0.02
asthe authors do in their paper, orresponding to an error of 2%, we nd a limit ofh
0
/d = 12.5
, whi h is exa tly the resultobtained experimentally bythe authors.
Thesameanalysis ondu tedwitha ir ularori e wouldleadto aslightlydierent result,but
the analysisgivesa rough andrapid estimation ofthe ee t of thevelo ity distributionwithin
the ori e. Thismay be helpful to de ide whether, or not, the velo ity distribution should be
usedin the potential owsolution. Ifnot, we mayuse losed-form expressions for the velo ity
eld(see equations (9)and (10)ofthis dis ussionfor the squareori e).
Referen es
Bat helor,G.K. (1967). An introdu tion touid me hani s. Cambridge University Press.
BelaudG.andLitri oX.(2007). 3Dvelo ityeldgeneratedbyasideori einanopen- hannel.
32th CongressIAHR, Veni e, 1-4 July 2007.
Belaud G. and Litri o X. (2008). Closed-form solution of the potential ow in a ontra ted
ume J. of Fluids Me hani s,599:299307.
Bryant D.B.,Khan,A.A.andAziz,N.M. (2008). Investigationof FlowUpstreamof Ori es.
J. of Hydrauli Engineering,134(1):98104.
Shammaa, Y., Zhu, D. Z. and Rajaratnam, N. (2005). Flow upstream of ori es and slui e