HAL Id: hal-01083975
https://hal.archives-ouvertes.fr/hal-01083975
Submitted on 18 Nov 2014
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, estdestinée 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.
vessel under physiological pressure gradient
Dima Abi Abdallah, Agnès Drochon, Vincent Robin, Odette Fokapu
To cite this version:
Dima Abi Abdallah, Agnès Drochon, Vincent Robin, Odette Fokapu. Pulsed magnetohydrodynamic
blood flow in a rigid vessel under physiological pressure gradient. Computer Methods in Biomechanics
and Biomedical Engineering, Taylor & Francis, 2009, 12 (4), pp.445-458. �hal-01083975�
Pulsed Magnetohydrodynami blood ow in a rigid vessel
under physiologial pressure gradient
Dima Abi-Abdallah†,AgnèsDrohon†,VinentRobin‡,andOdetteFokapu†
†BiomehanisandBioengineering,UniversityofTehnologyofCompiègne,Frane
‡LaboratoryofAppliedMathematis,UniversityofTehnologyofCompiègne,Frane (June2007)
Blood ow ina steady magneti eld has been of great interest over the past years. Many re-
searhershaveexaminedtheeetsofmagnetieldsonveloityprolesandarterialpressure,and
majorstudiesfousedon steadyorsinusoidalows.Inthispaper wepresentasolutionforpulsed
magnetohydrodynami blood owwith asomewhat realistiphysiologialpressure waveobtained
usingawindkessellumpedmodel.Apressuregradientisderivedalongarigidvesselplaedatthe
outputofa ompliantmodulewhihreeivestheventrileoutow.Then,veloityproleandow
rateexpressionsarederivedintherigidvesselinthepreseneofasteadytransversemagnetield.
Asexpeted, resultsshowedow retardationand attening. Theadaptability ofour solutionap-
proahallowedaomparisonwithpreviouslyaddressedowasesandalulationspresentedagood
oherenewiththosewellestablishedsolutions.
Keywords:Statimagnetield,magnetohydrodynamiinterations,Halleet,windkessel,lumped
model.
1 Introdution
TheinreaseinexposuretohighmagnetieldsausedbythewideuseofMag-
netiResonane Imaging(MRI)asastandardmedialproedure, hasraiseda
onern in theresearh ommunity and onstituted an inentive for studying
theeets ofmagneti elds onhumanphysiologyand its impaton patients
health. Espeially that, in striving to ahieve higher resolution and greater
spetralseparation,theMRIsannersstatimagneti eldskeepaugmenting.
Studies evaluating theeet of human or animal exposure to magneti elds
haveshownnomajorhanges,exeptforaninreaseofsystolibloodpressure
aswell asalterations of the eletroardiogram (ECG) signalmanifested asel-
evations of theT wave,allof whih aredueto blood ow.
Themovementofaondutinguid,suhastheblood,inanexternallyapplied
magneti eld is governed by the laws of magnetohydrodynamis. When the
body is subjeted to a magneti eld theharged partilesof theblood ow-
ing transversally to the eld getdeeted by theLorentz fore thus induing
eletrial urrents and voltages, aross the vessel walls and in the surround-
ing tissues, strong enough to be deteted at the surfae of the thorax inthe
ECG. Furthermore, the interations between these indued urrents and the
applied magneti eld an ause a redution of ow rate and thus a reative
ompensatoryinrease inbloodpressure inorder to retain a onstant volume
owrate.
Magneti eld interations withblood ow have been demonstrated by mul-
tiple authors throughout in vitro experiments [1,2℄ where pressure and ow
rateweremeasured, aswellasinvivostudiessuhas[3,4℄whereanimalECG
alterations have been observed,and [5℄where theeetson humanvitalsigns
were foundto onsist essentiallyinanarterial pressure inrease.
Theoretialmagnetohydrodynamibloodowalulationshave,however,been
addressedmuhearlierandgobakasfar astheearlysixties.Korhevskiiand
Marohnik [6℄rst proposedaveloityprole solution for blood ow between
two parallel plates under a onstant pressure gradient with a perpendiular
magneti eld, under the assumption that blood is newtonian. Later other
studies foused on ow in a rigid irular tube with non onduting walls
plaed ina transverse magnetield to oer a more realistimodelfor blood
owinvessels. Inthis ase, the most omplete solution ofthemagnetohydro-
dynami equations of a onduting uid was proposed by Gold [7℄. Setting a
onstant pressure gradient, Gold derived expressions for the veloity prole
as well as indued elds and voltages. Vardanyan [8℄ subsequently published
an approximate steadysolution where veloityprole and ow rate were al-
ulated by negleting the indued elds.More reent studies were essentially
based on these founding works, suh asKeltner et al. [1℄ where a omparison
wasestablishedbetweentheresultsofGoldandVardanyantoassesstheonse-
quenesofnegletingtheindutions.WiththesamehypothesisasVardanyan,
Sudetal. [9℄laterdealtwithasinusoidalpressuregradientthatmodeledabit
loser thepulsed nature of blood ow inarteries. The hypothesis of ondut-
ingwallswasnotintrodued untilKinouhietal.[10℄whoinluded indutions
in the vessel and the surrounding tissues in the steady ow ase in order to
evaluate theindued ECG superimposedvoltages.
In this work, we revisit the ow of blood as a newtonian uid, in a irular
rigidvessel,withnonondutingwalls,inthepreseneofatransverseonstant
magneti eld.Nevertheless, insteadoftakinga onstant pressuregradient or
a sinusoidal one, we apply a realisti pulsed pressure gradient derived using
a windkessel lumped model, where the ompliant module provides the input
owintotherigid vessel.Then, negleting induedelds,wesolve themagne-
tohydrodynamiequationsto obtainveloityprole andowrateexpressions.
Thefatthatour resolution method isbasedonFourier deompositionmakes
the solutions easily adaptable to steady or sinusoidal ases, thus allowing a
omparison withtheprevious well establishedstudies.
2 General equationsand solution
The ow of a onduting inompressible newtonian uid inthe preseneof a
magnetieldisdenedbyaombinationofMaxwell'sequationsononehand,
andtheNavier-Stokesequationinludingthemagnetiforeontheother,along
withtheonservation equation, aswell asOhm'slaw.
If we neglet the indued elds, the veloity prole an be solely dened by
theNavier-Stokesequation, where the magnetifore term is evaluatedusing
Ohm's law,
ρ ∂~u
∂t +
~u·∇~
~u
=−∇p~ +η∆~u+σ
~u∧B~
∧B~ , (1)
where
B ~
is the magneti eld,~u
,ρ
,η
,σ
are respetively the uid veloity, density,visosityand ondutivityand ∇p~
is thepressure gradient.Byassuming thattheowis unidiretional, axisymmetri withno swirl in a
Figure1. Flowmodelgeometry
Thevesselisrepresentedbyaylindrialondutwhereblood owsalongtheOzaxis,inthe
preseneofatransversesteadymagnetieldoriented intheOxdiretion.
irularrigidvessel,itsveloityanbewrittenas
~u = (0, 0, u(˜ r, t))
(g.1).Theuidpressure isthus a funtion of theposition z and time.The external on-
stantmagnetieldisappliedtransversallysuhas
B ~ = (B
0cos θ,
−B0sin θ, 0)
and the vessel is onsidered to have non onduting walls. The ow would
thereforebe governed bythe
Oz
projetion of (1)whihexpressedinylindri-aloordinates gives,
a2 ν
∂u(˜r, t)
∂t =g(t) +∂2u(˜r, t)
∂r˜2 +1
˜ r
∂u(˜r, t)
∂˜r −Ha2u(˜r, t) , (2)
with
r ˜ =
ar, wherea
represents the vessel radius,H
a= B
0a
qσ
η is the Hart-
mannnumber,
ν =
ηρ isthe kinemati visosity,and g(t) =−a2η
∂p(t, z)
∂z , (3)
withboundary onditionat the walls
u(1, t) = 0
.The proposed resolution method onsistsof aFourier deomposition, followed
byaHankeltransform.
Fourier series deomposition. The pulsed ow studied here is periodiwith
period T,inverse ofthe ardia frequeny.
u(˜ r, t)
andg(t)
arethus Tperioditimefuntions thatan be deomposed inFourier series suhas,
u(˜r, t) =
+∞
X
k=−∞
uk(˜r)eiωkt where:uk(˜r) = 1 T
Z T 0
u(˜r, t)e−iωktdt
g(t) =
+∞
X
k=−∞
gkeiωkt where:gk = 1 T
Z T 0
g(t)e−iωktdt
with:
ω
k= k
2πT .By replaingin(2) we get,
a2 ν
+∞
X
k=−∞
iωkuk(˜r)eiωkt=
+∞
X
k=−∞
gkeiωkt+
+∞
X
k=−∞
∂2uk(˜r)
∂˜r2 +1
˜ r
∂uk(˜r)
∂˜r
eiωkt
−Ha2
+∞
X
k=−∞
uk(˜r)eiωkt
⇔ a2
ν iωkuk(˜r) =gk+ ∆uk(˜r)−Ha2uk(˜r) ∀k∈Z . (4)
Hankel Transform. For a funtion
f (r)
dened over[0, 1]
, the zero orderHankeltransform isdened [11℄ as,
H(f) = (fn∗)n∈Z ; fn∗= Z 1
0
rf(r)J0(rλn)dr ;
withthefollowing properties,
H(∆f(r)) =−λ2nfn∗ (iff(1) = 0) and H(cste) = cste
λn
J1(λn) ,
where
λ
n are therootsofBessel funtionJ
0(x)
.Knowing that
u(1, t) = 0
⇔u
k(1) = 0
∀k ∈ Z,applying the Hankel trans-formon equation(4) yields,
u∗k,n= J1(λn) λn iωka2
ν +λ2n+Ha2gk . (5)
Solution. To derive the solution we must inverse the Hankel transform to
get the veloity's Fourier oeients and then establish the veloity prole
expression.
We know that if
H(f ) = f
n∗ , then the inverse transform givesf (r)
suh as,f (r) = 2
P∞n=1 J0(λnr) J1(λn)2
f
n∗.From (5)we derive,
uk = 2
∞
X
n=1
J0(λn˜r) λnJ1(λn)
gk
iωka2
ν +λ2n+Ha2 , (6)
and thereforeaveloity prole,
u(˜r, t) = 2
∞
X
k=−∞
∞
X
n=1
J0(λnr)˜ λnJ1(λn)
gk
iωka2
ν +λ2n+Ha2eiωkt . (7)
Theowrate will thenbe given as,
q(t) = Z Z
A
u dA= Z 2π
0
Z a 0
u(r, t)rdrdθ
= 4πa2
∞
X
k=−∞
∞
X
n=1
eiωkt λnJ1(λn)
gk
iωka2
ν +λ2n+Ha2 Z 1
0
J0(λnr)˜˜rd˜r .
Using thefatthat
R
x
nJ
n−1(x)dx = x
nJ
n(x)
( [12℄,p.137),we get,q(t) = 4πa2
∞
X
k=−∞
∞
X
n=1
eiωkt λ2n
gk
iωka2
ν +λ2n+Ha2 . (8)
3 Pressuregradientexpression
Intheliterature,studiesonerningpulsedowsinrigidtubesusuallyonsider
sinusoidalpressuregradients[11℄and[9℄.Inthiswork,inordertosolvetheve-
loityprole,weseektoimposearealistiphysiologialpressuregradient.For
this,we usethe3-element Windkessellumped modelwhere a ompliant mod-
ulerepresentsthelargearteriesandapureresistanerepresentstheperipheral
rigid vessels. We then derive a pressure gradient, governed by the ompliant
module, to be applied on arigid vesselbeyond thelarge arteries.
(a)2-elementWindkessel (b) 3-elementWindkessel
Figure2. Windkessellumpedmodels
Pv(t):leftventriularpressure,Pa(t):aorti pressure,Q(t):outputowrateoftheleftventrile,
Q1(t):inputowratetotheperipheralvessels,Ra:resistaneoftheaortaandthelargearteries,
C:omplianeoftheaortaandlargearteries,Rp:totalperipheralresistaneofsmallarteries,
arteriolesandapillaries
They rely on an analogy with eletri iruits where urrents represent ar-
terial blood ows and voltages represent arterial pressures. In suh models,
resistanes standforresistaneto ow(arterialandperipheral)resulting from
visous dissipation inside the vessels, apaitors represent volume ompliane
of thevessels,and indutorsrepresent blood inertia.
Intheirulatorysystem,thesmall aliberarteriesan beonsideredasrigid.
These arteries get at their input a ow rate imposed by the large ompliant
arteries and an be modeled usingpureresistanes [13℄.
Inthefollowingsetionwewillestablishamodelthatwillallowusto evaluate
this owrate andtherefore deduean expressionof pressuregradient along a
rigid vessel.
The Windkessel model, oneived by Otto Frank in1899 and inspired by air
hamber pumps used in re engines, is a quite simple onguration that de-
sribestheow attheheartoutputand intothesystemi arteries.Themodel
onsistsofanelastiaumulationhamberplaedinarigidondutpreeded
bya valveand followedbyaPoiseuillehydrauliresistane. Whenthevalveis
open(systole)partoftheventriularowaumulatesinthehamber,andthe
restowsintheresistane.Whenthevalveislosed(diastole)thebloodwhih
hadaumulatedinthehamberisforedoutthroughtheresistane.Thisele-
trial modelwasinitiallyoneived withtwo elements omprisinga apaitor
that representstheelastiityof large arteries and aperipheral resistanethat
stands for the resistane of small arteries and arterioles (g.2(a)). The input
pressureoftheiruitisthe leftventriularpressure, assuminganullpressure
atthevenaava.Adiodeplaedattheiruit'sentryplaystheroleoftheaor-
ti valve whihlets theow getthroughonly whentheventriular pressureis
superior tothatoftheaorta. This2-element modelwaslater transformedinto
a 3-element model (g.2(b))where an additional resistanewasintrodued to
takeinto aount theresistaneof theaorta andlargearteries [14℄ .Thelatter
model wasfound to produe quiterealisti pressure and ow rateurvesthat
orretlyreprodue experimental data[15℄and thus remains verywidelyused
In the following setion we adopt the 3-element windkessel model and om-
pute an expression for pressure along a pure resistane at the output of the
ompliant module. We rst dene a mathematial expression to model real
ventriular pressureinput suhas[15℄,
P v(t) = Pmax
2 (1−cos 2γt) 0≤t≤tp
0 tp≤t≤T (9)
with
γ =
tπp
.
Thedierential equationdening the3-element iruit an be written as,
dP a dt +P a
τ = Q(t)
C , (10)
where
τ = CR
p .Solving for
P a(t)
in eah ardia yle phase : diastole and isovolumetri ontration phases whereQ(t) = 0
, as well as the ejetion phase whereQ(t) =
P v(t)−P a(t)Ra
,whileensuringurve ontinuity between thephasesyields
theaortipressureexpression,
P a(t) =
Pse−t+Tτ−ts 0≤t≤t1 (isovolumetriontration)
Ke−t−t1Zτ +A(t) t1≤t≤ts (ejetion)
Pse−t−tsτ ts≤t≤T (diastole)
(11)
with,
A(t) = Rp
Ra+Rp
Pmax
2
1−2γZτsin 2γt+ cos 2γt 1 + 4γ2Z2τ2
;
Ps=P v(ts) ; K=P v(ts)e−t1 +T−tsτ −A(t1) ; Z= Ra
Ra+Rp
;
t
1 denotesthebeginning oftheejetionphasewhenP a
beomeslessthanP v
,t
s denotes the end of systole whenP a
beomes greater thanP v
, andt
p isthe instant at whih the pressure in theventrile drops to zero, and
T
is theardia yleperiod.
The ow rate in the peripheral resistane is given by
Q
1(t) =
P a(t)Rp
and the
pressure drop along a peripheral vessel of radius
a
and lengthL
would beobtained by multiplying the ow rate with a hydrauli Poiseuille resistane
8ηL
πa4 yielding dP(t)
L
=
−πa8η4Q
1(t)
,and thus,−∂p(t, z)
∂z = 8η πa4
Ps
Rpe−t+T−tsτ 0≤t≤t1 K
Rpe−t−t1Zτ +A(t)R
p t1≤t≤ts Ps
Rpe−t−tsτ ts≤t≤T
(12)
Inordertoderivetheveloityproleexpressionby(7)weneedtoomputethe
Fourier oeients
g
k ofg(t)
dened in(3) . AfterFourier integral alulationfor eah ofthethree phases,
gk = 1 T
Z T 0
−a2 η
∂p(t, z)
∂z e−iωktdt
= 8
T πa2Rp
Z t1
0
Pse−t+T−tsτ e−iωktdt+ Z ts
t1
Ke−t−t1Zτ +A(t)
e−iωktdt
+ Z T
ts
Pse−t−tsτ e−iωktdt
! ,
we get,
gk = 8 T πa2Rp
Ps 1 τ +iωk
e−T−tsτ
1−e−t1(τ1+iωk)−e−iωkT +e−ts(1τ+iωk)+Tτ
+ K
1 Zτ +iωk
eZτt1
e−t1(Zτ1 +iωk)−e−ts(Zτ1 +iωk)
+ Rp
Ra+Rp
Pmax
2
"
fk− e−iωkt[(−iωk−2γǫ) cos(2γt) + (2γ−iωkǫ) sin(2γt)]
(1 + 4γ2Z2τ2)(4γ2−ωk2)
ts
t1
#) ,
(13)
where
ǫ = 2γZτ
andf
k=
(
t
s−t
1k = 0
e−iωkt1−e−iωkts
iωk
k
6= 0.
4 Results
All numerial omputations were done using the parameter values shown in
table 1.
Figure3showsthepressureandowrateurvesomputedusingthe3-element
Windkessel model (g.2(b)). They agree very well with measured pressure
urvesshownin[16℄.Fromthese omputedresults we analulate otherar-
diayleparameters.Wendameanardiaoutputof
77cm
3/s
(4.62l/min
),a stroke volume of
62cm
3 and a systole duration oft
s= 30%T
, all oin-iding well with the normal range values (Comolet [17℄, for example, gives
C.O.
≈87cm
3/s
,S.V.
≈70cm
3 andt
s≈37%T
forP
max= 140mmHg
).Figure 4(a) shows the pressure gradient wave inthe rigid vessel withradius
a
,aswell astheowrates for dierent Hartmannnumbers. It ouldbenotedthatforsmall
H
athevariationdynamisoftheowrate(andveloity)isalotslowerthan thatof thepressure gradient, however for larger
H
a theow rateWindkesseliruitomponents
Ra 0.0334 mmHg.s/cm3
Rp 1 mmHg.s/cm3
C 0.77 cm3/mmHg
Bloodharateristis
η 4.10−3 P a.s
ρ 1050 kg/m3
σ 0.5 S/m
Vesselradius
a 0.3 cm
Cardiayleparameters
f req 75 bpm
Pmax 120 mmHg
tp 50%T s
Table1. Numerialvalues
Theresistaneandapaitorvaluesarehosenbased on[15℄andagreeingwith[14℄,blood
harateristisaretakenfrom[10℄andtheardiayleparametersarehosentomaththe
averagetypialvalues.
Figure3. Ventriularpressurewaveandaortipressureandowrateurves
Pressureandowratewavesomputed usingthewindkesselmodelovertwoardiaylesat75
bpm.TheventriularpressureP v(t)issetasinequation (9) ,theaortipressureP a(t)is
omputed by(11),andtheaortiowrate(ventrileoutput)isQ(t) = P v(t)−P a(t) Ra
urvestends to follow that of thepressure. The retardation of themovement
an also be learly depited, as theow slows down when the magneti eld
intensity inreases(g.4(b)).
Note that by applying the Poiseuille standard formula for stationary ows
q
pois=
πa8η4∆p∆z ,inavessel ofthesamealiberunderapressure gradient equaltothemeanvalueoftheomputedpulsedgradient,wewouldgetapproximatly
77cm
3/s
.Whihisthevaluewegetbyomputingthemeanvalueofthepulsedowratefor
H
a= 0
.Given that a Poiseuille prole yields a maximum veloity
U
max=
a4η2∆p∆z ≈544cm/s
,gures5(a)and 5(b)represent normalized veloityprolesasratiosto this value for
H
a= 0
andH
a= 2
respetively.(a) Pressuregradient and owrate inarigid
vessel: The pressure gradient is alulated by
(12),andtheowrateby(8)fordierentHart-
mannnumbersforapulsedowinarigidves-
selwhihharateristivaluesaregiveninta-
ble1
(b)MeanowrateasfuntionoftheHart-
mannumber:Thegraduationsontheleft
representaowratenormalizedasaratio
to a Poiseuilleow rate, whilethe grad-
uationsontherightrepresentaowrate
incm3/s
Figure4. Flowrateforvariouseldintensities
0
T
2T 0
0.5 1 0
0.2 0.4 0.6 0.8 1
Time Normalized velocity profile (Ha=0)
r/a
U /(UmaxPois)
(a) Ha=0
0
T
2T 0
0.5 1 0
0.2 0.4 0.6 0.8 1
Time Normalized velocity profile (Ha=2)
r/a
U /(UmaxPois)
(b) Ha=2
Figure5. Normalizedveloityproles
Theveloityisomputedfrom (7)andthenormalizationisdonerespetivelytothePoiseuille
Umax≈544cm/s
5 Comparison with otherow types
Inthissetionwedisusstheobtainedresultsbyomparingthemtootherwell
establishedases of blood ows,whether stationaryor pulsed,inthepresene