epapepapffect of swirling inlet condition on the _ow _eld in a stenosed

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

http://www.elsevier.com/authorsrights

Author's personal copy

Medical Engineering & Physics 36 (2014) 119– 128

Contents lists available at ScienceDirect

Medical Engineering & Physics

jou rn al h om epage: www.elsev ier .com/ locate /medengphy

Effect of swirling inlet condition on the flow field in a stenosedarterial vessel model

Hojin Haa, Sang-Joon Leea,b,∗

a Department of Mechanical Engineering, Pohang University of Science and Technology, San 31, Hyoja-dong, Pohang 790-784, Republic of Koreab Center for Biofluid and Biomimic Research, Pohang University of Science and Technology, San 31, Hyoja-dong, Pohang 790-784, Republic of Korea

a r t i c l e i n f o

Keywords:Spiral flowStenosisSpiral flowHelical flowHemodynamicsAtherosclerosisParticle image velocimetry (PIV)

a b s t r a c t

Blood flow in an artery is closely related to atherosclerosis progression. Hemodynamic environmentsinfluence platelet activation, aggregation, and rupture of atherosclerotic plaque. The existence of swirlingflow components in an artery is frequently observed under in vivo conditions. However, the fluid-dynamicroles of spiral flow are not fully understood to date. In this study, the spiral blood flow effect in anaxisymmetric stenosis model was experimentally investigated using particle image velocimetry velocityfield measurement technique and streakline flow visualization. Spiral inserts with two different helicalpitches (10D and 10/3D) were installed upstream of the stenosis to induce swirling flows. Results showthat the spiral flow significantly reduces the length of recirculation flow and provokes early breakout ofturbulent transition, but variation of swirling intensity does not induce significant changes of turbulenceintensity. The present results about the spiral flow effects through the stenosis will contribute in achievingbetter understanding of the hemodynamic characteristics of atherosclerosis and in discovering betterdiagnosis procedures and clinical treatments.

© 2013 IPEM. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Circulatory vascular disease (CVD) is the leading cause of deathin many developed countries. Atherosclerosis is one of the maincauses of CVDs, which lead to high mortality and morbidity. Carotidatherosclerosis notably causes cerebral ischemia and infarction(stroke) [1]. The World Health Organization reported that 17.5million people (30% of all global deaths) died from cardiovascu-lar diseases in 2005 [2]. Heart attack and stroke caused 7.6 and 5.7million of these deaths, respectively.

Blood flow characteristics in an artery is closely related tothe progression of atherosclerosis. Once atherosclerotic plaquesare developed in a blood vessel, the blood flow is significantlydisturbed by the local contraction of the vessel diameter. Thisdisturbance is characterized by high shear stress at the stenosisapex, flow separation, vortex shedding, and turbulent transition atthe downstream region of the stenosis. These hemodynamic envi-ronments influence platelet activation, as well as aggregation and

∗ Corresponding author at: Center for Biofluid and Biomimic Research, Depart-ment of Mechanical Engineering, Pohang University of Science and Technology, San31, Hyoja-dong, Pohang 790-784, Republic of Korea. Tel.: +82 54 279 2169;fax: +82 54 279 3199.

E-mail address: [email protected] (S.-J. Lee).

rupture of atherosclerotic plaque. Therefore, better understandingof the hemodynamic characteristics in a stenosis is important inidentifying better diagnosis and clinical treatments of atheroscle-rosis.

In vitro experiments using artificial flow phantoms are effectivefor investigating the fluid mechanical aspects of the circulatory sys-tem without ethical and safety problems arising from animal andhuman experiments. Giddens [3,4] investigated mean flow fieldand frequency contents as functions of Reynolds number (Re) inboth steady and pulsatile flows using in vitro stenosis models. Thesinusoidal shape of stenosis with 50% reduction in diameter (75%reduction in area) used by Giddens became a canonical stenosismodel. The mean and fluctuating flow fields in this model havebeen widely investigated as baseline flow [5–8]. Vétel et al. [9] car-ried out in-depth study of asymmetry and transition to turbulencein a smooth axisymmetric stenosis using both stereoscopic parti-cle image velocimetry (PIV) and time-resolved PIV technique. Jetdeflection toward the wall due to Coanda effect, flow asymmetrydownstream of the stenosis, and unsteady turbulent flow above acritical Reynolds number (Re) have been well characterized throughin vitro experiments.

Although most previous studies have employed fully devel-oped Poiseuille flow with a parabolic velocity profile as an inletcondition, evidence shows that blood flow in an artery has a spi-ral corkscrew pattern. For example, Stonebridge and Brophy [10]

1350-4533/$ – see front matter © 2013 IPEM. Published by Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.medengphy.2013.10.008


120 H. Ha, S.-J. Lee / Medical Engineering & Physics 36 (2014) 119– 128

reported the existence of a spiral blood flow in human femoralartery. Numerous reports also support that the spiral flow is a nor-mal physiological flow phenomenon in circulatory system [11–14].The three-dimensional (3D) complex squeezing motion of the heartand the tapered and curved geometrical configurations with rifledendoluminal surfaces of arteries are suspected to induce spiralflows. Despite abundant evidence of spiral flows, their exact rolesin arteries are not fully understood to date.

Understanding the detailed mechanism of spiral flow in the cir-culatory system requires an in-depth study of their fluid dynamiccharacteristics. Computational fluid dynamics (CFD) techniqueshave been used to investigate the spiral flow effects on the steno-sis and sudden expansion channel [15–17]. According to previousnumerical simulations, the spiral flow provides relatively uniformdistribution of wall shear stress, as well as reduces flow stagna-tion, separation, and instability. Experimental studies are relativelylacking compared with numerical simulations.

The main objective of the present study is to investigate theeffect of spiral blood flow in a stenosis model with 50% reduction indiameter by means of experimental flow measurement techniques,especially PIV technique. This research specifically focuses on thechanges of the recirculation length, location of turbulent transition,and turbulence intensity with regard to the swirling intensity of theinlet flow.

2. Materials and methods

2.1. Fabrication of stenosis model and spiral inserts

The stenosis model depicted in Fig. 1A was created using thefollowing cosine-form formula [18]:

r(z)R

= 1 − ıc

[1 + cos

(z�

D

)], − D ≤ z ≤ D (1)

where R and D are the radius and diameter of the normal vessel,respectively; r and z represent the radial and axial coordinates;parameter ıc denotes the percentage of the vessel constriction. Theıc of this model is 0.25, indicating 50% and 75% reductions in thediameter and cross-sectional area, respectively. The total length ofthe stenosis model is 220 mm (22D), where diameter D = 10 mm.The length of the upstream and downstream of the stenosis is 10D,and the length of the stenosis region is 2D.

Fig. 1B shows a schematic diagram of the spiral insert. The spiralinsert has the same diameter as the stenosis channel (D = 10 mm).The length of the insert is 5D. The spiral insert has a four-arm type ofspiral structure having 0.3D of inner core and four fan-shaped armswith 45◦ angle. Two spiral inserts (spiral types I and II) with differ-ent helical pitches were designed to change the swirling intensityof the flow. Helical pitches of spiral types I and II are 10D and 10/3D,respectively. The spiral insert was placed at the inlet of the chan-nel, after which it produced spiral flow 5D upstream of the stenosisregion.

The stenosis model and spiral inserts were fabricated using acry-lonitrile butadiene styrene (ABS) thermoplastic using a 3D printer(Fortus 400mc, Stratasys). Then, the silicone mold of the steno-sis model was made using a silicone rubber compound (Silastic3481, Dow Corning). The ABS stenosis model was removed fromthe silicone mold by slicing the mold open. The mold was thenrefilled with a low melting-point alloy metal (melting point 70 ◦C)to produce a metal replica of the original stenosis model. Thefabricated metal replica was lightly sanded and casted using a poly-dimethylsiloxane (PDMS) silicone compound. The metal replicawas then melted out from the PDMS stenosis channel. The finalmodel was washed with 7% nitric acid solution for 5 h and with DI

water for 30 min to remove any debris that adhered on the channelsurface.

2.2. Flow circuit system

Fig. 2A shows a schematic diagram of the experimental setupused in this study. A total of 5 L working fluid was prepared in anacryl reservoir; a 15 W centrifugal pump circulated the fluid at aconstant flow rate through the circuit. As a fluidic low-pass filter, a0.5 L air container was installed using a three-way connector tostabilize possible fluctuations of a flow rate [19,20]. A variable-area type flow meter (Visi-Float®, Dwyer instruments, Inc.) wasinstalled to measure the flow rate after properly re-calibrating withthe working fluid. The flow rate can be controlled by the internalfluid valve of the flow meter.

Silicone tubes 11 mm in internal diameter were connected tothe inlet and outlet of the PDMS stenosis channel model. The inletpart was connected to a 1 m-long straight tube coaxially alignedwith the stenosis channel to establish a fully developed laminarflow before entering the stenosis channel. The outlet tube was con-nected vertically because camera B (Fig. 2B) was positioned to takeimages of the cross-sectional flow through the outlet window. Theflow phantom was placed in the acryl container filled with index-matched working fluid to remove optical distortion at the surfaceof the model due to the difference of the refractive indexes betweenthe stenosis model and the surrounding fluid.

2.3. Working fluid

The blood mimicking working fluid in this study was preparedaccording to Yousif et al. [21]. First, a mixture of glycerol and watermixture (44:56 by weight) was prepared, and then 15% (by weight)of sodium iodide was dissolved. Based on the Abbe refractometer(ATAGO, Japan), the refractive index of the resultant working fluidis 1.4130 ± 0.0005. This refractive index matches that of the presentPDMS stenosis phantom well. In addition, the dynamic working vis-cosity of the fluid is 4.30 ± 0.05 cP, which lies within the range ofhuman blood viscosity (4.4 ± 0.6 cP) [21]. The working fluid wasseeded with silver-coated hollow glass spheres (Conduct-O-FillSH400S20 silver hollow, Potters Industries, Inc.) with mean diam-eters of dp ∼ 13 �m and then circulated through the flow loop. Theindex-matching enclosure was filled with unseeded working fluid.

2.4. PIV measurement

As shown in Fig. 2B, Q-switched double-pulse Nd:Yag laser(Gemini, New wave) generated a thin laser sheet to illuminatethe measurement plane. The 4.2-megapixel high-resolution 14-bitcharge-coupled device camera (PCO2000, PCO) was synchronizedwith the laser pulses to run at frame rates of up to 10 Hz. Up to75 instantaneous velocity fields were obtained from a set of imageacquisitions. Measurements were repeated six times to obtain atotal of 450 instantaneous velocity fields. The obtained velocityfields were statistically analyzed to obtain their mean and fluc-tuation components. Each instantaneous velocity vector U(t) [Eq.(2)] in steady flow can be decomposed into a time-averaged meanvelocity component U [Eq. (3)] and fluctuating velocity componentu’(t).

U(t) = U + u′(t) (2)

U = 1N

N∑0

U(t) (3)


H. Ha, S.-J. Lee / Medical Engineering & Physics 36 (2014) 119– 128 121

Fig. 1. Schematic diagrams of the stenosis model and spiral inserts. (A) Schematic design of stenosis phantom; (B) geometric shape of spiral insert (left) and cross-section(right) and (C) photographs of silicone stenosis phantom (left) and spiral inserts (right).

where N is the total number of instantaneous velocity vectors. Inaddition, root mean square (RMS) velocity fluctuation URMS is esti-mated to represent the fluctuation component [Eq. (4)].

URMS =

√√√√ 1N

N∑1

(U(t) − U)2

(4)

2.4.1. Horizontal plane measurementThe laser light sheet with 0.5 mm thickness illuminated the hor-

izontal center plane of the stenosis model. Camera A was used tomeasure the longitudinal direction of the flow. A cross-correlationPIV algorithm was applied to the acquired flow images to extractthe instantaneous velocity fields. The multi-grid interrogation win-dow scheme was adopted, and the size of the interrogation windowwas 96 × 96, 64 × 64, and 64 × 32 pixels, with 50% overlapping. Thetime interval �t was typically between 100 and 600 �s dependingon flow rates. The distance between two adjacent velocity vectorswas 32 pixels along the horizontal axis and 16 pixels along the ver-tical axis, which corresponded to 0.52 and 0.26 mm, respectively.

2.4.2. Cross-sectional plane measurementThe laser light sheet illuminated the cross-sectional plane of

the model. Camera B was used to measure secondary motion ofthe flow. The thickness of the laser sheet for imaging the sec-ondary motion of the flow was 2 mm. The interrogation windowswere 32 × 32, 16 × 16, and 12 × 12 pixels, with 50% overlapping.The time interval �t was typically between 0.1 and 4 ms. The dis-tance between the adjacent velocity vectors was 6 pixels, whichcorresponded to 0.28 mm.

2.4.3. Uncertainty analysisSee materials and methods section in Supplementary material.

2.5. Bubble-trapping test

See materials and methods section in Supplementary material.

2.6. Streakline visualization

See materials and methods section in Supplementary material.

Fig. 2. Schematic diagrams of the experimental set-up. (A) Flow circuit system and (B) PIV velocity field measurement system.



3. Results

3.1. Poiseuille flow through the stenosis

As baseline flows, fully developed Poiseuille flows at Re = 256,446, and 645 are given upstream of the stenosis model. Velocityfields of the flows through the stenosis are measured using PIV mea-surement technique. The time-averaged velocity and RMS velocityfluctuations are shown in Fig. 3. The flow through the stenosisconsists of four flow regions: laminar flow, accelerated jet flow,turbulent flow, and re-laminarization regions. The parabolic veloc-ity profile of the fully developed flow is changed to high-speedjet flow with a blunt velocity profile as this flow is acceleratedthrough the stenosis. The velocity profile almost returns to its ini-tial shape at X/D > 7. However, perfect recovery is not observeduntil X/D = 8. At all flow rate conditions tested in this study, thejet flow emanating from the stenosis is slightly deflected towardthe channel wall, causing asymmetry of the recirculation regionat the post-stenosis region because of the Coanda effect. The con-tour map of Urms (Fig. 3B) shows where the turbulent transitionof the flow occurs. At Re = 256, velocity fluctuations are very small(<30 mm/s) because the turbulent transition does not occur at thisRe. The steady laminar flow is maintained all over the channel. AtRe = 446 and 645, significant increase of velocity fluctuations (up to50% of its mean velocity) at the post-stenosis region occur becauseof the turbulent transition. This increment is recovered as the flowis stabilized in the re-laminarization region at X/D > 7. Slight flowinstability is observed along the shear layer separated from thestenosis apex due to Kelvin–Helmholtz instability. The combina-tion of the Kelvin–Helmholtz instability and the high measurementerrors of PIV technique causes abnormally high Urms at the apex ofthe stenosis. Therefore, the Urms at the stenosis apex regions areexcluded from further analysis.

3.2. Swirling intensity

In this study, six inlet conditions are generated by combiningtwo different spiral inserts (spiral types I and II) and three Re num-bers (Re = 256, 446, and 645). To analyze the swirling flow inducedat the inlet of the stenosis, the cross-sectional velocity fields aremeasured at X/D = −1. A typical result for the spiral type I insertat Re = 446 is shown in Fig. 4A. The cross-sectional velocity fieldshows swirling velocity components around the center axis. Thetangential velocity of the swirling flow is an order of magnitudesmaller than the axial velocity. To characterize the swirl intensity ofthe flow quantitatively, swirl number S defined as non-dimensionalangular momentum flux, is estimated as described by Kitoh [22].

S =2��

∫ R

0r2vxv�dr

��R3U2ref

(5)

where R is the radius of the channel, Uref is the bulk velocity, vx

is the axial velocity obtained from the horizontal plane PIV mea-surements with axisymmetry assumption, and v� is the swirlingvelocity components obtained from the cross-sectional plane PIVmeasurements. As shown in Fig. 4B, the swirl intensity S is highlydependent on the helical pitch of the spiral inserts (S ≈ 0.25 for spi-ral type I, S ≈ 0.6 for spiral type II). The effect of Re number on theswirl intensity is minimal.

3.3. Effect of swirling inlet condition on the flow through thestenosis

Fig. 5 shows the effect of spiral inlet condition on the flow fieldthrough the stenosis at Re = 256. The velocity vector field showsthat the parabolic inlet velocity profiles are slightly distorted as

the swirling increases. Compared with the velocity field for thePoiseuille flow, spiral type I causes a higher degree of jet deflectiontoward the wall. In the case of spiral type II, the flow jet emanat-ing from the stenosis apex is highly disturbed by the out-of-planevelocity components (1.5 < X/D < 3). The Re number is under the crit-ical Re for the turbulent transition, so any significant changes on theUrms are not observed, except for the slight increase at 2 < X/D < 3 inthe case of spiral type II (Fig. 4b).

At Re = 446, the effects of spiral inlet condition on the flow fieldthrough the stenosis are more significant. Compared with the jetlength of Poiseuille flow (X/D ≈ 6) (Fig. 6A), the spiral flow reducesthe length of the jet flow emanating from the stenosis apex. Thelength of the jet flow with spiral types I and II are X/D ≈ 3.5 and 1.5,respectively. The turbulent region of the flow, distinguished by thehigh-Urms region, approaches the stenosis apex as the spiral inten-sity increases (Fig. 6B). The shortened jet flow and early breakoutof the turbulent region are shown at Re = 645 (Fig. 7).

To analyze the effect of the spiral inlet flow on the breakoutof the turbulent flow, the radial averaged Urms at Re = 256, 446,and 645 are plotted and the locations of Urms maximum points arecompared in Fig. 8. At Re = 256, no significant variations of Urms areobserved because the flow remains laminar and turbulent transi-tion does not occur (Fig. 8A). At Re = 446 and 645, the flow exhibitsa clear peak point of Urms (Fig. 8A and B). Compared with Poiseuilleflow, the location of maximum Urms significantly approaches thestenosis apex when the spiral flows are increased at Re = 446 and645, as shown in Fig. 8D.

The early breakout of the turbulent flow by spiral flow influencesthe region of recirculation flow. The length of the recirculationregion is investigated using streakline flow visualization. Fig. 9Ashows typical streakline images of the flow passing through thestenosis at Re = 446 (streakline images for other flow conditionsare depicted in the Supplementary Data). The swirling motions ofthe flow at the inlet region are shown at spiral types I and II. Theswirling inlet flow generates high disturbance at the post-stenosisregion and reduces the length of the recirculation region. The effectof spiral flow on the reduction of the recirculation region is com-pared in Fig. 9B. According to the result, the effect of spiral flowon the reduction of the recirculation region is statistically signifi-cant (p < 0.01) except at spiral type I with Re = 446. The spiral typeII insert shows better performance on the recirculation reductioncompared with the spiral type I insert.

3.4. Variation of swirling flow along the stenosis

To investigate the variation of swirling flow during the turbu-lent transition, cross-sectional velocity fields at Re = 446 for spiraltype I are measured along the axial direction. The center-plane flowstructure through the stenosis is depicted in Fig. 10A and comparedwith the cross-sectional flow information. The laminar spiral flowis formed at the inlet and changes to turbulent flow at X/D ≈ 3. Asshown in Fig. 10B, the swirling flow induced by the spiral insertremains stable upstream of the stenosis (X/D = −1). The flow retainsits swirling component up to X/D = 1, whereas the swirling flowis focused around the center of the channel. The swirl center ofthe flow moves slightly downward because of the jet deflection atX/D = 2. Flow instability occurs on the interface between the asym-metric center core of the swirling jet flow and the surroundingrecirculation flow region. At X/D = 3, due to the turbulent transition,the cross-sectional flow is highly disturbed, with large flow fluctu-ations. The streakline image also shows random and irregular fluidmotion of the flow at the turbulent transition.

3.5. Bubble trapping at the post-stenosis region

Bubble infusion test is performed using the stenosis model toinvestigate the effect of the spiral flow on the material deposition



Fig. 3. Velocity fields of the Poiseuille flow through the stenosis model at Re = 256, 446, and 645. (A) Time-averaged mean velocity fields. (B) Contours of RMS velocityfluctuations. Only one quarter of the axial velocity vectors is presented for clarity.

at the stenosis. Fig. 11A shows the temporal process of bubble accu-mulation at the recirculation flow region behind the stenosis atRe = 446. For the case of Poiseuille flow, bubbles are easily cap-tured at the post-stenosis region because of the large size of therecirculation zone. Conversely, bubbles are not easily depositedpost-stenosis when the spiral flow is given because of the effectof spiral flow on the reduced recirculation zone and the high flowinstability at the post-stenosis region. In this region, the net direc-tion of the fluid shearing stress is exerted on the forward direction,so any bubble deposition is effectively inhibited. Fig. 10B showsthat higher swirling reduces the number of bubbles trapped at thestenosis channel.

4. Discussion

This study focused on investigating the effect of spiral inletflow on flow characteristics through a stenosis model with 50%reduction in diameter. Spiral inserts with two different helicalpitches (10D and 10/3D) were installed upstream of the stenosisto induce swirling flows, and corresponding flow fields were

experimentally analyzed using PIV velocity field measurementtechnique and streakline flow visualization. Specifically, the recir-culation length, location of turbulent transition, and turbulenceintensity depending on the swirling intensity were investigatedbecause of their pathological relevance.

Although many numerical simulation studies on spiral flowthrough a stenosis have been conducted, experimental studies arevery limited. Peterson and Plesniak [23] investigated the effect ofasymmetric mean inlet velocity profile and secondary Dean flowon the flow physics downstream of a stenosis model. Their resultsshowed that asymmetric inlet velocity profile reduced the region ofinfluence of the stenosis (∼30%) by forcing the stenotic jet towardthe channel wall via non-uniform radial pressure gradient. In theirresults, the curvature-induced secondary flow had a minor role inthe stenosis. The secondary Dean flow induced by tube curvatureseems to be similar to the present spiral flow. The spiral flow inthe present study produced a strong swirling flow that circulatedaround the tube axis. This swirling flow shortened the location ofturbulent transition by more than 70% compared with the normalPoiseuille inlet flow. This effect indicates that the swirling flow

Fig. 4. Typical velocity field and swirl intensity of the flow induced by spiral inserts. (A) Streaklines (left) and velocify field (right) at Re = 446. Insert of spiral type I is usedfor swirling generation. (B) Variations of swirling intensity. Data represent mean ± SD.



Fig. 5. Effect of spiral inlet condition on the velocity field of the flow through the stenosis model at Re = 256. (A) Time-averaged mean velocity vector fields. (B) Contours ofRMS velocity fluctuations. Only one quarter of the axial velocity vectors is presented for clarity.

causes more significant results compared with the asymmetric inletvelocity profile with the secondary Dean flow. Although the axialvelocity profiles in this study are also slightly deviated from theparabolic shape because of the inherent features of the spiral-insertdriven flow, the skewness of the velocity profiles are very minor.Therefore, its effect seems negligible compared with the swirlingeffect.

The spiral inlet flows presented two important fluid-dynamicchanges in the flow through the stenosis: early breakout of tur-bulent transition of the flow and reduction of the recirculationflow region. These resultant flow characteristics can cause a ben-eficial effect on the stenosis blood vessel. The early breakout ofturbulent transition induces the faster mixing and re-distributionof fluid momentum, which causes the forward shift of the flow

Fig. 6. Effect of spiral inlet condition on the velocity field of the flow through the stenosis model at Re = 446. (A) Time-averaged mean velocity vector fields. (B) Contours ofRMS velocity fluctuations. Only one quarter of the axial velocity vectors are presented for clarity.



Fig. 7. Effect of spiral inlet condition on the velocity field of the flow through the stenosis model at Re = 645. (A) Time-averaged mean velocity vector fields. (B) Contours ofRMS velocity fluctuations. Only one quarter of the axial velocity vectors are presented for clarity.

re-establishment (re-laminarization). This result indicates that thelength of flow disturbance caused by the stenosis is reduced bythe spiral flow, which eventually reduces the region of potentialdamage by the turbulence on red blood cells and endothelial cells[24]. The reduction of the recirculation flow region is related toplatelet aggregation and thrombosis at the stenosis. The recircula-tion region has relatively low shearing stress compared with thesurrounding region, so platelets and other thrombogenic proteins

frequently adhere on the recirculation region [25,26]. In this study,we hypothesized that the spiral flow can reduce deposition ofthrombogenic materials by shortening the length of the recircu-lation flow region. To support this hypothesis, the bubble-trappingtest was carried out as shown in Fig. 11. The results showed thatbubble deposition post-stenosis is significantly reduced by thespiral flows, which also reduce the recirculation flow region. Net-forward directional turbulent flow with high velocity fluctuations

Fig. 8. Variation of RMS velocity fluctuations at (A) Re = 256, (B) Re = 446, and (C) Re = 645; (D) Variation of the peak location of RMS velocity fluctuations. Data representmean ± SD.



Fig. 9. The length of the recirculation region at different spiral flows. (A) The streakline shows the recirculation region of the flow at the post-stenosis region (Re = 446). Theregion of the streakline image is marked as the gray dotted region of the stenosis model. (B) Length of the recirculation region at Re = 256, 446, and 645. *P < 0.01 comparedwith Poiseuill flow at the same Re as a control group. Data represent mean ± SD.

replaces the recirculation flow. Although the bubbles in this studydo not fully represent the dynamic behaviors of biological mate-rials, the significantly high shearing force effectively inhibits anymaterial deposition at the post-stenosis region. This analysis sup-ports the reduction of platelet adhesion in a glass-tube surface [16]and sudden expansion tube [27] when swirling flow is induced.

Since the existence of spiral flow is observed in vivo bloodflow, many clinical studies reported successful employment ofarterial grafts which induce spiral flow as an alternative of con-ventional non-spiral grafts. Jabrome et al. implanted both of plaincircular and spiral ePTFE grafts on their pig arteriovenous graftmodel [28]. While the normal blood flow in native arteries is spirallaminar flow, conventional vascular graft does not produce spi-ral laminar flow and thus it is suspected as one of the reasons formalfunction and failure of arteriovenous grafts. Therefore, intimahyperplasia formation is expected to be reduced when spiral flowgrafts are employed [29]. Following successful performance of spi-ral flow prosthetic grafts in animal studies, recently, first-in-manstudy of spiral laminar flow prosthetic bypass graft is reportedby Stonebridge et al. [30]. These studies show potential of spiralflow-enhancing prosthetic grafts for improving the patency rate ofarteriovenous grafts [30].

Recently, a different type of a helical design was introduced tostents and vascular grafts. It is characterized by a circular cross-section twisted along a helical curve like a spring coil. Thesehelical-shaped stents and vascular grafts were demonstrated toreduce in-stent restenosis [31]. Caro et al. [32] showed that even

small amplitude helical-shaped bypass grafts significantly reducethrombosis and intimal hyperplasia compared with conventionalshunts. While their design is also known to induce swirling flow,however, their results seem to be more related to the effect ofspring-type helical graft on intra-luminal mixing of blood pro-teins, and oxygen mass transfer between blood and vessel wall [33].Therefore, fluid-dynamic behaviors in their bypass grafts should beclearly distinguished from the present work since their geometricalconfigurations are different.

Stonebridge et al. [10,13] first reported the concept of spiralblood flow from an angioscopic examination of blood-flow pat-terns in femoral arteries. In their in vivo observations, the rotationalvelocity of spiral flow was estimated to be around one sixth offorward axial velocity [12]. Average ratios of the rotational veloc-ity to the axial velocity are estimated to be one tenth (spiral typeI) and one half (spiral type II) within the physiologically observ-able swirling range, so the fluid-dynamic changes induced by spiralflows through the stenosis investigated in this work are physiolog-ically relevant.

Paul and Larman [15] and Linge et al. [17] numerically simulatedthe spiral blood flow through arterial stenosis with steady and pul-satile flow conditions using CFD techniques. They employed thestandard �-� model for simulation of the blood flow at Re = 500and 1000. Their results show that spiral flow reduced the tur-bulent kinetic energy, turbulent intensity, and wall shear stressthrough the stenosis as it induced rotational stabilities in theforward flow. However, experimental results did not show any

Fig. 10. Cross-sectional velocity field of the flow through the stenosis model. (A) Schematic diagram of laminar spiral flow to turbulent flow at Re = 446 for spiral type I. (B)Variation of cross-sectional vector field at various axial locations. The corresponding streakline images are shown at the left inset.



Fig. 11. Bubble trapping at the post-stenosis region. (A) Temporal process of bubble accumulation post-stenosis at Re = 446. (B) Cumulative number of trapped bubbles.

significant reduction of the turbulent intensity by spiral flows, asshown in Fig. 8. In addition, the cross-sectional flow measurementshows that the swirling flow post-stenosis is easily disturbed bythe recirculation flow and turbulent transition (Fig. 9), whereas thenumerical simulation expected that the spiral flow is maintainedup to X/D = 20. These gaps are due to the inadequate performance ofconventional turbulence models for predicting turbulent transitionof the flow through the stenosis.

Sherwin and Blackburn [6] applied linear stability analysis anddirect numerical simulation to investigate 3D instabilities andtransition of steady and pulsatile flows through an axisymmet-ric stenosis. In their simulation, a Coanda-type wall attachmentand turbulent transition occurred at a critical Re of 722. However,the present study shows that the flow through the stenosis hashigh velocity fluctuations and the turbulent transition is causedat Re ≥ 446. This result coincides with the previous experimentalwork of Vétel et al. [9], which estimated the critical Re ∼ 400. Inthe numerical simulation of Sherwin and Blackburn [6] and Vargh-ese et al. [7], small geometric perturbation and upstream noise ofthe flow rate, which can be included in the experimental study, aresuspected to influence the turbulent transition of the flow.

The present PIV measurement technique provides only 2D pla-nar velocity data. In-depth study of flow structures, such as 3Drecirculation and vortex pattern, could not be fully understood. Infurther studies, stereo PIV and volumetric PIV, which measure 3Dvelocity vector fields, will be helpful for understanding the com-plex fluid structures. In addition, the effect of spiral flow through anasymmetric stenosis and the effect of pulsatile spiral flow throughthe stenosis also warrant future studies.

5. Conclusions

The effect of spiral flow in a stenosis model was experimen-tally investigated using PIV velocity field measurement techniqueand streakline flow visualization. The spiral inserts with two differ-ent helical pitches (10D and 10/3D) installed at the upstream of the

stenosis caused swirling flows. The results show that the spiral flowsignificantly reduces the length of the recirculation flow region andenhances the early breakout of turbulent transition. However, tur-bulence intensity variation depending on the swirling intensity wasnot observed. The present findings show that naturally or artificiallyinduced spiral flow has several beneficial effects in a stenosis ves-sel. In the future, studying spiral flow effects through the stenosiswill contribute in attaining better understanding of the hemody-namic characteristics of atherosclerosis and in discovering betterdiagnosis procedures and clinical treatments.

Ethical approval

Not required.

Acknowledgment

This work was supported by the National Research Foundationof Korea (NRF) under a grant funded by the Korea government(MSIP) (No. 2008-0061991).

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.medengphy.2013.10.008.

Conflict of interest

There are no conflicts of interest to declare.

References

[1] Timsit SG, Sacco RL, Mohr JP, Foulkes MA, Tatemichi TK, Wolf PA, et al. Early clin-ical differentiation of cerebral infarction from severe atherosclerotic stenosisand cardioembolism. Stroke 1992;23:486–91.



[2] World-Health-Organization. Prevention of cardiovascular disease: guidelinefor assessment and management of total cardiovascular disease risk. Geneva:Organization WH; 2007.

[3] Ahmed SA, Giddens DP. Pulsatile poststenotic flow studies with laser Doppleranemometry. Journal of Biomechanics 1984;17:695–705.

[4] Khalifa A, Giddens D. Characterization and evolution of poststenotic flow dis-turbances. Journal of Biomechanics 1981;14:279–96.

[5] Ojha M, Cobbold RS, Johnston KW, Hummel RL. Pulsatile flow throughconstricted tubes: an experimental investigation using photochromic tracermethods. Journal of Fluid Mechanics 1989;203:173–97.

[6] Sherwin S, Blackburn HM. Three-dimensional instabilities and transition ofsteady and pulsatile axisymmetric stenotic flows. Journal of Fluid Mechanics2005;533:297–327.

[7] Varghese SS, Frankel SH, Fischer PF. Direct numerical simulation ofstenotic flows. Part 1. Steady flow. Journal of Fluid Mechanics 2007;582:253–80.

[8] Varghese SS, Frankel SH, Fischer PF. Direct numerical simulation of stenoticflows. Part 2. Pulsatile flow. Journal of Fluid Mechanics 2007;582:281–318.

[9] Vétel J, Garon A, Pelletier D, Farinas M. Asymmetry and transition to tur-bulence in a smooth axisymmetric constriction. Journal of Fluid Mechanics2008;607:351–86.

[10] Stonebridge P, Brophy C. Spiral laminar flow in arteries? The Lancet1991;338:1360–1.

[11] Karino T, Goldsmith HL, Motomiya M, Mabuchi S, Sohara Y. Flow patterns invessels of simple and complex geometriesa. Annals of the New York Academyof Sciences 1987;516:422–41.

[12] Stonebridge P, Buckley C, Thompson A, Dick J, Hunter G, Chudek J, et al. Nonspiral and spiral (helical) flow patterns in stenoses: in vitro observations usingspin and gradient echo magnetic resonance imaging (MRI) and computationalfluid dynamic modeling. International Angiology 2004;23:276–83.

[13] Stonebridge P, Hoskins P, Allan P, Belch J. Spiral laminar flow in vivo. ClinicalScience (London, England: 1979) 1996;91:17.

[14] Uchida Y, Nakamura F, Tomaru T, Morita T, Oshima T, Sasaki T, et al. Predictionof acute coronary syndromes by percutaneous coronary angioscopy in patientswith stable angina. American Heart Journal 1995;130:195–203.

[15] Paul MC, Larman A. Investigation of spiral blood flow in a model of arterialstenosis. Medical Engineering and Physics 2009;31:1195–203.

[16] Zhan F, Fan Y, Deng X. Swirling flow created in a glass tube suppressed plateletadhesion to the surface of the tube: its implication in the design of small-caliberarterial grafts. Thrombosis Research 2010;125:413–8.

[17] Linge F, Hye MA, Paul MC. Pulsatile spiral blood flow through arterial stenosis.Computer Methods in Biomechanics 2013:1–11.

[18] Ahmed SA, Giddens DP. Velocity measurements in steady flow through axisym-metric stenoses at moderate Reynolds numbers. Journal of Biomechanics1983;16:505–16.

[19] Kang YJ, Yang S. Fluidic low pass filter for hydrodynamic flow stabilization inmicrofluidic environments. Lab on a Chip 2012;12:1881–9.

[20] Lee J, Rahman F, Laoui T, Karnik R. Bubble-induced damping in displacement-driven microfluidic flows. Physical Review E 2012;86:026301.

[21] Yousif MY, Holdsworth DW, Poepping TL. Deriving a blood-mimicking fluid forparticle image velocimetry in sylgard-184 vascular models. In: EMBC AnnualInternational Conference of the IEEE. 2009. p. 1412–5.

[22] Kitoh O. Experimental study of turbulent swirling flow in a straight pipe. Journalof Fluid Mechanics 1991;225:445–79.

[23] Peterson SD, Plesniak MW. The influence of inlet velocity profile and secondaryflow on pulsatile flow in a model artery with stenosis. Journal of Fluid Mechan-ics 2008;616:263–301.

[24] Ku DN. Blood flow in arteries. Annual Review of Fluid Mechanics1997;29:399–434.

[25] Bluestein D, Gutierrez C, Londono M, Schoephoerster RT. Vortex sheddingin steady flow through a model of an arterial stenosis and its relevanceto mural platelet deposition. Annals of Biomedical Engineering 1999;27:763–73.

[26] Bluestein D, Niu LJ, Schoephoerster RT, Dewanjee MK. Fluid mechanics of arte-rial stenosis: relationship to the development of mural thrombus. Annals ofBiomedical Engineering 1997;25:344–56.

[27] Zhan F, Fan Y, Deng X, Xu Z. The beneficial effect of swirling flow on plateletadhesion to the surface of a sudden tubular expansion tube: its poten-tial application in end-to-end arterial anastomosis. Asaio Journal 2010;56:172–9.

[28] Jahrome AK, Hoefer I, Moll FL, Houston GJ, Stonebridge PA, Blankestijn PJ, et al.An animal paired crossover ePTFE arteriovenous graft model. Annals of SurgicalInnovation and Research 2010;4:1–4.

[29] Jahrome OK, Hoefer I, Houston GJ, Stonebridge PA, Blankestijn PJ, Moll FL,et al. Hemodynamic effects of spiral ePTFE prosthesis compared with standardarteriovenous graft in a carotid to jugular vein porcine model. The Journal ofVascular Access 2010;12:224–30.

[30] Stonebridge PA, Vermassen F, Dick J, Belch JJ, Houston G. Spiral laminar flowprosthetic bypass graft: medium-term results from a first-in-man structuredregistry study. Annals of Vascular Surgery 2012;26:1093–9.

[31] Shinke T, Robinson K, Burke MG, Gilson P, Mullins LP, O’Brien N, et al. Novelhelical stent design elicits swirling blood flow pattern and inhibits neointimaformation in porcine carotid arteries. In: Poster Presentation, American HeartAssociation Meeting. 2008. p. 8–12.

[32] Caro CG, Cheshire NJ, Watkins N. Preliminary comparative study of small ampli-tude helical and conventional ePTFE arteriovenous shunts in pigs. Journal of theRoyal Society Interface 2005;2:261–6.

[33] Zheng T, Wen J, Jiang W, Deng X, Fan Y. Numerical investigation of oxygen masstransfer in a helical-type artery bypass graft. Computer Methods in Biomechan-ics 2012:1–11.

epapepapffect of swirling inlet condition on the _ow _eld in a stenosed

Documents

spiral blood

spiral inserts

pohang university of

atherosclerosis progression

swirling ows

authors institutionand

cases authors

republic of koreaa r