Shear Stress Investigation Across Mechanical Heart Valve

Zhang, Pei*; Yeo, Joon Hock†; Qian, Ping*; Hwang, Ned H. C.‡

doi: 10.1097/MAT.0b013e318157c093
Biomedical Engineering

The particle image velocimetry technique was used to study the shear field across a transparent mechanical heart valve model in one cardiac cycle. Shear stress was continuously increased until peak systole and high turbulent stress was observed at the orifice of the central channel and also around the occluder trailing tips. The peak Reynolds shear stress was up to 500 N/m2 at peak systole, which was higher than the normal threshold for hemolysis.

From the *School of Mechanical and Automation Engineering, Shanghai Institute of Technology, Shanghai, China; the †School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore; and the ‡Division of Medical Engineering, National Health Research Institute, Taiwan.

Submitted for consideration May 2007; accepted for publication in revised form July 2007.

Reprint Requests: Pei Zhang, PhD, School of Mechanical and Automatic Engineering, Shanghai Institute of Technology, 120 Cao Bao Road, Shanghai, China, 200233.

Article Outline

The flow field past a mechanical heart valve (MHV) is quite complicated due to the complex geometry, elastic walls, and moving boundary of the valve occluder. The classic point measurement techniques such as hot wire anemometry or laser Doppler anemometry (LDA) do not provide spatial disturbance and cycle-resolved full field velocity information.1 Particle image velocimetry (PIV), as a powerful planar measurement method, is suitable for analyzing time-dependent and turbulent flow structures around an MHV.1

Of special interest in MHV study is the link between flow dynamics and complications of hemolysis and/or thrombosis.2 The high shear stress and flow stagnation induced by MHV dynamics cause blood damage and affect coagulation. For typical exposure time through heart valves (about 1 ms), Sallam and Hwang3 showed that a turbulent stress above 400 N/m2 would cause hemolysis, and sublethal damage to red cells can occur at a turbulent shear stress level of 50 N/m2.

Many studies have investigated the forward flow through MHV as a source of hemolysis. Yoganathan et al.4 conducted the first LDA investigation of turbulent stress in the peak forward flow field. The peak downstream stress reached between 120∼200 N/m2 for different valves. Lim et al.5 used the PIV technique to obtain velocity vector fields and Reynolds stress mapping in the aortic root region. The bioprostheses showed the highest Reynolds stress reached 209 N/m2. Similar investigation of the porcine bioprosthetic valve was continued under pulsatile flow conditions,6 with downstream stress mappings obtained at different time steps of the systolic phase.

The leakage flow after MHV full closure also attracts much attention due to the high speed leakage jet and elevated turbulent shear stress. Yoganathan et al.4 found that under pulsatile flow conditions, the aortic tilting disk had a peak reverse velocity of 0.28 m/s and a turbulent shear stress of 68 N/m2 10 millimeters upstream of the valve. Baldwin et al.7 conducted two-dimensional LDA measurements proximal to the MHVs in an electrical left ventricular assist device. The peak turbulent shear stress was detected 1 mm upstream of the valves and 0.6 mm from the adjacent wall, with a peak value between 900∼990 N/m2.

Until now, knowledge about flow properties inside a MHV lumen was limited due to the opaque material of the MHV. Although the flow evolutions between two transparent occluders were already provided,8 the shear distribution under this situation still needs elucidation.

This work aims to investigate the turbulent shear field around a transparent MHV model in one complete cardiac cycle, thus identifying regions with possible hemolysis and/or thromboembolism.

Back to Top | Article Outline

Materials and Methods

Pusatile Flow Test Loop

The experimental setup is illustrated by Figure 1. A computer-controlled servo-motor (SEM Ltd, Orpington, Kent, UK) drove an 8 cm diameter piston to produce pulsatile flow in the test loop. The motion curve of the piston can be created, modified, and implemented by a Programmable Transmission System (Quin Company, Workingham, Berkshire, UK). An acrylic box, connected to the piston head, was filled with water and accommodated a collapsible left ventricular (LV) sac. Upstream of the LV, a left atrial chamber was connected to a commercial tilting-disk valve. An acrylic test section simulating the aorta was placed downstream of the LV, and a transparent aortic valve model was installed inside the test section. The afterload resistive and compliance elements placed downstream of the test section return back to the atrium and can be tuned separately to obtain desirable physiologic pressure and flow rate. Two pressure transducers coupled with a calibrated multi-channel amplifier (AM-PACK AP9991, Vivitro System Inc, Victoria, BC, Canada) were used to measure ventricular and aortic pressures immediately upstream of the aortic valve and five aorta diameters downstream, respectively. Figure 2 shows typical pressure waveforms recorded by an oscilloscope (TDS36, Tektronix Inc, Wilsonville, OR), which was roughly acceptable referring to the standard physiological pressures. The heartbeat was at 63 bpm with 40% systolic duration and the mean flow rate was 4.3 liters per minute. These parameters were set at the low limit of the physiological range to protect the transparent MHV model under prolonged testing conditions.

Back to Top | Article Outline

Transparent MHV Model

Separate parts of an acrylic MHV model, as shown in Figure 3, were constructed with computer numerical controlling (CNC) machining and Figure 4 shows an image of a complete valve model. The transparency of the valve material enabled PIV measurements without obstruction, especially the regions in-between the occluders. To approximately match those of an SJM Medical (St. Paul, MN) bileaflet valve, the parts’ length, radius, and thickness were checked by a caliper and the model’s opening and closing angles were calculated from PIV images.

Back to Top | Article Outline

PIV System

A TSI Insight PIV system (TSI Inc, Shoreview, MN) was used in this study. The double pulsed laser from two separate Nd:YAG systems (532 nm, max 50 mJ/pulse, MiniLase III, New Wave Research Inc, Fremont, CA) provided illumination for the flow field plane. CCD camera (1000*1016 pixels) was coupled with a Micro-Nikkor lens f/60 mm for the observation of the flow field. A typical flow field image is shown in Figure 5. The displacement of particle images captured between two light pulses was determined through the evaluation of two PIV recordings by cross-correlation. The velocity vector spacing reaches 1 mm under 50% overlapped Nyquist sampling.

When processing PIV images, a leaflet thickness of 1 mm was used to calibrate the velocity field. The corresponding vector files, created in Insight software, were loaded into Tecplot (version 8.0, Tecplot Inc, Bellevue, WA) to present the PIV data graphically.

Back to Top | Article Outline

Working Fluid and Seeding Particles

The working fluid was made up of 79% saturated aqueous sodium iodide, 20% pure glycerol, and 1% distilled water by volume. This composition yielded a kinematics viscosity of 3.4 cS, matching that of human blood at high shear rates. The refractive index of this fluid was measured as 1.49, the same as that of the acrylic test section, hence eliminating refractive deformation from the light passing through the curved liquid-solid interface. The seeding particles, 10-μm silver-coated glass spheres, were selected due to their good image contrast and suitable particle image diameter.

Back to Top | Article Outline

Triggering Setup

The frame rate of the CCD camera was 30 frames per second, which was grossly insufficient for analyzing the MHV closing process <50 ms; hence, a triggering method was developed. A photoelectric sensor (HPX-NT1, Yamatake Corp, Tokyo, Japan), shown in Figure 6, produced a signal when the MHV started its opening process. The zero instant was thus selected at the valve’s fully closed position. The sensor signal was then inputted to an Intel 82C54 Programmable Interval Counter/Timer (1600 Data Acquisition Board, Keithley Instruments Inc, Cleveland, OH), which produced an adjustable time delay in the signal. The signal with 10-μm-delay step was connected with a synchronizer for triggering data acquisition by PIV. Two consecutive PIV images were captured in one cardiac cycle and 500 cycles of images at this instant were saved for postprocessing.

Back to Top | Article Outline

Shear Stress Calculations for PIV Images

Because of the periodic nature of the flow field, PIV images were analyzed using an ensemble average technique, in which the average information was obtained by considering all the samples at the same instant of various cycles. Each instantaneous axial velocity ui and transverse velocity vi could be divided into mean values U, V, and high-frequency turbulence fluctuations ui′, vi′ as:

where i was the cycle sequence number and N was the total number of cycles.

The flow disturbance introduced by the MHV implantation may cause undesired strong fluid dynamics effects on the blood components. Thus, it is necessary to quantify the level and distribution of the fluid stress when investigating the flow property of a prosthetic heart valve. The Reynolds stress is normally much higher than the corresponding viscous stress.9 In particular, Reynolds shear stress (RSS) τxy demonstrated a strong positive correlation with hemolysis and platelet lysis.3 The measurement of RSS was dependent on the orientation of the reference axis and the maximum obtainable RSS (also known as major RSS) was computed as follows10,11:

Back to Top | Article Outline


Back to Top | Article Outline

Opening Process

At t = 0 as in Figure 7, the two occluders were kept at the fully closed position by the backflow. Meanwhile, the forward flow started to develop in the upstream region, with a magnitude around 0.1 m/s. At the instants of t = 5 ms, 10 ms, 16 ms, 20 ms, and 30 ms, the flow momentum inside the central and side channels was continuously increasing, with peak velocities up to 0.52 m/s, 0.55 m/s, 0.7 m/s, 0.79 m/s, and 1.19 m/s, respectively. For the stress distribution as in Figure 8, the magnitude of turbulent stress was also increasing in the opening process. Relative high shear stress was normally found around region I (proximal to the orifice of central channel) and region II (occluder trailing tips). Region I was also extending its dimension to the downstream with the opening of two occluders.

Back to Top | Article Outline

Fully Opened Stage

The flow momentum across MHV kept strengthening after t = 30 ms until it reached peak systole at t = 50 ms. Afterward, the flow field experienced a deceleration phase, while keeping the fully opened position unchanged for about 240 ms. At the instant of 50 ms, as in Figure 9, the flow velocity further increased, with a maximum value around 1.7 m/s observed at the inlet of the central channel and outlets of the side channels. Whereas at t = 220 ms and t = 270 ms, the three channels had their fluid momentums continuously reducing, and relative large flow velocity was shown instead at the middle-rear section of the central channel.

At the peak systole of t = 50 ms, strong turbulent shear stress was identified at Region I and Region II, with a magnitude reaching 501 N/m2 at Region I, as shown in Figure 10. After peak systole, major RSS decreased continuously, with peak values around 160 N/m2 and 15 N/m2 at the instant of 220 ms and 270 ms, respectively. When the central jet disappeared at t = 270 ms, relative strong shear stress was instead demonstrated at the downstream wake.

Back to Top | Article Outline

Closing Process

The closing phase started at the onset of the backflow, which pushed the occluder from the fully opened position to close. At t = 310 ms, as shown in Figure 11, the backflow in the central channel separated into three streams. One stream reversed back through the central orifice with a peak velocity of 0.2 m/s. The other two regurgitant streams were perpendicular to the occluder surface. The occluder closing motion in turn initiated the backflow inside the side channels. The backflow inside the central nozzle and side channels both strengthened with the closing process. The peak closing velocity was increasing, with a value of 0.36 m/s at the instant of 330 ms and 0.49 m/s at t = 350 ms, as in Figure 11.

For the stress distribution at t = 310 ms, as in Figure 12, relatively high shear stress was produced inside the central channel. At t = 350 ms, the maximum shear stress occurred at the exit of the central nozzle, with a value up to 42 N/m2.

Back to Top | Article Outline


In this work, the flow characteristic and stress field were acquired from the PIV investigation of a transparent valve model. The flow velocities demonstrated in the opening process, fully opened stage, and closing process were comparable to the findings of Subramanian et al.8 The turbulent shear stress was found to continuously increase until peak systole and the maximum major RSS was up to 500 N/m2, which was higher than the normal threshold for hemolysis. High shear was found at Region I, where the strong central jet separated from the valve surface and caused large turbulent mixings. Furthermore, when the jet penetrated into the central channel with the opening process, the length of Region I was shown extending forward due to the relative small occluder opening angle. It is believed that the larger the fully open angle, the less likely the above-mentioned phenomenon occurred. However, a large open angle may cause strong occluder motion upon its closure, which may result in a high shear stress and also an increased risk of cavitation occurrence. The determination of MHV open angle should be compromised between the two sides.

The occluder trailing tips were observed with flow separation and/or wake formation in the MHV opening process and fully opened stage. The elevated shear layer thus occurs, resulting in large local deformations. Although the shear stress was also increasing in the MHV closing process, its magnitude was less than the opening process due to the relatively weakened backflow. Large shear stress was demonstrated either inside the central channel, where the unevenly distributed backflow causes strong velocity gradients, or around exit of central nozzle, resulting from the reverse jet mixing with relative weak upstream flow.

The limitation of this work can mainly be attributed to the dynamic performance of the MHV model. The simplified hinge design and the densities of the transparent MHV model and of the working fluid may affect the flow field and shear distribution around the valve.

Back to Top | Article Outline


For the first time, this work provides relative complete knowledge of the flow property and turbulent stress field around a transparent MHV in one cardiac cycle. Regions with possible blood damage were identified at the orifice of the central channel, with a maximum stress about 500 N/m2 at peak systole, higher than the threshold for hemolysis. Enlarging the open angle may decrease the magnitude of the turbulent shear occurring at this region.

Back to Top | Article Outline


This research was supported by the Shanghai Science and Technology Committee, Project No. 06PJ14083, and the School of Mechanical and Aerospace Engineering in Nanyang Technological University (NTU), Singapore.

Back to Top | Article Outline


1. Lim WL, Chew YT, Chew TC, Low HT: Particle-image velocimetry in the investigation of flow past artificial heart valves. Ann Biomed Eng 22: 307–318, 1994.
2. Sallam AM, Hwang NHC: Human red blood cell hemolysis in a turbulent shear flow: contributions of Reynolds shear stress. Biorheology 21: 783–797, 1984.
3. Blackshear PL: Mechanical haemolysis in flowing blood, in: Fung YC (ed), Biomechanics, Prentice-Hall, 1972, pp. 501–528.
4. Yoganathan AP, Woo YR, Sung HW: Turbulent shear stress measurements in the vicinity of aortic heart valve prostheses. J Biomech 19: 433–442, 1986.
5. Lim WL, Chew YT, Chew TC, Low HT: Steady flow dynamics of prosthetic aortic heart valves: a comparative evaluation with PIV techniques. J Biomech 31: 411–421, 1998.
6. Lim, WL, Chew YT, Chew TC, Low HT: Pulsatile flow studies of a porcine bioprosthetic aortic valve in vitro: PIV measurements and shear-induced blood damage. J Biomech 34: 1417–1427, 2001.
7. Baldwin JT, Deutsch S, Geselowitz DB, Tarbell JM: LDA measurements of mean velocity and Reynolds stress fields within an artificial heart ventricle. J Biomech Eng 116:190–200, 1994.
8. Subramanian A, Mu H, Kadambi JR, Wernet MP, Brendzel AM, Harasaki H: Particle image velocity investigation of intravalvular flow fields of a bileaflet mechanical heart valve in a pulsatile flow. J Heart Valve Disease 9:721–731, 2000.
9. Tennekes H, Lumle JL (ed): A First Course in Turbulence. Cambridge, MIT Press, 1972.
10. Barbaro V, Grigioni M, Daniele CG, Boccanera G: 19 mm sized bileaflet valve prostheses’ flow field investigated by bidimensional laser Doppler anemometry, part II: maximum turbulent shear stresses. Int J Artif Organs 20:629–636, 1997.
11. Grigioni M, Daniele C, Avenio GD, Barbaro V: On the monodimensional approach to the estimation of the highest Reynolds shear stress in a turbulent flow. J Biomech 33: 701–708, 2000.

Cited By:

This article has been cited 1 time(s).

ASAIO Journal
Experimental Study on the Reynolds and Viscous Shear Stress of Bileaflet Mechanical Heart Valves in a Pneumatic Ventricular Assist Device
Lee, H; Tatsumi, E; Taenaka, Y
ASAIO Journal, 55(4): 348-354.
PDF (1329) | CrossRef
Back to Top | Article Outline
Copyright © 2007 by the American Society for Artificial Internal Organs