Artificial Organs

26(7):608613, Blackwell Publishing, Inc.

2002 International Society for Artificial Organs

Computational Fluid Dynamics Insights in the Design of

Mechanical Heart Valves
Stuart G.D. Kelly
WS Atkins Consultants Ltd., Bristol, U.K.

Abstract: Computational fluid dynamics (CFD) analysis

can provide detailed, three-dimensional predictions of
blood flow through mechanical heart valves, which can
help to optimize valve hemodynamics and reduce the potential for blood clotting. A number of CFD studies, considering both forward and retrograde flow through valves,
have been published. In this paper, a geometrically accurate CFD model capable of predicting the threedimensional, time-dependent flow through an open ATS
bileaflet valve is presented. A detailed picture of the blood

flow is obtained, including small-scale flow features in the

pivot regions. Results from the model can also be used to
investigate the opening position of the ATS valve leaflets.
Future work will be aimed toward improved models that
provide valuable design information while minimizing the
development time and computational resources required.
Such practical CFD models clearly have the potential to
reduce the costs, time scales, and risks associated with development of new heart valve designs. Key Words:
Computational fluid dynamicsDesignHeart valves.

Prosthetic heart valves present a wide range of

design choices and challenges, illustrated by the diversity in form, clinical efficacy, and commercial success of the many different valve designs implanted
since the first caged-ball valve in 1953.
In the case of mechanical-type heart valves, design
requires special attention to the closely interrelated
problems of material selection, mechanical strength,
durability, blood flow, and biological response. For
heart valves, the main biological response of concern
is the formation of thrombus at the valve site, which
may lead to thromboembolism or hinder the valve
function. Blood clotting is known to depend on the
three-way interaction of blood biochemistry, the
blood-contacting material, and the blood flow (1).
A detailed understanding of blood flow is therefore valuable in heart valve design, not only to optimize the overall hemodynamic performance of the
valve (e.g., minimal transvalvular pressure drop,
closing volume, regurgitant flow, and so forth), but
also to reduce the potential for blood clotting at the
valve.
Computer-based numerical flow simulation tools,

such as computational fluid dynamics (CFD), can

provide detailed, three-dimensional predictions of
fluid flow in passages of complex geometry. Such
complex flows occur in many medical devices, including heart valves. A major benefit offered by
CFD is the ability to evaluate preliminary designs
quickly, before committing to the expense of prototype manufacture and testing. When applied with
sound engineering judgment, CFD can therefore reduce the costs, time scales, and risks associated with
development of a new heart valve design.

LITERATURE REVIEW
Published computational studies of blood flow
through mechanical heart valves provide insights
into the fluid dynamics of existing designs, which
may suggest design changes to improve performance. Three-dimensional models, which can include the influence of secondary flows perpendicular
to the main flow direction, are particularly valuable.
Shim and Chang (2) developed a threedimensional CFD model of steady flow through a
tilting-disk valve while King et al. (3,4) and Krafczyk
et al. (5) considered pulsatile flow through a bileaflet
valve. All three models considered valves in the fully

Received March 2002.

Address correspondence and reprint requests to Stuart G.D.
Kelly, WS Atkins Consultants Ltd., 220 Aztec West, Bristol BS32
4SY, U.K. E-mail: Stuart.Kelly@wsatkins.com

608

CFD IN MECHANICAL HEART VALVE DESIGN

open position and predicted complex three-dimensional flow structures within and downstream of the
valve annulus.
More recent CFD studies by Wang et al. (6), Gao
et al. (7), and Kelly et al. (8) have also considered the
flow through fully open bileaflet valves with particular attention to the detailed geometry at the valve
pivots. In addition to global flow structures, these
studies demonstrated smaller-scale, complex threedimensional flows close to the pivots which are
highly dependent on the individual valve design.
Gross et al. (9) reported results from a model of
the Medtronic Parallel bileaflet valve, including
time-dependent inlet flow and prescribed leaflet motion, to predict flow structures in the hinge region of
the bileaflet valve throughout one entire cardiac
cycle. A separate submodel of the hinge cavity was
constructed to achieve detailed flow velocities in this
region which showed good agreement with experimental measurements.
The model of Hsu et al. (10) also applied a prescribed leaflet motion, in this case for a tilting-disk
valve during closing. At a partially closed leaflet
angle, the moving leaflet results were shown to predict higher blood shear stresses compared to a steady
flow solution at the same angle.
A range of different mathematical approaches
have been applied in these studies, including finiteelement, finite-volume, and Lattice-Boltzmann
methods. Structured and unstructured meshes have
been used. In general, comparison with experimental
measurements suggests that all of these methods can
provide reasonable predictions of mechanical heart
valve flow fields, provided that their relative
strengths and weaknesses are taken into account
when developing the CFD model.
CFD has also been used to investigate the leakage
flows through closed mechanical heart valves, with
particular attention to cavitation and shear stresses
which may damage the valve and/or blood. Because
of the small gap size relative to overall valve dimensions, such models generally simplify the valve
geometry as a two-dimensional problem (1113) although Makhijani et al. (14) created a three-dimensional model of a tilting-disk valve, including leaflet
motion, to investigate cavitation during closing.
Leakage flow will be viscous dominated, and therefore non-Newtonian blood flow properties are likely
to be important; however, the majority of leakage
flow CFD models reported to date have assumed a
Newtonian blood analog fluid. Nevertheless, some
agreement with experimental measurements has
been achieved in several of the studies.

609

CFD ANALYSIS OF ATS BILEAFLET VALVE

Purpose
The ATS bileaflet valve (ATS Medical Inc., Minneapolis, MN, U.S.A.) incorporates a new design of
leaflet pivot compared to previous bileaflet designs.
Whereas most existing pivot designs consist of a cavity on the valve housing ring in which a semicircular
peg on the leaflet locates and slides, the ATS design
replaces the peg and cavity with a convex hemisphere on the valve housing which locates in a cutaway section of the leaflet. As shown in Fig. 1, the
leaflet motion is constrained by leaflet stops on either side.
We developed a geometrically accurate CFD
model of the ATS valve to provide detailed bloodflow predictions, particularly within the valve housing ring. The methodology developed was intended
to be suitable for use as a practical tool to screen
potential new valve designs quickly and within realistic computational resources. Full details of the
model development have been presented in Ref. 8; a
brief description is provided here with some additional results.
To increase confidence in the calculation results, a
10:1 enlarged scale experimental model of the ATS
valve was also constructed. The experimental and
CFD models in combination provided a well validated, detailed picture of the flow field through the
ATS valve.
Method
The CFD model represents an ATS valve in a
cylindrical duct as shown in Fig. 2. Note that only
one-quarter of the valve is modeled, taking advantage of the two symmetry planes (XX and YY) to
reduce computation time. The valve housing geometry is defined in full detail, including the pivots and
leaflet stops, as shown in Fig. 3. The computational
mesh was refined downstream of the leaflets to capture unsteady flows in the leaflet wake. Timedependent flow conditions approximated the aortic

FIG. 1. The ATS bileaflet valve is shown.

Artif Organs, Vol. 26, No. 7, 2002

610

S.G.D. KELLY

FIG. 2. CFD model geometry

showing overall dimensions. The
defined symmetry planes XX and
YY are used for presentation of results.

valve of a patient with a heart rate of 75 beats per

minute and a cardiac output of 5 L/min.
Simplifications
The valve was modeled in a simple, rigid duct geometry primarily to allow ready comparison with the
experimental model for validation purposes. The geometry up- and downstream of the valve could be
readily changed for future analyses. The use of symmetry planes in the CFD model restricted its application to cases with a symmetric, parallel inlet flow.
Only the acceleration phase of forward flow
through the valve was calculated, assuming laminar
flow due to the stabilizing effect of the accelerating

FIG. 3. Mesh detail in pivot region of the ATS valve is shown.

Artif Organs, Vol. 26, No. 7, 2002

fluid. The valve leaflets were assumed to remain

fully open throughout the forward flow phase which
was considered reasonable because the opening and
closing motions occur quickly compared to the total
opening time.
For forward flow through a valve, when viscous
fluid forces are relatively unimportant compared to
inertial forces, the influence of non-Newtonian
blood viscosity variations will be small. Blood was
therefore modeled as a homogeneous Newtonian
fluid with dynamic viscosity 3.5 103 Pas and density 1,050 kg/m3.
Flow results
Velocity vectors at peak systole on planes XX (upper panel) and YY (lower panel) are shown in Fig. 4.
On plane XX, the jets through the central and lateral
orifices of the valve are visible, separated by a
low-velocity wake region. A large recirculation
zone downstream of the pivot regions is evident on
plane YY.
Figure 5 shows three-dimensional views of the inside of the valve housing ring, with the leaflet omitted for clarity, at three different points in the acceleration phase: one-third and two-thirds of the time
to peak systole and at peak systole (for the valve in
aortic position). Forward flow is from left to right on
the figure. The dark regions mark computational
cells where the axial velocity is reversed, indicating
positions of recirculating flow.

CFD IN MECHANICAL HEART VALVE DESIGN

611

FIG. 4. Velocity vectors downstream of the ATS valve are shown.

Stability of leaflet opening position

During our experimental testing, it was observed
that under steady forward flow, the leaflets of the
ATS valve opened only to around 80 degrees compared to the 85-degree design angle (i.e., the leaflets
did not open fully onto the leaflet stops). To investigate this observation, we used the CFD model to
calculate the magnitude and direction of the moment
exerted on each leaflet by the blood flow.
At peak systole, the peak moment, calculated
about the center of the pivot, was 4.1 106 Nm in
a direction tending to push the leaflet open. At first
sight this result suggests that the leaflet is being
firmly held against the leaflet stop. However, the
total force on the leaflet was 15.5 103 N, implying
a lever arm of only 0.26 mm. Considering that the

actual pivot location of the leaflet can move by 0.5

mm away from the pivot center as illustrated in Fig.
6, it can be seen that the actual moment direction
may depend on whether the leaflet is in contact with
the central or lateral leaflet stop. For the cylindrical
duct geometry considered, the CFD results therefore
suggest that the leaflet is only just held in the fully
open position by fluid pressure, if at all.
A possible reason for the relatively unstable open
position is that the pivot axis of the ATS valve is
slightly downstream of the housing ring leading edge
whereas alternative bileaflet designs often locate the
pivot axis upstream of the housing ring. Because the
pressure differences across the leaflets result from
the high-velocity blood jets, the position of the pivot
axis relative to the constriction of the housing ring is
Artif Organs, Vol. 26, No. 7, 2002

612

S.G.D. KELLY

FIG. 6. Schematic diagram showing two possible pivot centers for

the ATS valve leaflet.

also resolved smaller-scale flow features in the pivot

regions. This information provided the basis for a
qualitative assessment of the risk of clotting at the
ATS valve which was generally found to compare
favorably with other bileaflet valve designs as discussed in Ref. 8. The analysis did, however, identify
some areas where flow might be improved and provided some insights into the clinically and experimentally observed incomplete opening of the ATS
valve.
FUTURE DIRECTIONS

FIG. 5. Shown are positions of reversed flow within the valve

housing ring at three different times up to peak systole.

likely to be important. Reduced opening of ATS

valves has also been measured in vivo by cineradiography (15). In future work, it would be valuable to
investigate the influence of a more realistic downstream geometry on the leaflet forces and to compare leaflet forces with other bileaflet valve designs.
Value of results
In addition to a three-dimensional picture of the
global flow, our geometrically accurate CFD model
Artif Organs, Vol. 26, No. 7, 2002

For an implanted mechanical heart valve, a totally

comprehensive definition of the behavior of the
heart valvetissueblood system throughout one or
more complete cardiac cycles would require modeling of realistic cardiovascular geometry and flow
data; unsteady flow effects such as vortex shedding;
turbulence models applicable at low Reynolds numbers of around 5,000; blood represented as a nonhomogeneous, non-Newtonian fluid; and moving
boundaries and coupled fluid-structure interaction.
Continual advances in computer power and the
capabilities of commercially available CFD software
already provide the ability to include most, if not
all, of these features in numerical simulations; however, the time and computational resources required
can be onerous. In addition, as the complexity of a
computational model is increased, it becomes increasingly difficult to judge whether the solution is
reasonable without independent experimental validation.
It would therefore be valuable for future heart

CFD IN MECHANICAL HEART VALVE DESIGN

valve CFD research work to investigate not only
methods of making models more complex and realistic, but also the relative importance of each of the
features listed above, throughout the cardiac cycle,
to provide an informed and defensible basis for development of simplified engineering models as practical and cost-effective design tools. For example, a
comparison between quasisteady and pulsatile inlet
flows was presented in Ref. 8 which showed similar
qualitative flow results but significant quantitative
differences. Similarly, the comparison of fixed and
moving leaflets in Ref. 10 suggests that the leaflet
motion is important if the aim of the model is to
calculate shear stress magnitudes accurately.
Integration of CFD throughout the heart valve design cycle also deserves further attention. During
evaluation of concept designs, blood flow modeling
can be used to reduce costs and time scales associated with iterative prototype manufacture and testing. At the detailed design stage, CFD can quickly
investigate the effects of design changes on blood
flow to reduce the risk of unexpected knock-on effects which may otherwise only become apparent at
a later stage. Subsequently, when a final design has
been reached, flow analysis can be used to confirm
that design goals have been achieved. Regulatory
submissions can be strengthened by using CFD
analysis to support and explain experimental results.
The development of design processes that can fully
benefit from the integration of CFD, ideally alongside other valuable computational tools such as finite-element stress analysis (FEA), could lead to significant reductions in heart valve development time,
cost, and risk.
Acknowledgments: The ATS valve model was developed while the author was working at the Institute of Biomedical Technology, Ghent University, Belgium, and at
the Bioengineering Unit of Strathclyde University, U.K.
The support and guidance of Professors Pascal Verdonck
and Erik Dick was central to the work.

613

REFERENCES
1. Virchow R. Phlogose und thrombose im gefsystem. Gesammelte abhandlungen zur wissenschaftliche medecin. Frankfurt am Main: Meidinger & Sohn & Corp, 1856.
2. Shim EB, Chang KS. Numerical analysis of three-dimensional
Bjork-Shiley valvular flow in an aorta. J Biomech Eng 1997;
119:4551.
3. King MJ, Corden J, David T, Fisher J. A three-dimensional,
time-dependent analysis of flow through a bileaflet mechanical heart valve: comparison of experimental and numerical
results. J Biomech 1996;29:60918.
4. King MJ, David T, Fisher J. Three-dimensional study of the
effect of two leaflet opening angles on the time-dependent
flow through a bileaflet mechanical heart valve. Med Eng
Phys 1997;19:23541.
5. Krafczyk M, Cerrolaza M, Schulz M, Rank E. Analysis of 3D
transient blood flow passing through an artificial aortic valve
by Lattice-Boltzmann methods. J Biomech 1998;31:45362.
6. Wang J, Yeo TJH, Lim C-S. CFD Investigation of steady flow
in bi-leaflet heart valve hinge. Crit Rev Biomed Eng 2000;28:
617.
7. Gao ZB, Hosein N, Dai FF, Hwang NH. Pressure and flow
fields in the hinge region of bileaflet mechanical heart valves.
J Heart Valve Dis 1999;8:197205.
8. Kelly SGD, Verdonck PR, Vierendeels JAM, Riemslagh K,
Dick E, van Nooten GG. A three-dimensional analysis of flow
in the pivot regions of an ATS bileaflet valve. Int J Artif
Organs 1999;22:75463.
9. Gross JM, Shu MCS, Dai FF, Ellis J, Yoganathan AP. A
microstructural flow analysis within a bileaflet mechanical
heart valve hinge. J Heart Valve Dis 1996;5:58190.
10. Hsu AT, Yun JX, Hwang NHC. Application of an unstructured grid algorithm to artificial heart valve simulations.
ASAIO J 1999;45:5816.
11. Reif TH. A numerical analysis of the backflow between the
leaflets of a St. Jude Medical cardiac valve prosthesis. J Biomech 1991;24:73341.
12. Lee CS, Chandran KB. Numerical simulation of instantaneous backflow through central clearance of bileaflet mechanical heart valves at closure: shear stress and pressure
fields within clearance. Med Biol Eng Comput 1995;33:
25763.
13. Bluestein D, Einav S, Hwang NHC. A squeeze flow phenomenon at the closing of a bileaflet mechanical heart valve prosthesis. J Biomech 1994;27:136978.
14. Makhijani VF, Siegel JM, Hwang NHC. Numerical study of
squeeze-flow in tilting disc mechanical heart valves. J Heart
Valve Dis 1996;5:97103.
15. Aoyagi S, Kawara T, Fukunaga S, Mizoguchi T, Nishi Y,
Kawano H, Arinagi K. Cineradiographic evaluation of ATS
open pivot bileaflet valves. J Heart Valve Dis 1997;6:25863.

Artif Organs, Vol. 26, No. 7, 2002