A fluid-structure approach to optimize
the thrombogenic potential of
artificial heart valves
Filippo Piatti1, Francesco Sturla1,2,
Thomas E. Claiborne3, Danny Bluestein3, Alberto Redaelli1
1Dept. of Electronics, Information and Bioengineering, Politecnico di Milano, Milan, Italy
2Division of Cardiac Surgery, Università degli Studi di Verona, Verona, Italy
3Dept. of Biomedical Engineering, Stony Brook University, New York, USA
Materials and Methods
email@example.com CAE Conference 2014
Results and Discussion
Figure 1 –(A) Prototype of the polymeric aortic valve (Innovia LLC, Miami, USA). (B)
ViVitro gold standard testing benchmark (ViVitro Labs, BC, Canada)
The Inlet and Outlet reservoirs were characterized by physiological
pressure waveforms extracted from experimental testing, thus
obtaining comparable working conditions between the computational
model and the ViVitro benchmark.
Figure 2 –Schematic visualization of the ideal
optimization process implemented with the FSI
time 320 304
Figure 3 –Visualization of the velocity field contours on a cross-sectional plane
(upper panels) and of the FSI and experimental kinematic of the device (lower
panels), at four different time-points: (A) early systole, (B) systolic peak, (C) closing
phase, (D) stable diastole.
The research leading to these results has received fundings from the Cariplo Foundation Project,
Grant Agreement N°2011-2241.
Table 1 –Numerical verification variables
1. T. E. Claiborne et al. (2013) Journal of Biomechanical Engineering
2. D. Bluestein et al. (2010) Annals of Biomedical Engineering
(B) (C) (D)
Moreover, a quantitative
validation of the numerical
results was accomplished
through the evaluation of fluid
dynamic variables and the
comparison with device
performances, as highlighted
in Table 1.
Computational optimization process
Detailed numerical results
•Fluid dynamic performances
•Micro-scale particle tracking analysis
Revision of the design
•High velocity hotspots
•Particle tracking (blood
A particle tracking methodology was adopted to obtain the trajectories of
numerical ideal platelets. Numerical models were used to combine
stress and time (τ(t))so as to perform a quantification of the
thrombogenic potential of the device as well as of critical hot-spots.
Experimental procedures are commonly used to obtain an overall
quantification of the fluid dynamic and thrombogenic performances of
cardiovascular devices .
Meanwhile, the implementation of acomputational approach could
mimic the realistic operative conditions of a testing benchmark and
provide extremely localized and detailed information.
•Long lasting and high
•Physical prototypal device
•Kinematic and fluid
The computational models (Figure 2, left) and their finite element
discretization were implemented on ANSYS Gambit 2.4.6 (ANSYS,
Canonsburg, USA). The FSI simulation was performed with the explicit
solver LS-DYNA R6 (LSTC, Livermore, USA), coupling eulerian fluid
elements and lagrangian solid ones.
At this aim, this work presents an innovative tool that emulates a gold
standard testing benchmark (Figure 1-A) to perform a micro-scale
analysis of the performances of a prototypal polymeric heart valve
(Figure 1-B) in order to identify thrombogenic localized hot-spots for
further possible design optimizations.
In first instance, the numerical solution was compared with experimental
results and its reliability was verified: as reported in Figure 3, valvular
kinematics was qualitatively compared with the in-vitro mechanical
The thrombogenic evaluation was able to
highlight the commissural zones as high-risk
locations, due to vortexes propagation (figure 4-
A). Micro-scale analysis was used to extract the
stress-time history of selected dangerous
The proposed tool was able to accurately replicate realistic experimental
working conditions. Moreover, a micro-scale analysis provided detailed
information regarding the thrombogenic potential of particular zones,
potentially allowing for further design optimization of the device.
Figure 4 –(A) Trajectories of the Core and Commissural zones. (B) Stress-time
waveforms extracted from the domain
0 0,1 0,2 0,3 0,4 0,5