0
Flows in Complex Systems

Computational Fluid Dynamics-Based Design Optimization for an Implantable Miniature Maglev Pediatric Ventricular Assist Device

[+] Author and Article Information
Jingchun Wu

 LaunchPoint Technologies, Inc., 5735 Hollister Ave., Suite B, Goleta, CA 93117jwu@launchpnt.com

James F. Antaki

Department of Biomedical Engineering and Computer Science,  Carnegie Mellon University, 700 Technology Drive, Suite 4321, Pittsburgh, PA 15213Antaki@andrew.cmu.edu

Josiah Verkaik

 Launch Point Technologies, Inc., 5735 Hollister Avenue, Suite B, Goleta, CA 93117jverkaik@launchpnt.com

Shaun Snyder

 Launch Point Technologies, Inc., 5735 Hollister Avenue, Suite B, Goleta, CA 93117ssnyder@launchpnt.com

Michael Ricci

 Launch Point Technologies, Inc., 5735 Hollister Avenue, Suite B, Goleta, CA 93117mricci@launchpnt.com

J. Fluids Eng 134(4), 041101 (Mar 27, 2012) (9 pages) doi:10.1115/1.4005765 History: Received August 11, 2011; Revised January 03, 2012; Published March 27, 2012; Online March 27, 2012

Computational fluid dynamics (CFD)-based design optimization was applied to achieve the finalized design of the PediaFlow® PF4, a magnetically levitated rotodynamic pediatric ventricular assist device. It features a streamlined blood-flow path with a single annular fluid passage between the rotor and the stationary housing. The resulting impeller is composed of a first-stage mixed-flow section having four blades at the conical nose region followed by a second-stage fully axial-flow section with three blades within the annular gap region. A stator with three inwardly-directed vanes is provided at the conical tail region to recover pressure and straighten the flow. CFD predictions of head and efficiency characteristics agreed remarkably well with the validation experimental data: with overprediction of head by <7 mmHg over the entire operational range and a slight overprediction in best efficiency by ∼1%. The new optimized PF4 extended the maximum flow range of the previous PF3 device by more than 100% to over 2.3 liter per minute (LPM) for the same range of operating speeds, and doubled the maximum hydraulic efficiency to ∼27%. Evaluation of hemolysis was performed by a Lagrangian particle-tracking technique with analysis of regional contributions to the overall blood damage. The simulation revealed that hemolysis increases with an increase in both the flow rate and rotor speed but not necessarily with just an increase in flow rate at a constant rotor speed. At the flow rate of 1.0 LPM and a head of 138 mmHg, PF4 has a hemolysis index of 0.0032 compared to 0.0058 produced by PF3 at the same flow rate with a head of 48 mmHg. Numerical simulation of radial fluid forces performed by the CFD model with an eccentric rotor revealed the presence of negative fluid stiffness that was monotonically related to both flow and speed. Finally, conjugate heat transfer analysis predicted temperature rise adjacent to the motor to be inversely proportional to the length, but not exceeding ∼2 °C over the intended range of operation. In conclusion, CFD-based design optimization greatly expedited and facilitated the completion of the PediaFlow® flow path and contributed to the system-wide optimization to produce a miniature maglev pump with exceptional hemocompatibility.

FIGURES IN THIS ARTICLE
<>
Copyright © 2012 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Two generations of the PediaFlow® ventricular assist device: PF3 and PF4. Cutaway (bottom) shows critical internal components of PF4.

Grahic Jump Location
Figure 2

High-quality multiblock structured CFD surface mesh generated by custom-developed automatic elliptic generation system with boundary orthogonality and complete continuity at interfaces. (a) Entire PediaFlow® PF4 flow path. (b) Refined clustering O-grid around blade profiles. (c) Butterfly-type grid for cross sections of inflow cannula and outflow graft.

Grahic Jump Location
Figure 3

PediaFlow® PF4 flow validation fixture. (a) Experimental assembly. (b) Corresponding CFD model with high-quality multiblock structured surface mesh.

Grahic Jump Location
Figure 4

CFD model for coupled solid-fluid thermal analysis. Shell of titanium alloy (Ti6AI4V) was included with a thickness of 0.175 mm, and two motor lengths of 3.0 mm and 6.0 mm were modeled.

Grahic Jump Location
Figure 5

Predicted flow field at speed = 16,000 rpm. (a) Velocity vectors on mid-span blade-to-blade surface for first- and second-stage impeller (Q = 1.5 LPM). (b) Velocity vectors (Q = 2.3 LPM). (c) Pathlines through PF4 entire flow path (Q = 1.5 LPM).

Grahic Jump Location
Figure 6

Comparison of CFD predicted H-Q results (thick solid lines) with the experimental data for PF4 flow validation fixture (dashed lines). Also superimposed are the experimental data for an equivalent PF3 flow validation fixture (thin solid lines).Analog blood has a density of 1076 Kg/m3 and a dynamic viscosity of 0.0024 Pa-s.

Grahic Jump Location
Figure 7

Comparison of CFD-predicted η-Q results (thick solid lines) with the experimental data for PF4 flow validation fixture (dashed lines). Experimental data for an equivalent PF3 flow validation fixture provided for comparison (thin solid lines). Analog blood has a density of 1076 Kg/m3 and a dynamic viscosity of 0.0024 Pa-s.

Grahic Jump Location
Figure 8

Comparison of CFD-predicted performance for PF4 flow validation prototype (solid lines), and the maglev model (dashed lines). (a) H-Q Curves. (b) η-Q Curves. Analog blood has a density of 1040 Kg/m3 and a dynamic viscosity of 0.0035 Pa-s.

Grahic Jump Location
Figure 9

Computed mean blood damage through the entire PF4 pump model (based on 6000 particles) throughout entire flow range at different pump speeds (and pressure heads). Same hemolysis index converted from in vitro data for PF3 at flow rate of 1.0 LPM and pump speed of 16,200 rpm (48 mmHg) shown for comparison.

Grahic Jump Location
Figure 10

Portion of total blood damage contributed by different pump components over the entire flow range at different pump speeds and heads.

Grahic Jump Location
Figure 11

Pressure contours at different cross sections computed at a radial eccentricity of 160 μm (Q = 2.3 LPM and speed = 16,000 rpm)

Grahic Jump Location
Figure 12

CFD-predicted negative fluid stiffness as a function of flow rates and rotor speeds

Grahic Jump Location
Figure 13

Predicted temperature distribution on the longitudinal cut-plane (meridional section) at a flow rate of 1.5 LPM and rotor speed of 16,000 rpm. Heat is concentrated within the motor region and dissipated downstream along the shell into a very thin layer of fluid.

Grahic Jump Location
Figure 14

Computed temperature distribution on the interface of the titanium shell and flowing blood (flow rate = 1.5 LPM, rotor speed = 16,000 rpm): (a) 3.0-mm-long motor and (b) 6.0-mm-long motor

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In