Carnegie Mellon Logo

Prahlad Menon — 2011-12 Fellow

Prahlad Menon photo
  • Department: Mechanical Engineering
  • Advised by: Kerem Pekkan (Biomedical & Mechanical Engineering)
  • Project Title: Hemodynamic performance characterization of aortic outflow cannula jets for improved neonatal cardiopulmonary bypass
  • Prahlad Menon's Research Summary (Powerpoint Slide)

Project description
In the US, approximately 1 in 100 children are born with a clinically significant congenital heart defect (CHD), representing 30,000 children each year in the United States and 1,300,000 children worldwide. The palliative repair of these defects requires complex biventricular and univentricular surgical operations in newborns often smaller than 2 kg. A major component of these surgeries is the cardiopulmonary by-pass (CPB) operation which if prolonged can potentially lead to neurological complications and developmental defects in the young patient. Tiny aortic cannulae (2-3 mm inner diameter), with micro-scale blood-wetting features transport relatively large blood volumes (0.3 to 1.0 L/min) resulting in high blood flow velocities. These severe flow conditions are likely to result in platelet activation, vascular and blood damage. Clearly, there is a definitive need for engineering small yet hemodynamically efficient aortic outflow cannulae that can provide high blood volume flow rates but with low exit force and outflow velocity, for use in extracorporeal circulation (ECC) during neonatal CPB repairs.

The focus of this project is the cannula jet wake region and control of fluid flow through the design of more hemodynamically efficient internal cannula geometries. We aim to conduct high-resolution computational fluid dynamics (CFD) studies (approaching direct numerical simulation (DNS) resolution) of the cannula jet wake using our in-house cardiovascular fluid dynamics solver, and validate these simulations through particle image velocimetry (PIV) experiments. Therefore our specific aims are to: (i) conduct device-specific (device geometry is acquired through microCT scanning) outflow cannula jetstream characterization using CFD simulations. These results will serve as a baseline for design of hemodynamically efficient neonatal aortic outflow cannulae; (ii) Experimentally validate the in-house CFD simulations for single and multiple jet flows at laminar, turbulent and transitionary flow regimes; (iii) Using efficient high performance computing (HPC) simulations to design internal flow-control features for decreased blood damage and increased permissible flow rate. In order to speed up the CFD enabled design procedure, we will develop a CUDA C++ version of our in-house CFD code taking advantage of the compact GPU cyber-infrastructures for improved portability and shared memory parallel speed-up in. Our goal is to eventually use this tool for interactive surgical planning during time-critical neonatal CPB operations and also enable inverse optimization studies.

Research plan
CFD is a powerful tool for assessing local hemodynamic inside patient-specific anatomies as well as the design of hemodynamically superior medical devices. Device-related thrombosis is an important cause for concern in medical device design and CFD is useful to assess blood damage using empirical blood damage functions formulated based on time history of stress exposure, viscous energy dissipation and transient turbulence. Furthermore, CFD enables the evaluation and optimization of hemodynamic performance for minimal blood damage. Our group has applied this technique to design better surgical anastomosis templates and cardiovascular medical devices. The present proposal that focuses on characterization of cannula jet-flow regimes is a challenging problem for cardiovascular fluid mechanics since even symmetric confined single jet flows are known to develop asymmetric flow patterns from shear layer interaction with the confining walls, leading to instabilities and complex flow physics. Based upon our earlier experience with similar flow regimes [1], these simulations can require computational domains with close to 10 million nodes in order to resolve details of transient vertical flow structures, mixing jet shear layers, recirculation zones near impinging surfaces.