CFD Modeling and Venturi Design Parameters

The control model for our simulations was derived from Model #94 of the Vascular Model Repository (VMR), representing a healthy human aortic root with no intervention. In the baseline LAUD model (Model 0), a 12 mm conduit was added to connect the left atrium to the descending aorta, simulating the mechanical unloading function of the device. This conduit dimension was selected based on the smallest available size of the Medtronic Contegra bovine jugular vein graft, commonly used in cardiovascular surgical applications.

The LAUD’s performance was simulated using a constant inlet flow rate, representing steady-state pumping activity into the systemic circulation. To evaluate the effects of flow regulation and optimization, Models 1, 2, and 3 incorporated Venturi-based geometries within the descending aorta.

These Venturi configurations varied in throat-to-aorta area ratios:

• Model 1: 35% reduction

• Model 2: 50% reduction

• Model 3: 75% reduction

All Venturi structures shared the same dimensional framework:

• Converging section length: 3.43 cm

• Straight section length (for conduit connection): 2.00 cm

• Diverging section length: 6.87 cm

• Total Venturi length: 12.29 cm

To ensure a smooth transition and optimal integration of the LAUD conduit, a conical nozzle was modeled at the distal (aortic) end of the conduit, tapering from 12 mm to 7.5 mm to align with the Venturi throat geometry. This nozzle design was standardized across all three Venturi models to maintain consistency in comparative analysis.

These simulations enabled detailed examination of pressure gradients, velocity profiles, and shear stresses within the descending aorta, contributing to the iterative optimization of Cardiost’s LAUD design for maximized unloading efficiency and minimal hemodynamic disruption.

Simulation Methodology and Boundary Conditions

The Navier–Stokes equations, which describe the motion of viscous fluid substances, were used to mathematically model blood flow dynamics within all geometries. To solve these equations numerically, each model was meshed using SimVascular, generating unstructured finite element meshes suitable for transient flow simulation.

Once the meshes were generated, svPre—a SimVascular preprocessing module—was used to produce the required input files for the Navier–Stokes flow solver.

Standard physiological parameters were applied across all simulations:

• Blood density: 1,060 kg/m³

• Dynamic viscosity: 4 centipoise (0.004 Pa·s)

• No-slip boundary conditions were applied to all vessel walls, simulating realistic flow adhesion to endothelium

• Inflow boundary condition: A pulsatile flow waveform was applied at the ascending aorta inlet to reflect physiologic cardiac output variability

The flow rate waveforms used in the simulations (see Figure 2) correspond to different levels of cardiac output (CO) and were used to evaluate LAUD performance under a range of physiologic conditions.

These computational models provide a powerful tool for optimizing LAUD performance by simulating real-world hemodynamic interactions—supporting a data-driven, iterative design process grounded in clinical relevance.

Model 0 - LAUD, no Venturi stent

Pressure distribution

CO = 4.5 L/min, LAUD = 0.5 L/min