Numerical and experimental analysis of an axial flow left ventricular assist device: the influence of the diffuser on...

Thousands of adult cardiac failure patients may benefit from the availability of an effective, long-term ventricular assist device (VAD). We have developed a fully implantable, axial flow VAD (LEV-VAD) with a magnetically levitated impeller as a viable option for these patients. This pump's streamlined and unobstructed blood flow path provides its unique design and facilitates continuous washing of all surfaces contacting blood. One internal fluid contacting region, the diffuser, is extremely important to the pump's ability to produce adequate pressure but is challenging to manufacture, depending on the complex blade geometries. This study examines the influence of the diffuser on the overall LEV-VAD performance. A combination of theoretical analyses, computational fluid (CFD) simulations, and experimental testing was performed for three different diffuser models: six-bladed, three-bladed, and no-blade configuration. The diffuser configurations were computationally and experimentally investigated for flow rates of 2-10 L/min at rotational speeds of 5000-8000 rpm. For these operating conditions, CFD simulations predicted the LEV-VAD to deliver physiologic pressures with hydraulic efficiencies of 15-32%. These numerical performance results generally agreed within 10% of the experimental measurements over the entire range of rotational speeds tested. Maximum scalar stress levels were estimated to be 450 Pa for 6 L/min at 8000 rpm along the blade tip surface of the impeller. Streakline analysis demonstrated maximum fluid residence times of 200 ms with a majority of particles exiting the pump in 80 ms. Axial fluid forces remained well within counter force generation capabilities of the magnetic suspension design. The no-bladed configuration generated an unacceptable hydraulic performance. The six-diffuser-blade model produced a flow rate of 6 L/min against 100 mm Hg for 6000 rpm rotational speed, while the three-diffuser-blade model produced the same flow rate and pressure rise for a rotational speed of 6500 rpm. The three-bladed diffuser configuration was selected over the six-bladed, requiring only an incremental adjustment in revolution per minute to compensate for and ease manufacturing constraints. The acceptable results of the computational simulations and experimental testing encourage final prototype manufacturing for acute and chronic animal studies.