Introduction: Heart diseases continue to be the primary cause of death worldwide, and ventricular assist devices (VADs) have emerged as a promising therapy to treat these conditions. However, the reliability and biocompatibility of VADs still represent significant challenges in clinical practice. In this study, we present a numerical setup for simulating a centrifugal blood pump, which was employed to optimize the shape of a pump designed by the US Food and Drug Administration (FDA). The optimization process considered the effects of operating uncertainties on the hydrodynamic and hemocompatibility performance of the device to obtain an optimal configuration that minimizes variations. Our approach can be extended to more complex VAD geometries, enabling the design of more biocompatible and reliable blood pumps. This research contributes to the development of better VADs that can reduce the risk of postoperative complications for patients. These findings have important implications for improving the clinical application of VADs, which have the potential to benefit a larger number of patients suffering from heart diseases. We present our results at this biomedical conference to share our findings with the scientific community and stimulate further research in this area.
Materials and Methods:: The numerical analysis of a centrifugal blood pump aimed to evaluate fluid dynamic efficiency and hemolysis risk using computational fluid dynamics modeling. The OpenFoam software was used to simulate the FDA benchmark test case, and the Navier-Stokes equations were solved using a finite volume discretization method. The impeller rotation was modeled using a Moving Reference Frame approach, and turbulence was modeled using the standard k-Epsilon turbulence model. Hemolysis was modeled using an Eulerian approach based on the transport of a passive scalar representing the blood damage/hemolysis index. The proposed CFD model is robust and suitable for optimization, and it provides accurate enough predictions.
A multi-objective optimization was run using modeFRONTIER to obtain an improved pump configuration in terms of both hydraulic efficiency and blood hemolysis. Geometry changes to the impeller were made with Simcenter (NX), then a surface mesh was generated in ANSA. A series of points were intially sampled using a Uniform Latin Hypercube method, after which response surface models for both the hemolysis and efficiency metrics were trained. The optimization was performed using these response surfaces, using an ESTECO proprietery algorithm. Points along the pareto were then selected, validated on the full-scale model, and used for robustness analysis under uncertain flow rate conditions. The optimized pump configuration reduces the external battery size and increases lifetime, thereby increasing patient comfort. The resultant pump also maintains similar metrics under varying inlet conditions.
Results, Conclusions, and Discussions: The CFD model was validated by comparing it with both PIV experimental data and a higher fidelity model based on a transient (URANS) sliding mesh method. The results are in good agreement, showing that the steady MRF methodology provides accurate enough predictions and is also the most suited for performing multiple runs, as required by the optimization procedure.
For the analysis, a set of geometrical design variables controlling the shape of the impeller blades and vanes is considered, while keeping the housing dimensions fixed to maintain the device's overall size. In particular, the optimization focuses on the blade fillets, where high shear-stresses originate, as well as the blade height and length, without changing the rotor's maximum dimensions. Additionally, the effect of the rotor hub diameter variation is also investigated since it determines the characteristics of the flow approaching the blades.
Hemolysis is modeled using an Eulerian approach based on the transport of a passive scalar representing the blood damage/hemolysis index. Damage index analyses are conducted for various working conditions, and the results are in line with literature, showing that the proposed CFD model is robust and suitable for optimization.
Starting from these variables, a robust multi-objective optimization is run using modeFRONTIER to obtain an improved pump configuration in terms of both hydraulic efficiency and blood hemolysis. By maximizing efficiency, the VAD external battery size and lifetime are reduced, increasing patient comfort, whereas the minimization of hemolysis is paramount to limit the blood damage induced by the pump and assure its biocompatibility. The optimization is performed under uncertain operating conditions: in this study, keeping into account how small changes in the mass flow rate affect the performance. The pump hydrodynamic characteristics are highly sensitive to this condition. During the optimization loop, for each new iteration, additional points are sampled near the nominal design to estimate the responses statistics. This information is used to constrain the process and obtain a robust solution whose performance does not deteriorate due to uncertainties.
Acknowledgements: : The authors would like to thank Dr. Stefano Grossi, Head of the Electrophysiology Department at the Ospedale Mauriziano in Turin, for his suggestions on VAD devices. We would also like to thank Arric McLauchlan at Medtronic medical devices for his suggestions on factors to consider in the robustness study.
References: : P. Wu, “Recent advances in the application of computational fluid dynamics in the development of rotary blood pumps”, Medicine in Novel Technology and Devices, vol. 16, Dec. 2022.
R.A. Malinauskas, P. Hariharan, S.W. Day et al., “FDA Benchmark Medical Device Flow Models for CFD Validation”, ASAIO Journal, vol. 63(2), pp. 150–160, March/April 2017.
B.C. Good and K.B. Manning, “Computational modelling of the Food and Drug Administration’s benchmark centrifugal blood pump”, Artificial Organs, vol. 44(7), pp. E263-E276, July 2020.
M.S. Karimi, P. Razzaghi, M. Raisee, et al., “Stochastic simulation of the FDA centrifugal blood pump benchmark”, Biomech. Model. Mechanobiol., vol. 20, pp. 1871–1887, Oct. 2021.
