Experimental and numerical investigation of a scaling law for haemolysis in rotary blood pumps

Background

The rigorous optimisation of VADs for both hydrodynamics and haemocompatibility is highly desirable. Currently we have a method to optimise for hydrodynamics using Computational Fluid Dynamics (CFD) and surrogate modelling which enables millions of designs to be considered during the optimisation process. The method exploits the scalability of hydrodynamic pump characteristics to minimise CFD simulations and therefore, to perform a similar optimisation for haemolysis would require a universal haemolysis scaling law. The authors previously used numerical simulations to reveal the existence of such a scaling law and presented the work at ISMCS 2023Such a law has been revealed using numerical simulations and presented previously. The aim of this study was to determine if such a scaling law for modified index of haemolysis (MIH) could be validated with in vitro experiments.

Methods

CFD simulations have been carried out for 32 pump blade geometries as potential candidates for use in the NeoVAD paediatric LVAD. CFD simulations were completed across two different rotating speeds and 8 different flow rates per pump resulting in 512 simulations. From these simulations, HI was calculated using the power law integration method. Comparison of MIH curves across all designs provided insight for a novel haemolysis scaling relation.
In vitro experiments of the PediMag paediatric blood pump were conducted with 35±2% haematocrit bovine donor blood for 6 hours under four different haemodynamic conditions. The flow rates were 0.5, 2.0, 1.375, and 4.23 LPM, with corresponding differential pressures of 52, 80, 480, and 390 mmHg. The pump speeds were 2000, 2600, 5500, and 5500 RPM, respectively.

Results

A scaling relation between pumps at varying speeds was revealed such that the magnitude of MIH scales with (ω1/ω2)α , where ωi corresponds to rotating speed at condition, i, when plotted against dimensionless flow rate (Q/ ωD3, where D is pump diameter), and α is the power law coefficient for scalar shear stress. This relationship holds true for in vitro results of the PediMag (using α = 1.963 from Fraser et al. 2012), with scaled values falling within the estimated standard error. Mean values suggest a smaller value of α improves the fit. The scaling law also holds for experimental and computational results in the literature.

Conclusions

The novel haemolysis index scaling relation developed in this study allows the prediction of blood damage across pump operating conditions from simulation results of just one rotating speed. The method enables rigorous blade design optimisation to minimise haemolysis, an objective that has not been available previously.