A Practical Computational Hemolysis Model Incorporating Biophysical Properties of the Red Blood Cell Membrane
Purpose: Hemolysis is a key issue in the design of blood-handling medical devices. Computational prediction of this phenomenon is challenging due to the complex multiscale nature of blood. As a result, conventional approaches often fail to predict hemolysis accurately, commonly showing deviations of multiple orders of magnitude compared to experimental data. More accurate models are typically computationally expensive and thus impractical for real-world applications. This work aims to fill this gap by presenting accurate yet simple and efficient computational hemolysis models. Methods: Hemolysis modeling relies on two key components: a red blood cell model and a hemoglobin release model. In this work, we compare three red blood cell models: a common stress-based model (Bludszuweit), a simple strain-based model based on the Kelvin-Voigt constitutive law, and a more complex tensor-based model (TTM). Further, we compare two hemoglobin release models: the widely used power-law approach and a biophysical pore formation model. Results: We evaluate these models in two benchmark cases: the FDA blood pump and the FDA nozzle. In both benchmarks, the simple strain-based model combined with the pore formation model achieves absolute predictions of hemolysis within the standard deviation of experimental measurements. In contrast, stress-based power law models deviate by several orders of magnitude. Conclusion: The strain-based pore modeling approach takes into account the biophysical properties of red blood cell membranes, in particular their viscoelastic deformation behavior and hemoglobin release through membrane pores. This leads to significantly improved hemolysis predictions in a framework that can easily be integrated into common CFD workflows.