FDA's nozzle numerical simulation challenge: non-Newtonian fluid effects and blood damage

Abstract

Data from FDA's nozzle challenge-a study to assess the suitability of simulating fluid flow in an idealized medical device-is used to validate the simulations obtained from a numerical, finite-differences code. Various physiological indicators are computed and compared with experimental data from three different laboratories, getting a very good agreement. Special care is taken with the derivation of blood damage (hemolysis). The paper is focused on the laminar regime, in order to investigate non-Newtonian effects (non-constant fluid viscosity). The code can deal with these effects with just a small extra computational cost, improving Newtonian estimations up to a ten percent. The relevance of non-Newtonian effects for hemolysis parameters is discussed.

Figure 1. Nozzle geometrical specifications defined in the FDA CFD challenge.

Figure 2. Curves: blood viscosity values as a function of the effective shear rate for different rheological models.

Histogram: actual distribution of shear rates in the CFD simulation results presented in section Results and Discussion.

Figure 3. Relative error on mass flow rates, , obtained by integrating the axial velocity profiles (multiplied by the density) at different positions along the axis.

Figure 4. Axial velocity along the nozzle centerline for the three viscosity models.

CFD results (lines) are compared to experimental data represented by their means and confidence intervals (i.e.: ).

Figure 5. Axial velocity profiles (CFD results and experimental values) along radial cuts at different positions of the nozzle.

Figure 6. CFD predicted pressure drop values along the nozzle centerline for the three different viscosity models (lines), together with maximum pressure drop value derived from peak velocity value using Bernoulli’s equation (thick dashed horizontal line).

Experimental data are represented in light gray color because of their lack of reliability due to experimental errors, as pointed out in FDA’s report .

Figure 7. Blood shear stress profiles along radial cuts at different positions along the nozzle’s axis.

Lines represent CFD predictions and indirect experimental data is represented by their means and confidence intervals. Since shear stress values obtained from experimental data actually require to set a viscosity model, here we plot three different experimental data sets (Newtonian, CY and Casson), one for each viscosity model.

Figure 8. WSS values as a function of the axial position.

Lines represent CFD predictions for the three viscosity models considered (using the same color code as in previous figures), whereas experimental results (indirectly measured from velocity profiles and assuming each of the three viscosity models studied in this article) are plotted by their mean values and confidence intervals.

Figure 9. Flux-weighted NIH average values () using the different viscosity models, together with the analytical lower limit.

Thick curved lines correspond to the values obtained from computing the flow-weighted averages over radial cuts along the main axis using the NIH evolution field included as a PDE. Thin dashed lines represent the averaged values at outflow given in Eqs. (22), obtained from a post-processing analysis of the velocity field.