Optimisation of cfd haemolysis model coefficients to enable accurate prediction of haemolysis performance in vads

Alessandra Molteni, Katharine Fraser, Haitham Yousef, Kenny Low, Samuel Rolland, Graham Foster

Research output: Contribution to journalMeeting abstractpeer-review

Abstract

Introduction: Previously our CFD model was able to provide an estimate for
haemolysis that showed a good relative correlation to lab data, but it was inaccurate in absolute terms, one order of magnitude greater. This study aimed to investigate a numerical haemolysis formulation capable of an improved
absolute blood damage prediction across a range of operational conditions
and in VADs with significant geometrical differences.
Material and methods: The normalized index of haemolysis (NIH) was computed
using a scalar transport model and a particle tracking model (PT).
Mathematically, haemolysis was modelled as a power law function of shear
stress (τ) and exposure time (Δt). The model parameters are determined
from fitting the VADs experimental haemolysis values. Haemolysis experiments
were carried out at different flow rates and pump speeds. It was found
that the model coefficients are dependent on the operating conditions. Fitting
was carried out in two stages: fitting flow rate dependency followed by
fitting pump speed dependency. The fitted haemolysis equation, below, includes
a scaling factor that is a function of flow rate (m) and pump speed (Ω),
and it was used in conjunction with CFD PT, NIH = C’·(Ωn/m)·(Σ Δt·τβ/α)α. The scalar transport method used a formulation described by Taskin et al (2012)
with a modified source term (S), S = ρ·(Hb·C’·(Ωn/m)·τβ)1/α, Hb = plasma free hemoglobin.
Results: So far we had significantly improved the absolute haemolysis estimate
[g/100L] when modelling the Centrimag:
1 l/min; 2250 rpm- NIHLAB = 0.001817; NIHPT = 0.001775;
5 l/min; 2250 rpm- NIHLAB = 0.000300; NIHPT = 0.000299; NIHST = 0.000229;
5 l/min; 3300 rpm- NIHLAB = 0.003250; NIHPT = 0.003250; NIHST = 0.002723;
10 l/min; 2250 rpm- NIHPT = 0.000161; NIHST = 0.000098;
Discussion: This work is currently ongoing. The absolute haemolysis estimate
has been significantly improved compared to our previous CFD model. The
PT is predicting absolute haemolysis to within 5% of the lab value and the ST
within 25%. Next steps are to expand the geometries of pumps tested including
Calon new MiniVAD.
Original languageEnglish
Pages (from-to)380-381
JournalThe International Journal of Artificial Organs
Volume38
Issue number7
Publication statusPublished - Jul 2015

Fingerprint Dive into the research topics of 'Optimisation of cfd haemolysis model coefficients to enable accurate prediction of haemolysis performance in vads'. Together they form a unique fingerprint.

Cite this