CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Blood Damage Modelling via Particle Tracking in a Centrifugal Heart Pump (

scatman May 4, 2011 10:56

Blood Damage Modelling via Particle Tracking in a Centrifugal Heart Pump
Hi All! :)

First of all, I would like to say what an incredible help this forum has been to me already. The knowledge available and shared on this forum is outstanding, it's truly been an invaluable resource.

Now, down to the problem...

I'm investigating centrifugal heart blood pumps (VAD) as my final year engineering project using ANSYS CFX. My supervisor has already performed steady-state simulations (without particle tracking) of the pump I am considering, and the results of my simulations match his closely (also without particle tracking). I desire to model blood damage by introducing particle tracking to my simulations.

Reading the literature available, and other academic papers, one-way coupling (as opposed to fully coupled) is an acceptable implementation of particle tracking for this scenario, which allows me just to re-run my simulations with the results as the initial values (so that only the particle tracks are calculated).

The formula for quantifying red blood cell damage is related to shear stresses and the exposure times of these stresses on the cells, as presented below:

(dHb/Hb) = (1.8e-6)([tau]^1.991)([dT]^0.765)

If we consider one particle track from the inlet of the pump to the outlet, summing the above equation at each position of the particle from the inlet to the outlet will result in the total cell damage for that particle. Thus, the above equation needs to be calculated at each point of each particle track. Furthermore, each particle track needs to be considered separately in this fashion, since multiple particle tracks may cross the some point in space but have different histories and futures.

It would be very appreciated if someone could point me in the correct direction to solve my problem.

ps. I do not need help determining the shear stress, academic papers have suggested the approxiamation of "tau=2*Density*Turbulent Kinetic Energy" is acceptable.



stumpy May 4, 2011 15:58

I wouldn't use particle tracking at all for this. Instead, investigate using an Additional Variable using the transport equation option (i.e. it just follows the flow). Create a source term for the Additional Variable that is a function of the shear stress and timestep. The Additional Variable value represent the total cell damage. Have a search on this forum for using Additional Variables to track fluid residence time, it's the same idea.

ghorrocks May 4, 2011 19:17

I agree with Stumpy. Why do this using a Lagrangian particle tracking approach? The Eularian approach suggested by Stumpy will be much more computationally efficient. It will take a bit of setting up but it sounds like a good method.

scatman May 4, 2011 23:54

Thanks for the quick reply! I'll give your suggestion a shot and see how it goes.


scatman May 5, 2011 02:36

Glenn and Stumpy, I had a meeting with my supervisor today and he insists that a Lagrangian approach be taken. I have a feeling he wants the simulations to have the same method as other simulations in the field of ventricular assist devices, all of which i've seen use Lagrangian tracking.

So, although I realize it's not the most elegant solution, any help with using a Lagrangian approach to this problem would be greatly appreciated.

ghorrocks May 5, 2011 07:23

If I was you, until you find a good reason to do it using Lagrangian particle tracking I would do it using both approaches and you will see for yourself the advantages of the approach Stumpy has suggested. I don't find "because everybody else does it that way" a very convincing argument.

Particle tracking is demonstrated in several CFX tutorial examples. That is the best place to start.

Someeee January 7, 2018 21:53

I am also trying to do blood damage modelling. The blood damage index is function of shear stress and exposure time, as mentioned in this post previously. Can you please give details on how to do this using Eulerian approach without injecting particles? I am uncertain about creating a source term as mentioned in this post.

Any help will be greatly appreciated.

usamaperwez January 8, 2018 00:59

First of all make additional variable:
Variable type: Volumetric
Units: as per your problem
Tensor Type: scalar

Then Add expression:
For example like this "(0.26[s^-1]*(1+tanh(-(T*1[K^-1]-335.3)/5)))/2)"
It need to stable in terms of unit

Then add subdomain:
Tick Sources and add the expression name in that.

Here we go.

I basically did it for DNA Species and it worked excellently (it took me 3 months to model that).

All times are GMT -4. The time now is 16:11.