Numerical simulation of blood flow in flexible arteries using Fluid-Structure interaction

Numerical simulation of blood

flow in flexible arteries using

Fluid Structure InteractionMostafa Ghadamyari, B.Sc Project

Ferdowsi university of Mashhad, Iran

Summer 2013

Simulating blood - Issues

Unsteady flow:To be more accurate, the

flow is steady-periodic.

Blood viscosity

differs:Blood is a Non-Newtonian

fluid, Viscosity depends

on shear rate

Arteries material are

complex:Different layers, different

properties

Arteries wall deform:Inside flow Pressure -> Artery

expands or collapses ->

Change inside flow

Complex geometry:Arteries bifurcate and join

again.

Model definitionModeling blood:Non-Newtonian blood

Carreau model :Artery material: Elastic isotropic

Unsteady

flow model:1. Pressure cycle

2. Velocity cycle

Both cases will be

discussed

Modeling Flexible

walls :Fluid structure interaction :

Fluid and Solid equations are

solved coupled

Geometry model :Modeling part of the blood

system, with realistic

boundary conditions

1.Pulsatile Pressure flow model simulation

Outlet flow : 0 Pa (Static gauge pressure)

Inlet flow (Static gauge pressure)Fixed inlet & outlet wall

FSI Boundary

P inlet = 100 + 100*sin(pi*t) [Pa]

1mm or

2mm Wall

thickness

Mr. Shaik model (Part of PHD. Thesis):

Pulsatile pressure model – answer

We started with ADINA simulation

PHD Thesis model answer :

Centerline velocity

Ave. 0.9m/s,

Range : 0.05m/s

Ave. 0.3m/s,

Range : 0.2m/s

Pulsatile pressure model – answer

Ansys CFX :

Ave. 0.9m/s,

Range : 0.5m/s

Comsol answer :

Centerline velocity

Ansys Fluent doesn’t accept static pressure at INLET !

Only Total pressure can be defined -> Velocity is included.

Ave. 1.9m/s,

Range : 0.5m/s

Inconsistent pulsatile pressure model – Why ?

We found this while searching for ‘Why fluent doesn’t accept static pressure at inlet?’

2.Pulsatile velocity flow model simulation

Outlet flow : 100mmHg ~ 13332 Pa

(Static gauge pressure)

Inlet flow (Velocity inlet)Fixed inlet & outlet wall

FSI Boundary

Mr. Wangn, journal paper :

2.Pulsatile velocity flow model simulation

Adina:

- Very light ~ 300mb

- Simple GUI

- Fastest in our simulation (default

settings)

- The least problems in convergency in

our model

Comsol:

- Medium size ~ 4GB

- Smart GUI – Fastest in modeling

- Slowest in our simulation (default

settings)

- Special solver settings needed (Good

for professionals)

Software comparison (continued.)

Fluent:

- The most famous software

- Can be coupled to Ansys structural

(Using system coupling)

CFX:

- Famous software

- Can be coupled to Ansys structural

(direct coupling)

Ansys : Large ~ 6.5G, Complicated, Stable

Pulsatile velocity result – Velocity

Differences are due to :

- Different mesh sizes

- Different solvers

Outlet surface midpoint velocity ,

ADINA :

Ave. 0.3m/s

Max. 1.1m/s

COMSOL :

Ave. 0.35m/s

Max. 1.32m/s

CFX :

Ave. 0.33 m/s

Max. 1.05 m/s

Pulsatile velocity result – Pressure

ADINA :

Ave. 13.48 KPa

Max. 14.50 KPa

COMSOL :

Ave. 13.4 KPa

Max. 14.80 KPa

CFX :

Ave. 13.48 KPa

Max. 14.55 KPa

Results differ

less than 0.5%

Inlet surface midpoint pressure :

Pulsatile velocity result – Pressure

ADINA

- 0.45mm initial disp. (4.5%)

- 0.03mm cyclic disp. (0.3%)

Middle surface Top Point displacement

CFX

- 0.65mm initial disp. (6.5%)

- 0.03mm cyclic disp. (0.3%)

Study 1: FSI vs CFD

Outlet surface midpoint velocity, CFD (Green) vs FSI (Blue) :

Max. Velocity:Rigid -> 1.2m/s

Deformable -> 1.1 m/s

~ 10% error if walls were considered rigid

-> Pressure : 1.6% error

We performed a rigid wall analysis with ADINA and compare the results to

deformable wall case.

Study 2, wall thickness and displacement:

We halved thickness of the artery, displacement of middle artery plane Top

point :

2mm thickness case

- 0.45mm initial disp. (4.5%)

- 0.03mm cyclic disp. (0.3%)

1mm thickness case

- 1.1mm initial disp. (4.5%)

- 0.03m cyclic disp. (0.3%)

FSI could be even more important …

Max. Displacement :

4mm ( 40%)

Max. Displacement:

-1.5 mm (-15%)

This is a sample analysis of blood flow in a bifurcate , in this case :

V in = 0.3 m/s , P in = 0Pa (case1) & -2.13Pa (case2), wall thickness=0.5mm

Conclusion

Wall shear stress -> Low WSS -> More susceptible to Atherosclerosis

Study more complex geometries (bifurcates, …)

Newtonian vs. Non-Newtonian blood

Solver settings (specially Comsoll)

Continue this project …

Pressure pulse vs. Velocity pulse -> Pressure model should be used with

considerations

Rigid walls vs. Deformable walls -> 10% error in velocity profile

Halved thickness -> doubled displacement

Acknowledgement

Thanks to Dr. M. Pasandideh Fard. my supporter and advisor throughout my career

at Ferdowsi university of Mashhad, who gave me the opportunity to work on this

project and introduced me to the fascinating field of computational fluid dynamics…

Case show

Animation

Thank you !

Any questions?