Technical Presentation

by

Sandesh Rajput

23-Sep-13 1

Content

Introduction

Total Cavopulmonary Connection

Computational Modeling

Methodology

Results

k-ε Turbulence Modeling

Conclusion and Future Work

What Differentiate my Work ?

Other Computational Skills

2 23-Sep-13

Introduction

Congenital Heart diseases (CHD)

o defects in the structure of the heart at the time of birth

o 8 out of 1000 live births

o 2,00,000 children born with such defects and accounts for 10% infant

mortality rate in India

Single ventricular deficiencies

o only one effective or functional pumping chamber

Fontan Operation

o In 1971, Fontan and Baudet successfully by-passed right heart

o Tricuspid atresia ( complete absence of tricuspid valve )

3 23-Sep-13

Total Cavopulmonary Connection (TCPC)

Super Vena Cava (SVC) connected directly to right pulmonary artery (RPA)

IVC (Inferior ) can be connected to PA through right atrium or outside the heart

Objectives

o To study flow pattern fluctuations using numerical modeling of TCPC

o To measure the energy losses and determine the most energy efficient TCPC

model for improving long-term outcome of Fontan patients

4 23-Sep-13

Original Fontan Operation Intra-atrial conduit Fontan Extra-cardiac conduit Fontan or TCPC

Computational Modeling

23-Sep-13 5

LPA

IVC

ɸ20 mm

SVC

ɸ18 mm

RPA ɸ12 mm

Base Model

offset

2 cm Offset Model

Modeling was done using DesignModeler from ANSYS workbench

10 computational models prepared to study effect of offset, blockage, control plate

and bifurcation of inflows on the flow pattern and energy efficiency

Offset model prepared for offset value of 0.5cm,1 cm, 1.5 cm and 2 cm

Computational Modeling

6 23-Sep-13

22% Blockage Plate

Blockage Model Control Model Optiflo Model

In blockage model, 22% of area was blocked by introducing half sphere of radius 4mm

so as to study the effect of blockage on flow properties such as pressure

Optiflo models were based on bifurcation of IVC or SVC or both at a distance of 5 cm

and uniformly changing cross-section from circle at top to ellipse at the junction

In control model, a plate was introduced at the center to avoid head-on collision

Methodology - Meshing

ANSYS MESH was used to discretized the solution domain into elements so as to

apply fluid equation at each elements

The mesh consist of :

o 3D tetrahedral elements containing four nodes

o Triangular 2D elements with three nodes at the boundaries

o Number of elements ranged from 180,000 to 200,000

o Structured grid was chosen for its suitability to simple model

Mesh Quality

o Mesh quality was checked using Minimum Orthogonal Quality and Skewness

Inflation layer were used to capture the velocity gradient of the boundary layer

occurring from a no-slip condition of the domain Wall

23-Sep-13 7

Methodology-Simulation

Solver used for this CFD analysis, FLUENT (ANSYS Inc. ) , was selected

because of its reputation for solving Low Reynolds flows

Density-based implicit solver was used

Steady state solver was used in order to get pressure and velocity after the flow

field was fully developed

Second-order upwind scheme recommended by ANSYS for accuracy in solving

the discretized solution domain using finite volume method was used

Assumptions

o Blood vessels were assumed to be circular and uniform in diameter with rigid

wall

o Blood was assumed to be incompressible and Newtonian fluid

o Laminar and steady state flow condition assumed throughout the simulations

8 23-Sep-13

Convergence Criteria

o Fifth-order decrease in residual, which is a measure of the local imbalance of

each equation being solved, was the convergence criteria

The total pressure and volume flow were calculated using area-weighted average

at the inlets and outlets

Power Loss, W diss : It is difference between the total energy rate at the inlet and

at the outlet of the model i.e. W diss = W in −W out

23-Sep-13 9

Methodology

Results

similar patterns in the trends of energy loss as a function of RPA flow split ratio; least

energy loss at the equal flow split condition

The power losses in 3D model is more than in 2D model. This is due to relieving effect

of third dimension and 3D complex fluid structures

the power loss goes on decreasing with increasing offset values

Power losses are reduced by almost 9 % when the model is offset by 2 cm.

10 23-Sep-13

Power loss through offset models for Q =3 Lit/min Power loss comparison with previous studies

Power loss comparison in TCPC models

The power losses in Optiflo models are the least. This is because of the absence

of head-on collision at the connection

Application of turbulent modeling predicts more accurate power losses as

recorded in previous studies.

The power loss in Control Model is lesser than Base Model due to absence of

central vortex

23-Sep-13 11

0

1

2

3

4

5

6

Optiflo Half (IVC)Optiflow

Half (SVC)Optiflow

Control Case(0.5 mm)

Control Case (1mm)

Blockage Model Base Model Base Model(Turbulence )

Pow

er L

oss

(m

W)

TCPC Models

Velocity Contours

23-Sep-13

Base Model

Flared 0.5 diameter offset

12

Blockage Model

Optiflo Model

0.5 cm offset Model

Control Model

k-ε Turbulence Modeling

Simulation on Control model and Base model was done using turbulence

modeling to understand the effect of turbulence in the flow

RANS based k-ε turbulence model with 5% of turbulence intensity was selected

since no adverse pressure gradients were expected in the simulation

Realizable k-ε was used to improve performance since the flow involved

recirculation and separation

The Turbulent kinetic energy, k variable determines the scale of the turbulence

while ε, variable determines the energy in the turbulence

Turbulence causes dissipation of kinetic energy at the connection because of

which power losses are relatively higher than the laminar model

As a result of this, the power loss in control model is 30% lesser than in Base

model 23-Sep-13 13

Conclusion and Future Work

Power losses are reduced by increasing offset distance , providing flaring and

equal flow split between RPA and SPA

The bifurcation of SVC and IVC in Optiflo avoid head-on collision, and hence

this model is most energy efficient

The power loss predictions are more accurate when turbulence modeling is used

Even small power saved can improve long-term output of Fontan patients

Models presented here were not as geometrically complex as are in reality.

Further studies are required to study effects of compliant wall and complex vessel

shapes and sizes

The steady state flow limits the study of complex 3D fluid structures in numerical

TCPC models since inflow from IVC and SVC is never “pure” steady flow

Future studies should include pulsatile flow by using transient state solver

14 23-Sep-13

What differentiate my work ?

In summary, this work utilizes a powerful computational tool i.e. Computational

Fluid Dynamics to model and simulate the blood flow so as to improve the long-

term outputs of Fontan patients by increasing the energy efficiency of TCPC.

Experimental work was also done on Base Model to study flow pattern using Flow

Visualization techniques. This was done to improve our understanding of the

physical concepts of the model.

The energy losses of 5 to 6 mW predicted in the Base Model is on good correlation

with the results of previous experimental and computational studies.

Also, the results obtained from this study will be used to build mechanical heart

valve in the same laboratory

The power loss comparison of different TCPC models under same boundary

conditions is the highlight of this work. This will help future studies to concentrate

on the most energy efficient model and improve it in further. 23-Sep-13 15

Other Computational Skills

Familiarity with various CFD techniques such as Finite Difference Method ,

Finite Volume Method and Finite Element Method

Understanding of different turbulence modeling classes such as RANS, LES,

DES and DNS

Understanding of nature of Governing equations and their discretization :

o Mathematical forms and their applications

o Mathematical Nature ( types of PDEs)

o Physical Nature (Expected Solution)

Fundamental knowledge of Grid Generation

Analysis and understanding of accuracy and convergence behavior of the solution

Keen interest in Computational Wind Engineering applications in Wind Turbine

and Urban Environment

23-Sep-13 16