You are on page 1of 56

CFD AND IT’S APPLICATIONS

Presented By:
Mohit Nigam
Sc-B,
CFD Division,
HR Wing, NSTL
OUTLINE:
1. Introduction to CFD
(i) History of CFD
(ii) Need for CFD
(iii) Background

2. Finite Volume Methods


(i) Some Finite Volume Schemes
(ii) Discretisation methods in different dimensions
(iii) Difference Based Methods-Choice of Convective Fluxes
(iv) Upwind and Quick Schemes

3. Some Numerical results


(i) Methods of Solution using central difference
(ii) Numerical Results (Central difference and Upwind)

4. Application-CFD Analysis of Varunastra with Control Surfaces


(i) Comparative performance of different versions of Control Surfaces
(ii) Results
INTRODUCTION TO CFD
History of CFD:
(i) Started in early 1970’s.

(ii) Acronym for combination of physics, numerical mathematics and


computer sciences.

(iii) First application: Simulation of transonic flows based on the solution


of non-linear potential equation.

(iv) Solutions of 2-D and 3-D Euler equations achieved.

(v) Possible to compute fully viscous flows around underwater


configurations.
Need for CFD:
(i) To understand the physical events in the flow of fluids
around and within the designate objects.

(ii) These events are related to the action and interaction of


phenomena such as dissipation, diffusion, convection, shock
waves boundary layers and turbulence.

(iii) Most of these phenomena are non-linear and as a


consequence have no analytical solution which motivates
the need for the numerical solution of the associated partial
differential equations.
Navier Stokes Equation

Navier Stokes
equation
Background
Independent of the specific application under study, the following sequence
of steps generally must be followed in order to obtain a satisfactory solution.

1. Problem specification and geometry preparation.

2. Selection of governing equations and boundary conditions.

3. Selection of gridding strategy and numerical method.

4. Assessment and interpretation of results.


SOME FINITE VOLUME SCHEMES
With quadrilateral control volumes, the main distinction in the various finite
volume methods lies in the positioning of mesh point or nodal parameters
used in the discretisation. There are two main choices:

1) Cell centre Schemes 2) Cell Vertex schemes


Unknowns are associated Unknowns are associated
with the centres of the cell with the vertices of the cell
Discretization Method in one dimension
2-D

3-D
Difference based Methods
Let us consider the one-dimensional problem on a uniform mesh
  u ' b  x  u  '  S  x  on  0,1 ;  1.a 
u  0   uL , u  1  uR  1.b 

 
We integrate equation  1 over the control volume  x 1 , x 1 
 i  2 i 2 
around the node xi to derive the approximation
F 1 F
i i
1
 Si x  2
2 2

where x  x 1 x 1
i i
2 2

We suppose that Si is the average of S  x  over the control


volume.
Choice of Convective Fluxes 
Some commonly used schemes that use atmost the three values
U i , U i 1 ,U i  2 to calculate F 1 have the following form.
i
2

 simple upwind  U i 1 b
1
 3.a 
i
2

1
 central difference   U i  U i 1  b 1  3.b 
2 i
2

1
 QUICK   3U i  6U i 1  U i 2  b 1  3.c 
8 i
2
THE UPWIND DIFFERENCING SCHEME

This scheme takes into account the flow direction when


determining the value at a cell face:the convected value
of  at a cell face is taken to be the value at the upstream
node.
We consider the one-dimensional control volume
shown in the next figure.
QUICK SCHEME

The quadratic upstream interpolation for convective


kinetics  QUICK  uses a three point upstream
weighted quadratic interpolation for cell face values.
The value of  is obtained from a quadratic function
passing through two bracketing nodes (on each side
of the face) and a node on the upstream side.
Quadratic profiles used in QUICK Scheme
Some Numerical Results and Comparison of
the Schemes.

Consider the steady convection and diffusion equation


of property  in one dimension domain
d d  d 
  u     
dx dx  dx 
with the boundary conditions 0  1 and L  0 .
Let length L  1.0 m ,   1.0 kg/m ,   0.1 kg/m-sec.
3

We consider three cases:


The exact solution when u  0.1m / sec is
1  exp  x 
  x  1
exp  1  1

The exact solution when u  2.5m / sec is


1  exp  25 x 
  x  1
exp  25   1
Method of solution using Central difference :
Integrating the equation over the control volume and
using Gauss divergence theorem we get
d d  d 
CV dx  u  dx  CV dx   dx  dx
 d   d 
   u A  e    u A  w   A    A 
 dx e  dx  w
 d   d  
 F  e  w         
 dx e  dx  w 
For nodes 2, 3 and 4,
 E  P P  W
e  w 
2 2
 d  E  P  d  P  W
     
 dx e x  dx  w x

The node 1 and 5 require special treatment.


 d  2(P   A )
For node 1, w   A  1 and    .
 dx  w x
 d  2(B  P )
For node 5, e  B  0 and    .
 dx e x
Substituting these values, we have a system of
discretised equations at each node which can be
put in a compact form as:
a   a   a   Su
P P W W E E

with central coefficient


a  a  a  Sp
P W E
The following table gives the tabulated form of the coefficients
at the various nodes

Node a a Sp Su
W E
1 0 D-F/2 -(2D+F) (2D+F) A

2, 3, 4 D+F/2 D-F/2 0 0

5 D+F/2 0 -(2D-F) (2D-F)


B


F  u D
x
The matrix form of the equations at each node is given by
 1.55 0.45 0 0 0  1  1.1
 0.55 1.0 0.45 0 0    0 
   2  
 0 0.55 1.0 0.45 0  3   0 
     
 0 0 0.55 1.0 0.45 4  0 
 0 0 0 0.55 1.45  5  0 
The solution to the above system is
1  0.94211 
  0.80060 
 2  
3   0.62764 
   
4  0.41625 
5  0.15789 
Central Difference:
Case 1: u=0.1 m/sec with 5 grid nodes
Case 1: u=2.5 m/sec with 5 grid nodes
Case 3: u=2.5 m/sec with 20 grid nodes
Upwind Scheme:
Case 1: u=0.1 m/sec with 5 grid nodes
Case 2: u=2.5 m/sec with 5 grid nodes
Case 3: u=2.5 m/sec with 20 grid nodes
SAMPLE APPLICATION OF CFD RELEVANT TO
NSTL
CFD Analysis for Varunastra with Control Surfaces

Requirement: Computationally evaluate the performance of three


versions of control surfaces of Varunastra. (Fig. 1(a) ,1(b) and Fig. 1(c))

Mesh: A hybrid mesh with around 0.9 million hexahedral cells and 2.6
million tetrahedral cells was used to discretise the domain for solving all
the configurations. (Fig. 2(a), 2(b) and 2(c))

Mesh generation: Ansys ICEM CFD version 10

Solver: Fluent 6.2

Hardware: SGI Altix 350 (8 nodded parallel processing machine)


Objective
To study the hydrodynamic performance of the fin to
estimate the following:

(i) Lift generated by the foil sections of small and large rudders. This
will give the relative merit of the sections.

(ii) Lift generated by small and large rudders when attached to the
torpedo.

(iii) Moment of the C.G. of the torpedo caused by the deflection of small
and large rudders.

(iv) Actuator torque requirement for small and large rudders


Varunastra-Fig.1(a) Large control surfaces 1(b) Small control surface

Fig. 1(c) Varunastra with eight


rudders
Fig. 1(d)
Small
Rudder
Fig. 1(e)
Large
Rudder
Fig. 2(a)- Hybrid Mesh
Fig. 2(b)-Large rudder with 5 deg angle of deflection
Fig. 2(c)- Large rudder with 10 degree of deflection
Large
Small

Fig. 3 Comparison of Large and Small rudder sections


8000

7000

6000
Lift Force(N) for 4 rudders

5000

4000

3000

2000 Large
Small

1000

0
0 5 10 15 20 25
Angle

Fig. 4-Combined lift force for four rudders together (large and small) when
attached to the torpedo
0.250

0.200

0.150

0.100
Cl

0.050

Large
Small
0.000
0 5 10 15 20 25

-0.050
Angle

Fig. 5-Lift coefficient for large and small rudder when attached to the torpedo
60000

50000

40000
Moment of Body (Nm)

30000

20000

10000

0
0 5 10 15 20 25
Angle (Deg)

Fig. 6 Moment about Torpedo CG for deflection of large Rudder


45000

40000

35000

30000
Moment of Body (Nm)

25000

20000

15000

10000

5000

0
0 2 4 6 8 10 12 14 16 18 20
Angle (Deg)

Fig. 7 Moment about Torpedo CG for deflection of Small Rudder


140

120

100
Moment at Rudderstock (Nm)

80

60

40

Large Rudder
Small Rudder
20

0
0 5 10 15 20 25
Angle (Deg)

Fig. 8 Actuator Torque Requirement for Large & Small Rudder


Performance of eight control surfaces for
Varunastra
The effect of eight control surfaces placed at 45 degrees is examined.

Rudder Deflection Drag Force (N)


Angle (Degrees) Four Fins and Rudders Eight Fins and Rudders

0 6277 6540
5 6457 6676
10 6855 6976
15 7457 7659
20 8334 8531
12000

11000

10000

9000 T orped
o
R udder
8000
T otal
7000
Lift Force (N)

6000

5000

4000

3000

2000

1000

0
0 5 10 15 20 25

Angle (Deg)
Fig. 2 Lift Force Generated by 8 Rudder and torpedo
Fig. 9- Lift Force generated by 8 Rudder and torpedo
0.30

T orpedo

0.25 R udder

T otal

0.20
Lift Coefficient

0.15

0.10

0.05

0.00
0 5 10 15 20 25

Angle (Deg)
Fig. 3 Lift Coefficient of Rudder for different angles
Fig. 10-Lift coefficient of Rudder at different angles
45000

40000

35000
T orpedo

R udder
30000
T otal
Moment (N-m)

25000

20000

15000

10000

5000

0
0 5 10 15 20 25

Angle (Deg)
Fig. 4. Moment about CG of Torpedo
Fig. 11- Moment about CG of torpedo
140

120

100
Torque at rudder stock (N-m)

80

60

40

20

0
0 5 10 15 20 25

Angle (Deg)
Fig. 5 Actuator Torque Requirement
Fig. 12-Actuator Torque requirement
Results:
1. Large rudder is hydrodynamically better than small one and generates
higher lift. Large rudder produces 40% more lift force than smaller one at
20 degree deflection.

2. The actuator torque requirement to get 20 degree rudder deflection for


large rudders is 123 Nm with four control surfaces. Only 47 Nm is
required for same deflection of the small rudder with four control
surfaces. With eight control surfaces, the actuator torque requirement is
116 Nm.

3. When compared with available experimental results, CFD predictions


show reasonably good agreement and hence are reliable.

You might also like