Vous êtes sur la page 1sur 24

LID DRIVEN CAVITY FLOW

Computational Fluid Dynamics : Pressure Poisson Equation Solution for Lid Driven
Cavity Flow at Re=1000

DECEMBER 4, 2014
SHASHANK MISHRA
University of Cincinnati

ABSTRACT
The project deals with Pressure Poisson equation solution for Lid driven cavity flow. Simulation
results
are
for
Reynolds
no
1000.
Cavity
dimension
is
0.2MX0.2M.
Two different approaches are used for pressure calculation namely Multi-path Integration and
Pressure Poisson Equation. Since a Collocated grid is used for evaluating pressure at each node
instead of staggered grid, correspondingly an error term is introduced in the second order FDM
formulation to account for constraint introduced due to Divergence theorem for pressure field.
For multi path integration technique, integration along x direction across the bottom wall and
then evaluating pressure values at each vertical node is used for first result and then same was
redone with integration base line being vertical left wall and evaluating pressure at each node
along the horizontal line. Variation of pressure along the vertical and horizontal center line and
also along the walls is studied. The governing equation and boundary conditions are established
for the pressure distribution p. Next using the pressure Poisson equation obtained by the
analytical divergence and finite differenced divergence pressure is obtained. The results are
compared with the literature available and since pressure field values for Re= 1000 were not
available, results obtained are also compared with commercial Ansys Fluent software. Pressure
Poisson equation and Multi path integration method is solved using Finite Difference Method.
Stream Vorticity formulation is used to calculate velocity distribution using Implicit ADI
technique. Matlab codes are attached in the Appendix.

CHAPTER-1
1. PRESSURE DISTRIBUTION GOVERNING EQUATION SET UP
In order to set up the governing equation for the pressure distribution, the momentum equations
in two dimensions will be considered:

2 2
+
+
=
+ ( 2 + 2 )

2 2
+
+
=
+ ( 2 + 2 )

From these expression the partial derivative with respect to x for equation 1 will be taken and
the partial derivative with respect to y for equation 2 will be taken yielding the following
expression (Note that the transient term will be neglected for the project since a converged
solution has already been determined):

2
2
2

2
3
3
+
(
)
+

+
=

+
(
+
)
2


2
3 2

2
2
2
2
3
3

+ 2+( ) +
= 2 + ( 2 + 3 )


These two expressions will now be added to each other to yield the following expression:
2 2
2
2
2 2
2
2

+
+

(
+
)
+

(
+
)
+
(
)
+
(
) + 2(
)
2
2
2
2


3
3
3
3
= ( 3 +
+
+
)

2 2 3
The continuity states the following:

+
=0

This means that the derivatives with respect to x and y are also equal to 0 as follows:

2
2
2
2
2
2
3
3
(&)
( ) + ( ) = 0 (&)
+ 2 = 0 (&)
+
=
0
+
= 0 (&)


2
3 2
3
3
+
=0
2 3

After applying these variations of continuity to the pressure equation, the pressure expression
simplifies to the following:

2 2

+ 2 = 2 (
)
2


Applying the gradient term to the expression above and simplifying the right hand expression,
the finalized pressure expression is as follows:
=

CHAPTER 2
2. BOUNDARY CONDITIONS

Figure 1: Grid set-up example


The gird represented above, is a representation of the grid that will be utilized to calculate the
pressure within the square cavity. Before this calculation can be computed, the boundary
conditions for the system must be initialized. The lower left corner of the grid, will be assumed

to have a pressure of zero. From this starting point, the pressures along the wall of the cavity
can be calculated as we traverse counter clockwise.
As we traverse along the lower wall of the cavity, the change in pressure with respect to the ycoordinate will be zero. This simplifies the pressure expression that was determined in the
previous section to the following equation:

2 2
=
+
( 2 + 2)

And since there is no variation in u with respect to the the y-direction, the expression can be
simplified further. The simplified term will be referred to as F.

This means that the pressure can be calculated by taking the integral of F with respect to the xcoordinate as follows:

This integral can be determined by using a summation expression as follows:


= (, 1) = 1

2
=2

Since the del p / del x term can be calculated using the converged solution for each point in the
grid, the pressure along the x-direction can easily be calculated. Once this is completed, the right
hand wall can be calculated in a similar fashion. The expression for the right wall can be
represented as follows:

= ( , ) =

=2

+
1
2

Where del p/ del y is represented as follows:

2 2
=
+
( 2 + 2)

Once the process of calculating the pressures along the right had wall are completed, the top and
the left and hand wall pressures should be calculated in the counter clockwise manner as

mentioned before. At the completion of this process, the boundary condition for the cavity will
be initialized and the pressure within the rest of the cavity can be calculated.

Multipath integration scheme is used to calculate the values of pressure in the lid driven cavity.
The governing equation for multipath integration scheme is the x and y momentum equation.
The procedure it follows described below:

CHAPTER- 3
MULTIPATH INTEGRATION SCHEME
GOVERNING EQUATIONS

BOUNDARY CONDITION

RESULTS FOR MULTIPATH INTEGRATION


Pressure contour plots obtained by solving the X and Y momentum equation using multipath
integration method are presented below

PRESSURE POISON GOVERNING EQUATION

BOUNDARY CONDITIONS

10

RESULTS
PRESSURE FIELD
PRESSURE- POISSON EUQATION

Discussion: The contour plot of Pressure field using Pressure Poisson equation shows minimum
pressure region along the center line region. The minimum pressure core is shifted towards left
wall due to high value of Reynolds number. Given, a high value of Reynolds number, convective
terms predominates over diffusion terms. Diffusion terms are Elliptic in nature and information
flows equally in all the direction giving a circular core region characteristic to the pressure field,
11

but convective terms are Parabolic in nature and Characteristic line is parallel to the fluid velocity,
thus stretching the core. The end result is that minimum pressure core region is not circular, but
resembles elliptical profile and is tilted towards the left wall due to positive u velocity along the
horizontal direction.
Explanation of Local high region of Pressure at the Wall Corners: We see locally high region of
pressure at top wall corners (fig: 6). Reason for such distribution of pressure field is due to
singular points present at the corner. Corner points are shared by vertical and horizontal wall
and we can use either x- momentum equation or y- momentum equation. This ambiguity is
resulting into Numerical error and physically we dont have localized high region of pressure.
This error can be eliminated but that is not the scope of this project.
6

Variation of Pressure at Vertical Centre Line

x 10

0.5

-0.5

-1

-1.5

-2

10

15

20
25
Nodal Coordinate

30

35

40

Fig-7: Variation of Pressure along the Vertical Center line

12

45

Variation of Pressure at Left Wall


1
0.8

Pressure in N/M2

0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8

10

15

20
25
Nodal Coordinate

30

35

40

45

40

45

Fig-8: Variation of Pressure along Left Wall

Variation of Pressure at Bottom Wall


3000

Pressure in N/M2

2500

2000

1500

1000

500

10

15

20
25
Nodal Coordinate

30

35

Fig-9: Variation of Pressure along the Bottom Wall

13

Variation of Pressure at Right Wall


5500

Pressure in N/M2

5000

4500

4000

3500

3000

2500

10

15

20
25
Nodal Coordinate

30

35

40

45

40

45

Fig-10: Variation of Pressure along the Right Wall


Variation of Pressure at Top Wall
6000

5000

Pressure in N/M2

4000

3000

2000

1000

-1000

10

15

20
25
Nodal Coordinate

30

35

Fig-11: Variation of Pressure along the Top Wall

14

Discussion: The fluid is expected to recirculate along the center line twice that is the velocity of
the fluid will change from positive to negative. Pressure variation depicts this phenomena since
we have a maxima peak and a minima peak at the vertical center line. Similar behavior is observed
at the Left wall. Since we assumed a zero pressure at the origin which is coinciding with the
bottom most corner of the cavity, pressure variation at the bottom wall rises steadily till the end.
Solution 5: The value of Maximum Epsilon = -0.000001 and Average value of Epsilon = -0.000013
Comparison of Pressure Field from Pressure Poisson Equation & Multi Path Integration Method

Fig-12: Variation of Pressure along the Right Wall

15

Fig-13: Variation of Pressure along the Top Wall

Fig-14: Variation of Pressure along the Left Wall

16

Fig-15: Variation of Pressure along the Horizontal Center line

Fig-16: Variation of Pressure along the Bottom Wall


17

Fig-17: Variation of Pressure along the Vertical Center Line

Discussion: The behavior of pressure field when evaluated from Multi path integration technique
follows the same pattern. Although the values and some differences can also be observed. There
is peaks along the bottom center line pressure variation which should not be present. The reason
for the difference is while evaluating pressure field from Multi path integration technique, we
are not considering the constraint imposed on pressure field in accordance with Divergence
theorem. Therefore Multi path integration technique is similar to solving Poisson equation
without modified source term. Also the accuracy of Multi path integration technique is highly
susceptible to initial path of integration selected. Different pressure values and distribution
across the domain is observed while selecting x direction and y direction as initial path of
integration.

18

Solution: 6 Literature Comparison


Discrete values of Pressure at each nodes were not available for Reynolds no 1000. Isobaric
variation of pressure field was presented in a paper and our result is compared with the same.

Pressure distribution follows the same pattern.


Owing to lack of data available in the literatures, commercial Ansys fluent software is used to
compare our results.
Fluent Problem Setup

Fig 18: A Quad Element Mesh of Square Cavity

19

The mesh contains 40000 cells and 80400 faces. Convergence achieved after 1300 iterations for
laminar flow steady state solver.
Results:

Fig-19: Contour Plot of Pressure Variation


The contour plot from Fluent shows similar pressure variation when compared with numerical
results obtained from solving Poisson equation in Matlab. Low pressure region code is elliptical
in shape and is shifted towards the right wall indicating effect of convection terms.
Since pressure is assigned a 0 value at the origin in our numerical simulation results, absolute
values of pressure cannot be compared. Absolute values of pressure are not important since we
are solving pressure field for incompressible fluid.
There are infinite no of pressure field possible satisfying the flow conditions. Pressure difference
is the parameter which is crucial and hence delta pressure is only compared with Fluent data and
a very good match is achieved.
Pressure field discrete values are extracted from fluent and corresponding graphs are compared
with results from Matlab code.

20

Fig 20: Pressure variation along the top wall from Numerical Simulation

Fig 21: Pressure variation along the top wall from Fluent
21

Fig 22: Pressure variation along the bottom wall from Numerical Simulation

Fig 23: Pressure variation along the bottom wall from Fluent
22

Fig 24: Pressure variation along the Horizontal Center line from Numerical Simulation

Fig 25: Pressure variation along the Horizontal Center line from Fluent
23

Vous aimerez peut-être aussi