Sie sind auf Seite 1von 13

Finite Volume Methods for CFD - Project report

Sreevatsa Anantharamu

INVISCID RAMP FLOW

A two-dimensional implicit viscous compressible flow solver was developed. The code was run for the following
three cases:
Inviscid ramp at Mach 2.5
Circular cylinder at Mach 5.0
Mach 1.5 flow over a flat plate at freestream Re of 2 105
Full matrix point implicit method given in Wright et al. was implemented. The code also had an option
to run an explicit calculation using first-order explicit time integration. Fluxes for most of the cases were
constructed using modified Steger Warming method. For flow over half cylinder sonic glitch correction was
used. Details and results of each of the above three calculations is given below:

Inviscid ramp flow

Flow over a 15o ramp was examined the developed code. Inlet mach number was set to 2.5. Ambient values
of pressure and temperature was used at the inlet. Calculations were run on a 360 120 grid. The time step
for the explicit method was of the order 106 . The shock angle obtained was around 37.569o . The exact
shock angle evaluated from the M relationship is 36.9449. The pressure and temperature contour
obtained from the explicit method is shown in Figure 1 and Figure 2 respectively.

0.5

Grid : 360X120

Explicit CFL 0.9

0.4

456000
396000
336000
276000
216000
156000
96000

0.3
0.2
0.1
00

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9
Figure 1: Pressure contour at M = 2.5.

1.1 1.2

0.5

T
481
451
421
391
361
331
301

Grid : 360X120

Explicit CFL 0.9

0.4
y

0.3
0.2
0.1
00

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9

INVISCID RAMP FLOW

1.1 1.2

1.1 1.2

Figure 2: Temperature contour at M = 2.5.


Streamline plots is shown in Figure 3.

0.5

Grid : 360X120

Explicit CFL 0.9

0.4
y

0.3
0.2
0.1
00

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9
Figure 3: Streamline plot at M = 2.5.

The same case was also run with full matrix point implicit time integration scheme. The contours of pressure,
temperature and the streamline plot is shown in Figure 4, Figure 5 and Figure 6 respectively. The code
converged successfully even at a relatively large CFL of 20. Both explicit and implicit methods gave the
same result for the shock angle .i.e. 37.569o .

0.5

Grid : 360X120

Implicit CFL 20.0

0.4

456000
396000
336000
276000
216000
156000
96000

0.3
0.2
0.1
00

INVISCID RAMP FLOW

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9

1.1 1.2

Figure 4: Pressure contour at M = 2.5.

0.5

T
481
451
421
391
361
331
301

Grid : 360X120

Implicit CFL 20.0

0.4
y

0.3
0.2
0.1
00

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9

1.1 1.2

1.1 1.2

Figure 5: Temperature contour at M = 2.5.

0.5

Grid : 360X120

Implicit CFL 20.0

0.4
y

0.3
0.2
0.1
00

0.1 0.2 0.3 0.4 0.5 0.6


x 0.7 0.8 0.9
Figure 6: Temperature contour at M = 2.5.

2 INVISCID FLOW OVER HALF CYLINDER


The number of outer iterations that the implicit method took to converge was nearly 1/5th of that obtained
using explicit time integration scheme. The residual plot is shown in Figure 7. From Figure 7, the steep
convergence of the implicit method can be clearly seen.

10-1
-3

10

-5

10

-7

10

-9

Residual

10

10

-11

10

-13

Explicit
Implicit

1000
2000
No. of iterations

3000

Figure 7: Comparison of rate of convergence for the explicit and implicit method.

Inviscid flow over half cylinder

Invscid flow over a half cylinder of radius 10cm was examined at an inflow Mach number of 5. For the
explicit time integration method, favourable time step was of the order of 107 . Ambient values of inlet
pressure, temperature was used as inlet conditions. The flow field within the domain was initialised with
the freestream values of velocity, temperature and pressure. The convergence rate was low when compared
with the ramp case. I believe that it might be due to the stagnation point in the leading portion of the
cylinder. Sonic glitch correction with = 0.3a was used to prevent the occurence of zero eigenvalues near
the stagnation points. A bow shock was found in front of the cylinder. Pressure and temperature contour
plots is shown in Figure 8 and Figure 9 respectively. Note that all scalar field variables are symmetric about
y = 0. The streamline plot is shown in Figure 10.

2 INVISCID FLOW OVER HALF CYLINDER

Explicit CFL 0.9


0.4
P
3.26E+06
2.63E+06
2E+06
1.37E+06
740000
110000

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 8: Pressure contour plot at M = 5

Explicit CFL 0.9


0.4
T
1830
1525
1220
915
610
305

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 9: Temperature contour plot at M = 5

2 INVISCID FLOW OVER HALF CYLINDER

Explicit CFL 0.9


0.4

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 10: Streamline plot at M = 5


Flow computation was also carried out using the full-matrix implicit method. Calculation converged successfully at CFL of 5.0. However, at larger CFL the implicit method was unstable. The contour plots of
pressure, temperature and the streamline plots obtained from the implicit calculation can be found in Figure
11, 12 and 13 respectively.

2 INVISCID FLOW OVER HALF CYLINDER

Implicit CFL 5.0


0.4
P
3.26E+06
2.63E+06
2E+06
1.37E+06
740000
110000

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 11: Pressure contour plot at M = 5

Implicit CFL 5.0


0.4
T
1830
1525
1220
915
610
305

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 12: Temperature contour plot at M = 5

2 INVISCID FLOW OVER HALF CYLINDER

Implicit CFL 5.0


0.4

0.2

-0.2

-0.4
-0.2

0.2

0.4

Figure 13: Streamline plot at M = 5


A comparison of the rate of convergence can be seen in Figure 14. The implicit method converges about
four times faster than the explicit method.

3 SUPERSONIC FLOW OVER FLAT PLATE

10
10

-2

Implicit
Explicit

Residual

10-4
10

-6

10

-8

10

-10

10

-12

10

-14

10000
20000
No. of iterations
Figure 14: Convergence plot at M = 5

Supersonic flow over flat plate

Viscous flow over a flat plate at freestream Reynolds number of 2 105 is evaluated. The inlet Mach number
is 1.5. The plate is 1m long. The freestream temperature is set to 300K. The freestream pressure is set so
that a freestream Re of 2 105 is obtained. Viscosity is then evaluated from Sutherlands law depending on
the temperature.
A 150 70 grid was used for this case. Figure 15 shows the grid used for the calculation. The leading edge
of the plate is placed at x = 0.06m. The near wall y-direction spacing was kept at 104 m to obtain y + < 1.
Then a constant expansion ratio of 0.08 was used in the y-direction. Compound stretching function was
used in the x-direction so that the grids do not become oversly stretched. Also, mesh with aspect ratio close
to unity were used in the leading edge of the plate. The gradients near the leading edge are going to be
very large and they have to be captured as accurately as possible. The y + of the first wall spacing from the
obtained result was found to be 0.7.

10

3 SUPERSONIC FLOW OVER FLAT PLATE

0.25
0.2
y

0.15
0.1
0.05
00

0.2

0.4

x 0.6

0.8

Figure 15: Grid used for the flat plate case


The rate of convergence in the explicit case was so slow that it took hours and hours to converge. Therefore,
the simulation was terminated as it is not feasible for fine grids as used in this case. The implicit case
converged in 12000 iterations. A constant time step of 2 106 is used. The pressure, temperature and
streamline plots are shown in Figure 16 and 17 respectively. The streamline plot can be seen from Figure
19. The residual plot can be seen in Figure 20. The Mach number plot can be seen in Figure 18.

0.25
P
812
762
712
662
612

0.2
y

0.15
0.1
0.05
00

0.2

0.4

x 0.6

0.8

Figure 16: Pressure contour at Re=2 105 and M = 1.5

11

3 SUPERSONIC FLOW OVER FLAT PLATE

0.25
0.2
y

0.15
0.1
0.05
00

0.2

0.4

x 0.6

0.8

T
450.5
425.5
400.5
375.5
350.5
325.5
300.5

Figure 17: Temperature contour at Re=2 105 and M = 1.5

0.25
M

0.2

1.265
1.015
0.765
0.515
0.265
0.015

0.15
0.1
0.05
00

0.2

0.4

x 0.6

0.8

Figure 18: Mach contour at Re=2 105 and M = 1.5

0.25
0.2
y

0.15
0.1
0.05
00

0.2

0.4

x 0.6

0.8

Figure 19: Streamline plot at Re=2 105 and M = 1.5

12

3 SUPERSONIC FLOW OVER FLAT PLATE

10-5

Residual

10-7

10

-9

10

-11

10

-13

10

-15

2000

4000

6000

8000

10000

12000

No. of iterations
Figure 20: Residual plot at Re=2 105 and M = 1.5 for the implicit case

13

Das könnte Ihnen auch gefallen