CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   2D Hypersonic Inlet in FLUENT - Convergence Issues (http://www.cfd-online.com/Forums/fluent/105240-2d-hypersonic-inlet-fluent-convergence-issues.html)

Fraisdegout July 25, 2012 22:15

2D Hypersonic Inlet in FLUENT - Convergence Issues
 
5 Attachment(s)
Hello,

I am trying to simulate a simple hypersonic flow with FLUENT (Mach 5, altitude of 10,000 m) on an axisymmetric inlet and am currently having some convergence problems. I'm including as much information as possible so that you can have a very good idea of the problem.

Model (Picture 1): I generated the model in Catia V5 and exported it into GAMBIT as an .igs file. It contains the external ramp of the inlet, cowl, and throat area, all enclosed in the computational domain.


Mesh (Picture 2): I used GAMBIT 2.4 to generate the mesh for this model. I wanted to start with the simplest possible case by neglecting all viscous effects, therefore the mesh does not contain a mesh boundary layer. It is an unstructured grid to which I added a sizing function on every inlet wall (external and internal ramps, and cowl) to capture more accurate results in areas of interest. The details of the Sizing Function are as follow: "Start Size" = 0.6, "Growth rate" = 1.2, "Max Size" = 8.


BOUNDARY CONDITIONS (Picture 3): Before I describe the parameters I used in FLUENT, I wanted to show you how I set up my boundary conditions in GAMBIT, which could very well be a source of error for my problem. My intake is set up as a "Pressure Inlet", the inlet walls as "Wall", two "Pressure Outlet" BCs were adopted at the outlets, one for the throat portion and the other for the ambient portion outside of the cowl. Finally, and this is where I have had some doubts, I set up a "Pressure Far Field" at the top portion of the computational domain with Free stream conditions of Mach 5 at 10,000 m.
For the top portion, I have wondered if I should set it up as "Pressure Outlet" instead of "Pressure Far Field", in which case should I leave the "Backflow Direction Specification Method" as "Normal to Boundary" ? Please advise...


FLUENT 6.3 - Problem Setup: Here I will list every parameter I set up after importing the mesh file from GAMBIT.

1.Define>Models>Solvers> Density Based, Implicit, 2D, Steady, Absolute Velocity Formulation, Green-Gauss Cell Based Gradient Option, Superficial Velocity Porous Formulation.

2.Define>Models>Energy> Energy Equation checked

3. Define>Models>Viscous> Inviscid

4. Define>Materials> Air, Ideal Gas (Left all other values as default)

5. Define>Operating Conditions> Pressure set to 0

6. Define>Boundary Conditions>

a. Wall > Nothing needed here
b. Pressure Inlet > Momentum>
- Gauge Total Pressure = 13987176.3 Pa (Altitude = 10,000m, M=5)
- Supersonic/Initial Gauge Pressure = 26436.3 Pa
- Direction Spec. Method = Normal to Boundary
Pressure Inlet > Thermal>
- Total Temperature = 1339 K (Altitude = 10,000m , M=5)
c. Free Stream > Momentum>
- Gauge Pressure = 26436.3 Pa (Altitude = 10,000m)
- Mach Number = 5
- X-Component Flow Dir. = 1
- Y-Component Flow Dir. = 0
Free Stream > Thermal
- Temperature = 223.15 K (Altitude = 10,000m)
d. Pressure Outlet #1 (for outside of Cowl) > Momentum
- Gauge Pressure = 0 Pa
- Backflow Direction Spec. Method = Normal to Boundary
Pressure Outlet #1 > Thermal
- Backflow Total Temperature = 1339 K (Altitude = 10,000)
e. Pressure Outlet #2 (At Throat exit ) > Momentum
- Gauge Pressure = 0 Pa
- Backflow Direction Spec. Method = Normal to Boundary
Pressure Outlet #1 > Thermal
- Backflow Total Temperature = 1339 K (Altitude = 10,000)

7. Solve>Controls>Solution> First Order Upwind, CFL = 0.5, AUSM Flux Type

8. Solve>Initialize>Compute from Pressure Inlet>Init.

9. Solve>Monitors>Residual> Print/Plot Checked, Normalization = Scale checked, Convergence Criterion = Absolute, Continuity/X-velocity/Y-Velocity/Energy Absolute Criteria = 10^-6

RESULTS & CONVERGENCE ISSUE (Pictures 4 & 5 ):
From Picture 4, you can see that the residuals stagnate at 10^-5 and I cannot seem to obtain a convergence to 10^-6. We expect a series of oblique shocks to appear along the external ramp, and some shock-shock interactions around the throat areas and a possible shock train in the actual inlet throat. This is precisely what picture 5 is starting to confirm, but more iterations are needed to show better results.

The problem lies here...why are the residuals stagnating at 10^-5 ??? What corrections can be made for this run to converge ??

Thank you so much for all the time you took reading this, and I hope you can help me figure out this problem.

flotus1 July 26, 2012 01:34

My initial guess is that the mesh you are using is much to coarse to resolve the shocks you expect in your simulation, especially when using a first order upwind advection scheme.
Adaptive mesh refinement could be an option here, espacially since you are using tri-meshes.

The reason why your residuals do not drop below 1e-6 is that the numerical accuracy of your system is reached. The computation is actually converged. Running the case in double precision will allow the residuals to drop further, but this is not the reason why you don't get the results you expect.

Fraisdegout July 26, 2012 13:41

Thank you for your resonse, I'm going to try and refine the mesh then, in addition to using double precision. I will let you know of the results.

wolverine March 14, 2013 02:45

did u get answers... well if you dont mind can you send me your model??
 
thanks in advance

AshwaniAssam January 25, 2014 13:41

Quote:

Originally Posted by Fraisdegout (Post 373761)
Thank you for your resonse, I'm going to try and refine the mesh then, in addition to using double precision. I will let you know of the results.

It will be nice if you can post was your problem solved or not.

sunilncz April 16, 2015 11:12

hypersonic flow
 
hello all
I want slove bluent conical rockt head in fluent (m=6) , but i can not solve them , maybe help me?


All times are GMT -4. The time now is 18:34.