[Blooper]

I'm running into some problems with this simulation, and been trying to correct it for quite some time now. Having this said, I decided to post my overall development here and, if anyone can spot any error/misconception, it would be of great help. It's a large topic so I'll divide it into parts:

1) The problem itself:
This is an ASHRAE lab experiment to measure pressure drop on an internal flow expansion, and it consists of the following data:

Air flows through a 200mm diameter tube and is fully developed. The volumetric flow equals to 700 L/s and the density of said air is 1.184 kg/m³ (this leads to an inlet speed of about 22.3 m/s).

It then goes through a 300mm smooth expansion (length parallel to the flow) connecting the previous tube (the 200mm one) to a 300mm diamater tube. This means the angle of the expansion is obtained by: Tan(a)=(300-200)/300, so the angle "a" equals to the arcTan of 1/3 (about 19 degrees). It is also good to notice that, due to continuity and this low-mach flow (nearly incompressible), the outlet speed on the 300mm tube should be around 9.9 m/s.

2) The input data
I've chosen to go with a Navier-Stokes approach, using the SST turbulence approach. Using a 2D mesh, I also set the simulation to be axisymmetric (AXISYMMETRIC= YES).

Now, for given density (1.184 kg/m³), and a chosen temperature of 300K (leading to a viscosity of 1.853e-5), the Reynolds (using the inlet diameter and speed) is 284722.2882, this is accompanied by REYNOLDS_LENGTH= 0.2.

All the input for gas constants were left on the standard values. Now this leads to a speed of sound of 347.715 (a = SquareRoot[gamma*R*T]), thus leading to a Mach of 0.06408 (inlet speed divided by a).

I then proceeded to use the secondary inlet type, the mass flow (INLET_TYPE= ( MASS_FLOW )), and set the inlet parameters as required:
MARKER_INLET= ( inlet(boundary), 1.184(density), 22.3(speed), 1.0(direction), 0.0, 0.0 ).

The walls were then initially set as adiabatic walls (MARKER_HEATFLUX, 0 heat transfer), but the software crashes due to the inlet velocity profile (since the only option is constant inlet, it crashes with the 0-speed-on-wall condition). So I over-extended the inlet a bit, set a small portion of euler walls on the entrance, and then set the adiabatic walls for the flow to fully develop (I set a long length for it to develop).

Lastly, for the outlet marker, I needed a back pressure. This one was kinda funky for me, cause I usually set the back pressures at 0 to calculate the pressure drop on other softwares, but on this one you apparently must set the static pressure the air will have at that position. Since I want it to leave at the same temperature and density (300K and 1.184 kg/m³, respectively), I set the back pressure to P=rho*R*T=102244.

3) THE PROBLEM
All's well and done, and with a little work on the numerical parameters, the simulation ran smoothly. The results are what kept bugging me for a long time now.

First of all, it sets both inlet and outlet temperatures up at 381K (approximately). Sure, the temperature didn't vary, but why is it at 381K? I thought the FREESTREM_TEMPERATURE= 300K was a sufficient condition?

Second, the inlet density was correct, but the outlet density is around 0.93 kg/m³. Then again, this is probably correlated with the temperature issue.

Lastly, the inlet Mach (with constant profile) resulted in 0.0568. This is probably because the speed of sound ALSO varied (due to the temperature now being 381K). The outlet Mach is at around 0.0420, but then again, this is probably due to both the temperature and density variations, leading to a different speed of sound used on the Mach relation.

4) Final notes
When it all comes to it, I have a feeling that my problem lies on not setting some correct input for the temperature, which is what probably leads to the large density variation as well. If anyone knows what might be the issue, any help is appreciated. Other than that, I'll keep on trying to correct my calculations over here.

Thanks in advance for taking your time to try to help!

[Blooper]

I also noticed that on the output files, the mass-flow continuity isn't really respected between any 2 cross sections (using only the output data and calculating density*velocity*area). Of course there's something terribly wrong on my setup but I just can't seem to find what it is.

[fpalacios]

Originally Posted by Blooper

I'm running into some problems with this simulation, and been trying to correct it for quite some time now. Having this said, I decided to post my overall development here and, if anyone can spot any error/misconception, it would be of great help. It's a large topic so I'll divide it into parts:

1) The problem itself:
This is an ASHRAE lab experiment to measure pressure drop on an internal flow expansion, and it consists of the following data:

Air flows through a 200mm diameter tube and is fully developed. The volumetric flow equals to 700 L/s and the density of said air is 1.184 kg/m³ (this leads to an inlet speed of about 22.3 m/s).

It then goes through a 300mm smooth expansion (length parallel to the flow) connecting the previous tube (the 200mm one) to a 300mm diamater tube. This means the angle of the expansion is obtained by: Tan(a)=(300-200)/300, so the angle "a" equals to the arcTan of 1/3 (about 19 degrees). It is also good to notice that, due to continuity and this low-mach flow (nearly incompressible), the outlet speed on the 300mm tube should be around 9.9 m/s.

2) The input data
I've chosen to go with a Navier-Stokes approach, using the SST turbulence approach. Using a 2D mesh, I also set the simulation to be axisymmetric (AXISYMMETRIC= YES).

Now, for given density (1.184 kg/m³), and a chosen temperature of 300K (leading to a viscosity of 1.853e-5), the Reynolds (using the inlet diameter and speed) is 284722.2882, this is accompanied by REYNOLDS_LENGTH= 0.2.

All the input for gas constants were left on the standard values. Now this leads to a speed of sound of 347.715 (a = SquareRoot[gamma*R*T]), thus leading to a Mach of 0.06408 (inlet speed divided by a).

I then proceeded to use the secondary inlet type, the mass flow (INLET_TYPE= ( MASS_FLOW )), and set the inlet parameters as required:
MARKER_INLET= ( inlet(boundary), 1.184(density), 22.3(speed), 1.0(direction), 0.0, 0.0 ).

The walls were then initially set as adiabatic walls (MARKER_HEATFLUX, 0 heat transfer), but the software crashes due to the inlet velocity profile (since the only option is constant inlet, it crashes with the 0-speed-on-wall condition). So I over-extended the inlet a bit, set a small portion of euler walls on the entrance, and then set the adiabatic walls for the flow to fully develop (I set a long length for it to develop).

Lastly, for the outlet marker, I needed a back pressure. This one was kinda funky for me, cause I usually set the back pressures at 0 to calculate the pressure drop on other softwares, but on this one you apparently must set the static pressure the air will have at that position. Since I want it to leave at the same temperature and density (300K and 1.184 kg/m³, respectively), I set the back pressure to P=rho*R*T=102244.

3) THE PROBLEM
All's well and done, and with a little work on the numerical parameters, the simulation ran smoothly. The results are what kept bugging me for a long time now.

First of all, it sets both inlet and outlet temperatures up at 381K (approximately). Sure, the temperature didn't vary, but why is it at 381K? I thought the FREESTREM_TEMPERATURE= 300K was a sufficient condition?

Second, the inlet density was correct, but the outlet density is around 0.93 kg/m³. Then again, this is probably correlated with the temperature issue.

Lastly, the inlet Mach (with constant profile) resulted in 0.0568. This is probably because the speed of sound ALSO varied (due to the temperature now being 381K). The outlet Mach is at around 0.0420, but then again, this is probably due to both the temperature and density variations, leading to a different speed of sound used on the Mach relation.

4) Final notes
When it all comes to it, I have a feeling that my problem lies on not setting some correct input for the temperature, which is what probably leads to the large density variation as well. If anyone knows what might be the issue, any help is appreciated. Other than that, I'll keep on trying to correct my calculations over here.

Thanks in advance for taking your time to try to help!

Thanks for the detailed information,
First of all, in order to discard any problem, could you please run a truly 3D simulation with the SA model? With respect to the parameters and boundary conditions.
In internal flows the nondimensionalization is done in
void CEulerSolver::SetNondimensionalization(CGeometry *geometry, CConfig *config, unsigned short iMesh) (inside of solver_direct_mean.cpp) with the free stream values we set the initial conditions.

On the other hand, the inflow using characteristics is done in
void CEulerSolver::BC_Inlet(CGeometry *geometry, CSolver **solver_container,
CNumerics *conv_numerics, CNumerics *visc_numerics, CConfig *config, unsigned short val_marker) (inside of solver_direct_mean.cpp)

and the outflow in
void CEulerSolver::BC_Outlet(CGeometry *geometry, CSolver **solver_container,
CNumerics *conv_numerics, CNumerics *visc_numerics, CConfig *config, unsigned short val_marker)

same file.

I hope this helps,

Thanks for using SU2,

Francisco