Channel flow simulation using Converge CFD   (Converge CFD, Skill Lync, Easy)
Aim: To simulate laminar flow in a channel in 2D.
Introduction:
The laminar flow in pipe is a very simple problem which can be a good problem to start with learning a new CFD software. In this report, a channel flow is simulated for 3 mesh sizes. The analysis is in 2D, so the shape of cross section of channel won't matter. So, a cuboidal channel is created with length as 0.1 m and cross-section as 0.01 x 0.01 m. The image below shows the image of computational domain with its named boundaries.
Here are the details of case setup:
Mesh Sizes: (Element size along X, Y, Z axes) (in metres)
There is only one element along Z axis.
1) (0.00020, 0.00020, 0.01) ---> 50 elements along Y axis
2) (0.00015, 0.00015, 0.01) ---> 75 elements along Y axis
3) (0.00010, 0.00010, 0.01) ---> 100 elements along Y axis
Type of simulation: Steady state with only gas flow simulation.
Steady State monitor properties: (Used to stop simulation when difference between two monitor variable values in sample is lower than tolerance)
Monitor variable: Average Velocity at outlet calculated from mass average flow otput file.
Monitor start delay: 200 cycles
Sample size: 100 cycles
Relative Tolerance: 0.005
Boundaries:
Inlet: Inflow boundary condition with pressure inlet at 100001 Pa.
Outflow: Outflow boundary condition with pressure outlet at 100000 Pa
Top and Bottom: No Slip walls.
Front and Back: 2D faces. (It tells solver to not solve in Z direction)
Initial region pressure = 100000 pa
Solver Properties:
Solver: Pressure based 'SIMPLE' solver with 'step' as flux limiter.
Details: 2 minimum iterations and 20 max iterations per cycle. Tolerance = 0.001
Minimum Cycles = 0
Maximum Cycles = 15000
Minimum time step = 1e-9 sec
Maximum time step = 1 sec
| Case number | Convective CFL | Mach CFL |
| 1 | 5 | 50 |
| 2 | 5 | 500 |
| 3 | 1 | 250 |
(The first case had larger element sizes, so the solver uses larger time steps for each cycle based on CFL criteria. This caused the solution to converge without becoming unstable. The limiting criteria for deciding time step was found from 'time.out' file as 'CFL_mach'. It was kept as 50. Since, the flow physics of this problem is not in supersonic range, it was thought that if this CFL number is increased, faster convergence can happen as the solver would take even larger time steps. So this CFL number was increased from 50 to 500. This worked out well for second mesh but for third mesh it didn't work good. The solution was getting unstable.
It was found that the 'Convective CFL' number of 5 was causing problems along with 'Mach CFL' number of 500. The solver was taking larger time steps than required. Then the Convective CFL was reduced to 1 and Mach CFL was reduced to 250. This led the solution to become stable in third mesh case.)(This gives me the learning about which values to change and by what value.)
Results:
(The images are in sequence: mesh 1, 2, 3)
1) Mesh:
The image above shows a part of meshes used. The height of the image is y = 0.01m. Element sizes are mentioned in brackets.
2) Velocity Plots.
The laminar flow has a parabolic profile.
As the mesh is refined, the value of maximum velocity converges to 1.25 m/s. Also, the curve becomes more accurate in shape.
3) Mass average velocity:
4) Mass flow rate
The mass flows from inlet to outlet and it can be seen that mass doesn't get reduced or increased anywhere. This says that mass in conserved in flow. Also the first case shows mass flow less than 0.01 kg/s and rest 2 cases shows value higher than 0.01 kg/s.
5) Pressure plots:
The pressure is not seen reducing to 100000 Pa at outlet.
6) Cell count plot:
The cell count plot shows that mesh cells remained same over the cycles. This tells that software didn't create or destroy the cells by its own.
7) Velocity Contours:
The velocity contour plots show the entrance length or we can say development of flow from straight profile to parabolic profile.
8) Pressure Contours:
The pressure contours show a very important feature. The entrance length can't be calculated by looking at velocity contours but we can give an approximate answer of length looking at pressure plots. The variation of pressure is seen till approximately 70% of length or 0.07 m. When pressure stops changing very much, it can be said that the flow is converged to parabolic profile.
9) Velocity animation videos:
Conclusion:
The channel flow problem is simulated and it gave a basic idea of simulating steady flow simulations using Converge CFD.