Numerical studies were done on a square cylinder and half square -half circular (hshc) cylinder. It is observed that the drag coefficient for 'hshc' is higher than square. Cd obtained is 2.8 and 2.3 for hshc and square respectively. Please find attached drag coefficient for various 3D shapes from 'Fluid Dynamic drag' by S. Hoerner text book also. It is expected that hshc has higher drag, but can anybody explain what is the physics behind this higher drag for hshc.
Thank you.
Could you elaborate more on "numerical studies"?
Code and/or method?
Re number and, in case, turbulence modeling approach?
Steady/unsteady?
Domain dimensions and boundary conditions?
Also, do you have any reference on how the values in the experiment were obtained?
Hi Paolo, i did LES for square cylinder of dia 30 mm. Infinitely long square cylinder. Flow is from left to right as per the attached figure. Low subcritical flow regime of Re 20000. Velocity inlet was given. Cylinder is kept 8D from inlet strem and slip walls are defined 8D above and below the cylinder.
There used to ESDU Engineering Science Data Unit publication , you will find all the values there if my memory is still good under section 7100
Ouch... this opens a lot more questions. Let us assume that you are properly using a code properly equipped for LES. My onest answer is that I have no idea but, I have few more questions nonetheless:
Which SGS model? Do you use wall functions?
How did you time average the solution?
Grid spacings with respect to the max y+ found at the walls?
Also, are these one shot simulations? How do you know that the grid couldn't be better?
The LES is performed in a wall-confined domain? The spanwise direction is enough large? The grid at the wall is well refined?
However, the drag in body with rear curved wall is greater, this is the reason why the car are truncated.
Paolo Lampitella
Used Smagorinsky constant model and no wall functions were used. After the initial transient stage(~ 5 flow through times, domain length of ~ 30D)only, data was captured and averaged.y+ was kept
@Fillippo Mario
Yes, LES was done in a wall confined domain. Span length was limited and periodic boundary condition was given at the span ends. Other details are mentioned in the previous answer. Thanks for pointing out the example of car shapes. I understand that higher drag is possible for HSHR shapes than square, but not able to explain why is it so OR what causes it. Can yo provide some light onto it.
Joemon Jacob , sorry, I only realized what you were actually asking for when I read the last answer by Filippo, that's why I started that inquisition.
Do you have pressure and skin friction distributions? If I had to point at something, I would point at larger back area (and a not so different flow) or/and, if present (I don't know), larger scale unsteadiness (which in the end means higher energy lost) in the rounded back case.
But I feel like the rounded back case might be more susceptible to the Re number.
We have two dominant drag on such configurations namely pressure and skin friction drag. For square one pressure drag dominates because of truncated tail. And for hshc one skin friction drag will be more because of late flow separation at the rear of body. So adding up these two different drags for two different bodies you will be ending with more total drag for hshc shape because of dominance of skin friction drag.
Mohsen Jahanmiri Thank you for providing insight to the contribution of skin friction drag. It explains well. Thanks a lot.
Paolo Lampitella Thank you Paolo. I havent checked pressure and skin friction distributions yet. I think these distributions can explain the physics well.
- Badges
- Science topic
- Similar topics
- Force