[Brock17]

Hi,

I've posted before briefly explaining my University project which involves simulating flow around a cricket ball. To remind you, by angling the raised seam along the equator of the ball relative to the inlet, I expect a side-force to be generated tending towards the direction in which the seam is angled. This force is generated (for low(er) Reynolds / low(er) speed flows) by the seam-side boundary layer transitioning into turbulence and therefore delaying separation whilst the unimpeded boundary layer remains laminar.

As I posted previously, I attempted to simulate this flow using Menter-SST DES turbulence modelling, and got a small negative side-force generated as opposed to a significant positive side-force (positive defined as in the direction of the seam angle, as expected and described above). This is clearly incorrect, but I believe (please don't hesitate to correct me if i'm wrong) can be explained by the characteristics of the modelling used; this approach adopts a RANS (SST) approach in the near wall region, therefore assumes turbulent mixing in regions with a velocity gradient and therefore assumes all boundary layers are turbulent. My research tells me that the raised seam thickens and weakens the turbulent boundary layer (as opposed to transitioning a laminar boundary layer) and therefore the seam-side boundary layer separates earlier than the unimpeded side (hence the side-force generated is reversed - for cricket aficionados this is essentially how 'reverse swing' is generated at higher speeds).

I have, however, since attempted to use the Gamma ReTheta transition model, and I see a very similar pattern of results (a negative side-force is generated). After showing one of my lecturers the results, he plotted a turbulent viscosity scene and concluded that, in this instance, the issue was that the laminar boundary layer transitions at the seam (good) but then, somehow, 're-laminarises' (for want of a better phrase) along the surface (why this causes earlier separation than the unimpeded side and results in a negative force is beyond me). I suggested that using a higher velocity flow may cause the seam-side boundary layer to transition, but the results still remain the same.

My supervisor also advised attempting a Spalart-Allmaras model. As this is also a RANS model, i'm unsure how this result would differ from my initial SST solution; surely the SA model would assume turbulent boundary layers also when my eventual 'correct' solution relies upon transition?

My apologies for the lengthy post. As the transition model appears to be unsuccessful i'm unsure as to where I can go next with this investigation.

Thanks for reading (if you made it this far) and I would appreciate any feedback.

PS. 1) Using Star CCM+
2) My investigation specifically involves trying to use RANS / RANS based modelling; trying LES isn't an option.