[Brock17]

Hi everyone,

I'm an Aero Engineering student currently attempting to simulate flow over a cricket ball in an effort to gauge and measure aerodynamic side-force generated by the ball (or 'swing force' as it would be referred to in the sport).

I've attached the mesh scene for the ball model and surrounding domain. To briefly explain, theory and experimental work on cricket balls dictate that when the protruded seam is angled downward as shown (relative to the inlet, left) the ball should generate a force downward. This is because, at lower speeds, the flow on the 'seam-side' of the ball is tripped into turbulence by the seam and separates later than the uninterrupted flow along the non-seam (top) side of the ball.

I've ran the sim using unsteady RANS turbulence modelling (I appreciate the limitations of this method but it is a fundamental aspect of my research to use RANS by means of reducing the calculation expense) with an inlet velocity of 20m/s and a time step of 5E-05s.

In the earlier stages of the simulation, the output looked reasonable. Please find attached screenshots showing mean velocity and vorticity plots after 0.01seconds (a couple of flow through times over the ball geometry; diameter 0.072m). Although the force coefficient (plot also attached) varies, the result is positive which, alongside the velocity and vorticity plots, displays a result in-line with what i'm expecting. However, as the simulation continues to run, the force coefficient dives and eventually, settles and stabilises somewhat around a very small negative value (to reiterate, the result should be positive).

Forgive me if this speaks to an ignorance of how unsteady simulations work, but I'm confused as to why the flow in it's initial stages appears to be 'correct' but then changes so dramatically and stabilises at a point which isn't coherent with what i'm expecting?

Any advice would be hugely appreciated as this investigation forms a significant portion of my dissertation and cannot be continued until this is resolved.

Thanks and please don't hesitate to ask for any additional information.



PS: 1) I'm using Star CCM+.
PS: 2) I'm aware the axes in the screenshots provided aren't consistent. Misinterpreting the co-ordinate system and the direction of the force-coefficient monitor is NOT the issue.

[FMDenaro]

I don't know the details about your setting but I thing that the early stage has almost an inviscid behaviour and only after you get a physical onset of turbulence... and I don't think that URANS can be the best formulation to work with...

[flotus1]

The result is far from being "stabilized". A more precise term would be statistically steady. So there is no point in discussing it yet.
Estimate the frequency of the largest vortices using the Strouhal number. It is ~55Hz, so the cycle time is ~0.018s. You will have to simulate at least 10-20 cycle times before you can begin to interpret the simulation results.

[Brock17]

Originally Posted by FMDenaro

I don't know the details about your setting but I thing that the early stage has almost an inviscid behaviour and only after you get a physical onset of turbulence... and I don't think that URANS can be the best formulation to work with...

Apologies but i'm not certain I understand?

If it's any help I chose a turbulent viscous regime in my physics models.

[Brock17]

Originally Posted by flotus1

The result is far from being "stabilized". A more precise term would be statistically steady. So there is no point in discussing it yet.
Estimate the frequency of the largest vortices using the Strouhal number. It is ~55Hz, so the cycle time is ~0.018s. You will have to simulate at least 10-20 cycle times before you can begin to interpret the simulation results.

Thanks for the quick reply. I'll allow my sim to run for longer and see what happens. I do, however, seem to recall allowing a similar sim to run for longer (around 0.3s from memory) and the result persisted (oscillating very slightly about a very small negative value). That sim was ran at a different inlet velocity (30m/s) but exhibited the same overall result; an initial peak in positive side-force during the initial stages followed by a reduction that persisted.

As I say, i'll let the sim run for longer and evaluate it again then, but working under the assumption the result will persist as it stands at 0.1s in my initial post (as i'd suggest it will based on my recollection of the similar sim mentioned), i'm still none the wiser as to why the trailing wake is asymmetrical, as expected, at first but then tends back towards a symmetrical pattern (or very slightly over compensated) that you would expect to see behind a standard smooth sphere? I don't understand why the flow differential either side of the ball caused by the seam is initially exhibited but then dissipates?

Once again I apologise if this speaks to a lack of knowledge of how an unsteady sim operates.

Thanks again.

[FMDenaro]

First, compute the total kinetic energy into the whole domaine and plot it versus the time, that gives an idea of the development of a physical energy equilibrium as it must oscillate around an average value.
Second, I am not sure if you wanto also to simulate a transient from the rest...but to have a physical meaning in this, you need to start from a physical meaning initial condition. And still, URANS is not the suitable tool..