Wednesday, May 3, 2017

Fluid flow movies

I have begun to make movies of fluid flow at Re in the range 60-150, at the moment. The results are very interesting.

This first series is generally based off the Featflow benchmark geometry, except that the upper boundary has been lowered by 0.01m - to provide symmetric geometry.

I have a number of full sets, up to eight simultaneous flow fields, at various Re.

The source of the driving mechanism behind turbulence, can be seen in the full sets. The eye is an amazing analysis device.

At this stage, I will give copies of these movies to collaborating colleagues. In the longer term, I plan to distribute these movies, flow studies, code, research findings, visualization studies, papers, via a paid site. Please contact me directly.

Kind regards,
Desmond Aubery

Tuesday, April 4, 2017

The 'Eureka' moment

After 19 years of wondering why fluid flow over a cylinder becomes unstable in even seemingly gentle flows. After some 12 odd years of intensive research, including using a number of research & commercial CFD solvers.

After reading countless academic & research texts, the underlying physics behind the phenomenon has emerged. At long last.

I had had glimpses of the phenomenon over the years, but made sure to investigate all other directions, in order to eliminate dead ends, & locate the root cause mechanism.

The 'inner secrets' of the incompressible Navier-Stokes are, finally, coming out of hiding. Pandora's box.

A combination of three things has allowed the recent discovery acceleration:
1.  GMSH allows really decent meshes to be built.
2.  Elmer simulation solvers & GUI have made simulation turnaround fast. VTK data storage has saved much-needed storage space. The NS solver is pretty fast when run from the command line.
3.  VisIt visualization software has made analysis & visualization a breeze.

The research phase now moves in a new direction - exploring the new-found phenomena in great detail, across a number of scales. Thereafter, the focus will be to follow the phenomenon along increasing Reynolds number.

It would be a gross understatement to say that I am incredibly excited at this point.

Best regards,
Desmond Aubery

Tuesday, November 29, 2016

Unusual finding - No critical Reynolds Number

After many simulations, of water flow over a tube contained within a 2D duct, searching for a 'Critical Reynolds Number' above which the flow becomes unstable, the following astounding findings emerged:

1.  No critical Reynolds number exists.
2.  Instability always emerges, given a long-enough time frame.
3.  The time to reach instability increases as Reynolds number is reduced.
4.  Simulation times of 1-2 hours may be required at low Reynolds numbers.

The simulation results were found during exploratory runs during benchmarking tests that had been performed against a historic published Royal Society paper which analyzed vortex patterns. After these runs, I decided to keep on reducing incoming flow velocity & hence Reynolds number, in search of the ever-elusive Critical Reynolds number.

So, there you have it!

All the best,
Desmond Aubery

Friday, September 16, 2016

Progress update

The project has now moved into a benchmark comparison phase.

The benchmarks have been extracted from a 1930 paper published by the Royal Society, where the Karman-type vortex streets were investigated experimentally, in great detail. Early simulation results are very promising.

In addition, the project is also investigating accelerating flow fields.

Desmond Aubery

Monday, June 6, 2016

Strouhal number investigations

During the past few months, numerous simulations have been performed - investigating various local Strouhal number values. Basically, to determine the relative relationships between the terms of the acceleration.

This is a very under-reported part of fluid flow science, it appears.

Some very interesting findings are emerging. A number of go-no go zones are emerging. Rether intriguing.

Kind regards,
Desmond Aubery

Wednesday, March 16, 2016

Acceleration terms, waves & time-steps

In the past, I used a fixed time-step to geometric spacing, in order to ride of the apparent 'wave-front'. This time-step/geometric spacing formulation worked fine for all my early research. It was rock-solid stable & was based on the relationship between the acceleration terms.

A month, or so back, I began experimenting with alternative timing, which made allowance for the possible effect of the viscous term, on timing. This work showed the convection wave collapsing after a certain simulation time. This got me thinking about whether was such a thing as the 'right' time-step.

I've recently discovered a link between suitable time-step sizes & the Strouhal Number.

What has been emerging is that certain time-steps allow stable solution into wave-forms, while others appear to decay the wave-forms. In many ways, some form of quantisation seems to be at work for the stable wave-forms. A very interesting research route appears to be emerging.

Kind regards,
Desmond Aubery

Monday, February 29, 2016

Near-onset physics

A huge number of simulations have now been run, starting off at very low velocities & gradually climbing upwards towards the observable point at which flow instability occurs.

Simple observations:
1.  Well below the onset point:
1.1  Convective wave emerges for a short time;
1.2  Viscous effects rapidly 'gobble up' the convective wave;
1.3  Convective wave fades away.

2.  Near the onset point - from below:
2.1  Convective wave lasts for longer duration;
2.2  Viscous effects gradually 'gobble up' the convective wave;
2.3  Convective wave slowly fades away.

The current round of simulations has reached the point where the convective effects are beginning to co-exist with the viscous effects. The convection-dominated waves begin emerging from a soup of convection-viscous interactions.

The onset point is observed to be gradual, rather than a sudden all-or-nothing event.

Mr Turbulence has come to town...  :)

Best regards,
Desmond Aubery