Beyond Viscous Drag, Cavitation, Laminar Flow, & Reynolds

[genergy]

When using Computational Fluid Dynamic (CFD) modeling you are able to see incredible complexity in water energy transfer. The Thermodynamic energy transfer occurs at much higher velocity so it does not apply in most of the cases I am looking at. What are the other parasitic energy transfers that occur besides Viscous Drag, Cavitation (an inefficiency is the way I think of it), Laminar Flow, and the Reynolds Effect?
http://hyperphysics.phy-astr.gsu.edu/hbase/lindrg.html
When you look at the basic formulas on Hyper-physics website you only see Viscous Drag modeled.

I do not see much in the way of Energy Transfer Analysis anywhere on the web.

 

[genergy]

Look at a simple 2D CFD and you see turbulence, friction, and energy in a new way. 3D is even more impressive.

 

[genergy]

Velocity is a 6X Energy Drain. The Faster an object moves in water the more energy resistance: the energy drain is 6 times the V.

 

[genergy]

How do you engineer an optimum velocity when any increase in velocity means significantly more parasitic energy loss?

 

[genergy]

The resistance is not only when an object is falling it is ALSO WHEN IT IS FLOATING UP... negative floating...look at how fast a 20 percent buoyant sphere SHOULD be moving...with one quarter g upward acceleration ("-0.25) g.

 

[jedishrfu]

Thanks for the information on CFD!

 

[genergy]

https://www.cfd-online.com/Forums/main/11682-magazine.html
Seems like there is not a lot online but there are a LOT of software products!
http://www.ticktutor.com/top-10-computational-fluid-dynamics-cfd-softwares/

It is interesting to see what are described as "Fluid Inventions"...
https://www.resolvedanalytics.com/theflux/fluid-dynamics-best-inventions

 

[boneh3ad]

First, what is the "Reynolds effect?"

Second, most commercial CFD programs do a poor job of modeling things like viscous drag. This comes especially from their difficulty in modeling turbulence and the transition from laminar to turbulent flow. The computational requirements to do this accurately are enormous.

Third, the 6 times velocity relationship appears to be related to buoyancy and not a general rule. The factor of 6 arises in a very specific situation called Stokes flow, where the Reynolds number is very small and the flow is dominated by viscosity. In that case, drag is directly proportional to velocity (the factor of 6 is actually fairly unimportant). In most flows, the relationship is actually that drag is proportional to the square of velocity.

In fluid mechanics, energy dissipation is generally accomplished by the effects of viscosity and the rate at which it occurs varies depending on the situation. It has been fairly well-studied if you know where to search. You should look into "viscous dissipation."

 

[genergy]

"Reynolds Effect" is what I think of when I think of all that happens around the "Reynolds Number" ( https://en.wikipedia.org/wiki/Reynolds_number ) which seems to be a weird quantity when it is supposed to have different effects at different velocities. How can you have a constant quantity when it does things differently with different fluid motion?
My limited experience with CFD has been shocking! I hired someone to do a "simple" model. The engineer explained that he needed to use 800 computers to "run" the model (!) and then he told me the model crashed the computers! He fixed it by changing making the time variable slower by a factor of 20 before he could even run the simulation/model.
When I saw the video I could see fluid movement that makes perfect sense and is so complicated.
http://i.ytimg.com/vi/7RYvWKRpDII/maxresdefault.jpg

Viscous Dissipation seems to relate to higher speeds in fluid dynamics than what we work with.

"I would define viscous dissipation as the transformation of kinetic energy to internal energy (heating up the fluid) due to viscosity. This includes both turbulent kinetic energy and mean flow kinetic energy." Jonas Larsson
Does this seem like a reasonable definition?

Most of what I study occurs at slow speed -20 m/s in fresh and seawater.
I have not verified but I have been told by several Marine Engineers and Naval Architects that Stokes flow and small Reynolds number are the scenarios we are looking at. At fluid or object velocity in water above 20 m/s things start to change and turbulent effects could create unstable water effects.

I have not read or even seen anything talking about shapes of objects in water and how different material drag coefficients affect acceleration underwater.

 

[boneh3ad]

Viscous dissipation happens at any velocity. The degree to which it occurs depends on largely Reynolds number. Again, it is THE largest source of.energy dissipation in a typical fluid flow and is fairly well-studied.

Also note that 20 m/s is not some magical number where things change. It honestly seems quite fast for Stokes flow to be relevant, but it's Reynolds number that determines that, not velocity.

 

[cjl]

I agree with boneh3ad - at 20m/s in water, I'd really expect you to be at a high enough reynolds number for stokes flow to no longer really apply, but it will depend on your details of your scenario.

 

[Klystron]

800 distributed computers can be difficult to control outside community networks though fine if you mean number of processors.

Two improvements to your network might improve CFD computations. Agree with reducing repetition rate. Distributed computing means distributed data. Careful design ensures that processes do not wait for results from upstream computations. Time stamps can be applied and stored with intermediate results as a measure as part of data integrity and error checking. Adapt the repetition rate to the CFD error metrics. Runs slower but machines shouldn't crash.

Let's call a computation function and it fails! Function returns an error code to the control programs which make sure no data or intermediate results were altered, then "check the time". If downstream functions would be waiting, then the master might alter the sequence of function calls, stretching the frame as needed. Something analogous happens if processes fail.

Optimize input/output (I/O) interrupts within processes. Tendency is to read data, compute, write results. Distributing these functions over many processors allows the operating systems (OS) to balance operations, reading/writing with CPU intensive computations. In this context I/O might include memory operations across the network with actual read/writes to "storage".

[BTW I like "Reynolds Effect".]

 

[bigfooted]

For your own problem, you could compute the Reynolds number, which is the:
density (in your case that of water, so 1000 kg/m^3 if your object is under water, air if it is above the water)
times
velocity (20 m/s)
times
the length scale, in your case something like the diameter or length of the largest cross-section normal to the flow,
and you divide it by the viscosity (for water, approximately 0.001 Pa.s, so you could also multiply density x velocity x length by 1000)

if the number is 'large' (say 1000 or more), it will not be laminar. Note that for water, you already multiply by 1000, density is 1000 and your velocity is 20, so if you have a ball of 1 meter diameter, your Reynolds number is already 20,000,000.

I have not read or even seen anything talking about shapes of objects in water and how different material drag coefficients affect acceleration underwater.

Note that the material does not affect the drag, it is the surface roughness that matters.
Note also that it doesn't matter if it's in water or air, what matters is the Reynolds number.
If you can be more specific on what you are interested in, we could maybe refer you to some literature. air bubbles? submarines?

How do you engineer an optimum velocity when any increase in velocity means significantly more parasitic energy loss?

What do you mean with 'optimum'? In general, you either do some measurements or simulations of your problem of interest at different working conditions and you then extrapolate trends from that to determine an 'optimum'. For simple problems, you could compute the optimum by hand. For extremely difficult problems with lots of (geometric) variables that can be optimized, there are special mathematical methods, that are implemented in for instance the ANSYS software that you've used to simulate the low-head dam.

 

[Klystron]

"Reynolds Effect" is what I think of when I think of all that happens around the "Reynolds Number" [snip...]
"I would define viscous dissipation as the transformation of kinetic energy to internal energy (heating up the fluid) due to viscosity. This includes both turbulent kinetic energy and mean flow kinetic energy." Jonas Larsson
Does this seem like a reasonable definition?
...
I have not read or even seen anything talking about shapes of objects in water and how different material drag coefficients affect acceleration underwater.

The photograph of an open pool of water could indicate measurement difficulties and data problems. Temperature differences due to motion subsumed by evaporation and other effects. Pressure variation measures inconclusive in unrestricted fluid.

The subject "moving shapes in water" is extensively researched and documented. Navies around the world collect and calculate reams of data on this subject. Access to this data lies outside the scope of a science forum but boat and ship designers might also have access to results.

 

[boneh3ad]

800 distributed computers can be difficult to control outside community networks though fine if you mean number of processors.

...

I feel like you are trying to reinvent the wheel here. Massive parallelization of CFD codes so that they run efficiently on thousands of cores simultaneously is something that is actively pursued and that happens every day. In fact, CFD simulations are one of the key sciences driving supercomputer development.

[BTW I like "Reynolds Effect".]

You may like it, but the term means nothing to a fluid mechanician. I do and teach fluid mechanics for a living and had never heard that term and had to ask when the OP meant by it.

For your own problem, you could compute the Reynolds number, which is the:

...

if the number is 'large' (say 1000 or more), it will not be laminar. Note that for water, you already multiply by 1000, density is 1000 and your velocity is 20, so if you have a ball of 1 meter diameter, your Reynolds number is already 20,000,000.

This is a gross oversimplification. There is no magic Reynolds number at which point a flow is not laminar with the exception of a few very specific cases (e.g. pipe flow). It's not uncommon for the Reynolds number on, for example, and airplane wing to reach $O(10^6)$ or $O(10^7)$ before transition occurs.

Note that the material does not affect the drag, it is the surface roughness that matters.

I feel it is important to note that, while this is generally true, the relationship between surface roughness and drag is remarkably complex. The primary driving factor is the effect roughness has on the disturbance spectrum in the boundary layer and thus on laminar-turbulent transition, but that is a large, unsolved problem.

Note also that it doesn't matter if it's in water or air, what matters is the Reynolds number.

Also just a clarification here as well: this is true if you doing everything nondimensionally, i.e. with a drag coefficient. The surrounding fluid absolutely will affect that actual dimensional drag.

 

[Klystron]

That's what comes when a member knowledgeable in one field such as computing across data centers and designing same, comments in another field.

Cross-posting threads discouraged. My post attempted to address computational IT issues raised by the OP.

[Also provide networks for language experts. The sheer illogic of the term <dimension-less number> + "-effect" caught my fancy. Thanks --Norm]