Intuitively it seems that if a non-frictionless flat surface begins moving horizontally in a gaseous fluid then eventually the fluid above the flat surface would start to move in the same direction that the flat surface is moving. How could one set up a system that does this and at the same time avoids or minimizes turbulence? One thing to consider is the viscosity of the gaseous fluid. Less viscous gases are more prone to turbulence. Another consideration might be how slowly the flat surface begins moving and how slowly it accelerates. Also, and maybe most importantly, one must consider the source of the friction. Looking beyond the classic division of friction into static friction, rolling friction, sliding friction, and fluid friction, it seems to me that fluid friction could arise from many different causes such as electrostatic, electromagnetic, mechanical, etc. Although I said “flat surface”, no surface is completely flat when you look at it on a molecular or atomic level. (You can’t really “look at it.”) Indeed, it might be hard to pin down exactly where the surface is located. It would seem too me that different types of friction might cause turbulence in different ways.
In the spirit of the spherical cow, I propose a very simple imaginary model consisting only of a non-frictionless flat solid surface that extends infinitely in all directions in the x, y plane. This flat surface is covered by a pure, monoatomic gas that extends infinitely upward in the z direction. All components of this system are at a constant temperature and the gas is at a uniform density and pressure everywhere, at least at the start. I guess there would have to be no gravity, otherwise the gas would be at an infinite pressure where it meets the flat surface, right? Note that this simplified model could not exist in our universe. Perhaps if we can come to some conclusions about this over-simplified model we can create more sophisticated models that are possible in our universe.
Would it be possible to slowly accelerate the flat surface up to constant speed such that the gas immediately above the surface is moving at or near the speed of and in the direction of the flat surface without creating so much turbulence that the entire system fails?
The concept you’re looking for is Reynolds number.
It applies to any fluid, not just gaseous ones. In general, low Reynolds numbers mean a viscous fluid, which tend to be laminar (smooth). A high Reynolds number is the opposite: low viscosity and turbulent.
Yes. Exactly. How do we keep the Reynolds number low so that laminar flow predominates when the only initial source of movement is the non-frictionless flat surface.
The answer to that is in the definition of the Reynolds number; keep the fluid density low with relatively high dynamic viscosity, have small gradients of fluid velocity, and a small characteristic length. The latter is done by ensuring that you don’t haves sharp features and any changes in shape are gradual.
It is true that substances are not smooth at an atomic or molecular level, and there are certain regimes of flow and types of electrochemical interactions where that can actually come into play, but in general fluid flows in terrestrial conditions are mostly concerned with measurable surface finish. A complication is heat transfer between the fluid and surface which can sometimes radically change fluid behavior at the boundary layer, which gets into hypersonic conditions.
Welcome to thermofluid mechanics. It is a complex, technically luscious, and compelling frustrating subject to really comprehend in practical application, and one that I daresay that few people can claim to have actually mastered, even people with advanced degrees and decades of experience, and those who have cannot bear to watch virtually any space fiction and cringe when people talk about “skin friction” with respect to reentry heating. It is far from the most current but the Landau and Lifshitz text is comprehensive with respect to essentials.
Would y’all agree that the flat surface must be non-frictionless in order to cause the gas above it to move? Would you further agree that there are no non-frictionless, flat, solid surfaces in the real world?
There’s nothing like a frictionless surface in the world.
In fact, the usual assumption in fluid mechanics is the no-slip condition. The molecules adjacent to the surface are stationary with respect to the surface.
There are exceptions to the no-slip condition, but it’s a good approximation most of the time. And I’m not aware of any condition where it breaks down completely (i.e., the moving surface does not induce any fluid movement at all).
Disclaimer: It’s been 3 decades since I did my PhD in this stuff (on pattern formation in weakly non-linear fluid convection–I can hear everyone snoozing), and I left academic science for the better pay of software engineering not long after that, so I’m a bit rusty.
A fluid is always assumed to have zero velocity at a solid boundary, so, in that sense, all solid surfaces are assumed to be sticky, not frictionless. One can approximate a “frictionless” boundary by assuming the fluid has so little viscosity that the boundary layer is thin enough that it can be ignored. In an inviscid fluid (one with zero viscosity), the boundary layer has zero thickness, and you could think of the solid surface as being frictionless, but it’s really the fluid that has zero internal friction (which is effectively what viscosity is). Of course, there aren’t any truly inviscid fluids, but there’s always some regimes where that’s a good enough approximation.
So the fluid at the solid boundary has zero velocity in the reference frame of the solid. OK, that makes sense. In the hypothetical OP it is the solid that is moving. So in the reference frame of the nonmoving fluid away from the solid boundary, at least at the start, the fluid at the boundary has some velocity relative the rest of the fluid away from the boundary, right?
Right. I should’ve been more clear–told you I was rusty. The fluid has zero velocity relative to the solid surface. It’s a common boundary condition, and all solutions to the fluid equations, whether algebraic or numerical, must satisfy it.
Superfluids like liquid 4He is technically frictionless (both internally and externally). Of course it doesn’t follow other normal laws of thermofluid mechanics and only exists very near temperatures of T << 1 Kelvin. Other superfluid states exist in neutron stars or other exotic, non-terrestrial conditions. The no-slip condition is obviously an approximation of the complex real world tribological interactions at the solid-fluid boundary but in terms of fluid mechanics simulations the deviations are lost in the noise of other assumptions. It would essentially be impossible for a solid object to move through fluid without losses due to hysteresis without creating turbulence conditions.
To expand upon that, if you don’t assume this condition, you get nonsensical, unphysical solutions.
If I’m reading it correctly, the no-slip condition pretty much still holds: the average fluid velocity goes to approximately zero as you get close to the boundary. But there are zillions of tiny vortex loops arising from the microstructure of the surface, and unlike with viscous fluids, they don’t get damped out. Macroscopically, it looks frictionless but microscopically, a complicated web of vortices mediates the interaction between the fluid and the surface.
With zero viscosity it is ‘impossible’ to have a no slip boundary on a smooth surface. (Technically, you have to have no movement at the boundary for the equations to provide sensible answers but the thickness of the bounary layer is zero.) A rough surface wall will induce momentum normal to the flow of the fluid because of essential physics of the fluid interacting with a non-flat surface having to change direction but that isn’t due to any kind of friction in an incompressible, inviscid fluid. The micromechanics of real boundaries are more complex as discussed above and it is essentially impossible to avoid turblence in any sufficiently fast-moving fluid because of microscopic surface roughness, localized heat exchange, and even extremely tiny density changes caused by cavitation. Superfluids, under the laboratory conditions under which they’ve been. created, do all manner of counterintuitive and seemingly impossible phenomena that are not explicable via normal thermofluid dynamics, requiring quantum hydrodynamics to explain.
Nitpick: the superfluid transition occurs at ~2 Kelvin. IIRC, the CERN magnets are cooled to <2 Kelvin in order to use the superfluid for thermal management.
I suspect he was referring to thermodynamics, a related but more foundational discipline that is more abstract in nature. Introductory thermofluid mechanics (or just “Fluids” as it is commonly referred to) is actually a pedagogically easier course because it is largely application of conservation laws, and the only proof that you have to study is that of the Buckingham π theorem (dimensional analysis using nondimensionalization of parameters). More advanced study involving turbulence models, two phase flows, numerical simulation, et cetera is very complicated and mathematically challenging but the fundamentals only involve some basic partial differential equations that can largely be reduced to algebra for conceptual textbook-type problems. I scored an A in that course without much effort after struggling to really grasp thermodynamics, and frankly I’ve had more direct use of that knowledge in practice, although understanding the fundamentals of thermodynamics has certainly given me insight to see basic errors in assumptions and punch through scams.