It's good to see folk having a go at CFD but what information do you want CFD to provide?
The reason for asking is because the physics of sound radiation from a port is a lot more complex (and interesting) than that from a driver. This complexity is reflected in the sets of assumptions required for CFD to produce physically valid simulations. At a guess the simulations above seem to be steady state 2D with a steady state turbulence model (are they?). These are almost certainly not reasonable assumptions but will expand on why and how to improve the assumptions when I know a bit more about what you want from the simulations.
The reason for asking is because the physics of sound radiation from a port is a lot more complex (and interesting) than that from a driver. This complexity is reflected in the sets of assumptions required for CFD to produce physically valid simulations. At a guess the simulations above seem to be steady state 2D with a steady state turbulence model (are they?). These are almost certainly not reasonable assumptions but will expand on why and how to improve the assumptions when I know a bit more about what you want from the simulations.
They are 3D, with a plane cut for view. Incompressible steady state.
The model is acutally like this, with symmetry on both flat faces.
As said earlier, the only purpose was to show that high air speed in itself does not cause turbulence, but changes in air-speed do and it happens at the port mouths.
The model is acutally like this, with symmetry on both flat faces.
As said earlier, the only purpose was to show that high air speed in itself does not cause turbulence, but changes in air-speed do and it happens at the port mouths.
@Tenson That was basically what I was trying to say 🙂 😉
But if someone doesn't understand the context of that, it kinda sounded like that flares in general are a bad thing.
Which obviously isn't true.
A straight pipe with flares (not these enormous ones) is still better than no flares at all.
But you're absolutely right that there is some optimum there.
There is plenty of stuff about this if you dive into aerodynamics, or also things like propellers etc etc 🙂
The same or very similar principles apply.
But if someone doesn't understand the context of that, it kinda sounded like that flares in general are a bad thing.
Which obviously isn't true.
A straight pipe with flares (not these enormous ones) is still better than no flares at all.
But you're absolutely right that there is some optimum there.
There is plenty of stuff about this if you dive into aerodynamics, or also things like propellers etc etc 🙂
The same or very similar principles apply.
That's because we are in the magical voodoo world of non-linear thermodynamics.
And oh boy, if you want get engineers and scientists exciting as well as angry at each other, that IS one of the subjects 😀 😀 😀
With driver design there is a little bit of that, but most for PA/high-excursion applications, and also just to some extend.
3D is good but symmetry is not if you want the flow to be unsteady and turbulent. I presume you are running incompressible steady-state to learn how to use CFD? Steady-state turbulence models usually assume a high Reynolds number which we won't have in a typical speaker port. Some form of low Reynolds number turbulence model would be needed for an accurate simulation or, possibly, just resolving the flow.
Steady state turbulence models involve large assumptions in their formulation and can generate poor results in a fair few circumstances. The back and forth flow in a port is inherently unsteady and cannot be predicted in any meaningful way with the assumption of steady-state turbulence.
Do you intend to go on and simulate the flow in and out of a port and the sound generated?
I didn't plan to sim anything further, but it would be quite interesting to see a sim like that to learn from it, especially if in SimScale which offers free use.
Just had a look at simscale which seems to a web wrapper around open source software. They don't seem to have got round to wrapping any open source acoustic software that I can see. Anyone?
There seems to be a number of open source software packages around these days that could be used to simulate the sound from a resonating and possibly chuffing port. Not aware of an open source package that directly simulates low Mach number compressible flow with accurate acoustics but there seems to be plenty that could be used to simulate the unsteady incompressible fluid flow (i.e. the acoustic sources) and a few that could run a subsequent acoustic simulation with those acoustic sources. This split is reasonable because in our case the weak acoustic waves are not going to significantly influence the unsteady fluid flow through the port.
The exercise would require a reasonable desktop/workstation (or a willingness to pay for cloud computing), installing something like this (or equivalent), understanding the input and output files, understanding some of the physics/maths/modelling w.r.t. to acoustic sources, and a chunk of time. Not straightforward but likely doable for those comfortable using computers.
There seems to be a number of open source software packages around these days that could be used to simulate the sound from a resonating and possibly chuffing port. Not aware of an open source package that directly simulates low Mach number compressible flow with accurate acoustics but there seems to be plenty that could be used to simulate the unsteady incompressible fluid flow (i.e. the acoustic sources) and a few that could run a subsequent acoustic simulation with those acoustic sources. This split is reasonable because in our case the weak acoustic waves are not going to significantly influence the unsteady fluid flow through the port.
The exercise would require a reasonable desktop/workstation (or a willingness to pay for cloud computing), installing something like this (or equivalent), understanding the input and output files, understanding some of the physics/maths/modelling w.r.t. to acoustic sources, and a chunk of time. Not straightforward but likely doable for those comfortable using computers.
That's the whole problem
Pure laminair flow = easy
total random chaos = bit trickier, but still doable since it's mostly a matter of statistics
high mach flows = becoming tricky but still doable with some margin of error
low mach flows = oh boy.....
It's like the butterfly effect, a tiny problem or bump which doesn't seem to have any effect in the beginning, can start serious problems later down the line.
Even measuring to confirm these simulation models are actually accurate, is a pain.
SimScale uses OpenFoam, I think.
There is ABEC / AKABAK for acoustics modeling but it uses LEM for port internals and BEM for what happens outside.
There is ABEC / AKABAK for acoustics modeling but it uses LEM for port internals and BEM for what happens outside.
The problem (I keep repeating myself) is not the constant laminar/turbulent air flow. The problem ... ehm, no: the interesting bit is the oscillating air flow.
Doesn't that basically work as a low pass filter?
After a while the mass of the air can't keep up -> results is that output collapses.
Also meaning that turbulence effect will be less as well.
So say everything roughly below an octave from that point, would just follow the simulations, "more-or-less".
The previous simulations all showed just an output stream, but in this case we also need an input stream.
Although, that usually doesn't produce a lot of issues.
One other thing I can think of, is that a turbulence "swirl" that is (relatively) slow and still kinda going, is sucked back in.
Which on itself also makes the net effect less in the end.
I can't think of a way that it would make things worse.
btw, I also assume that these simulations don't take the standing waves inside the tube itself into account?
Because I can imagine that it creates a whole new set of problems as well.
OpenFOAM is probably the most widely used open source CFD software but there are plenty of others with their pros and cons.
ABEC isn't suitable because it doesn't interface with other software like OpenFOAM for the acoustic sources and/or support acoustic analogies. It is designed to be a self contained commercial Windows program (which is fine) rather than an open research code one can use to process data in order to create geometry, mesh, solve flow, solve acoustics, integrate sources, create plots, create movies, create audio,...
The common approach mentioned above is to split the fluid motion into a large incompressible part and a small compressible/acoustic part. The incompressible part is the slow moving (<10% the speed of sound) flow in and out of the port, displacing the air in the room to create sound with some of it rolling up into unsteady vortices to create a chuffing sound when the velocity in the port approaches 10% of the speed of sound. The acoustic part moves fast at the speed of sound and is created/absorbed at the boundary by the motion of the driver cone and within the air by vortices (which conventional acoustic BEM codes don't include). The acoustic resonances in the port are fully included. The only thing missing is the acoustic motion creating sources/sinks to alter the incompressible motion. At low Mach numbers in reasonably isentropic flows acoustic waves are weak (small pressure variations compared to the incompressible motion) and the source terms small higher order ones making them reasonable to neglect in our case.
Well we were also talking about absorbers, and we already went down to port performance 😀 😀
Hah! Although the split into incompressible and acoustic is widely used and works well for most flows in our case we have a problem: in order for the incompressible slug of air in the port to resonate on the spring of the trapped air in the cabinet that air needs to compress albeit not much. So we need a compressible code that behaves itself at low Mach numbers (most don't) and doesn't necessarily support acoustic waves. An interesting test case.