Bellows-Action Cooling Device Simulation

[pwilkins]

Is there a good way to simulate a bellows-action cooling device?

[pwilkins]

Autodesk Support, have you ever tried this?  I am thinking along the lines of a Nuventix device or the GE dual piezoelectric cooling jet.

[OmkarJ]

I think the forum visitors and Autodesk Personnel may be more interested in the question if you simplify it, give more information, post some images about your model/BCs etc. The esoteric terms you used (Nuventix device, bellow action cooling) may be very obvious to you since you encounter them often in your line of work, but others may be oblivious to them 🙂

[pwilkins]

No model yet.  Just trying to get an idea if it's even possible to simulate these devices.  Bellows refers to the sucking in and blowing out of air, like a fireplace bellows that you use to get your fire burning nicely.  You know, the air puffer thing.  Basically, these little devices use the same concept to generate a small air flow over hot spots in electronic assemblies.  It is a rapidly pulsed air flow, though, because it is a bellows device.  So I guess I have two questions:

1) Can a pulsed air flow be simulated?

2) Can the sucking in and blowing out (bellows) be simulated?

I can create a model with the same outlet size as these devices and just assign a constant flow rate equal to that listed on their datasheets, but I wanted to simulate the bellows phenomenon if possible.

[nhahn]

I've seen these devices (i.e. http://www.nuventix.com/technology/) and thought about this a bit, though I've never modeled one.  Have you asked the manufacturer how they'd simulate it in your application?  It can surely be done; the way you'd do it depends on if you're (a) designing the bellows or if you're (b) using it for cooling.  It sounds like (b) to me.

In that case, the easiest way to get time-accurate behavior might be set up your model so that the bellows device opening is an inlet surface (a boundary of the fluid volume).  Then apply a transient boundary condition (periodic or harmonic) such as velocity or volume flow.  Set the coefficients so that the flow oscillates between sucking and blowing.  I think you'd need incredibly small time steps, since they run at >40Hz,  and a lot of meshing to make this reasonably accurate though, and I'm not sure what it gets you unless you are interested in where all the tiny vortices are going.  So far as cooling a chip or something similar, the flow is effectively constant.

If as I suspect all you want to model is the enhanced cooling effect (as compared to a small fan of similar flow), then what you really want is some way to increase the effective turbulent mixing in that area, rather than specifically modeling the transient flow pattern in a time-accurate manner.  Maybe someone from Autodesk can chime in?  I don't know if you can locally increase this (i.e. increase the turbulence ratio or intensity on flow coming from one inlet)  as opposed to global turbulence properties.  This might be beyond the capabilities of this general-purpose solver.  Maybe you could fool it by having a bunch of tiny columns or a grid modeled to trip up the flow and make it highly turbulent inside the nozzle of the bellows device?

If you're trying to model the bellows device to design one, you'd have to use the motion capability and make the walls of a cavity move periodically or harmonically.  Again, very very tiny timesteps and lots of meshing.  That would be a pretty specialized analysis, and hard to validate, but not impossible.  Building and testing a device would probably be quicker and more accurate!

[nhahn]

One more thought: possibly the simplest way, but a bit of a hack. It won't get you the flow behavior but will allow thermal modeling.... just decrease the thermal resistance of the parts that are being cooled by the bellows, according to the metrics from the bellows manufacturer. They have various charts and text descriptions about how heat sinks being cooled by the SynJet compare to natural convection and normal fans.

[apolo_vanderberg]

Nick's response here is approprite and how we would go about it assuming you wanted to model its impact on the system, not the bellows itself.

With some bellows devices, you might have to be careful if trying to model it diretly. As we aren't modeling deformable bodies, the bellows (in the mental image of the air puffer for the fireplace) between the flat planks of wood, would not deform and thus modeling this might require some assumptions.