Rotary heat exchanger (heat wheel) modeling

[Anonymous]

Hello,

So for my master thesis I am analysing rotary heat exchangers. I am trying to create the model using Inventor and then transfer it to Autodesk CFD. The object would be rotating wheel of diameter 200-1000 mm., length of 200 mm. The matrix of the wheel is made of aluminum, it consists of small parallel chanells (for example 1,4mm x 4 mm ) formed by a sinusoidal waves of triangles. the thickness of the aluminum sheet would be 0,07 mm. So you can imagine how complex it would be to design this object accurately.

The heating wheel is constantly rotating at a speed of about 10 RPM. The principle of it is to preheat the incoming air to the build by the leaving air of the buildg. The dense aluminum matrix takes the heat from hot air and by rotating transfers the heat to cold air. The example is below:

So I have few questions:

1) Is it possible to make porous solid material (3D not 2D surface), as far as I know it is possible to make only surface (2D) distributed resistance material?

2) How should I define the region between rotating matrix and stationary part? Should I leave small gap and mesh obejcts separetaly? Or I can define it somehow so there would be no leakage?

3) What would be your advice on modeling such an object?

Thank you

[Jon.Wilde]

Hey,

Interesting project 🙂

To answer your questions

1) Yes, just apply to a solid 🙂

2) I guess leave as no leakage, keep it simple right?

3) Why not use a fan material, applied to the volume, I think that will give you what you need - although maybe not the resistive properties

Thanks,

Jon

[Anonymous]

Hey! Thank you for fast response 🙂

Ok so I can apply resistance to a solid. But the thing is that I can not define density and specific heat of the material (I can apply only conductivity) if I use resistance material. So there is no heat transfer from air to solid and from solid to air in this way.

Is there a way to create a material which could be porous (flow through) and keep it's parameters such as specific heat, density and conductivity?

[Anonymous]

Hello again,

So I managed to create a functional model (by modelling the heat exchanger geometry as it is). In this case simulation would probably take weeks to converge (as it is transient, rotating heat exchanger).

So is there a way to make this model (heat exchanger) more simple?

[Jon.Wilde]

I would only think this is going to possible as a small section really - can you share a CFZ (support) file? This contains the cfdst also 🙂

The issue with a small section is that you are only going to be able to look at the airflow within a small bit, or with a periodic model that assumes the same thing happens throughout 360 degrees. 

I am wondering though, if a small section might be enough?

What if you did this to understand how a portion worked, then ran the full model with only air flow and a static resistance to understand how the flow is distributed across the open faces to get an idea of how it would perform as a whole?

I cannot see a quick way to run the whole thing.

What exactly do you need from the simulation though? Might taking a small section and optimising it be sufficient? I could imagine it might be quite enlightening once you have some results.

Thanks,

Jon

[Anonymous]

I attached the .cfz file of the example (this is model with greatly reduced diameter - the real heat exchanger D=200-1000 mm, this model is reduced to D=15 mm). So this is already a small portion of the real object. The heat wheel thermal efficiency and pressure drop does not depend on the diameter, as long as there is same face velocity maintained. Although it does depend on the length of heat wheel, so I should keep the same length (for example 80 mm in the attached file) 

I am not sure if this case is considered as "the same thing happens throughout 360 degrees". This is the example of temperature distribution across heat wheel:

From the simulation I expect to find out the temperatures in outlets (thermal efficiency) and pressure drop across the heat exchanger.

I was thinking about "frozen rotor" function, as it would save some solution time. Would it be suitable for this case (does it account for the heat transfer between solid and fluid?)

Thank you for your concern!

[Anonymous]

Also in this case should I use rotating region or rotational velocity? As far as I know rotating region is used for turbomachinery, so in this case maybe it is better to use rotational velocity? Would it also save solution time?

[Jon.Wilde]

No way, you did it with motion 😄

This is awesome!

I think a rotating region (RR) might be a lot simpler and you can also still have the mesh enhancement (wall layers) on (you will need to turn them back on as they are off by default with motion). I am thinking about how this might work - you may get away with just assigning it to the outer solid part and I think it will take everything inside it along for the ride 🙂

You will need a far better mesh too (4-5 elements is an ideal through a small gap) but this is a great start.

Leave Auto Forced Convection off - you would really want to run flow and thermal together here I think?

Gravity could stay off also as it has near zero effect, then you can just use air-fixed.

Really nice work so far, let's see if a RR can work a little better.

Jon