# Discretization of heat exchanger

#### femaleengineer

Hi.

I hope my description of the problem will be sufficient.

I'm trying to make an analytical model of a heat exchanger.

The hex is a coil wound type, (helical coil) with several pipes in parallel to reduce the pressure loss, it is double. Meaning it takes it's turns (windings) around a centre core, and wraps around to the outmost shell (the two coils are separated by a wall).
To discretize this I'm in a bit of a doubt whereas the best way is to "unfold" the system and do it along the length.
Or to separate the two "layers" and take one winding at the time.

#### Woody

I have no Idea, but...

It seems to me there are two important criteria to consider
1) which method gives the best answer?
2) which is easiest to implement?

Personally I have always firmly cinched my belt, and then fitted my braces.
If the implementation is not too complicated, why not do both?

It is possible you will find one approximation is best for a certain set of circumstances,
while the other is better for other circumstances.

Similarity between the results from the two models will provide confidence in the result.
Differences between the two models will flag warning signs for deeper investigation.

Last edited:
2 people

#### benit13

That sounds like a very complex geometry. I wouldn't know how to model such a complex geometry. You can perhaps try a 3D finite differences method on a Cartesian grid in the first instance and set the resolution to be quite high so that you get a very fine grid. Then, if that fails, you might have to try something more exotic. Perhaps there are models in the literature using cylindrical polar coordinates?

1 person