[arun1994]

I need t0 make a sinus0idal m0ti0n 0f the pist0n . Please see figure.

I was t0ld t0 use a very small r0tati0n s0 that fl0w g0es int0 d0main.

what shud be the c0rrect b0undary c0nditions and domains, interfaces?
stuck for past 3 days

thanks a milli0n.

[arun1994]

inserting CCL file f0r ref...

[urosgrivc]

Why would you use rotation if that is not what you are modeling.
What kind of rotation?
Why would flow not go into the domain without rotation, that is nonsence.

make an expresion of displacement relative to (t) time: example-> sin(t)*userfactors
set this displacement to your moving wall.

You might have some problems seting up the GGI conections as your mesh will totaly seperate your domain into three bodies when it gets past the inlet and opening chanels.
or maybe not, be sure to set up nonoverlap boundary conditions corectly.

What have you set for inlet?
Be sure to think it all threw, if you have set pressure than it is probably ok but if you have set speed than things will get kinda wierd

[arun1994]

when I turn 0ff r0tati0n, I get " atleast 1 highly skewed element f0und. "

[arun1994]

dt = 1.24972e-5[s]
greitis = 5000[rad s^-1]
gretis = -3.6[m/s]*t+Y0
poslinkis = greitis *t
poslinkis cilindro = -sin(poslinkis )*0.0042[m]
slegio kitimas p = slegio kitimas(t)
su funcijos duomenimis = Function 1(t)

these are equati0ns I use. but in m0nit0rs, I find displacement like in pic.

[urosgrivc]

Ok so where is the problem.
And where in the model is your rotation and what is it set and used for.
Where did you switch it off?
I meen in the domain, not in the equations for motion.

Your monitor points are kinda interesting as this is not a smooth sinusoidal motion.
I dont know if these monitor points are usefull to you as you are monitoring mesh displacement.
i think it is beter to monitor walues on the botom face of the cilinder (speed or coordinate value)

[arun1994]

0ne 0f the m0nit0r p0ints is 0n the b0tt0m.

I switched 0ff r0tati0n for all d0mains. but any idea ab0ut skewness pr0blem?

[urosgrivc]

Your mesh is deformable, that means that the elements get compresed on the way up and return back to normal when going to the initial state where mesh was first generated.
When elements get compresed or streched out -> skewnes gets higher and can get to the point that new mesh is requiered ->(remeshing).
This hapens if you have large displacements or small gaps ocuring.
Check in post what is going on with the mesh

You can fix this:
(1) if you start with center position when generating mesh-> so it doesent just gets compresed and back to normal, but first gets copresed than normal than streached and repeats.
(2) You can make a mesh that is initiali a bit streched (if you are generating mesh when piston is at the botom) so elements are best quality when the piston is in the middle.
(3) Remeshing when skewnes gets high (takes a bit extra computational effort)

[arun1994]

I reduced time step value. A few iterati0ns ran and I get this n0w.

[urosgrivc]

What is your inlet Boundary condition? (something is wrong there)
I dont know much about your case but it looks it should be
(opening pressure tipe BC)
Im guesing here as I dont know the details of your sim. and what you are triing to achieve

[arun1994]

https://drive.google.com/open?id=0B-...WxJdkE0MzhPNlU

I have archive file here. I am trying to make IC engine 2 stroke sim.

inlet: 1.4MPa
0pening : 0MPa

[urosgrivc]

I think that exhoust will play some role in your simulation than.
as you would than move opening tipe bc further down the line.

What about your energy input or combustion, it gets quite complicated if you add combustion, but I dont know about this tipe of simulations so someone with more experties will have to help.

[arun1994]

I dunt use cumbusti0n 0r energy.just c0mpressi0n and expansi0n

[arun1994]

S0 shud I use 0utlet furtther?

[ghorrocks]

I have simulated many 2 stroke port valve engines in my time. The basic approach is quite simple:

1) The ports are one domain with a stationary mesh.
2) The piston is a domain with a moving mesh.
3) The stretching part of the piston mesh must be an extruded/swept mesh. Do not use tets.
4) The ports connect to the piston with GGI interfaces in transient rotor-stator mode.
5) You set the non-overlap fraction of the GGI interfaces to appropriate wall boundary conditions.
6) You generally put opening pressure boundary conditions on the ends of the inlet and exhaust manifolds. Note that the length of manifold is important, also the end detail. It reflects pressure waves and these are quite significant for most engines.
7) Obviously you need a fully compressible gas, so ideal gas is good. More complex material models can do done but will add to the already complex simulation.

Note: There is no need to rotate anything. You do not need rotating frames of reference.

[ghorrocks]

Also: the model works best when it starts from top dead centre. It is easier to stretch a mesh than squash it.

Also 2: As long as your piston mesh is good quality you won't need remeshing. It can handle quite large changes in aspect ratio as long as you start with a high quality mesh.

[arun1994]

Wow ! That was perfect. I have a few questions

1) What option should I be using for transient rotor condition ? None, automatic, value

2) What about GGI Interface fraction ? I do not get this. What parameters should I specify?

I have a general question about the interfaces... Why do we use "stationary" condition for interfaces? As I presume, the mesh should be moving in the piston region and the interface should have " connection " to the moving cells right? I mean, when the piston is compressing, the cells are also compressed and when the interface region cells remain constant, doesnt it create disharmony or between cells?( that they do not go together and meet not at edges )

3) About the exhaust, Do you think I should increase the length in this case?

Thank you so much Ghorrocks ! ANd also my slovenian friend. It means a lot to me !

[ghorrocks]

1) Pitch change = None.
2) GGI interface fraction - I am not familiar with this option.
2a) Stationary condition means it does not move. It means the GGI is only calculated once and never again. This is not appropriate for IC engines as the GGIs slide open and shut continuously so you need to constantly recalculate the interface.
3) You should model the inlet and exhaust manifolds to be the same as the real engine. Manifold length is important. You also need to end the manifolds correctly - this will usually be a bell mouth at the inlet and a pipe exit for the outlet. It all makes a difference, especially for high speed engines.

[arun1994]

I meant non overlap fraction for GGI . My bad sorry. what and where should I look for it ?

and for other answer, thanks

[ghorrocks]

The nonoverlap fraction is reported by the solver. As the port slides open and shut this will vary. When the port is shut it will go to 0%. So keep an eye on this and check that the value it reports makes sense. This is how to check that the ports are connecting up properly.