topic Re: how to simulate a long pipe in CFD Forum

how to simulate a long pipe

Anonymous — Mon, 08 Jun 2015 10:03:58 GMT

i have pipe about 3000 metres in length, it took too long time to iterate and build mesh (i must build mesh 200 mm size for acurrate geometry). so i did partition for that problem. my question: is the simulation accurate?

i tried to make the short pipe. the first pipe's outlet data result became next pipe inlet boundary condition. but when i compared with the same pipe without partition, the result simulation was really diferent.

Re: how to simulate a long pipe

Jon.Wilde — Tue, 09 Jun 2015 14:13:33 GMT

Hi,

I often run these in 2D (or axisymmetric) as they can be really mesh sensitive. There is a useful page on this here: Pipe Flow.

Is the model a straight pipe?

Kind regards,

Jon

Re: how to simulate a long pipe

Anonymous — Wed, 10 Jun 2015 03:06:43 GMT

Hi Jon, Thanks for your reply.

Ok, let's make it clearly. i have a ~3000 m pipe's length that lay down on seabed. as we know that seabed has a contour, so the elevation (or y axis) tends to different in every section. -not a straight pipe-

This is what i set, i'm sorry if there are something incorrect, i'm still beginner to use this software.

as i said before, the goal is very simple. i just want to know the accurate velocity in some points. i want to see the influence of gravity accelaration

Then, i'm getting confuse with the result. this is the image of inlet, -the rest of velocity contour is blue-

what setting that i missed, would you like to share with me what geometry, boundary condition and etc that i have to set to run this simulation. thank you Jon.

Re: how to simulate a long pipe

Jon.Wilde — Wed, 10 Jun 2015 06:42:04 GMT

I would say because this was run transiently for a second of real time 🙂
Lets start simple and add in gravity for a second run.

I would expect you to have either a high pressure or flow rate at the inlet, correct?
A P=0 at the outlet too.
Run steady state for 500 iterations and see what the results look like.
Advection scheme 5 (as per the page I shared) is likely best suited (solve ->solution controls->advection).

If possible, run a geometric 1/2 model rather than 2D which I agree isn't so sensible here. Takes half as long or you have scope for a finer mesh.

Can you shade the cut-plane by mesh so that we can see what it looks like?

Kind regards,
Jon

________________________________

Autodesk Limited
Registered Office: One Discovery Place, Columbus Drive, Farnborough, Hampshire GU14 0NZ
Registered in England and Wales, No. 1839239

Re: how to simulate a long pipe

Anonymous — Wed, 10 Jun 2015 08:59:38 GMT

thanks for advice Jon.

i set P=0 at the outlet and Vx = -5,98 m/s, Vy = -1,83 m/s at the inlet. I used free surface because i want to enable hydrostatic pressure/gravity as your advice. although there is an initial velocity, the flow is driven only by the gravity. so i think there is an increasing of velocity because of accelaration.

then, i did your advice to use steady state condition. adv 5, etc. but i couldn't see the change of velocity contour significantly. the contour change significantly only in the beggining of flow.

then what should i do?

Thank you very much

Re: how to simulate a long pipe

Anonymous — Wed, 10 Jun 2015 09:52:35 GMT

i'm sorry Jon. this is the second run, after adding gravity.

what do you think?

i still see a bit strange result

Re: how to simulate a long pipe

Jon.Wilde — Wed, 10 Jun 2015 09:57:19 GMT

Hi,

It's hard to tell, what looks wrong to you?

How does the mesh look?

Are you using the standard transient settings once you have enabled the free surface?

Thanks,

Jon

Re: how to simulate a long pipe

Anonymous — Wed, 10 Jun 2015 10:04:53 GMT

Re: how to simulate a long pipe

Jon.Wilde — Wed, 10 Jun 2015 10:15:10 GMT

Hi,

That looks OK - perhaps a little blotchy. Look at the results on the upper image without the mesh - are they smooth or are they driven by the edges of the elements?

We might need a finer mesh and a thinner boundary layer, but the question is - what looks wrong to you?

Thanks,

Jon

Re: how to simulate a long pipe

Anonymous — Wed, 10 Jun 2015 10:15:50 GMT

i'm sorry Jon, i just guess that the velocity in high degree (or almost vertical flow) show the significant change of velocity contour, but i have to study this part again.

i think i used standart transient setting, i have posted the image before

Re: how to simulate a long pipe

Jon.Wilde — Wed, 10 Jun 2015 10:19:16 GMT

Your image before had 250 inner iterations - not standard 🙂

I think the default is 3.

Also, it was set to run just one timestep, I see the approach but I would stick to the defaults for now.

Hope that helps!

Re: how to simulate a long pipe

Anonymous — Thu, 11 Jun 2015 04:18:19 GMT

Thanks Jon, it helps so much. My first question is about partition (i think it is imposible to run the whole pipe at once)

i need your advice, what should i do to make this partition result same with the non-partition result.

i ussualy use the outlet data in first pipe (with xy plot), to inlet data in seceond pipe.

Thanks

Re: how to simulate a long pipe

Jon.Wilde — Thu, 11 Jun 2015 12:55:14 GMT

Hi,

As the flow rate should be the same throughout, could you not just add up the pressure drops for each section?

Re: how to simulate a long pipe

Anonymous — Tue, 16 Jun 2015 07:42:15 GMT

Yes the flow rate should be same throughout. But i made the partition. i had many sections with various pipe's length, and you know the flow rate is diferent in each section (i used bulk calculator to calculate the velocity).

i think this is because the simulation predicted the entrance pressure, so the velocity becomes diferent because the difference of predicted pressure's value. as we know that length of pipe is one of the pressure drop's variable.

i need the simulation that resulting same velocity in every section, no matter how long the pipe section.

What is your advice? Thank you