CFD Pipe Flow - Results Diverge

I am modeling what is essentially a piping Tee (24"x30"x30") with 46 small holes drilled in the 36" side. The fluid is water. The inlet is the 24" side. This Tee is meant to diffuse water instead of dumping it into a tank directly from the pipe. The project is titled Tee Diffuser - Incompressible CFD.

The first run I performed produced results with unrealistically high velocities. I used a k-omega SST model. I defined the inlet as a volumetric flow rate with the outlets being pressure (0 psig). I also tried defining a uniform velocity inlet, but this produced similarly high velocity results.

I closed off all of the small outlet holes, so only the two 30" legs of the Tee were outlets. This produced results that make sense. It seems the 46 small holes are causing problems with the solver.

I am new to using SimScale, and CFD software in general, and am not sure how to proceed with getting results accounting for flow out of the small holes. Any help is appreciated.

Hi @rtrowbridge!

Very nice project you have there! @power_users and I will have a look at that and get back to you as soon as we have figured out the problem. Currently running a mesh followed by a simulation to see if we can avoid the problems you were mentioning.

Best,

Jousef

Hi @rtrowbridge!

I see you are using the Hex-Dominant Automatic meshing function. Normally I would typically use the Hex-Dominant Parametric meshing function for greater control at the points of interest.

With the automatic function it seems like the relative low quality of the mesh at the outlets (holes) is causing some problems for your results and I would recommend using the parametric function and refining the area near the holes to get a better resolution as currently it seems rather distorted.

I’ll mesh it out and see if it works, but in the meantime can you share how you verify your velocity? What would be considered the velocity that “makes sense” as you have mentioned?

I’ll play around with the results control to see if I am able to obtain out the values. For downloading your data and post-processing offline I think ParaView may be able to obtain our your flow rate and i’ll again get back to you on this.

Cheers,
Barry

2 Likes

Thanks for jumping in here @Get_Barried! Simulation finished yesterday evening and will have a look at the results later on. Some reference values would be nice to know if we are moving in the right regime or if we are completely off.

Cheers!

Jousef

2 Likes

Thanks so much for the help on this. The inlet flow rate is 45,000 gpm. I originally used this (173,250 in3/s) as the inlet condition. I then did a run with the inlet defined as a uniform velocity of 408 in/s, which is the average velocity for the flow rate in the 24" inlet pipe.

The average velocity out of the 30" outlets will be much lower than the inlet. Some of the holes near the middle of the Tee may have slightly higher velocities. I would prefer to get some flow out of the holes to limit the amount coming out of the 30" ends.

Hi @rtrowbridge!

Sorry I don’t exactly understand what you are looking for. Is the ideal criteria simply obtaining a flow rate at the 30" outlets to be lower than the inlet by more than half? Or the total flow rate out of the holes is higher than the flow rate out of the 30" outlets? Is there specific goal of the simulation in terms of numerical limits?

You did state the objective here but what do you mean by diffuse? A specific flow rate?

An example of a recent project I did was to deduce the average flow velocity within a floor of the building and compare that against an ideal flow velocity that had specific numbers. So its very clear what I need to compare against and allows me to have a good gauge if my results are way off the mark or not.

Cheers,
Barry

Hi @rtrowbridge & @Get_Barried!

Also finished my simulation - one with and another one without result control items. You may have a look at it and see if the results are fine or if you would have expected anything else.

Project Link: Tee Diffuser - CFD

Best,

Jousef

Thank you! These results are more like what I was expecting. It seems the inconsistent results I was getting were due to the poor mesh I had near the holes - as Barry suggested above.

Barry asked the objective of this Tee assembly. Here is a bit of background information. Currently a 24" pipe dumps directly into a basin. This stirs up solids that have settled in the basin. The operator wanted an economic way to minimize this. The Tee is meant to spread the flow across multiple points to lessen the disturbance on the bottom of the basin. 30" pipe was selected because it is available at site.

Thanks again.
Rick

2 Likes

Nice Rick (@rtrowbridge)!

If you need further help with the optimization of your Tee assembly please let me know. @Get_Barried, @1318980 as well as myself are here (almost 24/7 :wink: ) to help you with problems you may encounter.

Cheers and happy SimScaling! :slight_smile:

Jousef

1 Like

Hi @rtrowbridge!

Thanks for the input! I did a quick run myself and didn’t post it as my results seem a little off hence the need to ask for clarification on acceptable results.

Looks like Jousef got you covered! Let us know if you still have issues!

Cheers,
Barry

1 Like

Jousef,

We made slight modifications to the geometry to optimize the flow. I loaded this geometry into my project. It is geometry Trial_2. I meshed this similarly to how you performed the mesh in your project. The mesh is Tee_2. The mesh looks good. When I ran the simulation, I still got results that showed incredibly high velocities relative to the inlet. I did a few simulation runs with slight changes to the solver and had still had unreasonably high velocities throughout the model. The simulation runs under Tee_2_Smooth_Solver are set up just like the run you shared with me yesterday.

Can you take a quick look at the mesh and simulation run I mentioned above and see if I have made any errors setting it up? This is the first time I have ever done a CFD analysis, so I may be missing something obvious. I have been looking it over for a while and am stumped now.

Thanks so much for the help.

Rick

Hi @rtrowbridge!

You are currently running a simulation and I would like to see what your run produces. Did you delete the old simulation run? And do you have any reference value of what you would expect for the velocity? Maybe @Get_Barried or @1318980 have a good tip here!

Best,

Jousef

I just copied the project you shared with me yesterday and uploaded the new geometry to it. I meshed it and set up the solver by duplicating your work and reselecting faces, etc. That is the run I am doing now. The other results, which I mentioned in my previous post, are in my first project - Tee Diffuser - Incompressible CFD.

The water enters the Tee at an average velocity of 408 in/s. The average velocity out of the 30" outlets has to be much lower than that. Your results indicated this. My results are giving thousands of in/s out the 30" pipes, which is why I think I must have made an error somewhere.

Thanks,
Rick

Got you @rtrowbridge!

Can you tell me the name of the simulation? You have quite a lot in the project you have shared. Trying to find out what the cause of this might be.

Best,

Jousef

Sorry, the project is cluttered because I kept trying different meshes and simulations.

The simulation to look at is Tee_2_Smooth_Solver. In that run I tried to mimic the run you did that produced good results.

Hi @rtrowbridge and thanks a lot!

I will see what can be done here. Also keep in mind that turning some screws here and there just to get the right velocity does not make sense in every case but that’s how it sometimes works and might be the only approach possible. A lot of simulations are often trial and error and keep in mind that the “prototype hurdle” cannot been eliminated yet. :wink: The best is, as mentioned before, to have some reference values. From there you can then say ok the values are of by x.y% and then we could have a look at the model and which assumptions were wrong in the first place. A design option would of course be to increase the size of the small outlet holes inducing a bigger velocity drop at the both big outlets. Does that make sense so far? Regarding the turbulence model we also have to make sure that the values for the TKE (k) and turb. dissipation rate (\mathbf{\omega}) are correct/get adapted. @Get_Barried & @1318980, feel free to add your comments here.

Best,

Jousef

Hi @rtrowbridge!

Hmm I just checked your simulation and on the surface it does look the same as what @jousefm did with the exception of a few things. It seems like you’ve modified the starting geometry and removed one of the baffles in the center. That was the main thing I could see that was different other than a slight change in velocity input where Jousef set it as 418 in/s compared to your 408 in/s but it shouldn’t make such a huge difference. Other than that, the convergence graph and of course the resultant flow velocity out of the holes is very different from what Jousef got. It seems like your flow is accelerating out of the holes which is particularly strange.

I am currently remeshing your “new” geometry without the baffles and will re-run with Jousef’s parameters to see if I can get somewhat logical results.

Referring to what Jousef has mentioned and me earlier as well, some concrete values or percentage difference in the flow rate to reference off would be ideal. If instead the objective is to approach this by determining what designs will allow the flow rate to drop down to a significant enough level then probably calculating out the ideal flow rate out of the holes with certain assumptions would be able to give you that ballpark ideal figure you’re looking for.

Off the top my head I would assume you want laminar flow out of the holes? Not sure if its possible with the flow rate of this speed or how to compute out the Reynold’s number of the outlets considering that turbulence is involved. Any ideas on this Jousef and @1318980?

Get back to you once I’ve identified the key issue if any.

Regards,
Barry

1 Like

Thanks guys for the help. I ended up having some luck with this. I certainly would not have been able to get valid results without your help.

I re-meshed the new geometry (middle baffle removed and one end baffle relocated) and ran another simulation. I did this in project Tee Diffuser - Tests. The results are much more like I was expecting. The average velocity out of the 30" ends was about 120 in/s. Only a small percentage of the flow is going out the holes. If there were no holes, the velocity out the 30" ends would average 129 in/s. This makes sense. The holes really aren’t doing much, I believe because of the relatively high water velocity through to pipe.

I also uploaded another geometry, one that has been used at site, and simulated it. The results were as expected based on what was noted in the field.

Thanks again.
Rick

1 Like

Cheers @rtrowbridge! Glad things worked out!

Interesting conclusion. I would say probably increasing the number of baffles to reduce flow speed or increasing the length of the “diffuser” per say (section of the pipe where the holes are at) might help to achieve what you want.

Hope it helps!

Regards,
Barry