I am struggling with Venturi effect simulation. Basically, I am experimenting with CfdOF module, trying to simulate Venturi nozzle which will work as a pump sucking water. Air will be the pump’s driving pressure.
I went through the internet and I couldn’t find similar case, the closest one I have found was on this forum https://devtalk.freecad.org/t/cfdof-default-maxcomaxalphaco-in-controldict/69309/1 - I took some hints from this project, however I couldn’t achieve Venturi effect of pressure drop in the nozzle and trigger water suction.
I am attaching my project, I will appreciate any help or hints what I am missing or doing wrong.
I took a look on your simulation. In order to better understand your model, I created a 2D-drawing:
Nozzle_Cross_Section.pdf
On my computer, your simulation works. This is the result:
Nozzle_Air_Water_Distribution.JPG
I generated this picture using paraview. The picture is made with “slice”. You are performing a transient simulation, so the residuals look OK. So that you can better understand your results, you should reduce the “write interval” in order to get more images:
Nozzle_Write_Interval.JPG
I noticed that your model is very large and the calculation time is very long. I would recommend that you optimize the calculation times using the boundary condition “Symmetry”. You can also build a 2D model or an axially symmetrical model for simple preliminary tests.
I assume that you are not completely satisfied with the results. Some comments about your model:
You have a limited amount of water in your model. The water will be empty after some time.
By optimizing geometry and the speed of the air, you will get better results
The Venturi nozzle creates a pressure that is below the pressure of the atmosphere. In order for the water to flow, the pressure from the atmosphere must be able to act from another side. With your current arrangement you are creating a vacuum.
If you search for “Venturi effect” on Wikipedia, you will find many interesting examples:
Thank you for your response and valuable comments. Before moving to describing my changes I briefly describe origin of my CfdOF project and motivation.
Originally, I was experimenting with 3D prints and results were promising - a short video showing working nozzle in my professional test rig https://youtube.com/shorts/NiZ76fb50Xs
Nozzle on video has the same dimensions like nozzle in attached project, so it works (I know that pulsing pressure and water level differs form simulation, but major Venturi effect appears). Then I have added ChatGPT support to enhance nozzle design, however those enhancements were rather poor/random, but this is a different story…
So far, I have created small collection of nozzles (the biggest on the left is creation of ChatGPT, and it doesn’t work )
nozzles.PNG
As 3D prints are time consuming, so I decided to try simulations with CfdOF. BTW, thank you for your tutorial series on YouTube, it helped me a lot to start with simulations .
Regarding your recommendations and comments:
I introduced optimizing the calculation time using symmetry condition - it’s much faster now
I added open reservoir with water and keep it open and secondary inlet with atmospheric pressure, I am not sure how to simulate constant water level with atmospheric pressure
SecondInlet.PNG
Also I am not sure if I did that good because, even I setup initialization zone, I got strange bubbles at the beginning of simulation
alphaAir.PNG
I increased write interval to 0.05s
I am attaching zipped project, it got bigger than 1MB I don’t know why
POC_venturi_simulation_v2.zip
Many thanks for helping me!
Maybe the different story is interesting.
How can the AI compute a flow field?
And give a optimal design?
Whats your input?
As far as I understand the AI systems the computing method is to put
a filtering system (the “artificial intelligence”) over giant amount of data.
Not more.
It’s all about conversation and statistical predictions of next word, and if you are lucky it will tell you something right if not you will get complete garbage. Anyway it’s fun.
The same with programming, typical programming patterns it gets most of the time right, but some tasks more complex with longer context often is messed up.
Here is example of quick conversation about Venturi I did second ago https://chat.openai.com/share/85304383-034c-40c6-8b12-86b0bc27944a
It’s able to do some complex math, but I recommend to verify it closely.
I do not know that either. However, a sufficiently large water reservoir should be sufficient. In the simulation results you can then see whether the reservoir is running dry.
Hi Marek,
you can now try to optimize the layout. I see the following parameters that you can use to optimize:
diameter air inlet
diameter water inlet
angle water inlet
diameter after mixing air and water
To the individual points:
Point 3: In the arrangements I saw, the angle (“angle water inlet”) was in most cases 45° or more degrees. Your current design has a degree of 0°. Study the effect of higher angles.
Point 4: In most arrangements I saw, the diameter after mixing water with air was larger than the diameter at the mixing zone. The diameter in the mixing zone was always the smallest diameter. Study the effect of smaller or larger diameters.
You should determine characteristic values (pressure, speed, mixture) at certain points and write down these values together with the calculation variant. The “Steam Tracer” from Paraview can also be helpful.
So far, I have been experimenting mainly with the diameter of the mixing part and the depth of placement of the air inlet in the nozzle.
You are right, my design is missing an expanding nozzle element containing a mixture of water and air. Therefore, I will focus on adding that in the first place. Afterward, I will gradually incorporate other modifications.
Thanks again for spending time replying to my post
.