Learning Forum

[sebastiancg26]

Hi everyone!

Currently, I'm simulating the transient behavior in a reciprocating compressor. For this, I implemented the use of a UDF to open or close the compressor valves through Porous Zone (Changing the viscous resistance) based on a average pressure. This code works great, but the problem is that in the animation of total pressure, the behavior of that pressure is totally uniform, wich doesn't make sense. I would like to ask you if anyone of you knows what is causing this problem and how I can fix it?

[sebastiancg26]

[Rob]

Which solver licence are you using?

[sebastiancg26]

Ansys Student

[Rob]

Has the model run for long enough to see a change, and converged each time step?

[sebastiancg26]

Of course Rob, I've tried many things, but I still can't find any solution..

[Rob]

Can you plot some more data, eg velocity. Diagnosing from one plot with no idea about what else is set is difficult.

[sebastiancg26]

Sure. When I plotted velocity, I realized that the velocity values throughout the simulation were quite low, thus, the Dynamic Pressure values had the same behavior. For this reason, the dynamic pressure values aren't enough to modify the Total Pressure behavior. Next, I will leave the screenshots of Total pressure, Dynamic Pressure and Velocity plots.

[Rob]

And you're using superficial velocity in the porous media. Check static pressure too. I suspect the dP over the porous media is masking the results as the scale is very big.

[sebastiancg26]

This is the static pressure behavior.. practically, it's almost the same behavior of total pressure. The effect of dyamic pressure doesn't have much influence on the results of total pressure.. I don't know how to solve this, I would be very gratefull if you can help me with that..

[Rob]

Yes. Now replot with a range of 0Pa to about 20k Pa.

[sebastiancg26]

It had the same result..
Theese simulations are just an example, I have been simulating with only one boundary condition (Inlet pressure=100kPa), and the porous media is configured like this:
As I mentioned you, the idea is that the viscous resistance coefficent changes according the pressure in UDF (wich is already finished, and works well), however, now I have been running a series of test to determine the proper porous media setting..

[Rob]

Try a lower upper limit. Simply, the pressure field is being hidden by the dP over the membrane.

[sebastiancg26]

I already did it, but the result still remains the same.. I need to use "Alternative formulation" in porous media setting? or do I need to modify another parameter?

[Rob]

No need to use alternative. The solver is calculating the flow based on your settings. The porous coefficients are high, so the pressure loss all occurs at the membrane. Depending on the upstream boundary flow must pass through the membrane.

[sebastiancg26]

I need to establish a coefficient that guarantees a total flow (that simulates an open valve) and a coefficient that guarantees the blocking of the flow (that simulates a closed valve), without affecting the behavior of the pressure field in the cylinder. How I can do it? Thanks for all your comments..

[Rob]

Porous media will cause a pressure drop, depending on the external boundary conditions and other flow features it can't stop the flow. That's why the IC tools allow a crank angle activation of TUI commands: we switch surfaces from interior to wall.
We model oil reservoirs with porous media, with 3-400bar behind the system anything will move!

[sebastiancg26]

So.. Can I switch a surface from interior to wall depending on the pressure instead of the time?

[Rob]

Not without some horrible UDF scheme calls.

[sebastiancg26]

How can I do theese scheme calls?

[Rob]

With a UDF. Unfortunately if it's not in the UDF manual we can't offer any assistance. Simply, you're making a call to the solver based on a value that the UDF calculates, this isn't straight forward.

[sebastiancg26]

What a pity.. So, is there any other way to simulate a flow block?

[Rob]

Given it's a piston compressor why not base open/closed on the time?

[sebastiancg26]

This simulation is carried out through "Events"- The events are:
Delete Sliding interface at 0.2 [s] to closing inlet valve.
Create Sliding interface at 0.75 [s] to opening outlet valve.
Delete Sliding interface at 1 [s] to closing outlet valve.
Create Sliding interface at 1.8 [s] to opening inlet valve.
But, the pressure behavior remains the same, it's constant. Why this?

[Rob]

Have the commands executed? What does the TUI transcript report?

[sebastiancg26]

Yes sr. All events were executed correctly, in fact, I'll attach the resulting graph of the pressure in the cylinder..

[Rob]

If the graph is as above, what are you displaying to not see it in the contours?

[sebastiancg26]

Actually, I don't know.. however, if the simulation runs without events, it does show a pressure distribution. When the simulation runs with events, the pressure behavior seems constant throughout the area, as I showed you in the images in previous comments.. I clarify that the only modification I made between the two simulations was adding / removing the events. The rest of the configuration remains the same..
This a plot of the pressure behavior in a simulation whitout events..

[Rob]

Events are things changing, so will alter the result. If you display with a large scale/range on the plots subtle differences in the flow may be missed.

[sebastiancg26]

Here the same phenomenon is simulated, but the behavior of the pressure is very different, it seems very real. I think they simulate it in COMSOL, but how can I make it look like this with ANSYS?

[Rob]

That looks like they're using crank angle to drive the solution: exactly how you've done. However, as you've only shown one graph and an image at the "piston down" it's hard to see what is/isn't working.

[sebastiancg26]

These are the results produced by my simulation at the same time or crankshaft angle indicated in the previous example..

[Rob]

That looks like the interface creation is doing something weird. Try switching to SIMPLE and use PRESTO!

[sebastiancg26]

These are the results produced by my simulation with SIMPLE method and PRESTO, it's almost the same..

[Rob]

Almost certainly the interface then. The solver setting change was to confirm they weren't to blame. Focus the view around an interface and look at the velocity vectors. Using fluxes see how much mass is passing though an "open" valve.

[sebastiancg26]

[sebastiancg26]

Here are the pictures of the interfaces and the mass flow report through the valves..
I find it curious that the mass flow of the outlet is not the same as that of the inlet.

[Rob]

You also have moving mesh, are you accounting for the volume change too? Please can you post an image of the mesh? Or did you skip several vectors in the plots?

[sebastiancg26]

I use Layering method with a 2 mm quadrilateral mesh

[Rob]

Make the mesh a lot finer at the top (and overall) and see what happens.

[sebastiancg26]

That did not influence the effects.. I made a new 0.5 mm quadrilateral mesh, and I changed the step size to 0.0001 s

[sebastiancg26]

the turbulence model has something to do with it? I use k-w model..

[Rob]

Doubt it. Monitor flow in, flow out and piston volume. Also monitor the flow over the 3 interface pairs. We should see the latter give a zero result at various points, and if so the inlet/outlet ought to be around zero then too.