[lund8619]

I attempting to simulate the human ear canal, tympanic membrane (TM), and cavity. The canal and cavity are both fluids (air) and the tympanic membrane is a solid. A pressure spike is introduced at the inlet of the canal that decreases linearly with time.

ei. if(t>0.0[s]&&t<0.00003[s],(-2333333.333...[kPa/s]*t)+70[kPa],0.0[Pa])

I have successfully completed simulations with unreasonably high values for the Young's Modulus of the TM, but when I try to use a more realistic value (98kPa) for this, the solver runs until the pressure wave reaches the TM, at which the negative element is detected and the simulation ends.

I am self taught and still learning what information is useful when debugging and sharing my own simulations. I am willing to share my entire project in order to get a successful run of the model, so if more information is needed, just let me know. Attached are a gif of a successful run and the first and last portions of the .out file for a failed run.

Josiah

[mvoss]

Where is the neg. element? Try to save the mesh for every timestep up to the impact. Relaxation factors could be lower (like 0.5-0.1) in flow direction - same for the Convergence Target (1e-3 -1e-4)? Increase the coupling iterations. Convergence criteria for the meshmovement is crucial- i would suggest at least max 1e-6? Could be basically anything. But getting to position of the neg. element is the first thing to do.

[lund8619]

The negative element appears at or very near to where the TM meets the canal wall. As for the saving the mesh and the timestep, I have run with a step as small as 1e-9[s] outputting every 1e-7[s]. I do not know my relaxation factor or convergence criteria off the top of my head, but I believe they are the default values. I will look and post the values in the morning.

[mvoss]

Can you post some pictures?One overall and one zoomed in?

[lund8619]

I have gone in and looked up all the data you mentioned for this simulation.

-Time Step = 1e-7
-Output Frequency = every 10 timesteps
-Under Relaxation Factor = .75
-Convergence Target = 1e-2
-Max Coupling Iterations = 10
-Convergence Criteria = .001 (RMS)

From the looks of your suggested values, and the values used in my simulation, the problem may be solved. I will adjust the values and run again. Attached are the images you requested. Through the center of the model, user points are plotted for data observation during the simulation. The user point in that is out of line is the location of the negative element.

Thank you for your response,
Josiah

[mvoss]

You definetly Need to refine the mesh, before applying any of the suggested tweakings. Is the tm sealing the whole canal? Hard to tell from the pics.

[lund8619]

Yes, the TM completely spans the passageway between canal and membrane. What work can I do refining the mesh? I used the most fine settings when generating the mesh. (Average mesh quality is .804)