Learning Forum

[Andres.Mena]

Hello all!

I'm currently extracting several variables from my model where several bodies contact one another with a near-frictionless penalty-based contact (*CONTACT_AUTOMATIC_SINGLE_SURFACE). I'm trying to compare contact variables to an experimental reference where they report peak pressure, mean pressure, and contact area gathered using Tekscan pressure films. I have some questions that I hope someone else has experience with.

1. If two different parts are in contact, should both parts have the same contact pressure, force, and area readings? I have a force transducer on each of the two contacting parts, and I found that the forces from RCFORC agree pretty much exactly, but the results for pressure and contact area were close (~5 to 20% difference) and not exactly the same. Is this related to the mesh or element density, or could it point to something wrong in my contact setup? If it is just discretization, is there a recommended convention for which side to report or to average both sides?
   
   
2. I noticed that I have access to several "pressure" variables in LS-PrePost, so I'm wondering what each variable represents and which could be better suited to my purpose:
   a) "Pressure" from d3plot, which I guess could be hydrostatic pressure
   b) "Interface Pressure" from intfor, obtained using a transducer
   c) "Peak Pressure" from intfor, after setting NPRESU to 3 on *DATABASE_EXTENT_INTFOR

I compared these three variables across several simulations and the values were close, but not exactly the same. Is any of these options better to compare against a pressure film? Also, what's the difference between peak pressure and just taking the maximum value of interface pressure at a given state?
3. Related to the above, I also need contact area for the same comparison. Would anyone know how LS-PrePost computes the "Contact Area" quantity? Is there a threshold being used to sum elements or does it add all elements with nonzero values?

Thanks in advance for any guidance!

Best, 

Andres

 

[Nanda Veralla]

Hello Andres,

A few thoughts based on your observations:

The fact that the contact forces from RCFORC/transducers agree almost exactly is a very good indication that the contact definition is behaving correctly. For two contacting bodies, the resultant contact force should satisfy Newton's third law and therefore be equal and opposite. The differences you're seeing in contact pressure and contact area (5-20%) are generally not unusual for penalty-based contact. Unlike force, contact pressure is derived from local penetration, penalty stiffness, and segment discretization, so the reported pressure field can differ slightly between the two contacting surfaces. Mesh density and element size are often the main contributors. In practice, I would treat these differences as discretization effects unless they are significantly larger.

Regarding the pressure outputs:

1. Pressure (d3plot): Typically, the element hydrostatic pressure (mean stress) within the material and not a true contact pressure. I would not use this directly for comparison with Tekscan data.
2.  Interface Pressure (INTFOR): Contact pressure calculated at the interface. This is generally the most appropriate quantity for comparison with pressure-film measurements.
3. Peak Pressure (INTFOR, NPRESU=3): Intended to provide the maximum contact pressure measure on the interface. It may not be identical to manually taking the maximum contour value because of differences in storage, averaging, or nodal/segment-based processing.

For comparison with Tekscan, I would typically use:

• Peak Pressure: INTFOR Peak Pressure
• Contact Area: INTFOR/LS-PrePost Contact Area
• Mean Pressure: Contact Force ÷ Contact Area
• Pressure Distribution: INTFOR pressure contours

For contact area, as far as I understand, LS-PrePost computes it from the contact interface data stored in INTFOR. Conceptually, it sums the area of segments that are actively in contact rather than the full geometric area. Because the definition of "active contact" depends on the contact algorithm and local contact status, contact area is often the most mesh-sensitive quantity and may differ slightly between opposing surfaces. Overall, since your forces match closely and the pressure/area differences are relatively small, I would be more inclined to attribute the discrepancies to contact discretization and mesh resolution than to an issue with the contact setup itself. A mesh refinement study would be the best way to confirm this.

Regards,

Nanda

 

Have a look at our public help documentation: Ansys Help

For more exciting courses and certifications, check this out: Learning

 

[Andres.Mena]

Nanda,

Thank you for your thorough and detailed response! This clears up most of my doubts and I will use the INTFOR variables for my comparisons. Two follow-up questions if you don't mind.

I discarded the d3plot pressure since I understand it's a different quantity. I'm now trying to better understand the difference between Interface Pressure and Peak Pressure, since the two are not the same in my model (20 to 53% difference for max value for some parts), and both render slightly different contours in LS-PrePost (the number of unaveraged nonzero segments for both peak and interface pressures seem roughly the same, but peak pressure seems "scaled". Also, the max segment for peak pressure matches the max interface pressure segment. I can provide pictures of this if it's helpful). I had initially assumed Peak Pressure was a per-segment maximum over time of Interface Pressure, but the Peak values come out higher than the maximum Interface Pressure I see anywhere in my time histories for the same segment. Is this expected and is it okay to use any variable despite the difference?

2. On the two-part differences for contact area and pressures, I was previously thinking of averaging the two-part values, but do you think it would be reasonable to report the value from a single surface considering that a pressure film is physically placed on one surface only? I do understand the mesh refinement study is the right way to obtain one value and I'm planning on it. I'm mainly trying to cover my bases in case the element size needed for converging contact area  ends up finer than what my computer can realistically support.

Thanks again for your help.

Best,

Andres