Learning Forum

[peteroznewman]

I downloaded and solved Tech Demo Chapter 43 to explore the Archard wear model to see material removal as a function of the wear constant K, material hardness H, contact pressure P and sliding velocity.

I ran the TD43 (as is) which is an Axisymmetric model that includes Nonlinear Adaptive mesh code to automatically remesh the body when the element height is cut in half by wear.  The analysis is a two-step simulation with a compression force applied in step 1 where no wear occurs, while in step 2 the wear process occurs as a linear function of time in this static structural analysis.

Experimentally, test sample surface profiles are measured before and after wear operation using an optical profilometer. The profilometer software calculates the wear volume and wear depth from the acquired data.  Therefore a direct comparison between experimental measurements and simulation results would be wear volume and/or wear depth. 

I added two results to the simulation: a Volume result and a Volume Probe on the copper body and assumed Wear Volume = Initial Volume – Final Volume.

Volume Result
Step 1 showed a linear reduction in volume as the compressive load ramped on then a stair-step reduction in volume upon each remesh. Between remesh events, the volume increases slightly which is odd. 

Volume Probe Result

A large and sudden change in the Volume Probe happens when the first NLAD remesh occurred. Why does that happen? 

Quick Test

I suppressed the Nonlinear Adaptive mesh code, changed the End Time of Step 2 to 2.77 s which is when the first layer of elements wears down to a thin element height and added Step 3 to ramp the compression force down to zero.

From the Total Deformation at the end of step 3, the wear depth at the center is seen to be 1.05 mm.

Now the increase in Volume during wear in Step 2 is easily seen and furthermore, the volume after unloading is equal to the volume at the start of the simulation! 

Why does the volume increase when the elements are getting thinner? What is the correct way to measure wear volume?

[directcustomer]

Hi Peter,
Thanks for your patience. Let me try to run the simulation and will update you soon.

[Sampat Kumar]

Hi Peter,
I ran the model on my end and reached the following conclusions based on my observations.

In a wear analysis using Nonlinear Adaptive Mesh (NLAD), the mesh is periodically regenerated when elements become highly distorted. At each remeshing point, the existing mesh is discarded and a new mesh is created to better represent the current deformed geometry. State variables such as stress, plastic strain, wear, and contact status are projected onto the new mesh, rather than being preserved element by element.

Because post‑processing quantities such as Total Volume are recalculated on each newly generated mesh, small jumps or “stair‑step” changes in the volume result are expected at remeshing events. As seen in the attached image, this behavior is normal.

The slight increase in volume you observed is likely localized to specific elements. During remeshing, some newly created elements may be slightly larger while others may be smaller. This effect is purely a consequence of the mesh regeneration process. Overall, the total volume continues to decrease, which is consistent with the wear behavior.

The drastic volume reduction observed in the Volume Probe is valid, and I am able to reproduce the same behavior after the first remeshing point.

There may be some numerical artifacts associated with the Volume Probe result scoping. The probe tool scoping has some limitations when used with Nonlinear Adaptive Mesh (NLAD). The probe values may not always represent a smooth or fully consistent response when remeshing occurs.

I will further investigate the Volume Probe behavior in the presence of nonlinear adaptivity and will provide an update once I have more information.

Best Regards,
Sampat

[somaroy]

 

So the Archard wear commands compute wear depth at the contact interface, but this does not produce wear, as in geometric material removal? Is this why the solid body volume does not change unless mesh updating occurs? My understanding is that activating the Archard wear model computes wear, which is stored as a state variable rather than a volume change. Is there a way to extract this wear volume without implementing NLAD?

Archard’s wear equation states that the wear rate is proportional to contact pressure and sliding velocity. Why is the velocity exponent, n=0 in TBDATA commands?

If I change the hardness value (from H=1 MPa to H= 300 MPa for the copper ring) in the TBDATA commands (TBDATA, mat_id, K, H, m, n, 0), the wear volume is less, as one would expect. Why did the tech demo use a hardness value of only 1 MPa for copper? The user-defined materials do not have hardnesses specified. Wouldn’t the “unspecified” hardness of steel affect the stress distributions? I am assuming it has no effect on wear volume, because only the wear coefficient determines wear volume, and the hardness of the countersurface has no effect on wear volume.

 

[peteroznewman]

I believe the reason why the velocity exponent, n=0 is because in a Static Structural model, the relative sliding velocity vrel is 0, and when you raise 0 to the power of 0 the result is 1 so the wear rate can be calculated.

[peteroznewman]

Thanks for looking into this Sampat, really appreciate it.  The takeaway is to use a smaller tolerance to trigger more frequent remeshing so that the Volume output tracks the wear more smoothly.