Nonlinear static runs stuck at 0% with no Solver Log – repeated twice

Project Support
1

Hi SimScale Support / Community,

I am experiencing a repeated issue where a nonlinear static structural simulation remains at 0% and never generates a Solver Log.

I would like to determine whether this is a compute/job-start issue or whether something in the simulation setup prevents the solver from launching.

Simulation setup

  • Analysis: Nonlinear static structural
  • Geometric nonlinearity: enabled
  • Materials: hyperelastic
  • Contact: bonded
  • Element technology: Automatic
  • Mesh order: 1st order
  • Mesh size: 74,572 tetrahedra, 20,780 nodes
  • No Remote Force is used
  • Boundary conditions consist only of one Fixed Support and two Pressure boundary conditions
  • The loads are explicitly ramped with pseudo-time t rather than applied instantaneously
  • Automatic boundary-condition ramping is disabled because the pressure ramp is explicitly defined in the formulas

Run 1 – Nominal_Rest_Diagnostic_v1

  • Remained at 0% for approximately 47 minutes
  • No Solver Log was generated
  • I then requested cancellation
  • Cancellation itself took approximately another 30+ minutes
  • Final status: Canceled by user / Error 0%
  • Event Log stated:
    “The run was canceled but the cancellation didn’t finish in time. No core hours will be deducted from your account for this run.”
  • Final reported core-hour usage: 0.00

Run 2 – Nominal_Rest_Diagnostic_v2

  • The page was refreshed before starting the second run
  • The simulation setup was left unchanged
  • Again remained at 0% for more than 27 minutes
  • Again, no Solver Log was generated
  • I requested cancellation
  • Final elapsed time was approximately 50 minutes
  • Final status: Canceled by user / Error 0%
  • The same cancellation-timeout message appeared in the Event Log
  • Final reported core-hour usage: 0.00

The Event Log did not show a message indicating that the job was waiting for compute resources.

I also contacted the in-product Ray Support agent. It confirmed that:

  • standard Pressure boundary conditions applied to many faces do not normally cause this kind of long 0% delay;
  • the known preprocessing issue involving many faces mainly concerns Remote Force, which I am not using;
  • if a job is waiting for compute resources, this should normally be reflected in the Event Log;
  • there is no additional backend job-start status available to the user beyond the Event Log.

My main concern is therefore that in two independent runs the solver apparently never reached the stage where a Solver Log was created.

Could you please help clarify:

  1. Does the absence of any Solver Log after tens of minutes indicate that the solver/compute job never actually started?
  2. Is there anything in this setup that could block the solver during preprocessing before the Solver Log is created?
  3. Could this be a backend/resource-allocation issue even though the Event Log did not show a “waiting for resources” message?
  4. Would you recommend retrying the same simulation unchanged, or is there a specific setup item I should check first?

I have attached screenshots showing:

  • the final status of Run 1;
  • the final status of Run 2;
  • the mesh information showing 1st-order mesh, 74,572 tetrahedra, and 20,780 nodes.

Project link:
[PASTE PROJECT LINK HERE]

Thank you very much for your help.

image
image380×401 11.8 KB

image
image769×617 51.1 KB

image
image345×510 12.2 KB

image
image346×465 12.2 KB

3

Hi @kongye123,

When a nonlinear static simulation remains at 0% with no Solver Log, this may indicate a delay in cloud resource allocation or preprocessing, rather than a solver convergence issue.

Possible causes

  • Backend / Resource Allocation: The cloud instance may still be provisioning, preventing the solver from starting and generating a log.
  • Preprocessing: Large meshes or complex contact interfaces may require additional time to prepare before the solver starts.
  • Solver Convergence: Convergence issues generally occur after solver initialization and are usually accompanied by messages in the Solver Log.

What to check

  1. Simulation Tree: Ensure all mandatory settings are complete and no red status indicators remain.
  2. Mesh and Contacts: Check whether an excessively dense mesh or complex contact interfaces could be delaying preprocessing.
  3. Load Ramping: For nonlinear analyses, consider ramping your loads gradually to improve convergence once the solver starts. See Static Load in Nonlinear Analysis .
  4. Retry the Simulation: If the simulation remains at 0% for an extended period, try restarting it. If the issue persists, contact SimScale support.

Could you also please share your simulation URL so I can take a closer look at your setup and help identify the issue?

Thanks! Bryan

4

Hi,

On top of what Bryan mentioned above, due to the thousands of faces in the geometry, the mesh quality metrics seem pretty bad, as you can see from the last screenshot that you posted in your original message. E.g. the 99.99th percentile of tetrahedra aspect ratio was almost 2 million, which is incredibly high.

It’s very likely that the solution is very unstable in some regions of the mesh, due to a combination of mesh quality/physics/material properties. Is there any way to improve this geometry? Not having it all triangulated would already help quite a bit.

This is all off of the information that you posted in the thread, as the project does not seem to be shared.

Cheers

5

Hi Ricardo,
Thank you very much for pointing this out. The geometry was reconstructed from medical imaging data and exported as STEP after segmentation and processing, which is why the surfaces contain a large number of triangular faces.

I agree that the extremely high tetrahedral aspect ratio is concerning. Before rebuilding the geometry from scratch, would you recommend any specific geometry-cleanup or simplification approach that could improve the mesh quality while preserving the overall anatomical shape?

Thanks again for your helpful comments.

7

I guess it depends on how the geometry is created and what CAD tools are available.

E.g. this is a step file with a complex curvature where there are no individual triangles.

8

Thank you, that makes sense.
My geometry was originally reconstructed from medical imaging and then exported to STEP, so it seems that the STEP file may still preserve a highly faceted surface representation rather than smooth CAD surfaces.

In your experience, would you recommend rebuilding/fitting the surface into smooth CAD faces before meshing, or is there any practical way within SimScale or common CAD tools to simplify such a faceted STEP geometry without substantially altering the overall shape?

9

Hi Bryan,
Thank you for your reply. As this project contains unpublished academic research, I would prefer not to post the simulation URL publicly on the forum. Could you please let me know the best private channel to share the project URL with you for troubleshooting?
Thanks again for your help.

12

Hi Bryan,
I am following up once more regarding the nonlinear structural simulation that remains at 0% without a Solver Log. Since the project contains unpublished research, I would still prefer not to post the private project URL publicly. If there is no private support channel available for Academic projects, please let me know and I will proceed with alternative troubleshooting approaches.
Thank you.