NaN error in MoorDyn when simulating free decay and wave conditions

[YuYuJou]

Hi everyone,

I'm trying to simulate a floating MHK turbine designed by our team, and I’m currently testing both free decay and wave conditions. However, the simulation always fails with NaN errors very early in the run.

I've attached a screenshot of my MoorDyn input file and the corresponding error messages. From what I can tell, the issue might be related to the mooring system configuration, but I’m not sure what exactly is causing the problem.

Best regards,

[RyanDavies19]

Hi @YuYuJou

There's a couple things worth trying that might help. Does MoorDyn successfully finish the dynamic relaxation phase before beginning the simulation? If not try increasing your TmaxIC factor to something closer to 100 or 200 seconds. This ensures MoorDyn has enough time to find its equilibrium state before the simulation begins.

I also noted your bridle lines (line 1 and 2) are 22 m long, which is longer than the distance between points 1 or 2 and point 4 (~17 m). This means at the start the mooring line will be hanging in a catenary shape between those points. Because that point is free with no volume or mass, this means your system is defined out of equilibrium (which is fine, but requires a longer TmaxIC to reach steady state before simulation begins).

If that doesn't resolve your issue, I would suggest running with your fairleads as fixed (instead of vessel) to see if your MoorDyn configuration is stable without external forcing. It could also help to visualize your mooring system, which you can do with Paraview using the vtk files OpenFAST writes, or using a tool like MoorPy (see the visualization section of this user manual https://github.com/FloatingArrayDesign/MoorDyn/blob/dev/example/MoorDyn_standalone_demo.ipynb).

You can also always reduce your dtM until you no longer get nan's, though this will slow your simulation. This could be useful for getting a simulation to visualize when troubleshooting.

[RyanDavies19]

I'll also add that you can run MoorDyn on its own using the MoorDyn driver, which is very helpful for troubleshooting. Instructions for that can be found here: https://github.com/FloatingArrayDesign/MoorDyn/blob/dev/example/Running%20MoorDyn-F%20(OpenFAST).pdf

[YuYuJou]

Thank you for the suggestion. I tried changing the fairlead type to fixed as recommended. The simulation ran to completion, but I noticed that the 6-DOF displacements (platform motion) all became NaN after around 60 seconds.

Is this behavior expected in such a setup? Or does it suggest that there's still an issue with my MoorDyn configuration? I would appreciate any insights or further suggestions.

Thanks again for your help!

[RyanDavies19]

If your fairlead types are fixed, and you are running OpenFAST, then you are simulating a system with no mooring lines attached to the platform. In that case yes, the nan's would not be surprising.

Have you tried adjusting the tMaxIC values? Can you confirm if MoorDyn converges on an initial condition when running with your fairleads as vessel?

[YuYuJou]

When I set the fairleads to the vessel, the system did successfully converge. I also increased tMaxIC to 200, but the simulation still fails around 60 seconds. Do you think I should increase the line length further?

[RyanDavies19]

If the system converges before running, then I would suspect something in your OpenFAST set up is causing the numerical instability. I would suggest isolating different degrees of freedom in the system to see which cause the instability. You can also increase the number of segments in your lines to increase the model stability. Lastly reducing your dtM value will help increase stability.

I don't think you need to change your mooring design to get the simulation to run.

[YuYuJou]

I tried enabling only one degree of freedom at a time, and all six DOFs were able to complete the simulation successfully.
However, when all six degrees of freedom are enabled simultaneously — with all other conditions kept the same — the simulation detects a NaN halfway through the run.

[RyanDavies19]

@YuYuJou, I would suggest enabling VTK file generation when running your OpenFAST simulations so you can visualize what is happening with a tool like Paraview. That should hopefully help your troubleshoot where things are going wrong.

[YuYuJou]

I tried turning off the pitch degree of freedom while keeping the other five degrees of freedom turned on, and the simulation was able to run successfully.
What should I adjust to make the simulation stable with all six degrees of freedom turned on?

[RyanDavies19]

I don't know enough about your specific system to say how to fix your issue. Increasing the number of segments and decreasing the dtm value in MoorDyn will always help make the simulation more stable, at the expense of computational efficiency. With a low enough dtm (something like 1e-8) you should in theory be able to have the simulation be stable. Have you visualized your system in Paraview? That can provide insights into if things look off from what you expect. Have you tried using different MoorDyn output channels to see if the mooring system is behaving as expected (i.e. are the fairlead tensions in the expected range, fairlead motion and location as expected)? Have you tried disabling wave forcing and just look at loads on the platform and rotor from current?

All of these should hopefully help you track down where things are going wrong. It's also possible that your mooring system (single point bridle) could not be a very stable system for the loads you're subjecting the system to, which is causing large motions in the device and subsequently the mooring system.