
ID	0370
Class	BUG
Severity	1
Status	ready

Models: TrawlCable, MooringCable, Net/NetStructure

Found: 2026-10-12, feature wake-casting-trawl (measuring the effect of CastWake on the doors and clump weights), on eb26b53.

Evidence

examples/input/trawl_mooring_interaction/trawl_mooring_interaction.xml loads the maritime library (Ship/FrameTransform) and base math objects, neither of which is a dependency of this repository, so it cannot run from this repository's playpen. With those objects replaced by their constant outputs (mooringCable.PosA = "-984.5678,-921.3878,12", PosB = "812.5678,-1078.6122,1184", from the quaternion of the −5° mooring angle; the rest unchanged), the run (FhSim CLI, Release, Linux) logs within 0.03 s:

Net: Warning: a panel's relative water velocity was not finite, so its load was set to zero. The state is diverging;
No analytical Jacobian sparsity for this model: the whole 68740x68740 system is assembled as dense (StableSolver_i; the process grew past 8 GB).

Without the net's StateFile (CNetSTructureState_save.txt), with 200 warp elements and Euler_i at a fixed 1 ms step, the run is finite to 3 s and not finite from 3.5 s; the first non-finite outputs are CableS.BottomBridleEndForce and CableS.ClumpWeightPosition. At a 2 ms step it diverges earlier. The initial layout puts the beam at y = −1505 m (beam.TravelDistance = 1500) and the net's wing nodes near the origin, 1300 m from the beam with 2750 m of warp, so the rig starts far from equilibrium.

Effect

The only example of TrawlCable with MooringCable and a net does not run in 3.0.x; the CastWake variant trawl_mooring_interaction_castwake.xml (added by wake-casting-trawl) inherits this. Users have no runnable full-scale trawl input. Net cages are not affected.

Possible fix

Regenerate the net state file for the current Net/NetStructure, start the rig nearer its equilibrium (beam distance, net offset), pick an integrator and step that hold it (Euler_i at 1 ms or less, as the TrawlCable fixtures), and drop or document the maritime/base objects (constant mooring end points, as above).

Test that would prove it

A smoke run of the example (or a cut-down copy without maritime) for 30 s whose outputs stay finite, as SimObject.TrawlCable_Smoke does for the cut-down fixture.

Risk

Low: an input file only. The run takes long at full resolution (2000 warp elements per side).