FhSim  3.1.0
Marine systems simulation
Loading...
Searching...
No Matches
0460 — Net/Sphere drag coefficient is a constant, not a function of Re
ID 0460
Class KNOWN-LIMITATION
Severity 2
Status blocked

Models: Net/Sphere

Found: 2026-11-18, validation case A1 (fhsim_validation, cases/hydrodynamics/body_added_mass/A1_sphere_settling_speed)

Decision needed: the owner chooses whether the sphere gets a Reynolds-number dependent drag coefficient (for example a Morrison 2013 type correlation), or keeps the constant and documents the Re range in which it is valid.

Evidence

DragCoefficient defaults to 0.5, the default of environment::sphere_hydro::Sphere (fhsim_environment/include/fhsim_environment/hydrodynamics/SphereHydro.h:47: double cd = 0.5;). src/net/Sphere.h:69: "Drag coefficient of the sphere, not negative. (Default: 0.5 ...)". The law has no Re input.

Case A1 (tracked, tolerance 5 %) compares the terminal settling speed with the Morrison 2013 C_D(Re) correlation. Morrison C_D is 0.39 to 1.04 over the eight rows. Reference run (fhsim_validation/results/reference/validation_results_tracked.json, variant me_sphere): 1 of 8 rows pass. Settling speed error by Re:

Re 100 300 1e3 3e3 1e4 3e4 1e5 2e5
error +0.745 +0.394 +0.073 -0.035 -0.084 -0.094 -0.062 -0.081

Effect

Below Re 1e3 the sphere settles much too fast (constant C_D too low). Above 3e3 it settles 3 to 9 % too slowly. A float or weight in a net at low Re gets too little drag.

Possible fix

  1. Add a Re-dependent C_D (Morrison type) to the sphere law, with the constant as an option.
  2. Keep the constant and state its valid Re range in the model documentation and in case A1.

Test that would prove it

Case A1, variant me_sphere: all eight rows pass at 5 %.

Risk

Option 1 moves the results of nets that use Net/Sphere floats or weights, so regression baselines move and need an owner ruling.