Suggestion to modify the definition of structural damping #192

Currently, the definition of the structural damping is borrowed from BeamDyn. This implies that:

• It does not map 1:1 with any other software.
• Any change to the stiffness matrix forces to retune the damping coefficients.

Let's review how some wind turbine simulators model the structural damping.

• ElastoDyn: damping ratio for each modal coordinate (typically, 2 flap 1 edge).
• BeamDyn:
  • Modal damping.
  • Stiffness-Proportional Damping Coefficients. 6 parameters.
• QBlade:
  • Isotropic Rayleigh Damping. 1 parameter, optionally varying over the span.
  • Anisotropic Rayleigh Damping. 4 for the stiffness + 1 optional for the mass.
• Cp-Lambda and DYMORE from 2010 (should be version 4): same as QBlade Isotropic Rayleigh Damping.
• HAWC2:
  • damping_posdef - Rayleigh stiffness-proportional damping applicable to the classical Timoshenko beam element. 3 parameters (because the first 3 are related to the mass and should be set to 0). Theory described in M.H. Hansen - Anisotropic damping of Timoshenko beam elements.
  • damping_aniso_v2 - Rayleigh stiffness-proportional damping applicable to the FPM beam element. 3 parameters (because the first 3 are related to the mass and should be set to 0).

In my opinion, the definition of the structural damping in WindIO should be changed to the structural damping ratios for the clamped body in the undeflected configuration. This way, each software can tune its coefficient to match the target (measured) damping ratio. A couple of obvious questions are:

• How many modes?
• In which order?

We cannot expect that each software will model the same modes, nor that the eigen-solver will provide them in the same order. Therefore, a possible input could be:

- mode_name:     [ "flap_1",  "edge_1", ...,  "mode_n"]
- damping_ratio: [damping_0, damping_1, ..., damping_n]

This way, a software can model the first m low-frequency modes (m <= n), and the analyst can verify that each damping is applied to the expected mode. The first modes are always: flap 1, edge 1, flap 2, edge 2, while for higher-order modes a generic "mode_n" can be used.

Best regards,
Riccardo Riva