TurbSim SMOOTH spectral model — Richardson number as primary stability input and role of TI

Wind & Water Inflow Turbulence
1

Hi all,

I am currently setting up TurbSim for a series of aeroelastic simulations using OpenFAST for a research project on offshore wind turbine power prediction under varying atmospheric conditions. I am modelling a 15 MW offshore turbine (IEA 15 MW reference turbine) and want to explicitly represent atmospheric stability as a simulation input rather than relying on TI and wind shear alone.

Based on my reading of the TurbSim documentation and some forum threads, I believe the SMOOTH spectral model is the appropriate choice for this, as it accepts the gradient Richardson number (RICH_NO) as a direct input unlike the IEC Kaimal model. I have a few questions I’d like to confirm before proceeding:

1. Richardson number as primary stability input

My plan is to sample RICH_NO across a range from approximately −2.038 (very unstable, following Mohan 1998) to 0.1667 (which appears to be the TurbSim upper limit for stable conditions based on previous forum posts). Is this range correct for the SMOOTH model? Are there known numerical stability issues at the extremes of this range I should be aware of?

2. Role of TI when using SMOOTH

My understanding is that when using the SMOOTH spectral model, IECturbc is ignored and turbulence intensity is instead determined internally by TurbSim from the input RICH_NO and URef, with Ustar left as default. If this is correct, TI would effectively be an output of the simulation rather than an independent input — meaning I should extract TI from the simulated wind field after the fact rather than specifying it directly. Is this understanding correct?

3. Wind shear with SMOOTH

I plan to keep PLExp as an explicit input controlling the mean wind speed profile, with WindProfileType set to PL. Is there any interaction between PLExp and RICH_NO in the SMOOTH model that I should be aware of — for example, does the model internally derive a wind profile from Ri that would conflict with the specified PLExp?

4. Design of Experiments

My current DoE is 23 wind speeds × 50 LHS combinations of RICH_NO and PLExp per wind speed = 1150 simulations total. Does this seem reasonable for covering the offshore atmospheric parameter space adequately, or would you recommend a different sampling strategy?

Any guidance or pointers to relevant documentation would be greatly appreciated.

Thanks, Naomi

2

Dear @Naomi.Adeniji,

I’m no expert on the SMOOTH spectral model of TurbSim. I would suggest reviewing the TurbSim documentation for details: https://openfast.readthedocs.io/en/main/_downloads/cb14d3e2d3533d76e405d730fea19846/TurbSim_v2.00.pdf. Here are partial answers to your questions based on my understanding:

  1. I’m not sure.

  2. Yes, although my understanding is UStar in the SMOOTH model can be used to adjust the turbulence level in place of IECturbc.

  3. I’m not aware of any interaction, but I would generally expect higher shear for stable atmospheric boundary layers (ABL) and lower shear for unstable ABL.

  4. Yes, that sounds like a quite reasonable sampling of the parameter space.

Best regards,