Quote from David on 8. August 2026, 18:12

Hello,

Yes, the parameters you listed cover the main parts that need to be adapted. In addition, make sure that all other dimensional properties are scaled consistently, in particular the tower and blade geometry and structural properties, as well as the relevant hub, nacelle and drivetrain inertias and stiffnesses.

For the aerodynamic model, I would also check the resulting blade Reynolds numbers. Depending on the scaling factor and operating conditions, the relevant Reynolds number range may change significantly, so the blade polars may need to be revised as well.

Using geometric similarity is a reasonable approach for a comparative study, but it does not automatically ensure dynamic similarity. Quantities such as Reynolds number, natural frequencies, rotor speed and controller time scales do not all scale in the same way.

I would therefore first finalize the aerodynamic and structural scaling consistently and then retune the controller for the scaled turbine.

Best regards,

David

Quote from Kento Eiraku on 10. September 2026, 00:02

Hello David,

Thank you again for your previous advice regarding my geometrically scaled NREL 5 MW turbine model, in which the relevant linear dimensions, including the rotor radius, were increased by a factor of two. Hereafter, I refer to this model as the “2× scaled turbine.”

Based on your suggestions, I first worked on scaling the aerodynamic and structural models.

For the structural model, I modified the dimensions, masses, inertias, and stiffnesses of the blades, tower, hub, nacelle, and drivetrain according to geometric scaling laws. I also checked the natural frequencies of the scaled model.

For the aerodynamic model, I checked the range of Reynolds numbers under the expected operating conditions, since doubling the turbine dimensions changes the Reynolds number at each blade section. I then created a MultiPolar using polar data corresponding to multiple Reynolds numbers.

At this stage, I consider the aerodynamic and structural scaling to be provisionally complete, and I am now adjusting the TUBCon control parameters for the 2× scaled turbine.

To maintain the same tip-speed ratio as the original NREL 5 MW turbine at the same wind speed, I have modified the following parameters:

• Rated_Power: multiplied by 4, from 5 MW to 20 MW
• Omega_Rated: multiplied by 0.5
• TC_Q_Max: multiplied by 8
• TC_K_opt: multiplied by 32
• TC_Omega_Min: multiplied by 0.5
• TC_Omega_Min1 and TC_Omega_Min2: multiplied by 0.5
• TC_Omega_Max1 and TC_Omega_Max2: multiplied by 0.5

However, the parameters in Pitch_Control, Machine_Control, Supervisory_Control, and Event_Parameters are still mostly unchanged from the original NREL 5 MW settings.

As an initial validation, I ran simulations with uniform, steady wind speeds of 8, 11.4, 14, and 20 m/s. At all four wind speeds, the responses became stable after the initial transients had decayed.

At and above the rated wind speed, the rotor speed of the 2× scaled turbine converged to approximately 6.05 rpm, and its power output converged to approximately 20 MW. I also compared the steady-state pitch angles of the original and 2× scaled turbines and obtained nearly identical values:

• At 14 m/s: approximately 8.4°
• At 20 m/s: approximately 16.5°

The original NREL 5 MW turbine converged to approximately 12.1 rpm and 5 MW, while the 2× scaled turbine converged to approximately 6.05 rpm and 20 MW. Since the two turbines also had nearly identical tip-speed ratios and steady-state pitch angles, I believe that the current settings produce steady-state performance consistent with the intended scaling relationships.

I would be grateful for your advice on the following three questions:

1. In the current uniform-wind simulations, the original and 2× scaled turbines have nearly identical tip-speed ratios and steady-state pitch angles, and both converge stably to their respective target rotor speeds and rated powers. I also plan to compare their performance under wind-shear conditions using the values obtained after the initial transients have decayed. For this type of study, is it still necessary to modify the Pitch_Control settings for the 2× scaled turbine?
2. If the Pitch_Control settings should be modified, which parameters should be adjusted—for example, PC_Kp, PC_Ki, gain scheduling, or the rotor-speed filters—and what procedure or approach would you recommend for tuning them?
3. The main purpose of this study is to compare turbine performance during normal power-production operation. With this objective in mind, are there any important parameters in Machine_Control, Supervisory_Control, or Event_Parameters that should be modified for the 2× scaled turbine? If so, which parameters should be changed, and how should they be adjusted?

Thank you very much for your time. I would greatly appreciate any advice on which control parameters should be modified and how they should be tuned for the purpose of this study.

Best regards,

Kento Eiraku

Quote from David on 11. September 2026, 13:53

Hello Kento,

Your scaling approach generally looks reasonable. For the pitch controller, as a first approximation for the 2x turbine, the rotor-speed filter and notch frequencies could be reduced by about a factor of two. For the rotor-speed PID terms, dynamic similarity would approximately give Kp × 2, Ki unchanged and, if used, Kd × 4. Pitch-rate limits would similarly scale with approximately 0.5, while pitch-angle limits can remain unchanged.

The gain scheduling does not necessarily need to be scaled directly, but I would check it against the aerodynamic pitch sensitivity of the new rotor, particularly since your Reynolds numbers and polars have changed.

For normal power-production simulations, I would mainly adjust the relevant speed, torque and power thresholds according to the scaling. The event parameters are less important unless you are specifically investigating fault or shutdown cases.

That said, development of the TUB Controller has ceased, and the colleague who originally developed and maintained it has left our team. For future work I would therefore recommend using the open-source ROSCO controller:

https://github.com/NatLabRockies/ROSCO

ROSCO also provides tools for automatically tuning the controller for a given turbine, which would be particularly useful here:

https://github.com/NatLabRockies/ROSCO/tree/main/Examples/Tune_Cases

Best regards,

David