Dear @Jason.Jonkman
Following your reply, I can agree almost everything but the relationship between GenDOF and stiffness. I think I have a big misunderstanding about them. Why there is no stiffness when we set GenDOF=True? I think GenDOF=True only represents the generator can accelerate or decelerate.
Whether there is stiffness and torsional-damping in the drivetrain is decided by the DrTrDOF.
What is more, gamma and alpha are parameters I set, which can be explained as following:
Using this formula:
I get f_n=2.22Hz. (When substituing for K, J_w, J_e, those values are already converted to the high-speed shaft.) 2.22Hz is close to the 2.088Hz calculated by eigenvalue analysis.
From other reference, I saw the equation for calculting the frequency is different from yours, which is f_n = SQRT( (K/J_w + K/J_e)/2 )/(2*pi)=1.57Hz. Which equation is wrong?
Dear @Yinghan.Liu,
Thanks for clarifying. For the purposes of the natural frequency calculation, presumably you assume alpha = gamma = 0.
I don’t agree with the extra factor of 2 in the equation for the free-free mode of the drivetrain natural frequency. Can you clarify what values you are using for K, J_w, and J_e?
The drivetrain stiffness only plays a role when the azimuth angles of the generator- and rotor-side of the drivetrain differ (that is, when theta_w /= theta_e using your notation). When the drivetrain is rigid, there is no stiffness and the only mode left is the rigid-body mode. When both the generator and drivetrain DOF are enabled, both the rigid-body mode and the free-free mode of the drivetrain are accounted for in the model.
Best regards,
Dear @Jason.Jonkman
Thanks for your quick reply!
First, the calculation is based on NREL 5MW wind turbine. All the values are already converted to the high-speed shaft. For 5MW wind turbine, the parameters for K, J_w, and J_e are listed as follows:
The parameters need to be converted to the high-speed shaft, so we devide 97^2.
Second, for the equation of the natural frequency, how do you derive that? Since I not familiar with the structural mechanics, could you please give me some guidance?
Dear @Yinghan.Liu,
I agree with your values, except that I would normally expect you’d express all terms on the low-speed shaft side. So, instead of dividing K and J_w by G^2, I would multiple J_e by G^2. Regardless, the resulting natural frequency should be equivalent. Also, I’m not sure how you are deriving gamma, but again this shouldn’t affect the natural frequency. (Does your approach give the correct frequency if you set gamma = 0?)
You can find the derivation of the natural frequencies of a two-mass, one-spring system in any vibrations textbook. I also found a derivation from a simple Google search: homework and exercises - Two mass one-spring system natural frequency - Physics Stack Exchange.
Best regards,
Dear @Jason.Jonkman
Thanks for your guidance again!
First, the derivation of gamma is as follows:
I simply set the desired working point as the steady state. Why would you like to set gamma=0?
Second, when I set gamma=0, the natural frequency is 2.22Hz, still far away from 1.69Hz (the natural frequency). Why does that happen? I still cannot figure that out…
Third, thanks for the reference!
Dear @Yinghan.Liu,
Is the 1.69 Hz drivetrain natural frequency that you obtain with OpenFAST derived with a two mass-mass model (only generator and drivetrain DOFs enabled), or do you have more DOFs enabled, such as blade flexibility?
Best regards,
Dear @Jason.Jonkman
The 1.69Hz natural frequency is derived under the condition where almost all DOFs enabled.
Dear @Jason.Jonkman
I also try to derive the natural frequency when only generator and drivetrain DOFs enabled, but I failed, and the error hint is shown as follows:
Using the same code, the readlinear program can run successfully when I enable almost all the DOFs. That seems weird… Is that because I close all the DOFs that related to blades?
Dear @Jason.Jonkman
Another problem occured when I analyze the frequency response by FFT (fast fourier transform) in Matlab. By the equations shown as follows, for a three-mass model using the parameters of NREL 5MW wind turbine, the peak frequency (the frequency when the response magnititude reaches local maximum) should be 0.1029Hz and 10.2138Hz.
The three mass model is constructed as following:
For NREL 5MW wind turbine, I set the parameters to be (corresponding to the high-speed shaft):
The seperating point betweeen the rigid part of blade and the flexible part is derived by analyzing the stiffness of blades. Here I choose the seperating point to be the 7% point.
However, the simulation shows that the peak frequencies are around 0.7Hz and 1.9Hz, for the drive train torque, which is totally different from the values I get from the equations ( 0.1029Hz and 10.2138Hz. )
The DOFs set for the simulation is shown below:
I think for the three-mass model, the EdgeDOF and the DrTrDOF, the DrTrDOF should be enabled.
Could you help me figure out why the peak frequencies differ??? That has troubled me for a long time.
Thanks in advance!
Dear @Yinghan.Liu,
Enabling rotor flexibility will lower the frequency of the drivetrain torsion mode because the rotor will no longer act as a rigid body. But enabling rotor flexibility would not be equivalent to a three-mass model.
I’m not familiar with the error regarding new_seq_inp, but MBC3 does not apply unless you have states, inputs, or outputs associated with 3 blades in the rotating frame of reference.
Best regards,
Dear Jason,
I am still confused about the different frequency I get between the open-loop system and the natural frequency condition, which is described in previous posts. In order to make the open-loop frequency equal to the natural frequency, which DOFs should I enable? Or what conditions should I set for the open-loop system?
---- Replied Message ----
From | Jason Jonkman via NREL Forumnotifications@nrel.discoursemail.com |
| Jason.Jonkman
March 19 |
Dear @Yinghan.Liu,
Enabling rotor flexibility will lower the frequency of the drivetrain torsion mode because the rotor will no longer act as a rigid body. But enabling rotor flexibility would not be equivalent to a three-mass model.
I’m not familiar with the error regarding new_seq_inp, but MBC3 does not apply unless you have states, inputs, or outputs associated with 3 blades in the rotating frame of reference.
Best regards,
Moreover, in order to simulate the three-mass model in OpenFAST, which conditions should I set in OpenFAST ? For example, what DOFs should I enable? I want to verify my three mass model in OpenFAST simulation, verifying the peak frequency I derived in three mass model.
Thanks in advance!
---- Replied Message ----
From | Jason Jonkman via NREL Forumnotifications@nrel.discoursemail.com |
| Jason.Jonkman
March 19 |
Dear @Yinghan.Liu,
Enabling rotor flexibility will lower the frequency of the drivetrain torsion mode because the rotor will no longer act as a rigid body. But enabling rotor flexibility would not be equivalent to a three-mass model.
I’m not familiar with the error regarding new_seq_inp, but MBC3 does not apply unless you have states, inputs, or outputs associated with 3 blades in the rotating frame of reference.
Best regards,
Dear @Yinghan.Liu,
I’m not sure what you mean by “open loop” in this context, but to match the 2.22 Hz frequency of the two-mass model, you must disable all DOFs except the generator and drivetrain DOFs.
OpenFAST does not support the equivalent of a three-mass model without changes to the source code.
Best regards,
Dear @Jason.Jonkman
In OpenFAST, I notice that we don’t take the gearbox as a separate mass block to study. Why? I know that the rotational inertia and flexibility of the gearbox have been converted to the shaft, but if we want the model to be more specific, I think we should set a DOF named gearbox instead. Why don’t we do that? Looking forward to your reply! Thanks in advance!
Dear @Yinghan.Liu,
Sounds possible, but NREL is not currently funded to work on improvements to ElastoDyn’s drivetrain model.
Best regards,
