Sorry to bother again. I want to design a MPC controller in the FAST to reduce the fatigue load, but I met a problem. In some papers, the drivetrain can be modeled as:
the Ts is the torque on the intermediate shaft, and the volatility of Ts is considered to characterize the fatigue of the transmission system.
the tower can be modeled as:
the researchers use the Mt to describe the load of the tower.
And now I want to reduce the yaw fatigue load through the MPC controller, but I can’t find the control-oriented yaw model which can be used in the MPC controller(the yaw model that contains the index of yaw load and yaw angle), so do you have some clues?
I’m not sure I fully understand your question, but it looks like your drivetrain model has 3 states: variable speed, omega_r, drivetrain torsion, psi, and torsion rate, psidot. Through the linearization functionality of FAST / OpenFAST, you can generate a linear state-space model with these states. You can also generate a linear state-space model with other states and/or other outputs like yaw moment.
In my last post, my idea is to obtain a model which can describe the Tower-top / yaw bearing yaw moment(YawBrMzp) or Nonrotating tower-top / yaw bearing pitch moment(YawBrMyp). I want to use such a model to calculate the yaw moment directly to design a MPC controller. But unfortunately, I can’t find any related model, so it is grateful if you can show me your idea.
You can generate the linear model you describe by:
- Enabling ElastoDyn, AeroDyn, and ServoDyn modules;
- Enabling the generator and drivetrain DOFs in ElastoDyn (GenDOF = DrTrDOF = True; all other DOF = False);
- Including the tower-base fore-aft bending moment in the ElastoDyn OutList (TwrBsMyt); and
- After computing a steady-state solution and linearizing at a number of azimuth steps, azimuth-average the matrices and eliminate the generator azimuth state as a DOF (by applying the MBC3 scripts); this will result in a three-state model with the three states you describe and with the tower-base fore-aft moment as an output.
For yaw, do you want to include nacelle-yaw states in the model (if so, enable YawDOF in ElastoDyn, resulting in a five-state model) or just include the yaw moment as an output (if so, include TwrBsMzt in the ElastoDyn OutList)?
Thanks for your suggestion. But now I met another question. According to the “NRELOffshrBsline5MW_AeroDyn_blade.dat” file, the blade is divided into 19 nodes, but in the AD15 input file, we can only output 9 nodes in one time, if I want to output 19 nodes together, what should I do? if I cannot output all the blade nodes together, I want to decrease the NumBlNds to 9, which nodes I should delete in the “NRELOffshrBsline5MW_AeroDyn_blade.dat” ?
The case which I use is Test #25, and the blade properties is shown below:
I would not recommend changing NumBlNds for the NREL 5-MW baseline wind turbine model below 19, which was set to balance accuracy with computational expense.
Yes, you can output aerodynamic quantities from all nodes if you want. This is documented in the “Nodal Outputs” section of OpenFAST readthedocs here: openfast.readthedocs.io/en/main … input-file.
My goal is to conduct a yaw optimization considering both power and fatigue loads.
Therefore, in order to fully reflect the structural performance of yawed WTs, I need to formulate the combination of fatigue loads (by using DEL) at several components of a yawed wind turbine, including blade root, yaw bearing and tower base. However, I am not sure whether it is reasonable or not? And how can I achieve it?
Well, you can certainly use OpenFAST to calculate fatigue loads of the blade root, yaw bearing, and tower base. But I’m not experienced with the design of nacelle-yaw controllers to provide you detailed guidance on this multi-objective controls design problem.