Hello,
thank you for sharing your knowledge via this forum and supporting FAST users.
After several years with Flex5 i recently started using FAST. Since i need linear wind turbine models for control design (model predictive control)
i appreciate the FAST ability of linearization at certain operation points. The migration of a 2.1MW variable speed turbine to FAST is done and the linearization works fine.
Now i’m at the point to use the MBC3 transformation in Matlab. The version MBC3 1.00 creates a MBC averaged system matrix but i miss the B, C, D matricies avaraged.
The document in the following link hypothesizes there is a revision 1.01 with advanced “control-specific upgrades”.
nrel.gov/docs/fy10osti/44327.pdf
Is this revision available?
Thank you in advance!
Ingo Ibendorf
Dear Ingo,
I’m glad to hear that you’re making good use of FAST.
I recall that Gunjit Bir had made some changes to MBC3 before his departure from NREL, but I don’t think MBC3 v1.01 was ever completed. That said, I’m not aware of any problems with MBC v1.00. I suggest that you use MBC v1.00. You could quite easily change the code to compute the azimuth-averaged MBC3-transformed matrices if you need them.
Best regards,
Dear Jason,
thank you for the quick reply.
I’ve changed the Matlab file accordingly.
Another thing i was surprised about is that the feedthrough matricies D and Dd contain nonzero elements.
That means the pitch input (collective or individual) influences the blade and tower movements directly.
Same for the disturbance input: wind speed. Well i add a pitch actuator transfer function at the pitch input which does
not transfer steps and the wind speed won’t change stepwisely but i didn’t expect that.
Is this on purpose or did i miss any switch to enable an additional aerodynamic time behavior?
Kind regards!
Ingo
Dear Ingo,
In FAST, any output influenced by inputs or disturbances that is acceleration-dependent (e.g., accelerations or reaction loads) has a nonzero D or Dd term. For example, outputs such as OoPDefl1 will have a zero-valued rows in the D and Dd matrices (because OoPDefl1 only depends on the state, x), but outputs such as TipALxb1 or RootMyc1 will have nonzero-valued rows in the D and Dd matrices (because they are acceleration-dependent).
Best regards,