Openfast Linearization for RNA moments of inertia

Hi all,

I am trying to find the RNA characteristics (center of gravity, inertia moments) of the UMaine Semi from the last updated files of the on the IEA-15-240-RWT github, by following the procedures described on this forum ([url]http://forums.nrel.gov/t/openfast-2nd-order-linearization/2249/2] ; [url]http://forums.nrel.gov/t/ansys-model-tower/1722/4] ) using openfast v 2.4.

I manage to run the linearization, but in the linearization file I get the D matrix for floater motions is only composed of zeros…

I set all masses and inertias of floater and tower to 0 (for the tower to 0.001 to avoid error messages, directly in distributed properties in tower.dat file) and kept the originals of the RNA (yaw bearing, rotor, nacelle), disabled all DOFs except platform’s, and set flag of all features except CompElast in the .fst files to 0, to get only the inverse of the mass matrix of the platform in D matrix.

Is there a step I did not do correctly?

Best regards

Dear Benoit,

Which part of the linearized D matrix are you looking at? Are you looking at the 6x6 submatrix of D with inputs (u) that are the 6 loads applied to the platform and outputs that are the 6 platform accelerations?

Best regards,

Dear Jason,

My D matrix was already a 6x6 matrix, which I obtained by disabling all calculations features except elastodyn. My ouptuts were the plaforms DOFs (so not the acceleration), but my inputs were not the 6 loads applied to the platform but inputs for blade pitch, generator torque, etc.

Thanks for your answer, I realized I did not follow all the steps in the instructions. I changed it by enabling the linearization input option LinIputs to 2 and my 6x6 subset in the D matrix was not zeros anymore, I could get the desired mass matrix.

The mass matrix I have is non symmetrical, which quadrant of the matrix (lower left or upper right) should I use to know the coordinates of the center of gravity with respect to the reference point?

Best regards

Dear Bonoit,

I would expect the 6x6 mass matrix to be symmetric. If it is not, perhaps the asymmetry is the result of numerical round off. Are you generating your linearization file with fine precision, e.g., OutFmt = “ES17.9E3”?

Best regards,

Dear Jason,

Indeed, I wasn’t. I did it again with better precision, it helped for better matching of the diagonal coefficients, however my matrix is still non symmetrical. For example, my M(1,5) and M(5,1) parameters, related to the z coordinates, are respectively 1.4e8 Nm and 2.5e6 Nm, which I understand by the fact that there are two distinct points. Do you have another idea on why I have a non symmetrical mass matrix?

Best regards

Dear Benoit,

That seems quite different for significant mass offset like z. Perhaps try even more precision like OutFmt = “ES20.12E3”? I can’t think of anything else that would cause this problem. Perhaps sharing your *.lin file would help.

Best regards,

Dear Jason,

I will try that.

You can find my .lin file attached (in txt format, .lin is an invalid file extension) and a simple python script (in txt format also) if you use python, to directly show you the mass matrix I get. The load inputs for platform acceleration are from 361 to 366 and the starting col number for the D subset is 6485 in the .lin file.

Best regards
mass_matrix_test.txt (1.48 KB)
IEA-15-240-RWT-UMaineSemi.2.lin.txt (318 KB)

Dear Benoit,

I took a brief look at your file and see the problem. The platform accelerations that you are outputting have units of m/s^2 and deg/s^2. The latter needs to converted to rad/s^2 to ensure that the D matrix has consistent units before calculating M = D^-1.

You can also calculate M = B^-1 using the (7:12,361:366) subset of the full B matrix, which already has consistent units.

Best regards,

Dear Jason,

Thanks for the answer, that was exactly it, problem solved.

Best regards