I extracted Bmodes input file from an assumed tower geometry. By running Bmodes the first natural frequency of the tower resulted 0.365. Then, I built the tower FAST input file to run linearization mode of FAST and MBC. First, I disabled all the DOFs except the tower 4 DOFs to examine the eigenfrequency simulation of FAST. The simulation has been done in stationary mode as
But, the result shows 0.32 (hz) for the first two natural frequency of the tower, while I expected something around 0.365. Even by enabling the other DOFs in stationary or power production mode, the natural frequencies of the tower doesn’t change significantly.
tower.txt (21.8 KB)
CampbellDiagram.xls (216 KB)
For the same input data, BModes should give quite similar results as FAST when computing tower natural frequencies. For a fixed-base tower (not flexible foundation or floating offshore) Here are few things to keep in mind:
- You should ensure that the tower-top mass, center of mass, and inertias match between BModes and FAST, which is not trivial. To do this, what we’ve done before is linearize a FAST model with zero platform and tower mass and only the platform DOFs enabled to derive the 6x6 rigid-body mass matrix of the rotor-nacelle assembly, from which the tower-top mass, center of mass, and inertias can be derived for input to BModes.
- Ensure that the same distributed tower properties are specified in both BModes and FAST. Here’s how you would set up the BModes distributed data file to match FAST:
sec_loc = HtFract
str_tw = 0
tw_iner = 0
mass_den = TMassDen
flp_iner = Very small number (you can’t specify exactly zero for this input in BModes)
edge_iner = Very small number "
flp_stff = TwFAStif
edge_stff = TwSSStif
tor_stff = Very large number (FAST’s tower model neglects torsion)
axial_stff = Very large number (FAST’s tower model neglects stretching)
cg_offst = 0
sc_offst = 0
tc_offst = 0
- Derive the tower mode shapes for FAST by selecting the appropriate mode shapes output by BModes and fitting them with the polynomial needed by FAST. The ModeShapePolyFitting.xls spreadsheet provided in the FAST archive is useful for this.
- Linearize the FAST model for use in MBC3 with only the tower DOFs enabled.
I hope that helps.
Thank you so much for your helpful respond
I am trying your first method of using FAST linearization in order to calculate BModes input parameters.
I set the CalcStdy to False, all the DOFs except the platforms to False and change the TMassDen into a very small value (not zero).
The “fast_inp.lin” file then give me a mass matrix which has dimension of 6*6 containing, the total mass of Nacelle and rotor (M), inertia (I) and center of mass (x,y,z) as following:
M 0 0 0 Mz My
0 M 0 Mz 0 Mx
0 0 M My Mx 0
0 Mz My Ixx Ixy Ixz
Mz 0 Mx Iyx Iyy Iyz
My Mx 0 Izx Izy Izz
Where, moment of inertia of the top mass has been calculated at the base of the tower. Then, we should transform the moment of inertia to the center of mass and the top section of tower for applying in the BModes input file.
I just recheck the process of finding the BModes parameters.
Is that the right? doing the linearization with CalcStdy of False is true? because the enabled CalcStdy does not converge.
Yes, your approach sounds correct. However, your generalized 6x6 mass matrix of the rotor-nacelle assembly is missing some minus signs on the off-diagonal terms. The proper form is:
M 0 0 0 Mz -My
0 M 0 -Mz 0 Mx
0 0 M My -Mx 0
0 -Mz My Ixx Ixy Ixz
Mz 0 -Mx Iyx Iyy Iyz
-My Mx 0 Izx Izy Izz
Using the mass matrix derived from the undisplaced configuration is sufficient, so, I agree that setting CalcStdy to False is appropriate for this calculation.
I was following your suggestion for linearizing FAST to get the inertias at tower top that are input in BModes.
I can’t find a 6x6 matrix in the result.
Which DOF’s should I turn on?
Where can I find the mass matrix?
I’m using FAST 8.16
A similar question has been asked and answered in the following forum topic: http://forums.nrel.gov/t/how-to-find-the-paramter-of-the-first-bending-modal-mass-damping-and-stiffnees-coefficients/1594/2. Unfortunately, it is not possible in FAST v8 to expert the full-system mass matrix.