Tower Eigenfrequencies of NREL 5MW Turbine

Hi to everyone,

From modes I calculated the tower eigenfrequencies of the NREL 5MW-Turbine, which resulted in 0.40 Hz (First Tower EF) and 3.64 Hz (Second Tower EF). For all turbines I simulated up to now, there was a large peak at the first tower eigenfrequency in the tower-related channels of FAST, like YawBrTDyt or YawBrFyt.
For the NREL 5MW Turbine this is not the case, there is a peak at 0.30 Hz, but none at 0.40 HZ. So there must be a reason that the tower eigenfrequency shifted from 0.4 to 0.3 Hz. First I thought, this might come from a coupled platform / tower eigenfrequency, but all platform degrees of freedom are deactivated in the simulation (PtfmSgDOF = false, etc).

Does anybody have an idea, what causes the frequency shift?

Best regards,

1 Like

Dear Felix,

Since the FAST User’s Guide was last published, we now recommend that BModes be used in place of Modes. BModes employs a higher-fidelity model than Modes and also includes the option for tower-top (nacelle and rotor) inertia (in addition to a point mass) and tower-base (platform) DOFs. As described in the forum topic found here: Tower fore-aft modes shapes, the tower-top inertia can have a large impact on the mode shapes and natural frequencies of a tower. While Modes may be predicting a higher frequency, the correct frequency of the 1st fore-aft and side-to-side tower-bending modes when the tower-top mass/inertia are included is around 0.32 Hz for the land-based NREL 5-MW turbine. We’ve obtained very similar frequencies with BModes, FAST, and MSC.ADAMS.

I hope that helps.

Best regards,

1 Like

Dear Jason,

thank you very much for your support. I was not aware, that BModes should be used instead of Modes. So I tried to create a BModes input file from the existing parameters but got stuck somehow. I hope you can help me, how to get from FAST-parameters to BModes-parameters, without using ADAMS as suggested in the topic you linked:

BModes FAST Tower FAST Blade

radius TowerHt TipRad-HubRad
hub_rad TwrRBHt 0
precone 0 PreCone
tip_mass NacMass+YawBrMass+HubMass+3BldMass TipMass
cm_loc (NacCMxn
NacMass+(OverHang+HubCM)*cos(ShftTilt)HubMass+…) / “tip_mass” 0
cm_axial (NacCMzn
NacMass+(OverHang+HubCM)*sin(ShftTilt)HubMass+…) / “tip_mass” 0
ixx_tip (cm_loc^2+cm_axial^2)
“tip_mass” 0
iyy_tip 0 0
izz_tip NacYIner 0
ixy_tip 0 0
izx_tip 0 0
iyz_tip 0 0

sec_loc BlFract
str_tw StrcTwst
tw_iner StrcTwst ???
mass_den BMassDen
flp_iner FlpIner
edge_iner EdgIner
flp_stff FlpStff
edge_stff EdgStff
tor_stff GJStff
axial_stff EAStff
cg_offst EdgcgOf ???
sc_offst 0 ???
tc_offst 0 ???
With the parameters listed above, I get 0.393Hz for the NREL 5-MW turbine, instead of 0.32 Hz as you indicated. So what’s wrong here?

Dear Felix,

Your equations for the BModes input for the tower look correct, except for the inertia terms. The inertia terms should include the inertia of the nacelle and rotor. The rotor-inertia terms are not trivial to calculate by hand, but are very important for computing accurate tower mode shapes and natural frequencies. I’ve used the Aggregate mass tool in MSC.ADAMS to calculate the rotor/nacelle assembly mass, center of mass, and inertia of the NREL 5-MW turbine and the values are listed below. I suspect if you use these inputs you’ll obtain the proper natural frequency for the first tower-bending modes (about 0.32 Hz).

3.5E+005 tip_mass
-0.413775 cm_loc
1.96699 cm_axial
4.370E7 ixx_tip
2.353E7 iyy_tip
2.542E7 izz_tip
0.0 ixy_tip
1.169E6 izx_tip
0.0 iyz_tip

For the blade, your equations are correct except that radius should be set to TipRad and hub_rad should be set to HubRad.

It should be clear that BModes uses a different set of distributed beam properties than FAST. (E.g., the current version of FAST only considers flap and edge bending of straight pretwisted blades with no offsets of mass or stiffness, negligible sectional inertia, and no extensional or torsion stiffness.) If you are using BModes to generate mode shapes for FAST, you should specify distributed beam properties that mimic the beam model in FAST. Here are the appropriate settings to use for the blade:

sec_loc = BlFract
str_tw = StrcTwst
tw_iner = StrcTwst
mass_den = BMassDen
flp_iner = Very small number (you can’t specify exactly zero for this input in BModes)
edge_iner = Very small number "
flp_stff = FlpStff
edge_stff = EdgStff
tor_stff = Very large number
axial_stff = Very large number
cg_offst = 0
sc_offst = 0
tc_offst = 0

I hope that helps.

Best regards,

Dear Jason,

I just played around with the numbers you told me, but I was not successful to get the 0.32 Hz. I was able to move the frequency between 0 and 0.4 Hz depending on how “large” 1e12 or “small” 1e-5 I chose the inertia / stiffness. Could you please paste the BModes input files to this topic, so that I can check all parameters? Thank you very much in advance. I’ll be back in a week to work further on this topic.

Best regards,

Dear Felix,

I don’t have BModes input files for the land-based tower of the NREL 5-MW turbine that I can send you, but you can find BModes model of the NREL 5-MW turbine supported on a fixed-bottom offshore monopile with a flexible foundation treated as a 6x6 stiffness matrix at the mudline (i.e., a coupled springs representation) here: See the input file “CS_Monopile.bmi” and associated distributed tower data file “CS_monopile_tower_secs.dat.”

Best regards,

I have been working on simulating the NREL 5MW turbine using the foundation models for IEA-ANNEX XXIII subtask 3. I have been attempting to recreate the natural frequencies which were reported in the document OC3-Soil-Pile-Interaction Models_ReadMe.pdf with the coupled springs but have been unable to match them using the inertia properties as listed above and in … nopile.bmi

The values I get from Bmodes are:
0.251Hz SS 1st
0.253Hz FA 1st
1.327Hz SS 2nd
1.531Hz FA 1st

Those reported in Dr Passon’s report are:
0.2476Hz SS 1st
0.2456Hz FA 1st
1.5459Hz SS 2nd
1.5327Hz FA 1st

Do you have any suggestions as to the particularly low 2nd side-side frequency I am getting?
Jason, I noted that you reported higher values for the second modes for this task- around 1.7Hz; were these generated using ADAMS?

Kind regards,

Dear Rebecca,

I believe there is quite a bit of coupling between the second tower-bending modes and the bending of the rotor blades for this model. This coupling leads to an increase of the second tower-bending mode natural frequencies when compared to the natural frequencies of a model (like BModes) with a rigid rotor/nacelle assembly. We at NREL have predicted the natural frequencies from both FAST and ADAMS and both codes predict second tower-bending natural frequenices on the order of 1.6 - 1.7 Hz for this model (with rotor blade flexibility).

Best regards,

Dear Jason,

I just wanted to look into the topic again and tried to calculate the eigenfrequencies of the 5M-Tower. Using the bmi/dat-files from your directory [url]National Wind Technology Center's Information Portal | Wind Research | NREL, I get the following results:

tower base boundary condition bmodes 1.03.01 bmodes 3.00.00 1: cantilevered 0.2423 Hz 0.384 Hz 3: only axial and torsion constraints 0.2737 Hz Eigensolution failed: problem not +ve semi-definite 3838
So the eigenfrequencies strongly depend on the version of bmodes!

Has there already been any investigation, why that is the case?

Hi Felix,

Run the bmi/dat files in the directory [url]National Wind Technology Center's Information Portal | Wind Research | NREL with the BModes executable in that folder. This is a later version of BModes that has not been posted on [url]National Wind Technology Center's Information Portal | Wind Research | NREL.

We are looking into the differences between the two versions and will put the latest one on the NREL Design Codes site. For now, please use the one in the jjonkman/BModes folder along with the bmi/dat files there.

Sorry about that.


I just have a few questions about Bmodes and how to operate it correctly.
Im switching from modes to Bmodes because it is now advised.
But also because im interested in modelling turbine soil interaction, i had been assuming that the foundation and turbine were decoupled and in that way generate the mode shapes using modes however in Bmodes I see there is an option to include the foundation stiffness in a CS model form.

I have been trying to use Bmodes to get the tower mode shapes for both a fixed base and flexible conditions.

Fixed base, for offshore wind turbine,
I used the ‘CS_monopile.bmi’ modal and Bmodes.exe from
and simply put the tension to zero
and the mooring stiffness to zero
and hub_conn=1 (for a cantilever support)
my first five modal frequencies were,

Mode 1= 0.273744E+00 Hz (s-s)
Mode 2= 0.276006E+00 Hz (f-a)
Mode 3= 0.132889E+01 Hz (s-s)
Mode 4= 0.158898E+01 Hz (s-s)
Mode 5= 0.186703E+01 Hz (f-a)

Modes 1 and 2 are the fore-aft and side-to-side 1st tower frequencies however modes 3 and 4 both seem to be side-to-side as they contain only s-s disp, does this seem correct?
I then placed these modes shape (1,2,3,5) into the excel sheet and then into FAST (using the offshore monopile FAST files) and did a linearization run to calculate the modal frequencies I got back (for the tower modal frequencies)
0.277 Hz
0.278 Hz
2.2835 Hz
2.7394 Hz

The second tower modal of 2.7 seem a bit high.

The flexible foundation, for offshore wind turbine,
I used the ‘CS_monopile.bmi’ modal
and simply put the tension to zero
and in the mooring stiffness matrix I put in my own coupled spring stiffness for a pile foundation

I assuming that,
mass_pform is just the mass of the platform
and Platform mass inertia 3X3 matrix just correspond to the terms PtfmPIner, PtfmPIner and PtfmYIner in the platform file in FAST, ( do I have the matrix in the correct order)
[PtfmPIner 0 0
0 PtfmPIner 0
0 0 PtfmYIner]

I not sure what this term does (its not in the user manual) I just have it set to zero

When calculating the mode shape using the provided excel sheet
Im not quite sure about the ‘Scaling factor of y’
I’ve been leaving the first one at the suggest value
the second scaling factor it states should be
‘a user specified ratio of the max deflection to beam length for a deflected beam’
for this value im simply taking the biggest absolute value from column y (the deflection) and dividing by 107.6 (87.6+20) is this correct?
and should the sign of the scaling factor of y always be positive.

I’ve have being taking polynomial coefficients from the Normalized Projection Method when the foundation is included.

I was also wondering does anyone have a link to this document,
Memorandum: Derivation and Description of the Soil-Pile-Interaction Models, OC3-Derivation and Description of the Soil-Pile-Interaction Models.pdf,

Thanks in advance for any advice.

Just an extension from the question i have asked above
basically using mode shapes given in, NRELOffshrBsline5MW_Tower_Monopile_RF.dat, and carrying out an linearization analysis i get for the tower modal frequencies
0.28017 Hz
0.282 Hz
2.2504 Hz
2.3928 Hz

which seem right.
In the file NRELOffshrBsline5MW_Tower_Monopile_RF.dat the mode shapes in the Fore-aft direction and Side-to-side direction are different hence they must have been generated with Bmodes not Modes

Im simply trying to recreate these results using Bmodes myself before moving on

but as outlined above using the ‘CS_monopile.bmi’ modal and Bmodes.exe from
I get the tower modal frequencies
0.277 Hz
0.278 Hz
2.2835 Hz
2.7394 Hz

After some testing it appears the problem is with the second side-to-side mode shape I get from Bmdoes it gives the high second frequencies.
The Bmodes results seem strange also as Mode 3 and 4 both seem to be side-to-side modes


Use modes 1,2,4,5 (ss1, fa1, ss2, fa2) in your FAST analysis. Mode 3 is the tower first torsion mode, which is coupled to the side-to-side motions.

You don’t need the platform mass for a monopile. Yet, your platform mass matrix is in correct order.

n_secs_k_distr specifies the number of distributed stiffness points.

Regarding the ‘Scaling factor of y’ in the Excel spreadsheet, what you did seems fine.

Kind regards,


Hi Khanh
First off thanks for the response.
I was thinking the third mode might be a torsional mode however as it only contained s-s disp. that confused me i would have assumed a torsional mode would have contained both s-s disp. and f-a disp.

When i use the modes 1,2,4,5 as
ss1, fa1, ss2, fa2
(as you suggest) i still run into the same problem. (get too high a second s-s modal frequencies),

from Bmodes i get,
1st s-s 0.273744E+00 Hz
1st f-a 0.276006E+00 Hz
2nd s-s 0.158898E+01 Hz
2nd s-s 0.186703E+01 Hz

and then the s-s disp. and f-a disp into Bmodes I get the 6h order polynomial as,
s-s 1st mode
a2 0.8397
a3 -1.2928
a4 4.3545
a5 -4.2795
a6 1.3781

s-s 2nd mode
a2 36.6325
a3 -46.1807
a4 102.636
a5 -144.2992
a6 52.2114

f-a 1st mode
a2 0.8463
a3 -1.2997
a4 4.3707
a5 -4.2976
a6 1.3803

f-a 2nd mode
a2 122.6056
a3 -157.1805
a4 329.911
a5 -497.6733
a6 203.3373

and put these into NRELOffshrBsline5MW_Tower_Monopile_RF.dat and using FAST 7 to linearize the system i get the tower modal frequencies as,
0.27893 Hz
0.2793 Hz
2.2869 Hz
2.6874 Hz

The 2nd s-s tower modal frequencies (2.6874 Hz) doesn’t compare well to original frequency given by NRELOffshrBsline5MW_Tower_Monopile_RF.dat (2,3 HZ)

Any suggestion what im doing wrong?


Dear Michael,

As you have noticed, the tower mode shapes that are needed as input to FAST depend on the boundary conditions of the tower. That is, the mode shapes will be impacted by treating the foundation as rigid or flexible (or floating) or treating the rotor as rigid or flexible. BModes allows for the flexibility of the foundation, but not of the rotor-nacelle assembly (RNA).

When you compute the tower-bending natural frequencies from a FAST linearization analysis, are you disabling the FAST degree’s of freedom (DOFs) in the RNA (e.g., blade-bending DOFs)? If so, you should obtain quite similar natural frequencies between FAST and BModes. If RNA DOFs are enabled in FAST, the coupling between these DOFs and the tower/platform DOFs in FAST may lead to differences between FAST and BModes.

When we originally developed the mode shapes for the NREL 5-MW models, BModes was not available. Instead, we used the linearization functionality of a full-system ADAMS model to obtain the tower mode shapes. We derived these mode shapes with the proper tower-base boundary conditions (dependent on the NREL 5-MW model), as well as with the RNA DOFs enabled. Thus, the mode shapes included in the models NREL’s supplied for the NREL 5-MW turbiners may be a bit different than the ones obtained from BModes. Despite this limitation of BModes relative to ADAMS, we still suggest that users use BModes in place of Modes.

I’ve attached the Memorandum that you requested in your August 22 post.

I hope that helps.

Best regards,
OC3-Derivation and Description of the Soil-Pile-Interaction Models.pdf (385 KB)

Hi Dr Jonkman
First off thanks very much for your response and the attached document.

You are right when I disable the RNA DOFs, (and only have a 4 DOF system two tower modes in the s-s and f-a direction) I get the natural frequencies
0.27585 Hz
0.27768 Hz
1.654 Hz
1.8623 Hz
Which correspond very well to the natural frequencies recorded from Bmodes (given in my last post)

When the RNA DOFs are enabled in FAST the coupling does indeed produces different natural frequencies (for the tower) than those give by BModes
(again given in my last post) which is reasonable,
there seem to be a lot of coupling in the second tower s-s mode,
driving it up to 2.7Hz or so,
originally I just thought this seemed a bit high and assumed i was doing something wrong especially when i compared it to the tower natural frequencies given by NRELOffshrBsline5MW_Tower_Monopile_RF.dat,

Thanks from again for your response i think im using Bmodes correctly now.


I am about to try to build a monopile on which to put a 6.2MW wind turbine. We will need to build a completely new model of a wind turbine soon for certification of a commercial project. I would like to make sure that I understand how the tower modes are calculated and how the mean sea level is defined. I am also testing a modified version of FAST, including tower wind loads. The external aerodynamic loads ought to be applied only to the part of the tower between the mean sea level and the yaw bearing (!!). The model I have tested my new code on so far is the NREL 5MW onshore machine.

I am re-reading this forum to try to glean what I need to know about how such a model may be build, representing a wind turbine on an offshore monopile foundation.

Please can you tell me where I can get a copy of the following paper, which describes the contruction of a model of the NREL 5MW wind turbine on an offshore monopile foundation?

Jonkman, J., Butterfield, S., Passon, P., Larsen, T., Camp, T., Nichols, J., Azcona, J., and Martinez, A., “Offshore Code Comparison Collaboration within IEA Wind Annex XXIII: Phase II Results Regarding Monopile Foundation Modeling,” 2007 European Offshore Wind Conference & Exhibition, 4–6 December 2007, Berlin, Germany [online proceedings], BT2.1, URL: 206_Eow2007fullpaper.pdf

reference made from the following document

Kind Regards,

Mark Spring (Lloyd’s Register - Renewables Team)

Dear Mark,

Here is a link to the NREL-published version of that paper:

Best regards,

Dear All,

There has been mention in this forum about incorporating the functionality of BModes into FAST itself. At the beginning of a simulation, FAST could calculate (or recalculate) the structural modes (tower and blades separately) and incorporate new mode shapes and frequencies in the subsequent simulation.

Has this been done?
Is it still planned (given the modularisation program under way)?
Should we still be setting the logical inputs (CalcBMode and CalcTMode) always to FALSE?

Many thanks for an update on this issue.

Kind Regards,


Dear Mark,

We are currently working at NREL to develop a new module for FAST with nonlinear beam finite elements (FE) for improved blade modeling. This module is based on the geometrically exact beam theory (GEBT) instead of BModes. We will also include an option to derive mode shapes from this FE model for use in an improved modal method (including, e.g., torsion). More information will be forthcoming.

In the current version of FAST, “yes,” you should always set logic inputs CalcBMode and CalcTMode to False.

Best regards,