Nacelle Motions

Hi,

According to following figure for OC3 Hywind how I can measure nacelle translational displacement and velocity in FAST ? What about rotational displacement and velocity ? (relative to platform center of mass)

Regards,

Dear Mehdi,

For a turbine with a fixed nacelle-yaw angle, the nacelle motion and tower-top motion are the same. In FAST, you can output the tower-top translational and rotational displacements (relative to the undisplaced position) via outputs TTDspFA, TTDspSS, and TTDspAx (for the fore-aft, side-to-side, and axial translations) and TTDspRoll, TTDspPtch, and TTDspTwst (for the roll, pitch, and twist rotation angles).

The tower-top (nacelle) velocities and accelerations are only available in the absolute sense, i.e., relative to the inertia frame. The rotational velocities of the tower top are given by outputs YawBrRVxp, YawBrRVyp, and YawBrRVzp for rotations about the tower-top xp, yp, and zp axes, respectively. The translational velocities are only available via the nacelle inertial measurement unit (IMU) outputs, but you can locate the IMU at the tower-top (yaw bearing) by setting FAST input parameters NcIMUxn, NcIMUyn, and NcIMUzn all to zero. Then the translational velocities are given by NcIMUTVxs, NcIMUTVys, NcIMUTVzs for translations along the xs, ys, and zs axes, respectively. The rotational velocities are also available at the nacelle IMU via NcIMURVxs, NcIMURVys, NcIMURVzs, but these will include the motion of the nacelle-yaw angle. (See the FAST User’s Guide for a description of all coordinate systems.)

For more information on the outputs available from FAST, please see the OutListParameters.xlsx spreadsheet supplied with FAST.

Best regards,

Hi,Jason

I got 2 questions:

1.I noticed that Nacelle IMU motions are defined in accordance with the shaft coordinate system(Xs,Ys,Zs),but the angle between shaft system and inertia system is not constant since shaft system could yaw with nacelle and furl with the rotor,which makes coordinate transformation from shaft system to inertia system difficult. So I was wondering if there is a more direct way to output nacelle velocities and accelerations?

2.In many of sample inputs included in CertTest folder from FAST archive,IMU is not located at tower top but somewhere higher and more up-wind. I am curious about that if there is any principles I should follow when defining the IMU position? or why did they located the IMU there? Possibly a critical equipment there?

Thank you!

Dear Minxi,

Your understanding of the orientation of the nacelle IMU is correct. However, I’m not sure I understand your point about the transformation between the shaft system and the inertia frame. Any coordinate system fixed in the nacelle will move relative to the inertia frame. Please note that the nacelle IMU motion outputs are all absolute (relative to the inertia frame), but simply have there components expressed in the local coordinate system. If you don’t want to output motions within the nacelle, you can also have FAST output the motions of the yaw bearing or of nodes along the tower.

It is common for a nacelle IMU to be located on or near the main bearing of the drivetrain, which e.g. is where we’ve located the nacelle IMU in the FAST models of the NREL 5-MW turbine.

Best regards,

Hello Jason,

for the implemention of my subroutine I need the tower top deflection. I know it is an easy question, but since I havent found the mistake in my imlementation yet, I want to make sure I understand the output-parameter TTDspFA correctly . With “relative to the undeflected position” you mean relative to the yaw bearing CM position on the following picture right?

And TTDspFA is positive if it points downwind in the nacelle frame of reference, is that correct?

Most importantly:
Does the nacelle frame of reference move with the deflecting tower top, or does it stay in the undeflected position if the tower tow moves?

Thank you very much in advance,

David

Dear David,

You understanding of FAST output TTDspFA is correct–the output is relative to the undeflected tower centerline/yaw axis and is positive for downwind deflection.

The nacelle coordinate system does translate and rotate with the tower-top deflection. However, output TTDspFA is expressed in the tower-base (t) coordinate system, which does not translate and rotate with the tower deflection.

I hope that helps.

Best regards,

Hello Jason,

thank you very much, that already helped a lot! I got one more question though: As TTDspFA is the absolute deflection from the undeflected position, it is the sum of a static deflection (which varies for turbulent wind speed) and a dynamic deflection. If I have higher wind speeds the static deflection will be larger. The problem is: I only need the dynamic deflection of the tower top for my subroutine. Is it anyhow possible to extract the static deflection as an output? I could put the values of the static deflection for every time step into a lookup table and substract the static deflection from TTDspFA then to get the dynamic deflection of the tower top. I know how to calculate it for a forced vibration if I have the excitation function, but as I am simulating complex turulent wind models with turbsim I am a bit lost here. I also tried using a highpass filter on TTDspFA but that doesnt help in a lot of cases…

Thanks,

David

Dear David,

I agree that output TTDspFA is the sum of the static and dynamic deflection. However, the static contribution to the deflection is not a distinct output of FAST. The way I’ve seen it done before to separate the static and dynamic deflections is to separate the one output into two by apply a low- (or high-) pass filter to the data, as you’ve suggested. You could also create a look-up table of the static deflection; for turbulent wind cases, I would suggest basing the static deflection on the mean wind speed.

Best regards,

Dear Jason,

I build my model to analyse the dynamic response of the structure (just platform and tower). I am trying to modify source code (FAST 8.16) to create an interface between my model and FAST, however, I am a little confused which variable should be modified.

I need to extract the tower top forces and moments, use these data in my model and send the resulting tower top displacement and velocities including the rotational ones (6DOF) back to FAST, at each time step. ? May you help me to figure it out?

Best regard

Dear Enes,

Your new support structure model sounds a bit incompatible with the existing modules of FAST. From what you describe, it sounds like your new model could be used as an optional replacement for the SubDyn module. However, SubDyn is coupled to ElastoDyn by passing the motions (position/orientation, velocities, accelerations) from ElastoDyn to SubDyn and the loads (forces and moments) from SubDyn to ElastoDyn (while the SubDyn is normally connected to ElastoDyn at the platform at the tower base, without modification of the source code, the same interface can be used to connect SubDyn to ElastoDyn at the yaw bearing, as described in the SubDyn documentation).

It sounds like your new model needs the opposite exchange of information i.e. motions passed from your new model to ElastoDyn and the loads from ElastoDyn to your new model. It would be most straightforward to change your model such that you can receive motions and return loads; then, you could use the same coupling used between ElastoDyn and SubDyn.

Best regards,

Dear Jason,

First of all, I appreciated for your explanation. However, I need any quick but accurate way to impose aerodynamic loads to my model.

Equation of the model; [M+a]*qdd + [B]qd + [K+C]q = FHydro + FAero
M : Mass matrix
a : Added mass
B : Hydrodaynamic damping
K : Structural stiffness
C : Hydrostatic stiffness
FHydro : Hydrodynamic forces (simply wave diffraction forces)

In order to import FAero, I was thinking to modify some lines of FAST and create glue code which will;

1 - Initialize FAST without tower and platform DOFs
2 - Transfer the forces (FrcONcRt and MomBNcRt) calculated by ElastoDyn to my model.
3 - Calculate the response(q,qd,qdd) in the equation above and enhance the time step
4 - Impose the tower top motions where AeroDyn takes the TowerMotion%TranslationDisp, %Orientation, %TranslationVel
5 - Run AeroDyn and ElastoDyn again until FrcONcRt and MomBNcRt are calculated.

In brief, I intend to use FAST as a subroutine of my model. Does that approach make sense?

Thanks in advance,
Best regards.

Dear Enes,

I see a couple problems with your approach:

  • The loads you are transferring from ElastoDyn to your model will not include the inertia-induced loads from motion of the support structure.
  • While you are imposing tower motions in AeroDyn, the rotor motions AeroDyn is receiving from ElastoDyn will not include the effects from motion of the support structure so e.g. you won’t get the proper aerodynamic damping of the support structure.

Your model appears to be simple e.g. linear. With a little rearranging of the equations of motion, could you not update your model such that motions of the tower-top are input (instead of internally calculated) and forces at the tower top are internally calculated (instead of input)? This would then enable you to more closely couple your model to ElastoDyn, similar to the ElastoDyn-SubDyn coupling.

Best regards,

Dear Jason,

As I understand, I missed the effect of the support structure by disabling the DOFs of tower and platform, that will cause ElastoDyn not to consider inertia related loads. Thus, this approach needs to be reconsidered.

Actually, as you see, the model was firstly considered for the linear hydroelastic analysis of a large floating structure. I modified it to obtain time domain representation like HydroDyn does with frequency domain input. I was planning a replacement between HydroDyn and my model. However, since FAST only consider rigid body motions of the platform, I decided to form a weakly coupled solution at tower top.

I can rearrange my model to make it compatible with governing equation of FAST (M*qdd=F). So far, I haven’t looked at SubDyn module since I was studying on FASTv7, but as I understand it couples the solution on a junction.

Modifying the FAST source code such that it contains unlimited mode shape (or DOF) i.e. expanding the size of governing equation, and passing the forcing term and impulsive added mass to the governing equation would make more sense, am I right? However, I can’t foresee how long time the modification will take.

May you give any advice?

Best regards

Dear Enes,

I’m sorry, but I’m not really sure I understand your question.

I would take a look at SubDyn – perhaps it already has the capabilities you seek? The DOFs of SubDyn are limited only by the number of finite elements and resulting Craig-Bampton modes that are enabled.

Best regards,

Dear Jason,

I am sorry I couldn’t clarify my question. I will try to explain.

If I change my model as you said, I think it would be more proper to use HydroDyn module. As far as I understand, HydroDyn sends the impulsive added mass (6x6) and the platform forces (6x1) to the related rows and columns of the augmented matrix (NDOFxNDOF+1) of ElastoDyn. Since, my model contains elastic modes of the platform, I need to enlarge the augmented matrix to contains additional added mass and forces. However, I don’t know entire process of ElastoDyn. I wanted to ask if the modification that way is time consuming and if this coupling approach is reasonable.

I have reviewed SubDyn manual. It looks very useful. I will go deep into SubDyn module. Thanks for this reccomendation. But, I have one more question to be ensure: the linearization is not available when SubDyn is enabled, is it?

Thanks in advance.
Best regards.

Dear Enes,

In FAST v8, data is exchanged between HydroDyn and ElastoDyn only through input-output coupling relationships in this case motions are sent from ElastoDyn to HydroDyn and loads are sent from HydroDyn to ElastoDyn. HydroDyn does not directly pass a 6x6 impulsive added mass matrix to ElastoDyn.

Correct, it is not currently possible to invoke linearization of FAST v8 with the SubDyn module enabled.

Best regards,

Dear Jason,

Is that procedure about HydroDyn the same for the FASTv7?
I am very thankful for your continuous support.

Best regards.

Dear Enes,

In FAST v7, the coupling of HydroDyn to the structural model functioned differently. In FAST v7, indeed the 6x6 impulsive added mass matrix was passed from HydroDyn to the structural model through the interface.

Best regards,

Dear Jason,

I was thinking FAST 8 also use the same precodure. Thank you clarify this issue.

Best regards.

Dear Jason,

Additionally, my question in the post “Mon Apr 10, 2017” is still on the table, which is about the tight coupling of the flexible platform in FAST7 by editing augmented mass matrix? Do you have any recommendation?

Best regards.