SubDyn problem with larger substructures

Wind & Water Computer-Aided Engineering Software Tools
1

Hello everybody,

at first I have to appreciate you for your great complimentary software tools and for your voluntary support in this forum.

In order to analyse the behaviour of offshore substructures I’m using the FAST v8.03 modularization framework with SubDyn v0.04. Simulating a wind turbine with the monopile substructure provided in the CertTest folder works fine. However, this is the easiest case. When I substitute the monopile input file with the example input file provided at the end of the SubDyn Readme file (containing a kind of jacket) I’m getting the following warning messages at the beginning:

When I have a look at the FAST output file the results seem to be reasonable until the simulation time 0.3000 s, from then on every value is zero or NaN except of time and wind velocity columns. I tried to change some of the parameters in the SubDyn example file to those of the monopile input file but it didn’t work.

I interpret from the first warning message, that the displacement of the substructure at the interface point is too large, maybe due to the lower stiffness of the jacket compared to the monopile. But I am stumped for an answer to fix this problem. I hope someone can help me in this case.

Best regards,

Jan

2

Dear Jan,

We are aware of challenges and limitations with SubDyn and its interface to FAST within v8.03. We at NREL are busy working to resolve many of the problems and aim to have an improved version of FAST v8 with working models of several fixed-bottom offshore wind systems (including the OC3 monopile, OC3 tripod, and OC4 jacket, all with hydro-elastic interaction) by the end of the calendar year 2013.

While its probably preferrable to wait until our update is released, if you want to continue debugging your model, you could focus on potential problems, including:
*Choosing the appropriate time step for the coupled model
*Determining the impact of the jacket substructure on the mode shapes of the tower in ElastoDyn
*Determining how many Craig-Bampton modes need to be retained for accurate solutions of the jacket within SubDyn

Best regards,

3

Dear Jason,

thank you for your fast response.

I try further debugging and report my solution for this problem if there is one. However, it is sufficient for me to know that an improved version of FAST will be developed soon. So I promise that I will not rock the boat until the end of calendar year 2013 :laughing:

Best regards,

Jan

4

Dear Jason,

I have some questions:

  1. Can SubDyn be used to incorporate the pile-soil interaction?

  2. In FAST, if I use the distributed spring model to model the soil effect do I have to turn on all the 6 platform DoFs in time domain simulation since I encounter the same warning as discussed above. However if I turn off the platform DoFs, the warning did not appear and the results seem reasonable. Would these results be considered as good one?

  3. Would the output parameter TwrBsMyt be the overturning moment at the bottom of the pile or the loading of tower at the mud line?

Really appreciate!

Best regards,

5

Dear Lingling,

Here are my answers to your questions:

  1. SubDyn in FAST v8 cannot yet model pile-soil interaction. This on our to-do list, but will not happen this year.

  2. When I’ve modeled a monopile with a distributed-springs (DS) model in FAST v7, I only considered springs transverse to the pile (with no axial or torsional springs). So, I only enabled DOFs that would lead to transverse deflections; that is, I enabled platform surge, sway, roll, and pitch DoFs, and disabled platform heave and yaw DoFs. Disabling all platform DoFs will prevent transverse and healing motion of the bottom of the pile, which would not be realistic.

  3. In both FAST v7 and FAST v8, TwrBsMyt is the fore-aft bending moment at the tower-to-platform connection point–labled as “tower base” in Figure 20 of the FAST User’s Guide. Whether TwrBsMyt is at the bottom of the pile, mudline, or some other elevation depends on how you’ve set up your model (again, see Figure 20).

Best regards,

6

Dear Jason,

can you betray whether the new FAST v8 version will contain improvements for SubDyn regarding HydroDyn implementation or sparse matrix storage of system matrices?

We planned to implement the pile-soil-interaction by ourselves and willingly provide our solution, but we do not want to start our work on this until the new FAST version has been released to avoid double work on the code.

Best regards,

Jan

7

Dear Jan,

Our next release of FAST v8 (including the capability to model hydro-elastics of fixed-bottom multimember substructures) is now scheduled to be no later than June 30. We apologize for the delay. We’re running an extensive verification effort and we’ve had to debug many unexpected problems.

We’ve just started discussions about sparse matrix storage, which we plan to apply within SubDyn and e.g. the Jacobians needed by the FAST glue (driver) code. However, these modifications will not be completed in time for the June release.

Best regards,

8

Dear Jason,
Can new version SubDyn (SD_v1.01.00a-rrd) set a thrust to RNA like this?

Best regards,
Jason. Lai

9

Dear Jason,

When SubDyn is run in standalone mode (uncoupled from FAST), the motion (fixed offset or time history of displacement, velocity, and acceleration) of the transition piece (TP) can be specified directly from which the response of the substructure (motion and loading) can be derived. If a desired load at the TP is known in place of the displacement, the equivalent displacement is easy to derive because SubDyn is linear–i.e., K*d = F → d = [K^-1]F, where d is a 6x1 vector of desired displacements (3 translations, 3 rotations), F is a 6x1 vector of desired loads (3 forces, 3 moments) and K is the 6x6 equivalent substructure stiffness matrix, as found in the SubDyn summary (.SD.sum) file–called KBBt.

I hope that helps.

Best regards,

10

Dear Jason,
Thanks for your reply.
If I exert a thrust (such as 2000kN or 4000kN) to RNA structure, I can use your suggest to get the deflections at tower base.
After that,how can I get the deflections at RNA?

ps: I can get K value the 6x6 equivalent substructure stiffness matrix at transition piece (TP).

Best regards,
Jason. Lai

11

Dear Jason,

The equation I typed, d = [K^-1]*F, gives the deflection of the TP, based on the loads (F) applied at the TP. The location of the TP is up to you. If you prefer to model the entire support structure (tower + TP + substructure) within SubDyn, you can choose to locate the TP in SubDyn at the tower top. Then, d = [K^-1]*F will give the deflection of the tower top, based on the loads (F) applied at the tower top.

Best regards,

12

Dear all,

I am trying to analyse a larger jacket substructure with FAST8 v8.09. As my structure does not have a wind turbine on top, I have set AeroDyn and ServoDyn to 0 in the FAST input file. I am using the tower ElastoDyn file but I have disabled the Blade and Tower DOFs and set the mass and inertia of the Hub and Nacelle to negligible values in order to limit their effect.
When I ran the analysis I get the following error after SubDyn is computing the internal modal eigenvectors:

“Error allocating memory for the Pivot array for Jacobian LU decomposition array.”

I have looked at the echo files and the source files but I could not figure out the source of the error; the echo files indicate that both SubDyn and ElastoDyn are read correctly and hence I assume the error is probably related to the coupling of the files rather than the individual files.
Any help will be appreciated.

Thank you

Regards,

George

13

Hi, George.

That error is typically caused because the memory requirements for your model are larger than your computer can handle. Make sure you’re using the 64-bit version of the executable (FAST_x64.exe). If possible, use a machine with more RAM. If that still doesn’t run, you may have to simplify your SubDyn model (or HydroDyn model, if you’re using it).

14

Hello,

I’ve been trying to find any information about MpropseID2,
however I cannot find it.
What is this property referring to?

Regards,

Kan Ito

15

Dear Kan,

I think you mean SubDyn input parameter MPropSetID2? Input parameter MPropSetID2 specifies the cross-sectional property identifier for the end node of the member. Allowing differing cross-sectional properties between the start and end of the member allows for the modeling of tapered members. See the section 3.3.6 Members of the SubDyn User’s Guide and Theory Manual for more information.

Best regards,

16

Hi,
I have a question related to the output data from FAST, namely, the nodal displacements of the substructure nodes comig from SubDyn module and the displacements in each section of the tower coming from ElastoDyn module.

I am conducting a structural study in ANSYS software of a tripod-shaped structure, fixed to the seabed. My 3D CAD model, which I’m simulating, is composed by a substructure and a tower.

For the structural analysis I am using as input the outputs from FAST. For the substructure nodes I am using the output data from SubDyn module, ie, three displacements, MαNβTDxss, MαNβTDyss, MαNβTDzss and three rotations MαNβRDxe, MαNβRDye, MαNβRDze (I converted the local rotations into global rotations). For the substructure-tower interface, FAST calculates the outputs- IntfTDXss, IntfTDYss, IntfTDZss, IntfRDXss, IntfRDYss, and IntfRDZss, so I used these values.
For the tower I am using the displacements and rotations from ElastoDyn module, ie, TwHt1TDxt, TwHt1TDyt, TwHt1TDzt, TwHt1RDxt, TwHt1RDyt, TwHt1RDzt, etc. I divided the tower into 9 nodes.

When I ran the static analysis, I noticed that there is a displacement discontinuity near the first tower node (Figure 1). The stress is quite high in that region (Figure 2).

When I did the plot of the last 50 seconds of FAST analysis, I noticed that local tower displacements, for the first node, were lower than the interface displacements (Figure 3 and Figure 4 to see the X and Y displacements, for example). The interface region and the first tower node are placed near each other, so I do not understand these results.

After that, I ran FAST with the OC3 Monopile structure and the results in X and Y are similar.

Questions:
1 - Is there any incompatibility between the displacements coming from SubDyn and ElastoDyn?
2. Do the tower displacements take into account the substructure displacements?
3- Do the local xt, yt and zt axes have the same orientation as the SS coordinate system axes?

Best Regards,

Gabriel.

Figure3.JPG
Figure3.JPG795×624 31.1 KB

Figure 2.JPG
Figure 2.JPG817×732 42.6 KB

Figure1.JPG
Figure1.JPG800×717 37.8 KB

17

Figure 4:

Figure4.JPG
Figure4.JPG814×635 40.5 KB

18

Dear Gabriel,

The ElastoDyn output TwHt1TDxt represents the elastic deflection of the tower relative to undeflected position of the (rigid) tower expressed along the x-axis of the tower-base coordinate system (the tower-base coordinate system moves with the transition piece through the substructure flexibility in SubDyn). However, this tower deflection does not express the rigid-body motion of tower-base that as achieved through the substructure flexibility in SubDyn. The small value of TwHt1TDxt near the tower base simply expresses the small amount elastic deflection between the tower base and the nearby output node. If you want to see the actual position of the tower output nodes relative to the inertia frame, you can use ElastoDyn output TwHt1TPxi. (Likewise for y and z).

Best regards,

19

Jason,

Thank you very much for your input.

Best Regards,

Gabriel.