Response in Long term simulations and different MoorDyn time step

Hi everyone

I was studying how the length of the simulation impacts the mooring tension amplitude and run a 37000 [s] simulation. While analysing the results I saw an sudden change in the tension forces of the cables at around 16400 [s] (simulation with Mooring time step integration dtM=0.001):
dt1.png

I’ve reun the same simulation with dtM=0.0001 and this behaviour almost disappeared but it converged to the values after the change of the last simulation:

dt3.png

However the response of the platform experiences an initial large displacement that fades with time for the second simulation in surge and sway motions which dos not appear in the other degrees of freedom:

surge.png

Is this a transient effect? Do I need to run a very long simulation with dtM=0.001 in order to get the correct results or is something else the cause of this? All the simulations only consider hydrodynamic actions from waves generated from a JONSWAP spectrum with Hs=6 [m] and Tp=10 [s].
Thanks in advance

Francisco

I’ve found this thread in which the same problem seems to be occurring: [url]http://forums.nrel.gov/t/feedback-on-moordyn-v1-01-02f-numerical-instability/1816/1]

I am now running the same files with the OpenFast_v1.0.0 to see if it solves the problem.

UPDATE: It did not solve the issue.

Hi Francisco,

Thanks for providing the plots that show the problem, and also checking with the similar problem noted by a user last year. What you’re showing isn’t a behavior I’m familiar with, and I haven’t been able to think of what could cause it within the MoorDyn model. To help eliminate possible causes, do you think it would be informative to run the simulations with regular waves or perhaps no waves at all? (You could start the platform with an initial displacement or add in wind excitation if having platform motion is necessary for the issue to occur.)

Another line of investigation could be to look at the behavior in your existing results in detail (which I’m guessing you may have already done). Would you want to post zoomed-in plots (~60 s time span) of the fairlead tensions and platform DOFs in both low- and hi-tension states, and also for both dtM sizes?

Matt

I am still running aditional simulations with regular waves and without any waves but I can already provide the zoomed details you asked for:


I will also share the results of the other tests I am currently running.

I’ve managed to get good results with the revision [d] of FAST_V8.16 (as was achieved in the link sent before). I thought this revision was also present in OpenFastV.1.0.0 but it seems that it is not, as I was not able to reproduce good results with it:

It also happens with regular wave:

I did not find the cause, but must something that was updated in Fast_V8.16 but not in OpenFast.

Dear Francisco,

I know this is an old thread, but the source of this problem has now been found and fixed in OpenFAST. The problem was related to the storing of the time variable in MoorDyn in single precision, instead of double. This problem has now been fixed in a recent pull request OpenFAST: github.com/OpenFAST/openfast/pull/486. This problem was also recently discussed on our forum–see: http://forums.nrel.gov/t/question-1-for-turbulence-fields-points-outside-of-field/69/1.

Best regards,