Unexpected Hub loading part II

[Aonghus.OConnor]

Hi all,

I've been running some simple test cases and I've come across an unexpected AeroDyn hub load.
I've started with the Cert Test 25 as a base model and then modified it in the following way:

1. Analysis duration of 100 seconds
2. Time step of 0.0125
3. All modules disabled except:
ElastoDyn
a. All degrees of freedom disabled except 6 platform DoFs
b. No initial conditions - zero rotor speed.
InFlowWind
c. Steady wind input, 8m/s no shear profile (PLexp=0)
AeroDyn 15
d. TwrPotent = 0
e. TwrAero = False
HydroDyn
f. Regular wave; WaveHs = 4, WaveTp = 14
g. WaveDir = 30
MoorDyn
h. UnstrLen = 845.35 ( For all 3)


I ran this analysis in the 64bit double precision version of FAST 8.16 and plotted the hub load:

double_precision.PNG (77.73 KiB) Viewed 4444 times

I then ran this analysis in the 64bit single precision version of FAST 8.16 and plotted the hub load:

single_precision.PNG (72.18 KiB) Viewed 4444 times

I have a question and an observation about the above output:
1. Is there any explanation for the sudden jumps in rotor hub load? (There is no sudden jump in displacement of the platform nor any warning/error that a problem has occurred)
2. Except for the sudden jumps, the trend between the single and double precision versions is the same but the times that these jumps occur differ.

Thanks in advance!
Aonghus

[Jason.Jonkman]

Dear Aonghus,

I would normally expect the single and double precision versions of FAST to compute nearly identical output, except when enabling BeamDyn, whose numerical sensitivity usually necessitates double precision for numerical accuracy. So, it is a bit disturbing to me that your single and double precision solutions are giving quite different solutions. This suggests to me that there is a numerical issue with your simulation set-up.

I do see that you have set-up your simulation such that the rotor is not operational (spinning). However, you haven't indicated that you've disabled the wake/induction model (by setting WakeMod = 0) or unsteady airfoil aerodynamics model (by setting AFAeroMod = 1). We normally recommend that the wake/induction model and unsteady airfoil aerodynamics model be enabled only for cases where the rotor is operating (as these models were developed for such cases). I can't say with confidence that the enabling of these models is the cause for the difference between the single and double precision versions, but I would start by disabling these models (perhaps independently) for this case to see their influence on the solution.

I hope that helps.

Best regards,