wind simulation with Windsim and TurbSim

Dear all, I would like to use the same seed in matlab (rng) and in turbSim (NREL Code) to simulate the same time Wind signal and make a comparison.

for windsim, I use: rng(0)

for Turbsim I use seed=0.

But It does not give the same result.

Many thanks in advance

Nicolas

[VBBV]

windSim and TurbSim are turbulent wind generation tools which dont use same stochastic wind models. The former is based on CFD turbulence models and also has lot of additional functionalities (a software package for wind resource analysis!!!) besides turbulent wind modelling, while TurbSim is based on the Veers & Shinozuka wind model, So even though one may use same seed inputs, the outputs may still differ. Hope this helps & Happy Christmas

[E. Cheynet]

For windSim, are you referring to the function used here?

I suppose that the random number generator use different algorithms. See https://se.mathworks.com/help/matlab/ref/rng.html#d126e1424506

[E. Cheynet]

@VBBV I am not sure whether Nicolas refers to the Windim CFD software or the windSim function on Matlab File Exchange that uses a text-based approach to generate a turbulence correlated wind field like in Turbsim

[Nicolas Delepine]

Thanks Etienne for your comment. I am refering to WindSim function on Matlab file exchange.

please find enclosed a comparison of the temporal wind velocity variation at hub height from the 2 codes: to obtain the same signal in both cases, the only possible solution would be to recomile TurbSim and using the same vector of the seed value from windSim ? Or does there exist an other solution, with a different code ?

Many thanks in advance

Nicolas

[E. Cheynet]

Hi Nicolas,

The random number generators used in Turbsim are described here: https://www.nrel.gov/docs/fy06osti/39797.pdf

In matlab, there are different possible algorithms: "twister" | "simdTwister" | "combRecursive" | "multFibonacci" | "philox" | "threefry"

See https://se.mathworks.com/help/matlab/ref/rng.html

So it could be tricky to have the same sequences of random numbers between the two codes, especially since the sequences of random numbers are different on each "nodes" where velocity data are simulated

Best regards,

Etienne