Turbsim WARNING with long analysis time

Dear all,
I would like to run FAST simulations with extreme wind model and anlysis time of 1h; TurbSim can produce such files but shows this warning:

WARNING: EWM parameters are not valid for simulations longer than 10-min (or something like that)

Could you please explain such statement? Is this something related to errors in the wind produced? Which problems I will face running simulations longer than 10-min?

Thank you in advance.

Cheers,
Stefano

Dear Stefano,

The Turbsim User Guide states that:

This I believe is based on the EWM models described in the IEC 61400-1 standard which defines parameters for 10 min average wind speeds. You can refer to Turbsim User Guide and IEC standards for further clarification.

I am not sure why you would need to run a 1 hour EWM model. If you need a larger statistical sample, I would suggest running several 10 min runs with different seeds. I believe NREL even has some PERL scripts that will automate this for you.

Best Regards,

Subin

Dear Subin,
thank you for the quick reply. My fault, I missed that part in the TurbSim manual (also, I was looking to an old version of it). But your answer rises other questions :slight_smile:

I know I could do 6 simulations with different seeds (and the script to automate that is fairly simple) but as long as the standard allows for it (and it does) I would rather prefer to manage long simulations and less files to copy/paste/backup/archive. The IEC 61400-1 ed3 at 7.5 states that 6 10min runs (or alternatively 60 min continuous simulation) are required for turbulence wind.

So I was investigating the possibilty of 60 min sims and found the warning.

But looking at the formulae for ETM and EWM in the IEC standard I can’t clearly see the differences (and enventually errors) between 10 or 60 minutes wind time-series, once the Vref has been defined (and that is the case because the IEC define it given the turbine’s class)

Regards,
Stefano

Dear Stefano,

I have no experience with wind turbulence modelling and am not qualified to answer your questions. Bonnie or someone else at NREL will be able to provide more insight into the nuances of TurbSim in particular or turbulence modelling in general. Having said that, I believe the V50 and V1 that the IEC EWM model gives formulae for, are the 10 min averages which has a probablitity of occuring once every 50 years and 1 year respectively. If you use these values as average values for a 60 minute simulation I would think that the results would not be indicative of the probabilities that are being modeled. Of course this is just my interpretation. Someone with better understanding of these models will have to either confirm or refute this.

Best Regards,

Subin.

Dear Stefano,

Subin is correct in saying that the EWM model implemented in TurbSim contains values for 10-minute simulations; the IEC 61400-1 Ed. 3 standard lists all the values for this model using 10-min statistics. So, if you are trying to implement 60-min simulations, the warning produced by TurbSim is to let you know that your 10-min statistics will be incorrect. For example, let’s say you use the EWM model where the extreme 10-minute average is 50 m/s. Generating a 60-min time series of turbulence in TurbSim will give you a 60-min average of 50 m/s, which means statistically you will get 10-min averages that are larger than 50 m/s.

The 61400-3 (offshore) standard uses the EWM model from 61400-1, but–unlike the 61400-1 standard–it requires 60-min simulations, and so it also specifies equations to approximate 1-hr statistics from the 10-min statistics:

V50,1-hour = 0.95 * V50,10-min
sigma,1-hour = sigma,10-min + 0.2 m/s

It seems to me there are two options:

  1. you can generate six 10-min simulations with different seeds using the EWM model in TurbSim
  2. you can generate one 60-min simulation using NTM turbulence in TurbSim, specifying the power-law coefficient, wind speed, and TI you get using the equations above.

Dear Bonnie,
thank you for the answer. As test I will try to run NTMs with modified parameters as you suggested.
Btw which is, in your opinion, the best (in terms of accuracy and errors) way to go, 1 or 2 ?

Best regards,
Stefano

It is my understanding that most people who want to apply the 61400-1 standard use option (1): six 10-min simulations.