S-N curves for FOWT reference


I’m in a team working on a controller for the OC3 hywind (spar) floating turbine and we want to perform the load analysis with MLIFE to obtain the damage equivalent loads and, most important, the lifetime of the main components of the turbine.

To do this, we need for each component the Wohler exponents and the ultimate loads (which are related to the S-N curves).
I read in the FAQ (nwtc.nrel.gov/FAQ#TypeLMF) that a way to obtain an ultimate load is to multiply the maximum value of the load by a factor like 20.
I do not think this method is a good way to obtain realistic lifetime of the components because ultimate loads obtained by this method are correlated to the stress components endure during the simulations. For me, the ultimate loads should only take into account the material of the components.

Could you share ultimate loads values and Wohler exponents (or S-N curves) for the following components of the OC3 hywind FOWT:

  • Tower base loads (TwrBsMxt, TwrBsMyt)
  • Hub and rotor loads (LSShftMxs, LSSTipMys, LSSTipMzs)
  • Yaw Bearing loads (YawBrMxn, YawBrMyn, YawBrMzn)
  • Blade root loads (RootMxb*, RootMyb*, RootMzb*, and aggregate loads on x and y)

Also, do you considere that the choice we have made for the load components are relevant?

Best regards,
Benjamin Maréchal

Dear Benjamin,

I agree, which is why the FAQ explains that, “[ideally], LUlt would be based on a strength (e.g., FEA) analysis of the cross section of the component in question.”

However, the OC3-Hywind system is a conceptual design and I’m not aware of a strength analysis that’s been. So, I don’t have ultimate strength loads to share with you. For the Wohler exponent, we typically use 3, 4, or 5 for steel components and 8, 10, or 12 for composite components in the absence of more specific information.

I agree with your selection of outputs, except that I would normally look at the shaft moments in the rotating frame and at the main bearing (i.e., LSSGagMxa, LSSGagMya, LSSGagMza). Likewise, I would look at the yaw-bearing loads in a coordinate system that does not yaw with the nacelle (i.e., YawBrMxp, YawBrMyp, YawBrMzp).

Best regards,

Hello Jason,

Thank you for answering quickly.
For the controller we are developing, it is very important to be able to evaluate the lifetime of the turbine.

I understand that no strength analysis on OC3-Hywind is available.
Are you aware of a wind turbine for which the ultimate strenght loads (or the S-N curves) are available?
The best case would be a floating turbine of at least 5 MW.

We also have another question regarding the total length of the simulations for fatigue assessment.
In the article Simulation-Length Requirements in the Loads Analysis of Offshore Floating Wind Turbines, it is stated that “approximately ten 10-min simulations for each wind speed bin should be used to obtain converged statistics”.
For the fatigue assessment we perform, we use a full distribution of environnemental conditions, which represents a very large number of cases. We don’t think we need to do 100 minutes of simulations for each case.
Which cumulative length of simulations would you advice to perform a proper fatigue assessment?

Thank you for your time and Happy New Year,
Benjamin Maréchal

Dear Benjamin,

I’m not aware of any floating wind turbine model (FOWT) available in the public domain, which has the ultimate strength loads or S-N curves available for the tower, shaft, and blade root (that doesn’t mean there aren’t any). You may need to make your own assumptions for your own purposes.

I can’t recommend a specific cumulative length of simulation to perform a proper fatigue assessment for FOWT. I think more work is required on this topic before general statements can be made. I suggest you do a convergence study based on your own FOWT and metocean conditions. Certainly, though, for a given condition, fatigue loads converge more quickly than ultimate loads (the paper you reference covers both ultimate and fatigue loads).

Best regards,

In the post above on Sat Dec 22, 2018, 3:31 am, I knew how to set the S-N curve’s parameter in Mlife which is SNslopeLst. However, I am still confused about how to set TypeLMF and LUlt in Fig.1. In Mlife User’s Guide, I seemed not to find it. So could you please introduce some details and relevant materials? And in Mlife User’s Guide, TypeLMF is a string, but in Fig1, it is a number.

By the way, I seemed not to enter the FAQ (nwtc.nrel.gov/FAQ#TypeLMF). The website now is Fig2. Maybe it changed the address?

I can’t find some information about fatigue only Mlife tools.
Thank you!

Best regards

Dear Dezheng.Zhu,

As mentioned in the MLife User’s Guide, TypeLMF can be set to explicit fixed mean value or the keywords “AM” or “WM” can be used to specify the aggregate mean (AM) across input files or weighted mean (WM) based on the specified Weibull distribution.

The old FAQ was lost with the loss of the old NWTC Information Portal, but the specific FAQ regarding TypeLMF and LUlt was posted on our forum–see my post dated Jun 14, 2020 in the following forum topic: http://forums.nrel.gov/t/mlife-fatigue-del-calculation/661/29.

Best regards,