Does aeroelastic coupling in OpenFAST consider blade elastic vibration velocity?

Dear all,

I’m new to OpenFAST and I’m trying to perform aeroelastic coupling analysis with BEMT and GEBT on NREL 5MW Baseline wind turbine with steady, rated operation condition, i.e. wind speed 11.4 m/s, rotating speed 12.1 rpm without considering the effect of wind shear, tower shadow, yawed inflow, etc. The result is to compare with Section 2.4 and 3.1 of the paper “Li Z, Wen B, Dong X, et al. Aerodynamic and aeroelastic characteristics of flexible wind turbine blades under periodic unsteady inflows[J]. Journal of Wind Engineering and Industrial Aerodynamics, 2020, 197: 104057.

The input file for OpenFAST v3.5.0 is mainly based on the test case of “5MW_Land_BD_DLL_WTurb” in OpenFAST Github repository. I slightly modified the wind speed to make steady inflow. The simulation time is 30s with a time step of 0.0002s.

What I expected, as the paper reports, is that the aerodynamic loads can be significantly reduced by considering the aeroelastic coupling effect. As shown in the subplot (a) of Figure 1, when considering the aeroelastic coupling effect, the aerodynamic load F_N reduces with up to 2000 N/m at the blade span of around 54 m. For the deformation the blade tip flapwise deflection is reported to be around 4.5m.

Figure 1. Aerodynamic load distribution on the blade from the reference paper

However, what I obtained through AeroDyn and ElastoDyn+BeamDyn in OpenFAST is that the tip deflection is around 5.3~5.6m, much larger than the result of the paper. Also, the output Fx in AeroDyn at Node 16 (54.67m) of the blade span is around 6800-7400 N/m, which is also much larger compared with the result of flexible blade in Figure 1.

I compared the output of AeroDyn in both standalone mode (corresponding to rigid blade) and in OpenFAST (flexible blade), I found that the Vrel in both modes are almost the same, which may explain the deviation of the result in OpenFAST and the paper.

As the paper says, when calculating Vrel in aeroelastic analysis, the velocity of blade including the corresponding elastic vibrations should be considered, but maybe OpenFAST doesn’t consider it? Or maybe the difference between OpenFAST and other aeroelastic coupling methods is due to other reasons?

I’m sorry for such a long question, but I would appreciate it if anyone can help me solve this. It has puzzled me for a while.

Best regards,
Tongzhou Zhang

Dear @Tongzhou.Zhang,

I’m not familiar with the paper you reference, but OpenFAST does include the structural velocity from blade vibration in the calculation of Vrel and the aerodynamic loads. That said, in steady, uniform flow, I would guess the blade is not vibrating that much to have a large effect.

Can you share the results you are obtaining from OpenFAST that are equivalent to the figure from the paper you are referencing?

Best regards

Dear Jonkmann,

Thank you for your reply. Due to the limit of new users to the forum, I can pose only one figure in the question. Here I will show the rest of the results from the reference paper and OpenFAST.

For the aerodynamic load, the Fx from AeroDyn, corresponding to F_N in Figure 1, is shown in Figure 2 below. At Node 16 (54.67m of the span) where Fx is at maximum, the aerodynamic force is much larger than that in Figure 1. Also, Fx at Node 16 in the whole time history is plotted in Figure 3. The load is ranging from 6700 N/m to 7400 N/m, which is also much larger than corresponding F_N (around 6000N/m) in Figure 1.

Figure 2. Fx along the blade span in AeroDyn

Figure 3. Fx at Node 16 from AeroDyn

Figure 4. Tip flapwise deflectionin reference paper

For the deformation, the reference paper reports that the blade tip flapwise deflection in steady inflow is only 4.4 m (as shown in Figure 4), while what I obtain in OpenFAST is around 5.3~5.6m as shown in Figure 5.

Figure 5. Tip flapwise deflection from BeamDyn

Based on your reply, I still have several questions:

  1. I understand that the tip flapwise deflection is periodically changing due to the effect of gravity, but why is aerodynamic load changing in steady inflow? Is it due to the tip loss and hub loss, or some other reasons?

  2. The reference paper also mentioned the simulation in your work “Definition of a 5-MW Reference
    Wind Turbine for Offshore System Development” where the blade tip deflection is reported as 5.47m, matching pretty well with OpenFAST. The paper states that " The results of Jonkman et al. (2009) were obtained from FAST-AeroDyn, in which a linear modal beam model (ignoring the effects of the geometrical nonlinearity) is adopted for the structural analysis.". But that was in 2009 when BeamDyn hasn’t come into world. Why the result using BeamDyn doesn’t change much? BeamDyn uses GEBT and is expected to consider the geometrical non-linearity so I expect the tip deflection can be reduced.

Sorry for the late reply due to the time-lag. I’m sincerely looking forward to your and other users’ reply.

Best regards,
Tongzhou Zhang

Dear @Tongzhou.Zhang,

Regarding (1), I would guess the oscillations in blade deflections and aerodynamic loads are driven by a combination of gravity and shaft tilt, the latter of which induces skewed flow. You could avoid the former by setting Gravity = 0 and could avoid the latter by eliminating shaft tilt and disabling the tower degrees of freedom.

Regarding (2), I would expect steady-state deflection at rated wind speed (around 11.4 m/s) to be around 5.5 m. This is the deflection you are calculating, but differs from the paper you reference. But again, I’m not familiar with this paper and can’t explain their results. Moreover, the statement made in the paper about the FAST-AeroDyn results is incorrect. The structural model of old versions of FAST without BeamDyn, which became the ElastoDyn structural module of OpenFAST, is not a linear beam model. Instead, this structural formulation accounts for geometric nonlinearities, e.g. as discussed in the following forum topic: Coupled blade modes in FAST. I would generally expect ElastoDyn and BeamDyn to predict very similar blade deflections for the NREL 5-MW baseline wind turbine, except for a small amount of blade torsion that BeamDyn can capture that ElastoDyn can’t.

How does your model compare without and without flexibility, as shown in the paper?

Best regards,

Dear @Jason.Jonkman

I will examine the aerodynamic normal force at node 16 (54.67m of the blade span) here since the normal force is at maximum here and shows the most difference with the reference paper.

By running AeroDyn in standalone mode (which can be regarded as without flexibility) and obtaining result from OpenFAST (regarded as with flexibility), the Fx, angle of attack (Alpha), relative inflow angle (Phi), pitch and twist angle (Theta) and Vrel in AeroDyn at node 16 is shown in the following figure in corresponding order.

The difference of Theta in rigid and flexible blade may be due to the aeroelastic coupling effect, i.e. the torsional deformation is considered in aerodynamic load. However, the differnece of Vrel and other angles are small (up to 1 deg) so that according to the formula:


and lift coefficient C_l is around 0.114*alpha for NACA64_A17 airfoil in the linear range of alpha, thus the normal force Fx can be reduced with up to:


which matches the maximum difference in the first figure of this reply.

Maybe it is really because that the vibration is not so much, as you have mentioned?

Best regards,
Tongzhou Zhang

Dear @Tongzhou.Zhang,

I’m not sure I understand your question, but your results make sense to me.

Best regards,

Dear @Jason.Jonkman ,

Thank you for your reply. This nominal operating case is just for a validation. My ultimate goal is to carry out reliability analysis regarding tip deflection considering aeroelastic effect under the uncertainty of wind speed. I will carry on my research and not stick here.

Best regards,