I would like to ask a question anyone who have tried to perform a stress analysis using FAST results.
FAST runs a simulation with dynamic excitation which is time varying analysis.
Thus, it seems like the stress structural analysis for blades or a tower also need to be done with dynamic analysis.
If static stress analysis is performed with applied loads, I think it would produce wrong results because it neglects motion of the structure which eventually neglects inertia term of equation of motion.
If anyone who have experiences to find strain and stress of blades or tower using FAST results throughout simulation time, please suggest me a way to obtain strain and stress values of any local point of blades or a tower.
Wind turbine analysis typically follows a two-step calculation process, where tools such as FAST are used to predict the dynamic response of the complete wind turbine system, including the coupled aerodynamics, structural dynamics, and controller dynamics. The outputs from FAST are the loads (shear forces, bending moments, etc.) throughout the various components of the turbine, such as the blades, hub, drivetrain, and tower. These outputs are then used to feed FE-based models of individual components for calculatations such as stress, strain, and buckling. Because the loads output by FAST include both the applied (e.g., aerodynamic) loads and the inertia loads from structural dynamics, there is no need to capture the dynamics within the FE-based model.
As you explained, the outputs from FAST can be used to feed FE-based models of individual components for design purposes. But as you know the FAST outputs for blades are internal reactions at different span locations. I want to ask that what is your suggestion about a way to convert these internal reactions into their equivalent external loading to be used in FE models?
So far I have used the method as follows. I divide the blade into a number of segments and evaluate the internal reactions at segment end points from FAST outputs and then try to calculate the external loading at segment mid points by writing equilibrium equations for each segment. Is this correct in your opinion?
With reference to your post Stress Analysis of Blades and a Tower titled “Re: Stress Analysis of Blades and a Tower”, I want to understand if you are calculating external load from internal reaction forces as per below process. I also need to calculate the same hence I am checking with you.
Read Spn1FxLb1 at station n from Aerodyn = 100kN (assume)
Read Spn2FxLb1 at station n+1 from Aerodyn = 200kN (assume)
Therefore the applied load from above reaction forces = 100-200 = -100KN or 100 Kn in opposite direction.
Is this process correct and also do you do the same for moments also?
First of all please note that Spn1FxLb1, Spn2FxLb1, etc. are FAST outputs and not from AeroDyn. We prefer FAST outputs over AeroDyn because FAST reaction forces contain all of the effects involved in the problem including aerodynamic and inertial interactions.
In order to calculate external loads from internal reactions, you need to satisfy the equilibrium conditions for each segment of the blade. The complete set of equilibrium equations in the x-z and y-z planes (i.e. flap- and edge-wise directions respectively) are as follows.
Where rFj’s and rNj’s are distances of forces from the rotor axis of rotation. Note the sign convention of FAST.
But before performing the above calculations you should first rotate the Spn1FxLb1, Spn2FxLb1, etc. Force and Moment vectors to coincide with the direction of “b1” system at the blade root using the structural twist angle of any section. Please refer to FAST user guide (p. 10) or the following forum topic: [url=http://forums.nrel.gov/t/discrepancy-in-loads-between-blade-root-and-local-spans/1091/1] for a description of the local blade coordinate system.
Thanks a lot for your detail explanation. I can calculate now the external load Fx from internal reaction. Since I have kept structural twist (StrctTwst) as zero, I may not have to do coordinate transformation (do you agree ?).
Now, the external load which is acting at the aerodynamic center cannot be applied there as there is no FEM node there as the blade is hollow. So do you recommend distributing the Fx, Fy, Fz, Mx, My, Mz uniformly on the airfoil nodes of that section? How have you applied these forces on the airfoil?
a) If Structural twist is zero then I need not transform Spn1FxLb1, Spn2FxLb1 to coincide with the direction of “b1” system at the blade root
b) Since there are no nodes at the AeroCenter (as blade is hollow), we should distribute the Fx, Fy, Fz, Mx, My, Mz on airfoil nodes uniformly or on FEA nodes on boundary
If the structural-twist is zero, than the local blade coordinate system at the root (xLb/yLb/zLb) is identical to the blade coordinate system (xb/yb/zb), with no transformations needed. However, at other stations along the blade, the deflection of the blade can cause a small difference between the local blade coordinate system (body-fixed to the deflected blade at each cross section) and the blade coordinate system (which is constant along the blade, independent of deflection). So, while you no longer need to correct for the structural twist, you may still have to consider the deflection, or accept an error, which may be negligible if the two stations are nearby each other or the deflection is small.
I’m not sure I understand your second question (“b”).
Thanks for the clarification on twist angle and transformation. My point “b” is that referring to airfoil below, the forces (lift, drag and their resultants in normal and tangential directions) are acting on aerodynamic center O as per image below. But there is no material at this point as we have shear webs passing through the hollow of the blade. So in FEA tool, there will not be any node at aerodynamic center. So to represent these forces should we distribute them evenly on the airfoil boundary nodes?
I’ve not had personal experience of taking the lumped load output from a beam-type model and applying that to an 3D FEA model to find stress/strain. I would guess that applying the lumped loads directly would result in unrealistic stress concentrations. I’ve read the following paper where they recommend to linearly distribute the load along the chord, but I cannot say whether that is a typical approach. Hopefully someone else with more experience in this area can comment.
Berg, J. C.; Paquette, J. A.; and Resor, B. R. “Mapping of 1D Beam Loads to the 3D Wind Blade for Buckling Analysis.” 52nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference. 4-7 April 2011, Denver, Colorado. AIAA 2011-1880.
Excuse me, I have a question to ask you, please help me, thank you~
For the Blade 1 Local Span Loads of NREL 5MW, which and where is the acting location of the outputs of ElastoDyn, such as SpniMLzb1, SpniFLxb1, SpniFLyb1, SpniFLzb1 for the blade segments i ? In my understanding, the location is located at corresponding aerodynamic blade analysis node, the center of the blade segment i, along the local blade-pitch axis. Is my understanding correct? However, I find it is strange that the blade-pitch axis passes through airfoil section at 0.25 chord. That is, at root, both AeroCent and blade-pitch axis are 0.25 chord rather than 0.5 chord.
The local blade span outputs from ElastoDyn are the reaction loads at the corresponding structural output node dictated by ElastoDyn inputs NBlGages and BldGagNd. In the NREL 5-MW baseline models provided by NREL, NBlGages = 1 and BldGagNd = 9, so Spn1 corresponds to the 9th structural analysis node. The locations of the blade structural analysis nodes are written to the ElastoDyn summary file and node 9 is at 50% span.
The structural analysis nodes in ElastoDyn lie along the pitch axis. But the pitch axis does not need to be at 25% chord. The aerodynamic analysis nodes can be offset from the pitch axis. As of FAST v8 and OpenFAST, AeroCent is no longer an input in ElastoDyn. When AeroDyn v15 is enabled (CompAero = 2), the aerodynamic analysis node locations are specified in AeroDyn and can be offset from the pitch axis.
Dear @Jason.Jonkman
Thank you for your reply. In fact, I want to know the location (in each airfoil) of the structural analysis nodes about the local blade span outputs from ElastoDyn. It seems that the structural analysis nodes in ElastoDyn lie along the pitch axis, at 25% chord of corresponding airfoil?
Regards,
No, 25% chord is not assumed. Other than when aeroacoustics or AeroDyn surface visualization is enabled, the location of the leading and trailing edge of an airfoil is not information that is needed by OpenFAST. The structural analysis nodes lie along the pitch axis in ElastoDyn. And the aerodynamic analysis nodes (origin of the lift/drag/pitching aerodynamic loads) are specified in AeroDyn v15 as offsets from the pitch axis. The leading and trailing edge are only defined when specifying the airfoil geometry in the airfoil data file, only used for aerodynamic visualization and aeroacoustics.
Dear @Jason.Jonkman
I’m still confused about the position of the pitch axis of NREL 5MW baseline WTB.
I want to map the pitching moment MZ of each blade analysis node (FAST ElastoDyn outputs: Spn1MLxb1,Spn2MLxb1,…Spn9MLxb1 ) at the pitch axis to the 3D FEM. Thus, I think I need to know which point in the airfoil is the pitching moment calculated based on?
