Dear Dr. Jason.
FAST gets the mode shapes of the blade from the BModes program, and the mode shapes are entered into the FAST simulation. However, the interactive effect between tower/rotor and blade is not considered in the mode shapes obtained from BModes. Thus, natural frequencies and mode shapes of the blade obtained from FAST time domain simulation show sort of differences from the BModes results when all DOFs were turned on in FAST simulation. When all DOFs except for the blade mode are turned off, the mode shape and natural frequencies of the blade are almost the same as the BModes results. I followed your guide in your post (http://forums.nrel.gov/t/nrel-5mw-blade-mode-shape/1092/1) in order to be the same conditions between FAST and BModes.
Doesn’t the difference of mode shapes caused by the interactive effect between blade and tower/rotor considered in the FAST simulation? In other words, I think that the mode shapes that the interactive effect between blade and tower/rotor are considered has to be entered to the FAST for more precise simulation. Could you check whether or not my thought is reasonable?
I understand your concern. To summarize, the tower and blade mode shapes are inputs to the structural model of FAST v7 and the ElastoDyn module of FAST v8 are and typically derived isolated from each other, but in reality, the mode shapes should depend on the coupling between the rotor, nacelle, and tower. Moreover, for the blade mode shapes, the mode shapes will change with operational conditions, such as rotor speed and pitch angle. This is the biggest weakness of the modal-based method.
Within the structural model of FAST v7 and ElastoDyn module of FAST v8, while the mode shapes are not changed based on coupling or operational conditions, the mode shapes are used as shape functions in a nonlinear model that does consider coupling between the rotor, nacelle, and tower. Moreover, the loads calculated and output by FAST are solved using the load-summation method. With these features, my experience has been that while the prescribed mode shapes input to FASTshould be reasonable, they do not need to be exact for accurate response predictions. Similar questions have been discussed elsewhere on this forum.
Dear Dr. Jason.
I understand that very precise mode shape doesn’t need to be entered for the response. However, I want to use the FAST program for the health monitoring, and for this, I want to obtain more accurate natural frequency and mode shape in the simulation.
You mentioned that the non-linear FEM model can consider the interactive effect between blade and tower/rotor. Does it mean that if MSC Adams code is coupled to the Aero-dyn program, it is possible to consider the interactive effect in the operating condition?
You should be able to get accurate natural frequencies and mode shapes from the FAST simulation.
MSC.ADAMS (or an equivalent multi-body dynamics code) does not have the same limitations of the structural model of FAST v7 or the ElastoDyn module of FAST v8 because it does not depend on prescribed mode-shape inputs. But again, my experience is that the modal-based method of FAST is sufficient for most cases.
Now I am estimating the natural frequency and mode shapes using the frequency domain decomposition(FDD), which is one method of operational modal analysis(OMA). For this, the acceleration of some points of blade or tower in time domain simulation is required. Actually, the mode shapes are already prescribed from BModes program and they are used in the time domain simulation. As a result, the response results will reflect the prescribed mode shapes. This is meaningless in OMA.
Also, prescribed mode shapes don’t consider the coupled effect between blade, tower, and other sub-systems. When I turn off the tower elasticity options and turn on only the blade elasticity option, I can obtain the similar natural frequencies and mode shapes of the blade from FDD. But if the blade is coupled to the tower elasticity, the difference becomes sort of large. Of course, if the coupled effects are considered, the blade dynamics are very complicated and various. I know that FAST program shows the all the blade dynamic aspects and the results are similar compared to the other FEM program results. However, I wonder that the blade mode shapes are exactly estimated when I use the uncoupled mode shapes in time domain simulation and applied those acceleration results to OMA.
In this regard, I think that FE based methods, such as MSC. ADAMS and BeamDyn, are more appropriate than modal based method for the OMA. Could you check my idea?
I’m not sure I fully understand your question enough to answer it.
Keep in mind that the prescribe blade mode shapes input to FAST are used as shape functions in the nonlinear beam model, so, the actual deflection of the blade will depend on the superposition of the shape functions that are excited (the mode shape superposition is weighted based on the value of the corresponding degrees of freedom).
You obviously sound concerned about using ElastoDyn alone; you may want to compare the results of an ElastoDyn alone versus ElastoDyn plus BeamDyn or another structural model to verify whether ElastoDyn alone can be applied to your problem.