When running a time domain simulation and using the 3D turbulent time varying wind option, users sometimes find an error where the wind file is too small. In particular, a user may receive an error message that states that the 3D wind file is too small in either the longitudinal, lateral and/or vertical direction.
At the start of the simulation, Bladed will compute an offset distance to shift the wind file to ensure that the ‘box’ covers the wind turbine structure at the start of the simulation. However, some potential issues may still arise:
• The turbine structure may deflect significantly during the simulation. This may occur due to a poorly defined structural model or some model instability. In this scenario a portion of the turbine may exit the box causing the error.
• If the offset computed by Bladed is not sufficient then a portion of the turbine may not be covered by the wind file ‘box’ at the start of the simulation and a termination error message may occur. If the rotor plane is yawed at the start of the simulation then the offset may not capture this and so the ‘box’ will not cover the whole turbine. Alternatively, the turbine may have become numerically unstable such that large unphysical deflections occur causing it to exit the wind field.
There are two possible ways in which a user can try to overcome this issue:
• In “Calculation Parameters”, specify a large value for “Start time for turbulent wind”.
• Check the stability of the structural model. To understand whether a model is numerically unstable you can run a more simplified time domain calculation by:
1. Running constant wind rather than 3D Turbulent Wind.
2. Start outputting results from the start of the simulation by setting the "Time to start writing outputs" option to 0
3. Set the time step for which results are written to a small value such that the outputs can be viewed early before the simulation terminates
4. Then check the blade tip flapwise and torsional deflection. In many instances the model goes unstable because a high frequency component is not being resolved by the integrator. The solution could be to reduce the time step to capture the vibration but this will slow down the simulation time. Alternatively, remove the high frequency vibration from the simulation on the basis that it will not be excited.