CACTUS (Code for Axial and Cross-flow TUrbine Simulation) is a turbine performance simulation code, based on a free wake vortex method, to study wind turbines and marine hydrokinetic (MHK) devices.
I am using CACTUS to simulate VAWT behaviour with secondary tip rotors (HAWT). These tip rotors rotate 50x faster than the primary blades, thus in order to accurately represent the effect of the secondary rotors one must simulate them at a range of phase angles. Attempts to rotate the tip rotors 60deg per time step require 310 time steps per rotation. However, when attempting to run CACTUS at such a large number of time steps (even without secondary tip rotors) leads to torque/power coefficients greater than 1 after 8 revolutions. In these situations, transient behaviour is highly erratic and shows little sign of converging.
I have played around with vortex core radius settings; namely significantly reducing the vortex core radius, however, this does not improve the situation. Timestep filtering does result in coefficients<1, however, the results seem very artificial and non-physical.
I am wondering whether I am encountering some obvious/known limitation of the model, or if anyone else has encountered this issue. Tips are much appreciated
Hello,
I am using CACTUS to simulate VAWT behaviour with secondary tip rotors (HAWT). These tip rotors rotate 50x faster than the primary blades, thus in order to accurately represent the effect of the secondary rotors one must simulate them at a range of phase angles. Attempts to rotate the tip rotors 60deg per time step require 310 time steps per rotation. However, when attempting to run CACTUS at such a large number of time steps (even without secondary tip rotors) leads to torque/power coefficients greater than 1 after 8 revolutions. In these situations, transient behaviour is highly erratic and shows little sign of converging.
I have played around with vortex core radius settings; namely significantly reducing the vortex core radius, however, this does not improve the situation. Timestep filtering does result in coefficients<1, however, the results seem very artificial and non-physical.
I am wondering whether I am encountering some obvious/known limitation of the model, or if anyone else has encountered this issue. Tips are much appreciated