[Jerry1]

Hi All,

Just a general question. I am simulating a wind turbine to model the trailing wake and it's structure. I will be comparing this to experimental results to verify the model. The experimental work will be carried out in a wind tunnel. However, when I model the simulation using air, at a freestream velocity of 10m/s I have to simulate the turbine with very high rpm values, (in order to match experimental tip speed ratios). I want to use a timestep that allows for 1 degree of rotation per timestep, (highly recommended in literature). However, this results in a very small timestep and as a result, this leads to longer computational time. However, if I swap the medium to water, I could reduce the timestep. This would reduce the computation time needed for my simulation. I would plan to select a reduced rpm value by matching Reynolds numbers across the blade between the air and water cases. Would this be suitable? I have tried to find any validation of this technique but I am unable to find any. I just need advise. I think that using water (a more viscous medium), would result in the wake dissipating earlier than it should and this would make the model invalid to compare to experimental work. Has anyone come across this technique? I have seen it used to mimic pressure distribution over blades etc but not to model the wake struture and characteristics.

[FMDenaro]

Originally Posted by Jerry1

Hi All,

Just a general question. I am simulating a wind turbine to model the trailing wake and it's structure. I will be comparing this to experimental results to verify the model. The experimental work will be carried out in a wind tunnel. However, when I model the simulation using air, at a freestream velocity of 10m/s I have to simulate the turbine with very high rpm values, (in order to match experimental tip speed ratios). I want to use a timestep that allows for 1 degree of rotation per timestep, (highly recommended in literature). However, this results in a very small timestep and as a result, this leads to longer computational time. However, if I swap the medium to water, I could reduce the timestep. This would reduce the computation time needed for my simulation. I would plan to select a reduced rpm value by matching Reynolds numbers across the blade between the air and water cases. Would this be suitable? I have tried to find any validation of this technique but I am unable to find any. I just need advise. I think that using water (a more viscous medium), would result in the wake dissipating earlier than it should and this would make the model invalid to compare to experimental work. Has anyone come across this technique? I have seen it used to mimic pressure distribution over blades etc but not to model the wake struture and characteristics.

be careful, using air you have rho=O(1) kg/m^3, using water you have O(10^3) kg/m^3. This is only partially compensated by increased viscosity. Therefore, for constant reference velocity and lenght, the Re number is greater!

[Jerry1]

If i was to use water I would reduce the freestream velocity to 6m/s. This with the changed rpm of the turbine would give me the same reynolds numbers over the turbine blades. I'm just entertaining this idea to see if I can reduce computational time. So far I have only come across literature that use this technique when they are validating the forces over blades with experimental results. So far I cannot find any work on this technique being used when the trailing wake is investigated. I think this method is probably unsuitable as the higher viscosity value of water will more then likely dampen out the wake alot earlier than the experimental work, therefore making the model unsuitable for validation.

[FMDenaro]

provided you use the same values of Re, St, ecc. the physics of the flow is the same

[Jerry1]

are you suggesting that in order to reduce the prandtl number of water to have it in the same region as air, just to run the simulation using water but have water at a higher temperature. This will reduce my dynamic viscosity and reduce the overall prandtl number