I am trying to implement the BEM method in MATLAB for wind turbine modelling. I have few questions, which are listed below.
First of all, I modelled the NREL 5MW offshore wind turbine in MATLAB. My algorithm is just like the Aerodyn, but with different high axial operation correction-Spera Corrections with ac=0.37-which corresponds to Wilson et Lissaman 1974. And to validate the BEM, i compared the Cp outputs with that of Wt_perf (the results that you post on Tue Mar 06, 2012 7:09 am).
I set the initial conditions for induction factors to zero.
I set the Error tolerans to 0.005.
I added Prantl Tip and Hub Loss models.
I compared my results with that of WT_Perf
1.My results seem almost fine when compared, but there is a little difference, especially in the high axial operation region, the turbulent state region, where Glauert correction is applied. And the difference is especially increasing at low velocities around cut-in velocity, high PRM. I am trying to get the almost the same result as WT_perf. As it is clear in figures, the results are almost same in the windmill state, the region where there is no correction. Do you have any idea for this difference? What may be the problem?
I attched my results. These results change according to the error tolerans value. But i could not get the results of the Wt_perfs in the high axial correction region.
By the way, my model output does not include the torque and thrust from the blade weight, pure aerodynamic torque and thrust. I don’t know Wt_Perf also gets torque, so power from the weight of the blades, either. From the results achieved and the results you put on the form, Wt_perf does not consider the weight effect because in low axial operation range, the results are same.
- Apart from above, as unfortunately we know, there are more than one correction in the literature for high axial operation region. As I mentioned above, I utilized Spera’s with ac=0.37 corresponding to the one from the Wilson and Lissaman, which, I think, Wt_perf utilizes in older versions according to an article that I read. But I am not sure whether this is the same correction model for the Wt_perf 3.00?
From 1994 to 2012, There are many changes in Wt-perf. There is also a change that may be important for this difference written by Dr. Bulh as follows.
‘‘I replaced the equations to calculate the element Ct, Cp, and Cq with equations. Pat Moriarty gave me. It improved the correlation to the rotor Cp.’’ Please see these cahnges and the sentecence written by Dr. Bulh the link:HARP_Opt/ChangeLog.txt at master · NREL/HARP_Opt · GitHub
*I also realized that every correction formula, such as Glauerts or others, have square root term in their equation. During iteration, the term inside the square root gets negative value. Therefore, we get complex inflow angle, complex angle of attack, so Cl and Cd data from the corresponding angle of attack cannot be read. For interpolation in Matlab, i am using interp1 command, it says that it cannot work with complex number. But this problem, i think, is not the problem due to interp1. Therefore, i am putting a command like if the term inside sqaure root is less than 0 (negative), take the term very close to zero. With this way, i am a kind of changing the initial conditions. Is there any other mehtod that i can apply?
- I also read about a paper about the iteration process of what Wt_perf used. Initially, as far as i am concerned, iteration for axial and angular factors were separately, that is the tangential induction iteration were nested inside the axial induction iteration in the previous version it says. In which version it is like this? My questions, WT_Perf V3.00 uses an iteration method just like Aerodyn? Which uses one iteration loop for both axial and tangential induction factors?
And lastly, what do you think about my results? Is it fine enough to move on simulink model etc to design controller after adding inertia, gear box ratio etc.?
Mustafa SAHIN, PhD Candidate
METU Aerospace Engineering, Ankara, Turkey.