Construction of nacelle power-curve following Markov's theory

The standard procedure to constuct a power curve is the lEC standard. However, this model contains some flaws. In fact, by averaging the power inside a bin speed the power obtained isn't precise, since the power is directly affected by the turbulence intensity of the wind speed. Moreover, it requires long time to evaluate the influence of different settings with this method. The solution proposed in this thesis is to implement a novel method to construct a power curve, which is by following the Markov's theory. This method calculates the stationary power with a stochastic approach, which is more precise and should take less time than the standard procedure. Nonetheless, the construction of a power curve with the meteorological-mast (MM) anemometer isn't always feasible. In addition, the distance between the MM and the turbine can reduce the correlation of the wind speed read at the MM anemometer and the one which hit the blades of the turbine. Thus, to correct that, the construction of the power curve should be implemented on the nacelle anemometer. The final objective of this master thesis is to implement a Markov power-curve program, which will calculate the stationary power with the nacelle anemometer at each speed bin. Furthermore, this thesis will evaluate and optimise the parameters to obtain the best power curve following this novel method. Finally, the evaluation of the measurement time required to construct this type of power curve will be executed. Albeit the power curves obtained following this new method aren't stable, this thesis proposes the median to calculate the conditional moment, since the results are less affected by the turbulence intensity. Besides, it also proposes to average the wind speed over a period of two minutes, when using the nacelle anemometer, to eliminate the fast fluctuations of the wind speed due to the blade passage. Moreover, the utilisation of the data averaged over different periods of time doesn't change the result of the power curve. However, the time required to construct a Markov power-curve isn't affected by the averaging time of the data. Nevertheless, this novel method is a little bit faster than the lEC procedure. Further improvement should be done in the power-curve program. The software amelioration should make the results more stable and more accurate. In addition, the required time to construct a Markov power-curve might decrease again. The first recommendation is to verify with more data sets the rotor-position filter effect, when using the nacelle anemometer to construct a Markov power-curve. The second recommendation could be to correct the second minimum problem when calculating the stationary power with the minimal potential. A method to correct that will be to optimize the relaxation-time interval to determine the drift.

Level thesis:degree_level

masters

Grantor dc:publisher

École de technologie supérieure

Year dc:date

2007

Author dc:creator

• Poulin, Jonathan

• Anémomètres. Opérateurs de Markov. Turbulence atmosphérique. Éoliennes. Rotors. courbe, iec, nacelle, norme, puissance

Language dc:language

en

Identifier

INSERT-YOUR-LIBRARY-ID-HERE248

OAI record oai:identifier

https://espace.etsmtl.ca/cgi/oai2?verb=GetRecord&metadataPrefix=oai_etdms&identifier=oai%3Aespace.etsmtl.ca%3A248