Bestemmelse af ekstrem respons og svigtsandsynligheder for vindmøller under normal drift ved brug af kontrolleret Monte Carlo simulering

Estimation of extreme response and failure probabilities of wind turbines in normal operation using controlled Monte Carlo simulation

Sørensen, Søren ; Nissen, Jonas ; Jørgensen, Jacob

4. term

Structural and Civil Engineering, Master

2009

192

Dette projekt undersøger alternative metoder til at bestemme de største belastninger (ekstremlaster), som vindmøller kan opleve under normal drift. I dag beregnes ekstremlaster ofte ved at ekstrapolere ud fra få maksimumværdier fra nogle få 10-minutters simuleringer, men denne fremgangsmåde har ukendte statistiske usikkerheder. Første del gennemgår ekstrapoleringsmetoden i IEC 61400-standarden og belyser de tilhørende usikkerheder og praktiske problemer. Metoden anvendes også på en referencevindmølle til at estimere ekstremlaster. Derpå opstilles to enkle strukturelle modeller – en simpel forskydningsramme og en simpel vindmøllemodel – som bruges til at evaluere alternative metoder. De alternative metoder, Russian Roulette and Splitting with Distance Control (RR&S), Importance Sampling (IS) og RESTART, præsenteres som mulige erstatninger for den nuværende ekstrapolering. RESTART giver ikke lovende resultater, mens RR&S og IS viser lovende resultater på begge simple modeller og reducerer beregningstiden betydeligt. Til sidst implementeres RR&S-algoritmen i den aeroelastiske kode FAST. Her giver algoritmen lovende estimater af ekstremlaster, når vindmøllens kontrolsystemer er deaktiveret, men den forbedrer ikke beregningstiden, når kontrolsystemerne er aktiveret.

This project explores alternative ways to estimate the largest loads (extreme loads) that wind turbines experience during normal operation. Today, such loads are often derived by extrapolating from a few peak values taken from short, typically 10-minute simulations, a practice that carries unknown statistical uncertainties. The first part reviews the extrapolation method in the IEC 61400 standard and highlights its associated uncertainties and practical issues. The method is also applied to a reference wind turbine to estimate extreme loads. Next, two simple structural models—a basic shear frame and a simple wind turbine model—are introduced to evaluate the alternative methods. The alternatives, Russian Roulette and Splitting with Distance Control (RR&S), Importance Sampling (IS), and RESTART, are presented as replacements for the current extrapolation approach. RESTART does not produce promising results, whereas RR&S and IS show promising performance on both simple models and significantly reduce computation time. Finally, the RR&S algorithm is implemented in the aeroelastic simulation code FAST. The algorithm yields promising extreme-load estimates when the turbine’s control systems are deactivated, but it does not speed up computations when the control systems are active.

[This abstract was generated with the help of AI]

wind turbine ; monte carlo ; importance sampling ; controlled monte carlo ; extreme response / load ; russian roulette and splitting ; FAST ; oscillating systems ; Sample control function ; FORM ; variance reduction ; vindmølle ; monte carlo ; importance sampling ; kontrolleret monte carlo ; ekstrem respons / laster ; russisk roulette og splitting ; FAST ; oscillerende systemer ; kontrolfunktion ; FORM ; variansreduktion

Download
View record in AAU Student Projects