Links

Tools

Export citation

Search in Google Scholar

Experimental Vision Studies of Flow and Structural Effects on Wind Turbines

This paper is available in a repository.
This paper is available in a repository.

Full text: Download

Question mark in circle
Preprint: policy unknown
Question mark in circle
Postprint: policy unknown
Question mark in circle
Published version: policy unknown

Abstract

I nærværende afhandling udvikler vi to moderne kamerateknologier til at studere vindmøller med, dels med stereovision til at måle deformationer med og ved hjælp af Operational Modal Analysis (OMA) at studere dynamikken på en lodret-akslet vindmølle (VAWT) dels med en Background-oriented schlieren metode (BOS) for at studere tiphvirvler fra en 500 kW horisontal-akslet vindmølle under drift. Afhandlingen introducerer stereovision og OMA, og efterfølges af to praktiske implementeringer. I det første forsøg, der er udført på en VAWT vinge, har vi udviklet billedbehandlingsværktøjer til at analysere tidsserier af markører, som forskydes i tid og rum i billedplanen. Efterfølgende gør en udtænkt kovariant Sub Space Identification (COV-SSI) metode os i stand til at fortolke korte måleserier. Vi identificer de fire første egenfrekvenser og sammenligner resultaterne med klassisk modal analyse (EMA) og med finite element method (FEM) beregning. I et andet forsøg studerer vi deformationer på en stillestående, 3-bladet, 1-kW VAWT rotor med vindpåvirkning opnået i Polytecnico di Milano’s vindtunnel. Vi anvender en forbedret billedbehandlingsalgoritme, dog med den begrænsning at et af bladene med den anvendte 2-kamera opstilling ikke kan ses i billederne. I OMA afsnittet har vi udviklet data driven SSI (DD-SSI) og Frequency Domain Decomposition (FDD) koder for at studere dynamikken. De modale VAWT former er identificeret med OMA og valideres med simuleringer og med klassisk modalanalyse. Resultatforskelle forklares ud fra aero -og strukturdynamiske randbetingelseseffekter, hvirvelafløsning og bardun effekter. Usikkerheder for målte deformationer evalueres i de to nævnte eksperimenter ved fejlophobningsloven, og flere løsninger præsenteres for at mindske usikkerheden. I den sidste del af afhandlingen er BOS metoden blevet brugt til at studere tiphvirvler fra en Nordtank vindmølle i drift i naturlig vind. BOS baseres på synlig detektering af hvirvelkernens luftmassefyldegradient, og metoden behøver ikke kompliceret udstyr eller seedet flow, hvilket gør det til en bekvem metode for at studere flow i større skala. Udfordringen ligger i at kunne måle den lille brydningsindeksændring som følge af det lille Machtal. Spørgsmålet behandles i sidste kapitel, og ved at designe eksperimentet ud fra en foreløbig vurdering af tiphvirvlens egenskab, såsom størrelsen, tæthed og det maksimale trykfald henover kernen, kan tiphvirlen estimeres ved at tilpasse model og observationer. ; In the present thesis, two modern vision technologies are developed and used to study wind turbines: 1- Stereo vision to study vibrations and dynamics of the Vertical Axes Wind Turbine (VAWT) via operational modal analysis (OMA) 2- Background-oriented Schlieren (BOS) method to study the tip vortices that are shed from a Horizontal Axis Wind Turbine (HAWT) blades The thesis starts with an introduction to the stereo vision and OMA and is followed by two practical implementations of the basics derived in the introduction. In the first experiment, we developed the image processing tools to extract the displacement time series from stereo images taken from a VAWT blade subjected to the random vibrations. For analysing the time series, we devised an averaged approach of the covariance-driven stochastic subspace identification (COV-SSI) method. The method enables us to involve short measurement sets in OMA. Therefore, the first four natural frequencies are identified and agreed fairly with classical modal analysis (EMA) and finite element simulation (FEM). The second experiment is conducted on a VAWT rotor in the wind tunnel in a more controlled and designed condition, and the displacement time series are obtained using a more elaborated image processing algorithm. In OMA part, we developed the data-driven stochastic subspace identification (DDSSI) and frequency domain decomposition (FDD) codes for studying the dynamic behaviour of the turbine. The structural modes of the VAWT obtained with OMA are validated with the simulation and EMA, and then, the differences are explained with the aerodynamic effect and boundary conditions. The other frequencies obtained by OMA are interpreted via vortex shedding phenomena and guy wire effects. In the fifth chapter, the uncertainty of the displacements obtained in the two experiments mentioned above, is evaluated using the law of error propagation and several solutions are presented to decrease the uncertainty in the stereo vision experiments. In the last chapter of the thesis, the BOS method has been used to study the tip vortices behind a Nordtank horizontal axis wind turbine based on the density gradient in the vortex. The BOS method does not need complicated equipment such as special cameras or seeded flow, which makes it a convenient method to study large scale flows. However, the challenging part in the current case is the small refractive index change due to the small Mach number in the flow behind the HAWT. This issue has been addressed in the last chapter by designing a proper experimental setup according to the preliminary estimation of the tip vortex. The changes due to the vortex are modelled, and the tip vortex properties such as vortex size, density distribution and the maximum pressure drop in the vortex core are successfully estimated by comparison between the model and the experimental observations.