Inclusion of Dynamic Inflow in Model Predictive Control of Wind Turbines
Translated title
Inklusion af dynamisk inflow i model prediktiv regulering af vindmøller
Authors
Steffensen, Henrik ; Overgaard, Anders ; Jørgensen, Martin Østergaard
Term
4. term
Education
Publication year
2014
Submitted on
2014-06-03
Pages
152
Abstract
Denne afhandling undersøger, om dynamisk inflow bør indgå i modelprædiktiv regulering (MPC) af vindmøller. Dynamisk inflow betyder, at vinden ind i rotoren ændrer sig over tid og påvirker aerodynamikken, mens MPC er en reguleringsmetode, der bruger en model til at forudsige og optimere den fremtidige drift. Designmålet er at balancere to hensyn: høj elproduktion og lav tårntræthed. Projektet er foreslået og delvist vejledt af Vestas Wind Systems A/S i Aarhus, Danmark. Resultaterne bygger på simuleringer af NREL 5 MW-vindmøllen, en trebladet, landbaseret upwind HAWT, i FAST-software. Vi analyserer systemet for at afklare, hvad der bør adresseres i et MPC-design, der inkluderer dynamisk inflow, og giver en første vurdering af de potentielle forbedringer. Træthed vurderes med rainflow-tælling, og der beregnes en Damage Equivalent Load (DEL). Vi opstiller en samlet model med aerodynamik, drivlinje, tårndynamik, pitch-aktuator og generator og deler den i to varianter: en med kvasi-stationær aerodynamik og en med en enkel model for dynamisk inflow. Hver model bruges til at designe en MPC, og de to reguleringer sammenlignes via Pareto-fronter, der viser afvejningen mellem elproduktion og tårntræthed. Resultaterne viser tydelige forbedringer lige over mærkevindhastigheden, hvor dynamisk inflow er mest udtalt: MPC med dynamisk inflow reducerer træthed med mere end 20 % og mindsker samtidig den gennemsnitlige absolutte afvigelse fra effekt-referencen med cirka 5 %.
This thesis examines whether dynamic inflow should be included in Model Predictive Control (MPC) of wind turbines. Dynamic inflow refers to time-varying changes in the wind entering the rotor that affect aerodynamics, and MPC is a control method that uses a model to predict and optimize future operation. The design focuses on balancing two goals: high power production and low tower fatigue. The project was proposed and partly supervised by Vestas Wind Systems A/S in Aarhus, Denmark. Results are based on simulations of the NREL 5 MW wind turbine, a three-bladed, onshore, upwind HAWT, using the FAST software. We analyze the system to identify what an MPC with dynamic inflow must address and provide an initial assessment of the potential benefits. Fatigue is quantified with the rainflow counting method and summarized as a Damage Equivalent Load (DEL). We build an overall model that includes aerodynamics, drivetrain, tower dynamics, pitch actuator, and generator, and split it into two variants: one with quasi-steady aerodynamics and one with a simple dynamic inflow model. Each model is used to design an MPC, and the two controllers are compared using Pareto fronts that show the trade-off between power performance and tower fatigue. The results show clear improvements just above rated wind speed, where dynamic inflow effects are most pronounced: the MPC with dynamic inflow reduces fatigue by more than 20% and also reduces the mean absolute error of power relative to the reference by about 5%.
[This abstract was generated with the help of AI]
Keywords
Documents