Reduced order wind turbine aeroelastic modelling for condition monitoring & fault detection

Reduced order wind turbine aeroelastic modelling for condition monitoring & fault detection

Mangat, Naunidh (TU Delft Electrical Engineering, Mathematics and Computer Science)

Holierhoek, J.G. (mentor)
Meng, F. (mentor)
Watson, S.J. (graduation committee)
van Zuijlen, A.H. (graduation committee)

Delft University of Technology

Electrical Engineering | Sustainable Energy Technology

2019-02-27

Wind energy has emerged as a promising alternative to fossil fuel energy sources over the last two decades partly due to considerable reductions in the cost of power production. A common trend towards cost reduction has been to build larger and lighter wind turbines. These produce more power per unit and use less material. Another strategy for cost reduction could be to reduce the Operations & Maintenance (O&M) costs of wind turbines by using optimal O&M strategies. O&M costs account for approximately one-fourth of the overall energy cost for the full lifetime of a wind turbine. They assume even greater significance in the wake of growing interest in offshore wind farms since offshore wind turbines operate in harsher environments compared to their onshore counterparts and are more difficult to access. Condition monitoring has been proposed as a novel preventive maintenance strategy wherein sensors are employed to collect data related to the functioning of an operational wind turbine. This data can be processed to get meaningful information about component health of the wind turbine. An aeroelastic model can be used to simulate the fault-free response of the wind turbine for given operating conditions. By comparing real-time information from the condition monitoring system and aeroelastic model, it may be possible to develop routines which can detect developing faults in the wind turbine components. This forms the basis of a model-based condition monitoring system (MOD-CMS). For purposes of model-based condition monitoring it is required to have an aeroelastic model which is computationally fast to be capable of running real-time aeroelastic load simulations, and is highly accurate in order to detect faults. Furthermore, it is also desirable to have a linear aeroelastic model since this can also be used for controller and state observer design. A state observer can help estimate states of the wind turbine which are not easily measurable. This thesis reviews state-of-the-art aeroelastic tools to get a better understanding of their limitations. A reduced-order aeroelastic model is developed using the aeroelastic module in STAS WPP (State Space Analysis of Wind Power Plants) which is an open-source aero-hydro-servo-elastic tool in the Matlab/Octave environment. The linear reduced order aeroelastic model is verified by running test cases in the frequency domain for the NREL 5MW Baseline wind turbine. The accuracy of the linearized model is demonstrated by performing a stability analysis study for the NREL 5MW Baseline wind turbine. 
Furthermore, different coupling and numerical integration schemes are studied to develop a time marching simulation tool. Two main approaches are proposed. The first involves time marching of the monolithic, strongly coupled non-linear aeroelastic model using a multistep predictor-corrector integration scheme. In the second approach, a partitioned, loosely coupled version of the linear aeroelastic model is implicitly integrated over time. In the present work a tool based on the first approach of integrating the non-linear aeroelastic model is developed and subsequently verified by running simple power production test cases. 
 In conclusion, this thesis discusses and implements a framework of strategies which can be implemented to reduce the order of a high-order aeroelastic model. It suggests coupling and numerical integration schemes to utilize this reduced-order model in a time marching simulation tool.

Aeroelasticity
Reduced order model
Wind Energy
Structural Dynamics
Condition Monitoring
Fault Detection
aeroelastic instabilities
model order reduction

http://resolver.tudelft.nl/uuid:69a1c261-39ad-4b6b-b280-715f1fe94a1e

Student theses

master thesis

© 2019 Naunidh Mangat