Despite its rapid increase, large-scale wind farm to grid integration remains a technically challenging issue. A detailed or full order model (FOM) of wind farm plays a key role in understanding this problem and finding appropriate solutions. However, the detailed model is extremely complex and not convenient for large-scale simulation. This thesis aims to develop a simple, accurate, and manageable reduced order model (ROM) of wind farm.


In the first part of the thesis, linear model order reduction (MOR) of wind farm which can capture oscillatory dynamics of FOM during a small disturbance is considered. Effectiveness of five linear MOR techniques, namely balanced truncation (BT), alternating direction implicit (ADI)-based BT, rational Krylov (RK), iterative rational Krylov algorithm (IRKA), and subspace accelerated multi-input multi-output (MIMO) dominant pole (SAMDP) algorithm, are investigated on practical-sized wind farms with 90, 120, and 210 type-3 or doubly-fed induction generators (DFIGs). Merits and demerits of each method are discussed in detail. The ROM of wind farm is validated against the FOM in term of frequency domain indices and waveform agreement at the point of common coupling (PCC). Further, the use of IRKA method is extended to perform MOR of a multi-terminal direct current (MTDC) system connected to two large-scale wind farms with each having 48 DFIGs.

The second part of the thesis presents a method to develop a computationally efficient dynamic model of wind farm suitable for a large disturbance simulation. The method based on a trajectory piecewise linear (TPWL) approximation uses single and multiple training trajectories to develop a nonlinear ROM. Simulation results using a small demonstration wind farm system
with 2 DFIGs and a large practical wind farm system with 120 DFIGs are discussed to demonstrate the effectiveness of the model in capturing dynamic behaviour of wind farm following large disturbances.