Research On Electromechanical Transient Equivalent Modeling Of Dfig-based Wind Farms

With the global energy and environmental issues becoming prominent,wind power generation has accounted for high proportions in power grids because of its advantages of clean,flexibility,sustainability and so on.Large-scale wind farms,composing of dozens or hundreds wind turbines,have been formed with complex operating conditions.If every wind turbine is modeled individually,not only the simulation time and the model size will be enlarged,the effectiveness and accuracy of the calculation results will also be faced with great challenges.Therefore,it is critical to develop the effective equivalent model of the large wind farm,which can balance both the computation burden and the model accuracy,for wind power integration studies.Taking doubly-fed induction wind turbine for example,this paper researches the electromechanical responses of the wind turbines under balanced low voltage ride-through(LVRT)process.And through the field tests,theoretical deduction and simulation calculations,this paper has studied the modeling of the wind turbine,the equivalenting of the electrical network and the multi-and single-machine aggregating of the wind farms.Due to the advancement of wind power technologies and the improvement of grid codes,the existing wind turbine models are difficult to reflect the current LVRT characteristics of wind turbines,especially the post-fault power recovery performances after the fault clearance.Through extensive field tests,the complete LVRT processes for on-site wind turbines are analyzed and the universal active and reactive power waveforms are presented.The corresponding controllers for the gridand rotor-side converters are designed based on the modeling of the pre-fault steady state,fault-duration process,fault recovery period and the post-fault steady state.By the automatic updating of the controller reference values,the LVRT characteristics of the simulation model can be tracked well with that of the on-site ones.Thus,the modeling method of the complete LVRT process is proposed and validated by the field tests.Moreover it can be extended to full-power renewable energy generation units.The wind turbines are connected with each other series or paralleled by the internal electrical network of the wind farm.As a result,it is important to calculate its dynamic equivalent parameters to establish a high-precision wind farm model.Taking a four-machine wind farm with two rows and two columns as an example,a fast analytical equivalent parameter calculation method for the electrical network is proposed,and extended to arbitrary wind farms with complex topologies,based on the equal node injection currents and powers.Comparing with the two existing calculation methods(one is deduced by the principle of the equal network power loss,and the other is deduced by the principle of the equal weighted voltage differences),it is found the power losses of the proposed method have the highest correspondence in approximating that of the detailed wind farm model.Thus,the effectiveness of the analytical equivalent parameter calculation method for the electrical network is validated.The clustering of wind turbines is another prominent challenge in the research of wind farm equivalence resulted from the time variability and difference of the wind turbine operating conditions.Taking an on-site wind farm as an example,through the analysis of the electromechanical LVRT responses of the inside wind turbines operating at different wind speeds,two clustering indicators(the wind speeds corresponding of the starting and ending points of the maximum power tracking zone)are found by their aggregation characteristics.And thus,the inside wind turbines can be clustered into three groups to establish a three-machine equivalent model for a wind farm.Moreover,by analyzing the three-machine equivalence effect of typical wind distributions,it is found that wind turbines work in the maximum power tracking zone have to be aggregated into two more groups and the clustering index can be the single-machine equivalent wind speed of wind turbines working in the maximum power tracking zone.Thus,a practical four-machine equivalent strategy for the wind farm,which can avoid the complex computations of the clustering indexes changing with wind scenarios,is proposed.Simulation results show that the proposed four-machine equivalence method has good adaptability to wind scenarios and voltage drops.In order to further simplify the complexity of the wind farm equivalent,the feasibility of the single-machine equivalent to accurately characterize the actual wind farms has been discussed and demonstrated in this paper.Although the traditional single-machine representation requires small amount of calculations,it is hard to characterize the difference of dynamic behavior among individual wind turbines.A real wind farm is simulated with extensive wind scenarios,and it is found that wind turbines operating at different wind speeds have different steady power during the fault and take different recovery time after the fault clearance though they recover at the same rate.Taking a two-machine wind farm as an example,the analytical power expression of the wind farm is derived for the complete LVRT process,and thus the equivalent error of traditional single-machine model is found to be resulted from two aspects.One is the mismatch of the steady power during the fault,and another is the mismatch of the recovery rates after the fault clearance between the detailed and the equivalent model.Thus,the single-machine equivalent method by calibrating the power LVRT behaviors is proposed and extended to complex multi-machine wind farms.Further,from the perspectives of the obtaining of the voltage parameters and the calibrating of the power recovery rates,a practical simplified strategy is given and the corresponding implemention schemes are designed.Through the simulation analysis of the on-site wind farm and actual wind scenarios,the effectiveness of the proposed methods and the generalizability of full-power renewable energy stations has been demonstrated.