Principles of modeling large wind parks

Transcription

1 Doctoral school of energy- and geo-technology January Kuressaare Estonia Principles of modeling large wind parks Raivo Attikas doctoral student.c. Tallinn University of Technology Department of Electrical Power Engineering raivo.attikas@pv.energia.ee Kalle Kilk doctoral student.c. Tallinn University of Technology Department of Electrical Power Engineering kalle.kilk@pv.energia.ee Abstract In this paper the main aspects of modeling large wind parks for electrical calculations of power systems are presented. An overview of most common designs for wind turbines is given. Acceptable simplifications for steady state loadflow analysis are introduced. Requirements for the dynamic wind turbine models were also presented in this paper and an overview of basic wind turbine components was given. Wind farms aggregated modeling versus reduced modeling was observed by the authors. Keywords Large wind parks steady state transient stability shaft system aerodynamic model blade-angle control converter aggregated model reduced model. Introduction The share of wind turbines in generation structure has been growing fast in several countries. It was possible to treat the wind generators as negative load when they formed a little part of power system and to ignore their specific static and dynamic behavior in larger scale system studies. With increasing wind penetration also grows the need to model the wind parks in more detail. ain configurations of the wind turbines Wind turbines can be divided into variable speed and fixed speed turbines. ost of the variable speed turbines employ a full-rated power converter an electronic interface for grid connection while the generator itself is an ordinary or permanent magnet synchronous generator. This solution gives an opportunity for better optimization of converting wind energy into electric energy at different wind speeds as the efficiency of the turbine is heavily influenced by the speed ratio of blade tip. Fixed speed wind turbines use mostly squirrel cage induction generators which are directly connected to the grid. These turbines are simpler by design and also cheaper. Therefore a large number of such devices are currently in operation all over the world. Yet a growing market share is for induction generators with controllable rotor current that have widened speed range compared to squirrel cage induction generators. In fairly steady conditions the power extracted from the air stream by the turbine blades can be calculated as follows [1]: 1 3 P * * * R * * C p where = air density R = radius of turbine blades = speed of moving air stream C p = coefficient of performance for the composite airfoil or Betc coefficient. Air density and wind speed cannot be controlled and the radius of the blades is fixed. Therefore C p is the means for torque control. C p is not a constant for a given airfoil but is dependant on the tip-speed ratio i.e. the ratio of the speed of the tip of the blade to the speed of the moving air stream. In the simplest turbines the blades are designed so that C p falls dramatically at high wind speeds. This method of aerodynamic torque control is known as stall regulation and is used on conventional squirrel cage induction machines with fixed-pitch blades. ore sophisticated method of aerodynamic torque regulation is based on the fact that C p can also be changed by adjusting the angle of attack of the blades. Blade pitch adjustment allows the energy capture to be optimized over wider range of wind speeds. Often a wound-rotor induction generator with a mechanism for controlling the magnitude of the rotor current through adjustable external rotor circuit resistors is used combined with pitch regulation of blades. Even more sophisticated rotor current control scheme can be employed in a doubly fed induction generator where rotor circuit is supplied with current from a four-quadrant voltage source currentregulated power converter. 18

2 Rotational speeds of large wind turbines are partly limited by maximum blade tip speed and therefore are relatively low. A gearbox is needed to mach the generator speed to the blade speed. As a result a mechanical system with low and high-speed sections is obtained. o a mechanical equivalent circuit used for grid studies comprises of two masses hub with blades and the electrical generator/gearbox combination interconnected by a flexible shaft. ain objects of modeling large wind parks for power system studies. The nature of most high power wind parks with large numbers of small individual turbines interconnected by medium-voltage network requires equivalencing and simplification. For some purposes the park can be treated as a single entity whereas in some studies this treatment could not allow to represent the behavior of the park in the full range. The analytical studies of large power systems observe the time frames from milliseconds to steady state. Therefore the models must reflect the behavior accurately over the entire bandwidth. till some details of the mechanical systems or energy conversion process may not be represented if they have no impact on electrical performance. teady state modeling of wind parks. For steady state power flow calculations a wind plant consisting of several wind turbines can be represented as a single generating unit at the physical connection point with power grid []. A real power setting in the range from minimum to maximum power generation can be used to analyze power flows in the network. Complications arise when trying to determine the equivalent reactive capability of the wind park. While fairly standard and well known for conventional power plants this characteristic has not been so well introduced for wind parks. ostly the problem is that net reactive power of a wind plant is a function of large number of elements within the plant like turbine reactive compensation reactive losses in collector lines and step-up transformers and auxiliary reactive power compensation equipment. odern grid codes that determine technical requirements for connecting wind parks to power systems define quite strict rules for reactive power control systems. Therefore in most cases a simplification may be done for everyday operation planning. As the large wind park connecting to main grid has to have a central controller for controlling both net active and reactive power it can be assumed that any reactive power between minimum and maximum required limits is available to the system. The control mode for reactive power can be either Q=const or U=const. where either reactive power setting or a desired voltage setting to be kept with the help of available reactive power is pregiven. For both control modes a respective control mode can be employed in power flow simulation program giving the equivalent generator scheduled voltage or fixed reactive power value. odeling of large wind parks for investigation of transient stability Large offshore wind farms will have in future the problem of integration into the existing transmission system as they are situated far away from the load centers and are of significant power in relation to the planning values for interconnection to the network nodes in transmission system. For reasons of economy the network interconnection is often designed without redundancy. This means that the nominal power of the transformer to the transmission system is in the range of the nominal power of all wind generators of a wind farm. hort circuit faults are in power systems unavoidable and cause a temporary instability of power transfer. In systems with weak network interconnection the problem of fast voltage collapse is prominent. The main aspects having a possible impact on transient stability issues are [3]: 1. Wind resources are usually at different locations than conventional power stations. Hence power flows are considerable different in the presence of a high amount of wind power and power system are typically not optimized for wind power transport.. Wind generators are usually based on different generator technologies than conventional synchronous generators. 3. Wind generators are usually connected to lower voltage levels than conventional power stations. Other aspect especially the fluctuating nature of wind power have not been seen to be relevant to transient stability problems because wind speed variations are too slow compared to the time frame relevant to transient stability (one to ten seconds). However because of limited predictability of wind speed system with high amount of wind power usually require higher spinning reserve than conventional power systems which adds inertia to the system that has influence on transient stability. Wind turbine model for transient stability investigation The following requirements can be pointed out for the dynamic wind turbine models to be used in investigation of transient stability [4;5]: The dynamic wind turbine models must predict the values of the electric power the reactive power and the voltage profile sufficiently accurately. It is also important for the dynamic models to predict sufficiently accurate values which are monitored by the protective relay system. These monitored values are both the electric values (voltage machine current grid frequency etc) and the mechanical values (the generator rotor speed for example). 19

3 The model to be used for local wind turbine sites shall be equipped with a supplementary model that disconnects the compensating capacitor the generator the converter and order stop of the wind turbine. Use of supplementary relay models is probably not always necessary in case of large offshore wind farms because exceeding the protective relay settings will indicate that the solution is not acceptable. The relay setting data and accuracy of the relay model are important. Because variable-speed wind turbines with converter-controlled generators are often tripped at the converter blocking the converter representation is very important. The block diagram of the dynamic wind turbine model which is applied to investigate transient stability is shown in Fig. 1 [5] Blade Angle Control P E G External ystem ignal Incoming Wind Aerodynamic Rotor odel P haft ystem odel G P H Electric Generator odel Geneator Converter Control I G Q E P E Protectiv Relay odel haft ystem odel U f Fig 1.1 Dynamic wind turbine model to be applied in investigations of transient stability. Here P P H and P E are the mechanical power the generator rotor shaft power and the electric power respectively Q E is the reactive power I G is the machine current U is the terminal voltage f is the grid frequency G and are the generator rotor speed and the turbine rotor speed respectively and denotes the pitch angle The model contains the blocks with representation of the basic wind turbine components [5;6]: 1. The aerodynamic model of the turbine rotor. This model will give a coupling between the speed deviation and the mechanical power produced by the wind turbine at a constant wind. Also this will give coupling between the turbine mechanical power and the pitch angle.. The shaft system model representing possible tensional oscillations in shaft system. This is important for representing interaction between the wind turbines electrical and mechanical parameters. 3. The electric generator transient model. 4. The blade-angle control and servo model if it is a variable-pitch wind turbine. The blade angle control of wind turbines can be organized by two different ways: 4.1. Pitch-control where the mechanical power P is reduced when the global pitch angle increases. This principle is mostly applied in variable-speed wind turbines. Fixedspeed wind turbines can also be equipped with pitch control as well but not commonly. 4. Active-stall-control where the mechanical power P is reduced when the global pitch angle decreases. This control principle is commonly applied in fixed-speed wind turbines. 5. The converter and its control if the wind turbine generator is converter-controlled. 6. The protective relay system model where it is necessary. The protective system of wind turbines monitors several parameters and orders disconnection of the wind turbines at registering of abnormal operation in power grid. Wind farms aggregated modeling versus reduced modeling Aggregated models with representation of: 1. each wind turbine on the farm. its no-load capacitor 3. the transformer connecting the wind turbine generator to the internal network of the wind farm 4. with sufficiently detailed representation of the internal network. The aim of modeling aggregated wind farms from transient stability point of view are following: Investigation of the mutual interaction between several electricity-producing wind turbines within a large wind farm. The question is whether there is a possible risk of self-excitation of the large wind farm with a large number of the wind turbines having identical or different parameters and subjected to given control features. In this case the term of self-excitation relates to uncontrollable mutual oscillations between the wind turbines excited by a disturbance. Wind turbines response to a fault occurring in the wind farms internal network. Investigation of relay settings due to trip of a number of the wind turbines as the result of the fault occurring (in the wind farm internal network or in the entire power grid). In general the wind turbines can be with different parameters and relay settings and also at different operational points. The aggregated model allows simulations of the wind farm under consideration of an irregular wind distribution. For example irregular wind distribution 130

4 appears when the wind turbines are shadowing each other from incoming wind. Reduced wind farm model represents a group of the wind turbines by one wind turbine model (given as a re-scaled equivalent). From transient stability point of view the aim of modeling reduced wind farms are following: Investigation of the collective response of the wind farm to a short-circuit fault occurring in the power grid. Voltage stability investigation of the power grid interaction with large wind farms and its dynamic reactive compensation. When we are reducing wind farm model to the rescaled equivalents the element of inaccuracy will always be included some phenomena are unavoidably neglected. The reduced wind farm model does not distinguish between details of the specific wind turbines within the given farm or specific operational conditions. A number of considerations shall be fulfilled when the large wind farm is combined into a reduced equivalent [5]. The power capacity of the reduced equivalent shall be equal to the sum of the power capacities of the individual wind turbines IJ. This condition is expressed by N I J The power supplied by the reduced equivalent to the grid P shall be equal to the sum of the power supplied by the individual wind turbines P IJ. This condition is expressed by P N P I J Further the dynamics of the induction generators of the wind turbines are given by the sloops of the P-Qcharacteristics of the induction generators. The following condition shall be fulfilled. dq dp N dq dp I J I J where dq IJ /dp IJ is the sloop of the induction generator of the wind turbine indexed (ij) and dq/dp denotes the sloop of the P-Q-characteristic of the reduced equivalent of the wind farm. Taking into account that the P-Q-characteristics of the induction generator is approximately a parabola with zero-sloop and no-load operation Q QNL ap. Here Q NL is the reactive absorption at no-load and a is the coefficient. Induction generators are no-load compensated. Therefore the reactive absorption of the wind turbines seen from the grid becomes proportional to the electric power squared. The technical requirements state that the large wind farm is reactive neutral in the connection point. Therefore the aggregated model and the reduced equivalent of the large wind farm will be initialized to the same point with respect to the reactive power exchange with the grid which is zero. Because the wind turbine shafts are relatively soft the pre-twisted shaft systems accumulate nonnegligible amount of potential energy. The pretwisted shafts relaxation (during the grid fault) leads to more acceleration of the generator rotors and influences on the voltage behavior. Hence the potential energy accumulates in the pretwisted shafts of all the wind turbines within the farm W s is by the following relation. N 1 W WI J NT K where W IJ is the potential energy accumulated in the pre-twisted shaft the wind turbine indexed (ij) denotes the average value of X. In case of the single-machine equivalent the potential energy W will be expressed by the following W T K 1 K NT Here T is the mechanical torque predicted by the reduced equivalent. Conclusion In this paper the main aspects of modeling large wind park were presented. It was stated that representation of several turbines as one equivalent generator is valid enough for load flow calculations. Determining the reactive capability of this equivalent generator is simplified with modern wind park central control systems. Wind turbine modeling for transient stability investigation is very complex and interesting challenge. As it is shown in this paper wind turbine must be modeled sufficiently in detail for representing accurate response to the short circuit fault. In that case wind turbine components as aerodynamic model of the turbine rotor shaft system model transient electric generator model bladeangle control converter and protective relay are need to be modeled for investigation transient stability. Wind turbine can be modeled differently according to the grid or to the wind park point of view. From grid side wind farm can be modeled as a reduced model where a large number of the wind turbines is given by a re-scaled equivalent. Aggregated model of wind farm is used when transient stability problems inside wind farm are investigated. 131

5 References 1. Wind generation technical characteristics for the NYERDA wind impacts study EnerNex Corporation R.. Zavadil 003. AWEA electrical guide to utility scale wind turbines AWEA Impact of large scale wind power on power system stability Ch. Eping J. tenzel Analysis modeling and control of doubly-fed induction generators for wind turbines Andreas Petersson Analysis of dynamic behavior electric power systems with large amount of wind power PhD Thesis Vladislav Akhmatov Advanced modeling of double fed induction generator wind turbine under network disturbance. eman F.Lov J.Niiranen