ICSET 2008 Application of Interval Computation Technique to Fixed Speed Wind Energy Conversion System R.K. Thakur1, and V. Agarwal2, Senior Member, IEEE Abstract—Any wind energy conversion system (WECS) experiences uncertainty in wind velocity, tip-speed ratio, power co-efficient and stiffness of the wind turbine shaft. Also, the magnetizing inductance of the induction generator varies during self-excitation process. This makes the analysis and control of WECS, a difficult proposition. This paper presents an interval analysis based computation method for the calculation and analysis of various machine parameters and variables in WECS. The usefulness of the proposed method in the design aspects of a WECS is discussed. The realistic (dependent) uncertainties in the equations are solved to analyze the behaviour of a selfexcited induction generator under balanced operating conditions. Experimental results are included to highlight the behaviour of the induction generator voltage during transient phase. energy is, for many reasons, one of the most promising renewable energy sources. The wind energy can be harnessed using a Wind Energy Conversion System (WECS), as shown in Fig. 1, comprising a wind turbine, an electric generator, a power electronic converter and the corresponding control system [1]. This is a highly interactive, complex, multiple input multiple output system which exhibits non-linear, non-minimum phase dynamics. The objective is to convert the wind energy (at varying wind velocity) to electric energy while maximizing the power and safety of operators [2]. Wind is one of the most abundant renewable sources of energy in nature. It is an erratic, uncertain and uncontrolled variable. It is a function of many factors such as temperature, pressure, geographical topology, terrain, region etc. The available wind energy is converted into mechanical power by wind turbines [3]. The power can be captured at two different tip-speed ratios. Accordingly, there are two possible regions of turbine operation, namely the high- and low- speed regions. High-speed operation is in general bounded by the speed limit of the machine. Lowspeed region is not bounded by the speed limit of the machine, but in this region the system has non-linear, nonminimum phase dynamics [4]. To analyze WECS we need a computation method which can yield results with the user defined accuracy and the method of calculation should be reliable. In this context, it is noted that the interval IND Manuscript received April 30, 2008. R. K. Thakur is with the College of Military Engineering, Pune, India and is presently pursuing Ph.D. in System & Control Engineering, IIT-Bombay, Powai, Mumbai-400 076, India (e-mail: [email protected]). V. Agarwal is with the Department of Electrical Engineering, IITBombay, Powai, Mumbai-400 076, India (e-mail: [email protected], Fa x: +91-22-25723707). Pwind Pmech Gear Box Induction generator Capacitor bank Fig. 1 Schematic diagram of a wind energy conversion system (WECS) The paper is organized as follows: Section II provides the modeling of a WECS. Section III describes the interval computation technique. Section IV presents analysis and simulation of induction generator at different operating condition. Experimental results are included in section V. The major conclusions of the work are included in Section VI. II. MATHEMATICAL MODEL OF A WECS In this section, mathematical model of a wind energy conversion system comprising wind turbine and its associated parameters and an induction generator are developed. A. Wind turbine [2 - 4] The wind power is converted into mechanical power as per the aerodynamic power-speed characteristics of the wind turbine. The mechanical power is a function of wind speed (V), rotor speed (ωr) and the tip-speed ratio (Ȝ), and is given by the following equation: Pmech = 12 ρ Ad RC p (λ )V 3 (1) Wind speed varies drastically and hence averaging is used to determine the correlation between wind speed and mechanical power. So, the calculation of mechanical power is not unique rather it is valid over a range. Also, power is proportional to cube of the wind velocity. Hence, average value of wind speed is used to reflect the total energy available for harnessing, whereas actually the total energy available for harnessing varies over a range. Tip-speed ratio (Ȝ) is a function of linear velocity of tip of the blade and the wind velocity and is given by the following equation: 1093 c 2008 IEEE 978-1-4244-1888-6/08/$25.00 Pel computation technique is able to provide reliable results. I. INTRODUCTION W Vt Wind Turbine λ = ωr R V (2) As the wind velocity varies from top to bottom of the swept area of the solid disc of rotating blade, the tip-speed ratio value, Ȝ also varies accordingly and hence can be reflected by a range of its maximum and minimum values. It is an important parameter which determines the quantum of energy harnessing from the wind and is used for the maximum power point tracking. The power co-efficient (Cp) is a non-linear function of the tip-speed ratio (Ȝ), pitch angle (ȕ) and blade design. It is given by the following equation: ª π (λ − 3) º C p = (0.44 − 0.0176β )sin « » − 0.00184(λ − 3) β ¬15 − 0.3β ¼ (3) Fig. 2 D-q axis equivalent circuit of SCIG at no load corresponding to: (a) d axis (b) q-axis. It is sensitive to dirt on the blade surface. In most of the studies, Cp is regarded as a constant for all rotor speeds which is aerodynamically not true and its variation at different rotor speeds must be accounted for during investigations. Hence, it is more appropriate to consider a range of Cp and not a constant value. This parameter is maximized for harnessing the maximum energy from wind. Electric power, Pel = η Pmech (4) where, η is a function of ωr. For a stationary reference frame (Ȧ = 0) and before excitation, all terminal voltages are considered to be zero and the generator equation is given by the following equation [6, 7]. ªv qsº §Rs +pLs +1/ pC pLm 0 0 · ªi qs º «vds » ¨ 0 Rs +pLs +1/ pC pLm ¸ «i ds » 0 «v qr»=¨ pLm −ωr Lm Rr +pLr −ωr Lr ¸ «i qr » ¸ «v » ¨© ωr Lm pLm ωr Lm Rr +pLr¹ «i » ¬ dr¼ ¬ dr¼ (6) B. Mathematical modeling of induction generator The d-q axis equivalent circuit of a Squirrel Cage Induction Generator (SCIG) is shown in Fig. 2. The equivalent circuit represents the dynamic characteristics of SCIG [3, 4]. The following assumptions are made in the modeling of SCIG: The rotor and stator resistances and inductances are assumed to be constant; saturation in rotor and stator iron cores and space harmonics arising from saturation of stator and rotor teeth are ignored. The voltage equation for SCIG is obtained from the d-q axis equivalent circuit for a rotating reference frame. It gives the dynamic model of the SCIG in terms of the machine parameters and direct and quadrature axis current and voltage variables respectively. It is given by the following equation: ªv qsº §Rs +pLs +1/ pC Ȧ Ls pLm Ȧ Lm · ªi qs º «vds » ¨ -Ȧ Ls Rs +pLs +1/ pC Ȧ Lm pLm ¸ «i ds » «v qr»=¨ pLm (ω−ωr)Lm Rr +pLr (ω−ωr)Lr¸ «i qr » ¸ «v » ¨© (ωr −ω)Lm pLm (ωr −ω)Lm Rr +pLr ¹ «i » ¬ dr¼ ¬ dr¼ (5) The model (5) describes the dynamic characteristics of the induction generator in rotating reference frame. The electromagnetic torque is given by: 3 p Lm Tel = ( λ ds λ qr − λ qs λ dr ) 2 2 K (7) The dynamic equation of motion is written as: ω r = 1 J (Tm ech − Tel ) (8) where, J is the equivalent inertia constant of wind turbine and induction generator combined. III. INTERVAL COMPUTATION TECHNIQUE Interval computation technique [8] is a mathematical analysis method requiring a computer – manual computing will be too tedious. The successful and accurate implementation of interval analysis depends significantly on the understanding of the relevant mathematical analysis in the various applications in engineering and sciences. It translates the theorems and techniques of the application into sharper versions. It transforms the qualitative reasoning to a quantitative aspect. It can be used for the solution of large linear and non-linear equations with interval coefficients. It gives error bounds for singularly perturbed systems of ordinary differential equations. 1094 The squirrel cage induction generator is described by a large system of ordinary differential equation with saturated hysteresis non-linearity. For self-excitation process in a squirrel cage induction generator, a minimum air gap flux linkage is required [9]. This value of minimum flux linkage is dependent on machine parameters and capacitance values for stable steady-state operation. The self-excitation process is associated with bifurcation theory i.e. the emergence of two distinct solutions when a critical value of parameter is reached and exceeded. At the critical parameter value, a radical qualitative change takes place in the system dynamics i.e. either buildup or collapse of the steady-state operating points. SCIG self-excitation exhibits a saddle-node bifurcation in which the terminal voltage builds-up from zero equilibrium point and the critical parameter is magnetizing inductance value [10]. The self excitation process in an SCIG is described by a non-linear differential equation of the following type: . X = f ( X , Λ) (9) where, X represents the set of state variables and Λ represents the parameters of the machine which governs the dynamic behaviour of the SCIG system. By setting the derivative term of X equal to zero, the equilibrium point of SCIG is determined which yields the following equations: f ( X , Λ ) = 0 → A(Λ ) X = 0 (10) In the beginning the state variables are zero, so the trivial equilibrium point is X = 0 . The other feasible solutions of the system exist when the determinant of ( sI − A) is set to zero [9, 10]. The resulting equation yields expressions relating the rotor speed to the angular speed of generated voltage in terms of machine parameters, magnetizing inductance and excitation capacitance values. The describing function of (6) yields a characteristic equation of order 6. The coefficients of the characteristic equation are in the form of some range with their infimum and supremum values. The coefficients are itself the polynomials of these variables with structured uncertainty which makes otherwise the solution further difficult. Also, there exists truncation and rounding off errors in the computation using existing method. This leads to an erroneous result which gives unreliable values of the design parameters. Thus, interval computation technique is used to solve such equations for reliable solution. det [sI-A] = C 2 Rr + ω r X 1 )) 2 2 6 X1 s 2 5 + 2C X 1 X 2 s + (C + 2CL2 X 1 ) s 3 4 + 2 2 (X2 + X 1 (2 Rs 2 (2( Rs ( Rr X 2 + L2ω r X 1 )C 2 2 + ( L2 X 2 + Rr X 1 )C ) s + (C Rs X 3 + 2C ( Rs Rr L2 2 +ω r L2 X 1 ) + 2 2 L2 ) s + 2( Rs CX 3 + Rr L2 ) s + X 3 where, L1, L2, X1, X2 and X3 are + 2 Rr X 2 (11) L1 = L s + L m ; L 2 = L r + L m ; X 1 = L1 L 2 − L2m ; X 2 = R s L 2 + R r L1 ; X 3 = R r2 + ω r2 L22 . Equation (11) is solved using interval method for a pair of complex poles on the imaginary axis of the s-plane. The solution of the characteristic equation gives all possible values of the design parameters at a given operating point. The describing function for the same induction generator is also obtained with the resistive load. The characteristic equation with the resistive load yields an equation of order 6. Solving this 6th order equation by interval method for the condition of self-excitation process yields the capacitance value required at different loading conditions. IV. RESULTS AND SIMULATION ANALYSIS An induction generator with the following specification [7], is simulated using Matlab/Simulink software: 3-φ, Yconnected, 220 V, 4.8A, 60Hz, parameters in pu are, Rs = 0.0946, Rr = 0.0439, Xs= Xr = 0.0865, Xm (max) = 2.12. The rotor speed of induction generator and machine parameters other than magnetizing inductance are assumed to be constant during self-excitation. The magnetizing inductance is set to a value between 0 and the unsaturated value which ascertains the structured uncertainties and this translates the system dynamics into a sharper version enabling the quantitative analysis of the highly non-linear system. It also ensures the condition to capture all possible solutions. The solution of (11) for the above mentioned induction generator under the stated assumptions yields the capacitance value and the corresponding magnetizing inductance value. The results are summarized in Table 1. TABLE 1 VALUE OF MAGNETIZING INDUCTANCE CALCULATED BY INTERVAL METHOD AT RESPECTIVE SPEED=300 RAD/SEC Sr no. 1 2 3 4 CAPACITANCE Range of Magnetizing Inductance value, H VALUE AND ROTOR Range of Capacitance Value, ȝF 0.2678 - 0.27174 37.89 - 40 0.1483 - 0.1496 68.27 - 70 0.1164 - 0.1168 89.06 - 90 0.0983 - 0.09845 104.4 - 110 Clearly, the use of higher capacitance value drives the induction generator into deep saturated region. This condition of the system results in transients with large magnitude during self-excitation process. The plots showing the results obtained from the calculation method with and without load are shown in Figs. 3, 4 and 5. Clearly, Fig. 3 shows that the use of large capacitance value drives the induction generator into deeper saturated region which causes the transients with large magnitude. It is clear from the graph in Fig. 4 that an increase in load resistance requires a corresponding higher capacitance value to induce the self-excitation process in different operating conditions in the induction generator. It is evident from Fig. 5 that as the 1095 load resistance is increased the system is driven into deep saturated region causing increased transient amplitude in the voltage build-up. results for respective capacitance value are summarized in Table 2. The magnetizing inductance values obtained from simulation are close to the value obtained using the interval computation technique corresponding to the respective capacitance value. Thus, the method of calculation using interval computation technique is verified. The method is found to give the solutions for machine parameters which are accurate to the user defined value. Thus, it can be concluded that the proposed calculation method is reliable. TABLE 2 VALUE OF MAGNETIZING INDUCTANCE OBTAINED FROM THE SIMULATIONS AS SHOWN IN FIGS. 6(a), 6(b), 6(c), and 6(d) Sr no. Magnetizing Inductance Capacitance Value, ȝF value, H Fig. 3 Plot of magnetizing inductance value versus capacitance value. 1 2 3 4 0.265 0.145 0.115 0.098 40 70 90 110 Fig. 4 Plot of load resistance versus the corresponding minimum capacitance value (a) Fig. 5 Plot of magnetizing inductance versus load resistance The information regarding load resistance and corresponding capacitance and magnetizing inductance values in Figs. 3, 4, and 5 are useful in the design of different types of controller such as adaptive controller, deadbeat controller, model reference based predictive controller, quantitative feedback theory based controller, sliding mode controller, supervisory control and data acquisition systems etc. The results obtained from the proposed calculation method, and listed in Table 1 are verified through the simulation of the induction generator. The saturation non-linearity in the induction generator is included in the simulation by adopting five segments, piece-wise linearization of magnetizing inductance in the function block of Simulink. The magnetizing inductance value is obtained from the simulation of the induction generator for C = 40 μF. The simulation result is shown in Fig. 6 (a). The magnetizing inductance value is 0.265H. Similarly for other capacitance values: C = 70μF, C = 90μF, and C = 110 μF, the induction generator is set to self-excitation process and the simulation results are shown in Figs. 6(b), 6(c) and 6(d). The magnetizing inductance values obtained from simulation 1096 (b) (c) (d) Fig. 6 Plots showing the variation of magnetizing inductance for different capacitance values: (a) 40μF (b) 70μF (c) 90μF (d) 110μF Further the simulation results also verify the fact that the use of large capacitance value drives the induction generator deeper into saturation and thus change the transient characteristics of system dynamics with increased transient amplitudes and faster speed of response. V. EXPERIMENTAL RESULTS In this section, experimental results on induction generator with the following specifications are presented: 3-φ, Yconnected, 220 V, 5.1A, 50Hz, parameters are, Rs =5.7ȍ, Rr = 4.6ȍ, Ls= 0.021H, Lr = 0.01987H, Lm (max) = 0.319H. The induction generator of the above specification is driven by a dc motor at different speeds. The stator terminals of the induction generator are connected with the 3-φ, capacitor bank of desired value to induce self-excitation in the induction generator. The experimental results of the voltage build-up during self-excitation are shown in Figs. 7, 8 and 9. The figures are obtained using Agilent’s Infiiniium oscilloscope, with a multiplying factor of 20 on vertical axis. It is observed that with the rotor speed increased from 301.5 rad/sec to 314 rad/sec, keeping the capacitance value fixed for C = 70 μF, there is negligible transient amplitude in Fig. 7 whereas there is noticeable transient amplitude during selfexcitation in Fig. 8. The speed of response has improved as is obvious in Fig. 8 in comparison to Fig. 7. In Fig. 9, the speed of response has further improved with the capacitance value increased from C = 70 μF to C = 90 μF keeping the rotor speed constant to 314 rad/sec but with larger transient amplitudes. Thus, it is experimentally verified that the transients during the self-excitation process depend largely on the capacitance value used and magnetizing inductance value of the induction generator. The speed of response improves with an increase in the capacitance value but that increases the transient amplitudes. Thus, the knowledge of all possible solutions in the search region of capacitance value and magnetizing inductance value permits the trade-offs between the specifications of transient characteristics of the induction generator. Fig. 7 Output voltage: Capacitance value=70 μF, Rotor speed =301rad/sec, X-Axis:1small division = 0.5sec.,Y-Axis: 1small division = 5*20 volt Fig. 8 Output voltage: Capacitance value=70 μF, Rotor speed =314rad/sec, X-Axis:1small division = 0.5sec., Y-Axis:1small division = 5*20 volt Fig. 9 Output voltage: Capacitance value=90 μF, Rotor speed =314rad/sec, X-Axis:1small division = 0.5 sec.,Y-Axis: 1 small division = 5*20 volt 1097 VI. CONCLUSION A direct method based on interval analysis of computation for solution of characteristics equations of induction generator has been proposed. This method predicts the range in which lies the magnetizing inductance value and capacitance value required to induce the self-excitation in an induction generator. The use of the interval of magnetizing inductance leads to an accurate prediction of whether or not selfexcitation will occur for various capacitance values and captures all possible solutions. The proposed method is fast and can directly be used for transient analysis investigation of voltage build-up. It also predicts the range of capacitance value in which there is no solution and so self-excitation will not occur. With the decrease in magnetizing inductance value, the terminal voltage increases, representing a stable operation. ACKNOWLEDGMENT The authors wish to thank Prof. P. S. V. Nataraj, IITBombay, for his help during the course of this work. List of symbols: Pm ech =M echanical pow er λ = T ip-speed ratio R = R adius of the w ind turbine V =W ind velocity Pel = Electrical pow er Ad =D isc area ρ =A ir density η =Efficiency Vt = T erm inal voltage R s , R r = Per phase stator and rotor (re ferred to stator) resistance L s , L r = Per phase stator and rotor (referred to stator) inductance. L m = P er phase m agnetizing inductance (referred to stator). C = Per phase term inal excitation capacitance iqs , i ds = S tator quadrature and direct axis currents iqr , i dr = R otor quadrature and direct axis currents v qs , v ds = S tator quadrature and direct axis voltage REFERENCES [1] M.G.Simoes, F. A. Farret, “Renewable energy systems design and analysis with induction generators”, CRC Press, New York, 2004. [2] E.S. Abdin and Wilson Xu, “Control design and dynamic performance analysis of a wind turbine induction generator unit”, IEEE Trans. EC, vol. 15, no. 1, pp. 91-96, March 2000. [3] C. Grantham, D. Sutanto, and B. Mismail,“Steady-state and transients analysis of self-excited induction generators”, Proc. IEE., vol. 136, pt. B, no. 2, pp. 61-68,. 1989. [4] N.R.Ullah and T. Thiringer, “Variable speed wind turbines for power system stability enhancement”, IEEE Trans. EC, vol. 22, no 1, pp. 5260, March 2007. [5] D. Seyoum, C. Grantham and F. Rahman, “The dynamics of an isolated self-excited induction generator driven by a wind turbine”, Proceedings of IECON, vol. 2, pp. 1364-69, Dec. 2001. [6] D. Seyoum, M.F. Rahman and C. Grantham, “Terminal voltage control of a wind turbine driven isolated induction generator using stator oriented field control”, IEEE Applied Power Electronics Conference (APEC), Florida, Feb. 9-13, 2003, pp. 846-852. [7] Li.Wang, Ching-Huei Lee, “A novel analysis on the performance of an isolated self-excited induction generator”, IEEE trans. EC, vol. 12, No. 2, pp. 109-117, June 1997. [8] R.E. Moore, “Methods and Application of Interval Analysis”, SIAM studies in Applied Mathematics, 1979. [9] O. Ojo, “Dynamics and system bifurcation in autonomous induction generators”, IEEE trans. Industrial Applications, vol. 31, no. 4, pp. 918924, Jul/Aug. 1995. [10] O. Ojo, “Minimum air gap flux linkage requirement for self-excitation in stand-alone induction generators”, IEEE trans. EC, vol. 10, no. 3, pp. 484-492, Sept. 1995. v qr , v dr = R otor quadrature and direct axis voltage ω r = Angular speeds of rotor ω = Angular speeds of synchronous, reference fram e p = D erivative w ith tim e β = Blade pitch angle Tel =Electrom agnetic torque Tm ech =M echanical torque λ ds , λ qs = Stator direct and quadrature axis flux linkage λ dr , λ qr = R otor direct and quadrature axis flux linkage L1 = Ls + Lm L2 = Lr + Lm 2 K = L1 L 2 − L m A = System m atrix Λ = G enerator param eter X = S tate variables 1098
© Copyright 2026 Paperzz