Single-phase wind-BES microgrid with seamless transition capability

A new control strategy for a wind-BES (battery energy storage) microgrid is presented, which controls the power transfer and the mode of operation seamlessly from an islanded mode (IM) to the grid connected mode (GCM) and vice versa. In GCM, the DC link voltage regulation, the load reactive power compensation and harmonics current mitigation are performed by the load side voltage source converter (LVSC) using a double sinusoidal signal integrator (DSSI) based control algorithm. The maximum power from the wind generation is extracted by operating the generator side voltage source converter (GSC). At the grid fault or grid disturbances, this microgrid operates in IM and LVSC is controlled in the voltage control mode. A synchronizing controller is utilized to control the operation of synchronization. The grid current and point of common coupling (PCC) voltage distor-tions are kept within the limit as per the IEEE-519-2014 and the IEEE-1547 standards. The system operation is demonstrated through test results at several conditions such as, changing wind speeds, varying loads and seamless transition.


INTRODUCTION
The contribution of renewable energy sources (RESs) in the power generation is increasing to meet the growing demand and to replace the existing conventional sources of energy, i.e. fossil fuels, which are towards depletion [1][2][3][4][5]. RESs are freely available in nature, and clean source of energy. The RESs with a battery energy storage (BES), forms a microgrid (MG), which works in an islanded mode (IM) and the grid connected mode (GCM) along with seamless transition with continuously supplying the loads without interruption [6]. A power interruption to the loads, may lead to damage, malfunctioning and important data loss etc. In RESs, depending on its output supply, i.e. AC or DC, one or more power converters are required to interface the MG to the utility grid. Along-with the power generation, the MG has the capabilities of supplying an active power to the utility grid. It improves the power quality (PQ) in terms of active filtering and compensates the reactive power demand of the load [7,8].
The wind energy for the power generation is gaining popularity due to continuous improvement in the generator and power converter technology. The same is encouraged by the This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2020 The Authors. IET Power Electronics published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology world leaders towards the future of green and clean energy. Due to this, the injection of wind-generated energy to the grid is increasing exponentially. The large penetration of RESs into the grid raises the concerns of stability and power quality in the existing structure. These issues are addressed by the MG by providing uninterruptible supply to the non-linear loads, while improving the power quality of the utility grid. Apart from this benefit offered by the MG, it has several issues to be addressed, which are (1) control and its performance, (2) power management and performance at highly non-linear loads and, (3) seamless transition capability. These restrict the integration of MG to the distribution network. MGs, while operating in GCM, have less stability and operational issues [9]. However, these are major concerns for IM. Therefore, MG must have capabilities of switching the mode of operation from IM to GCM, i.e. from the voltage control mode (VCM) to the current control mode (CCM). However, the transition should be seamless, i.e. it should not interrupt the load power. The BES with the wind generation, reduces the variability of the microgrid [10]. To smooth power fluctuations due to intermittent nature of wind with the proper control of BES is reported in [11]. Power quality and stability improvement of the wind-BES microgrid supported by STATCOM is also presented. It requires additional circuit for STATCOM [12], which increases the complexity and cost of the system. In this work, the load side converter itself acts as DSTATCOM, which supplies the reactive power demand and harmonics and improves the power quality of the microgrid in the grid-connected mode of operation.
The MG utilizing wind energy for power generation, has to deal with an intermittent nature of the energy source, i.e. the behaviour of the wind energy is dynamic. Therefore, it is required to control two parameters in the wind turbine driven system, which are (1) maximum power point (MPP) tracking of the wind power and, (2) instantaneous torque produced by the generator. The MPP is achieved by varying the generator speed by a search process. Several techniques are reported in the literature to obtain MPP for harvesting the optimal power [13][14][15]. Some of them are listed as tip speed ratio, non-linear state space, power signal feedback, fuzzy logic, neural network and perturb and observe (P&O). Due to its independency on the system parameters and simple implementation, a P&O technique is found suitable for this application. The sensorless implementation reduces the complexity, while improving the overall performance of the system. The wind generator is controlled using the vector control (VC) method based on the information of the rotor speed and position. These two can be estimated by using with or without sensor-based scheme. However, sensorless scheme improves the reliability of the system and reduces its cost.
The extensive use of power converters in MGs as interfacing device and power transfer, along-with non-linear loads connected in the system, leads to poor power factor (pf), grid current harmonics, PCC voltage distortion and disturbance in frequency due to changing loads. To mitigate these PQ issues, several control schemes are reported in the literature [16,17]. Some of these are, an adaptive vectorial filter [18], delayed signal cancellation [19], frequency adaptive moving average filter [20], and multilayer resonator notch frequency control [21]. These techniques are either complex or use multiple blocks for their operation, which result in an increase in the computational complexity and affect the performance of the system. Therefore, the control must have simple structure and strong harmonics rejection within a relatively broad-frequency spectrum. A double sinusoidal signal integrator (DSSI) based control is presented for the mitigation of grid current harmonics. It is used to estimate fundamental quadrature component of the grid voltage and the load current. To trace the phase and frequency of the grid and the PCC (point of common coupling) voltage, an SOGI-based PLL technique is utilized. The MG performance is observed at varying wind speeds, varying loads and transition from GCM to IM and vice versa. The significance of the work is given as follows.
(i) This microgrid harvests the optimal power from the available wind energy by regulating the rotational speed of the PMSG using the sensorless vector control, which minimizes the cost of the system with the elimination of mechanical sensor. Moreover, the accuracy of the system is improved, and less maintenance is required. The cost and size of the system required are also reduced. (ii) The active power is transferred to the grid and a unity power factor is maintained, which improves the power quality of the grid. (iii) The microgrid utilizes SOGI-PLL [22] based technique to evaluate the phase angle and frequency of the grid voltage and the load voltage. It effectively traces the phase angle and frequency of the grid and the PCC voltages for seamless transition. (iv) It is shown through test results that the system changes the mode of operation seamlessly without interrupting the loads. The BES connected in the system supports the load by exchanging the energy by charging and discharging, while maintaining the DC link voltage at predefined value. (v) The change of mode of operation is smooth and fast, which makes the system reliable. The organization of the paper is as follows. Section 2 presents the architecture of the system under study. In Section 3, a brief of DSSI is presented and also control of GSC (generator side converter) and LVSC (load side voltage source converter) are discussed. Test results are presented in Section 4. In the end, Section 5 presents the conclusion of it. Figure 1(a) illustrates a wind-BES microgrid, which consists of a wind turbine, a permanent magnet synchronous generator (PMSG), two VSCs, a DC link capacitor (C dc ), a bidirectional converter (BDC), the grid, the BES and non-linear loads.

SYSTEM ARCHITECTURE
The converter connected to PMSG is termed as GSC, which ensures the optimal power harvesting from the available wind by controlling the PMSG speed as per the MPPT algorithm. The LVSC operates in the grid connected mode (GCM) and an islanded mode (IM) depending on the grid condition. It also provides seamless transition under transfer of mode of operation. These converters are connected on a common DC link (V dc ). A RC filter (R f , C f ) is used for mitigation switching ripples and an interfacing inductor (L i ) is utilized for smoothening of the LVSC current. An STS switch is utilized to connect or disconnect the microgrid from the utility grid depending on the grid condition. A power management scheme is presented in Figure 1(b). The battery is not utilized for power exchange; it , the control of wind generator and the battery are decided. It can be divided into following parts, which are, (1) when SOC min < SOC b < SOC max , the control is further divided into two parts, (a) if P PV > P L , the battery charges with the excess power and PMSG is operated to harvest optimal power, (b) if P PV < P L , the battery discharges to support the load, while PMSG is operated to harvest an optimal power. (2) If case 1 is not met, the control is further divided into two parts, (a) if SOC b ≤ SOC min , the load shedding is performed and (b) if SOC b ≥ SOC max , i.e. the battery is fully charged, then the PMSG is operated in off MPPT mode to generate the required power only.

CONTROL ALGORITHMS
The control of the MG is divided into two sections: one to extract the maximum power from the wind generation by the GSC control and other is the LVSC control. The LVSC control is further divided into two parts, i.e. GCM and IM. The detailed discussion of controls are given here.

Maximum power extraction
The P&O based algorithm is used to track MPP at varying winds [15]. For its operation, it does not depend on the device parameters. Under diverse conditions, the P&O algorithm performance is satisfactory. The output of MPPT algorithm implemented, is ω ref . The governing equations are as follows.
where, Δω is the difference between detected and previously noted PMSG speed. Figure 3(a) illustrates the flow diagram of the MPPT algorithm.

Double sinusoidal signal integrator (DSSI) based control algorithm
A dual sinusoidal signal integrator (DSSI) is used to extract fundamental components of the non-linear load current, which are 90 • phase apart from each other [23]. Figure 2(a) illustrates a schematics of DSSI. i Lα and i Lβ are fundamental in-phase and quadrature fundamental components of i L . The transfer functions of DSSI are given as [23], where, k 1 and ω 0 are gain and estimated grid frequency, respectively.
The amplitude-frequency characteristics of TF α (s) and TF β (s) are given as [23], It can be seen from Equation (3), that the amplitudefrequency characteristics of quadrature components are same, i.e. it provides same attenuation for each harmonic component in i Lα and i Lβ . By selecting the appropriate value of k 1 , the DSSI outputs desired fundamental quadrature components. Figure 2(b) illustrates the magnitude-frequency characteristics of TF α (s) and TF β (s) at different values of k 1 . From Figure 2(b), one can see that the low value of k 1 makes the DSSI has a better filtering capability, but a high one of k 1 does not. On the basis of these plots, an optimal value of k 1 can be 1.414, as this value makes the dynamic response fast and well-damped; meanwhile, it makes the DSSI has good filtering capability. Furthermore, it should be worth noting that a gain k 1 = 1.414 implies a damping factor ξ for the second-order system of Equation (2) equals to 0.707, which results in an optimal relation-

3.3
Comparative analysis Figure 2(c) illustrates a comparative analysis of between this DSSI-based filter and an SOGI filter. A signal consisting of fundamental, third, fifth and seventh harmonics components, is processed through DSSI-and SOGI-based filtering techniques.  Figure 3(c) depicts the control block of GSC. The GSC is controlled using sensorless vector control to harvest an optimum energy from the available wind energy. A back-EMF (electromotive force) based technique is utilized to estimate the PMSG rotor speed (ω est ) and position (ϴ est ). The detailed block diagram of the estimator is depicted in Figure 3(b). The reference rotor speed (ω ref ) is estimated by the P&O based MPPT technique by processing ω est and the wind power (P w ). The difference between ω ref and ω est , is processed through the PI controller to estimate reference q-axis current (I q ). The d-axis current (I d ) is set to zero, which improves the system dynamics. The generator reference currents (i ga * , i gb * , i gc * ) are estimated by transforming I q , I d by using dq-abc transformation. The reference and sensed currents are processed to generate GSC gating pulses through the hysteresis controller.

GSC control
The effectiveness of the used estimator is presented in Figure 3(d). The wind speed is changed from 10 to 12 m/s. It is observed that during the transition, the estimator speed output is smooth, but there is sudden change in the sensed speed at the transition point, which is clearly visible in Figure 3(d). Otherwise, at every point, the sensed speed and estimated speed are nearly equal. The similar behaviour is observed in the sensed and estimated rotor position. This analysis validates the estimator accuracy.

LVSC control
The LVSC control is divided into three parts, (1) GCM, (2) IM and (3) synchronizing control. In GCM, LVSC is controlled to supply the power generated from the wind energy, while mitigates the grid current harmonics, compensates the reactive power demand of loads and improves the pf of the grid. In IM, the LVSC is controlled to generate the PCC voltage (v L ), while meeting the load demand. The synchronizing control is used to synchronize the PCC voltage and the frequency to the grid voltage (v s ) and frequency (f s ). These controls are given in following subsections.

LVSC control in grid connected mode
In GCM, the LVSC is operated to achieve multiple objectives, such as grid current harmonics mitigation, maintaining unity pf and supplying reactive power to meet the load demand. Figure 4(a) illustrates the LVSC control in GCM and IM. The governing equations, while operating in GCM are given here. The 90 • shifted voltage signals (v sα , v sβ ) are obtained by processing the grid voltage (v s ) by DSSI. The in-phase and quadrature unit templates (u v and u q ) and peak amplitude (V sp ) are obtained as, The wind feed forward term (G PV ) and LVSC loss component (G loss ) are evaluated as, To determine weight (G L ) for i L , the in-phase portion, i Lα is sampled and hold at every 0th place of u q . The magnitude of reference grid current (G sw ) is valued as per the equation, which is given as, The reference grid current is evaluated as, The LVSC switching pulses are produced by processing i s and i s_ref through the current controller.

Synchronizing controller
The synchronizing controller monitors the conditions of the grid voltage (v s ) and PCC voltage (v L ). Depending on the outputs of amplitude checking and angle error checking block, it generates '1' or '0', i.e. to close the STS switch to synchronize with the grid or to open the switch to isolate the MG from the utility grid, respectively as depicted in Figure 4(c). The phase angles of v s and v L are estimated using the SOGI-based PLL.
As the grid appears into the system, the phase angle matching controller starts updating the phase of v L as depicted in Figure 4(b). When the phase and voltage conditions are met, it sets S to '1', which closes the STS switch and changes the LVSC control from VCM to CCM. Otherwise, the signal S is set to zero, which isolates the MG from the utility grid and changes the LVSC operation from CCM to VCM. The DC link voltage is maintained at constant value by the bidirectional converter. Figure 4(a) illustrates the bidirectional converter control. The converter is operated to maintain the DC link voltage (V dc ), when the MG is operating in IM. It operates in buck mode to absorb the excess power and in boost mode to supply the deficit power to meet the load demand. The governing equation is given as follows. where, I b * and I b are reference and sensed battery current. D is the duty ratio, which is after passing from a PWM generator, produces switching pulses for bidirectional converter.

RESULTS AND DISCUSSION
The performance of a wind-BES microgrid is demonstrated on a laboratory prototype. The MG is tested at numerous conditions such as sudden outage or recovery of the grid, its operation at changing wind speeds and loads. The applicability of DSSI-based current control, voltage control and power man-agement are shown through test results recorded on DSO and power analyser (Fluke). Figure 5(a) illustrates the photograph of the laboratory prototype. Figure 5 illustrates the system performance in an islanded condition. The system is operating at rated condition, i.e. the power generated by PMSG is 2.2 kW at V w = 12 m/s and the non-linear load connected here draws 1.05 kW as depicted in Figure 6. The LVSC operates in the voltage control mode. It Performance under mode change from islanded to grid con-

Islanded operation
is observed that the load voltage (v L ), load current (i L ) and the battery current (I bat ) are stable. The generator currents are sinusoidal. I bat is shown negative, i.e. the battery is charging and the bidirectional converter is operating in buck mode. The THD of v L is 3.0% and the voltage magnitude is stable at 220 V, 50 Hz as illustrated in Figure 6. V dc is stable at 400 V. Figure 7 illustrates the performance of the system at mode change from IM to GCM. To synchronize with the grid, a  switching pulses of LVSC, which is operating in VCM, as per the changes made by the phase angle matching controller. As the magnitudes of v s and v L are already within the limits. Once the phase angle error falls within the limit as set in the synchronizing controller, it sends '1' to the STS switch. This closes the switch to synchronize the microgrid with the grid as depicted in Figure 7(b). The grid current appears when the microgrid is synchronized and connected to the grid. The LVSC changes its control from the voltage control to the current control and starts feeding the grid, if the generated power is in excess. The load is supplied continuously without any interruption. The synchronization occurs seamlessly without any transient. Figure 8 illustrates performance of the system under sudden outage of the grid. Figure 8(a) shows that the PCC voltage is continuously available, i.e. the load is supplied continuously without any interruption as depicted in Figure 8(b). The moment the grid disappears, the synchronizing controller sends a signal '0', which changes the LVSC control from CCM to VCM. At the same time, the STS is opened and disconnects the microgrid from the utility grid. It can be seen from Figure 8(b), that the grid current becomes zero. The transfer of mode from GCM to IM is observed smooth.

Grid connected operation
The system performance is observed at various disturbances such as changing wind speeds and varying loads. The performance of the DSSI-based control is found satisfactory. The detailed discussion is presented here under various disturbances.  Figure 9 illustrates the performance of system at normal operation. The system is running at rated condition. The DC link voltage is maintained stable at 400 V. It is observed that the battery current is at 0 A, i.e. it does not charge or discharge in GCM. The excess power after meeting the load demand is fed to the grid. The current and power supplied to the grid are 5.08 A and 1.12 kW, respectively. The THD of i s is 4.4% as depicted in Figure 10, which is well below 5% as per the IEEE 519-2014 standard. The load demands 1.07 kW and draws a current of  Figure 11. This shows the efficacy of the DSSI-based control. Figure 12 illustrates the performance at variable wind speeds V w . V w is changed from 12 to 7.2 m/s. The reduction in the wind speed results in the reduction in the rotor speed, generator currents and so the generated power from the available wind

FIGURE 14
Performance at no load energy. As the load is constant and the wind speed is reduced. The generated power is insufficient to feed the load, the remaining power is supplied by the grid. The changes in various parameters are shown in Figure 12. It is also observed that the grid current is in-phase opposition with v s , when V w is 12 m/s, i.e. the generated power is more than the load demand. The grid current is in-phase when the wind speed is reduced and the generated power is insufficient to meet the load demand, i.e. the grid feeds the remaining power to the load. The change in magnitude and phase of the grid current, are shown with the variation of wind speeds. Figure 13 illustrates the performance at varying loads. The load is turned off and turned on. The effect is seen in i s as it decreases when the load appears in the system. The grid voltage and the grid current are in phase. The battery parameters and wind parameters are stable. The total generated power (2.16 kW) is fed to the grid. i s supplied to the grid is 9.72 A, while THD of i s is 3.7% as depicted in Figure 14.

CONCLUSION
In this work, a thorough analysis of wind-BES based microgrid has been made in detail. The presented control has given satisfactory results for seamless transition at the grid synchronization or the grid outage. It has also been found that the load is supplied without any interruption. The P&O algorithm has effectively harvested the maximum power from the wind. Test results show the performance of the microgrid operation in GCM with several disturbances such as changing wind speeds and varying loads, and found acceptable.