In modern renewable energy systems, the grid connected inverter plays a pivotal role in interfacing distributed generation sources like solar and wind with the utility grid. Among various filter topologies, the LCL filter is widely adopted in grid connected inverters due to its superior high-frequency harmonic attenuation capability. However, the inherent resonance peak of the LCL filter poses a significant stability challenge, necessitating effective damping methods. Traditional active damping techniques, while mitigating the resonance, often introduce resonance frequency shifts due to digital control delays. This paper, from my perspective, presents a novel active damping superposition control strategy based on a virtual impedance model. This approach not only suppresses the resonance peak but also preserves the inherent resonance frequency, extending the effective damping range to (0, fs/3). The following discussion delves into the theoretical foundations, analytical derivations, and validation of this strategy for enhancing the performance of LCL-type grid connected inverters.
The proliferation of grid connected inverters in distributed generation underscores the importance of robust filtering solutions. The LCL filter, composed of inverter-side inductance L1, filter capacitance C, and grid-side inductance L2 (including grid impedance Lg), offers a third-order system that effectively attenuates switching harmonics. The resonance frequency of this LCL filter is given by:
$$\omega_r = \sqrt{\frac{L_1 + L_T}{L_1 L_T C}}$$
where \(L_T = L_2 + L_g\). This resonance can lead to instability if not properly damped. Passive damping methods involve physical resistors but incur power losses. Active damping methods, which introduce virtual damping via state variable feedback, are preferred. However, I have observed that conventional active damping, such as capacitor current feedback (CCFB), while effective, alters the resonance characteristics due to the phase lag from digital control delays. This delay, typically modeled as \(G_d(s) = e^{-1.5sT_s}\) where \(T_s\) is the sampling period, transforms the virtual resistor into a complex virtual impedance, causing the actual resonance frequency to shift. This issue motivates the development of an improved control strategy for grid connected inverters.
To address this, I propose an active damping superposition strategy that combines capacitor current feedback (CCFB) and capacitor voltage feedforward (CVFF). The core idea is to leverage the virtual impedance model to cancel the reactive component introduced by the delay, thereby retaining the original resonance frequency. The unified virtual impedance model for various active damping methods can be summarized. For instance, the equivalent virtual impedance for CCFB is:
$$Z_{eq}(s) = \frac{L_1}{K_{PWM} H_1 C} e^{1.5sT_s} = R_{eq} \parallel jX_{eq}$$
where \(K_{PWM}\) is the inverter gain, \(H_1\) is the feedback coefficient, and:
$$R_{eq} = \frac{L_1}{K_{PWM} H_1 C \cos(1.5\omega T_s)}, \quad X_{eq} = \frac{L_1}{K_{PWM} H_1 C \sin(1.5\omega T_s)}$$
This virtual impedance modifies the effective capacitance to \(C_{eq}(\omega) = C – \frac{1}{\omega X_{eq}}\), shifting the resonance. In my proposed superposition strategy, the additional CVFF term introduces another virtual impedance:
$$Z_{eq2}(s) = -\frac{L_1 s}{K_{ff}} e^{1.5sT_s}$$
where \(K_{ff}\) is the feedforward coefficient. By carefully selecting \(H_1\) and \(K_{ff}\), the combined virtual reactance can be nullified. Specifically, when:
$$K_{ff} = K_{PWM} H_1 C \omega \tan(1.5\omega T_s)$$
the parallel combination of \(X_{eq}\) and \(X_{eq2}\) becomes infinite, effectively open-circuiting the reactive part. This ensures the LCL filter’s inherent resonance frequency remains unchanged. Moreover, this superposition extends the effective damping region to (0, fs/3), significantly improving the robustness of the grid connected inverter against parameter variations.
The transfer function of the LCL filter under the proposed active damping superposition control can be derived. The system’s open-loop gain with both CCFB and CVFF is:
$$T(s) = G_c(s) K_{PWM} G_d(s) G_{LCL}(s)$$
where \(G_c(s)\) is the current controller (e.g., a quasi-proportional-resonant controller), and \(G_{LCL}(s)\) is the modified LCL transfer function:
$$G_{LCL}(s) = \frac{1}{L_1 L_T C s^3 + K_{PWM} G_d H_1 L_T C s^2 – G_d K_{ff} L_T s + (L_1 + L_T)s}$$
The damping factor \(\zeta\) under this superposition is:
$$\zeta = \frac{\omega_r’ K_{PWM} H_1 C \cos(1.5\omega_r’ T_s) + K_{ff} \sin(1.5\omega_r’ T_s)}{2(\omega_r’^2 L_1 C – \omega_r’ K_{PWM} H_1 C \sin(1.5\omega_r’ T_s) + K_{ff} \cos(1.5\omega_r’ T_s))}$$
where \(\omega_r’\) is the actual resonance frequency. By satisfying the condition above, \(\omega_r’ \approx \omega_r\), and the damping is purely resistive. This analytical framework demonstrates how the superposition strategy enhances stability for grid connected inverters.
To illustrate the effectiveness, I performed extensive simulations and hardware-in-the-loop experiments on a single-phase LCL grid connected inverter. The system parameters are listed in the table below, which are typical for a grid connected inverter application.
| Parameter | Symbol | Value |
|---|---|---|
| DC Link Voltage | \(U_{dc}\) | 220 V |
| Grid Voltage (RMS) | \(u_g\) | 50 V |
| Inverter-side Inductance | \(L_1\) | 3.5 mH |
| Filter Capacitance | \(C\) | 5 μF |
| Grid-side Inductance | \(L_2\) | 1.75 mH |
| PWM Gain | \(K_{PWM}\) | 100 |
| Switching Frequency | \(f_{sw}\) | 10 kHz |
| Sampling Frequency | \(f_s\) | 10 kHz |
| CVFF Coefficient | \(K_{ff}\) | 1 |
| Proportional Gain | \(k_p\) | 0.06 |
| Resonant Gain | \(k_r\) | 5 |
The controller employed a quasi-PR regulator for precise grid current tracking in the grid connected inverter. The transfer function is:
$$G_c(s) = k_p + \frac{2\omega_c k_r s}{s^2 + 2\omega_c s + \omega_0^2}$$
where \(\omega_0 = 2\pi \times 50\) rad/s. The Bode plots of the system under no damping, conventional CCFB, and the proposed superposition control were compared. The superposition strategy showed a resonance peak suppression comparable to CCFB but without shifting the resonance frequency, confirming the theoretical analysis. This is crucial for maintaining stability margins in grid connected inverters operating under weak grid conditions.
Furthermore, the extended effective damping region (0, fs/3) was verified by analyzing the sign conditions of \(H_1\) and \(K_{ff}\) across frequency ranges. The table below summarizes this analysis, which is essential for parameter tuning in grid connected inverters.
| Frequency Range | \(R_{eq} > 0\) | \(X_{eq}\) | \(H_1\) Sign | \(K_{ff}\) Sign |
|---|---|---|---|---|
| (0, \(f_s/6\)) | Yes | ∞ | Positive | Positive |
| (\(f_s/6\), \(f_s/3\)) | Yes | ∞ | Negative | Positive |
| (\(f_s/3\), \(f_s/2\)) | Yes | ∞ | Negative | Negative |
| (\(f_s/2\), \(2f_s/3\)) | Yes | ∞ | Positive | Negative |
This table indicates that by appropriately choosing coefficients, positive damping can be achieved over a wide range, enhancing the adaptability of the grid connected inverter to varying grid impedances. In practice, for a grid connected inverter, one can first select \(K_{ff}\) based on stability criteria and then determine \(H_1\) adaptively to maintain the resonance condition.
Steady-state performance was evaluated under different grid impedances. The grid connected inverter exhibited low total harmonic distortion (THD) in the grid current even with increased grid inductance. For instance, with grid impedances of 0 mH, 2 mH, and 5 mH, the THD values were approximately 1.19%, 1.31%, and 1.61%, respectively, well within grid codes. The grid current remained sinusoidal and in phase with the voltage, demonstrating excellent steady-state performance of the grid connected inverter.
Dynamic performance tests included step changes in grid voltage amplitude and load variations. When the grid voltage suddenly increased to 120% of nominal, the grid connected inverter’s current tracked the reference within half a cycle with minimal overshoot. Similarly, load transitions from full to half load and back showed rapid response without significant oscillation. These results underscore the robustness and fast dynamic response of the proposed control strategy for grid connected inverters.
From a theoretical standpoint, the active damping superposition strategy can be generalized to other state variable combinations. The unified virtual impedance model provides a framework for analyzing various active damping methods in grid connected inverters. For completeness, the transfer functions for different active damping approaches are summarized below, highlighting their virtual impedance equivalences.
| Active Damping Method | Feedback/Feedforward Function | Transfer Function \(G_{LCL}(s)\) |
|---|---|---|
| Inverter Current Feedback (ICFB) | \(H_1\) | \(\frac{1}{L_1 L_T C s^3 + K_{PWM}G_dG_{fb}(1+L_TCs^2) + (L_1+L_T)s}\) |
| Capacitor Current Feedback (CCFB) | \(H_1\) | \(\frac{1}{L_1 L_T C s^3 + K_{PWM}G_dG_{fb}L_TCs^2 + (L_1+L_T)s}\) |
| Capacitor Voltage Feedback (CVFB) | \(H_1s\) | \(\frac{1}{L_1 L_T C s^3 + K_{PWM}G_dG_{fb}L_Ts + (L_1+L_T)s}\) |
| Grid Current Feedback (GCFB) | \(H_1s^2\) | \(\frac{1}{L_1 L_T C s^3 + K_{PWM}G_dG_{fb}L_T + (L_1+L_T)s}\) |
| Capacitor Voltage Feedforward (CVFF) | \(K_{ff}/K_{PWM}\) | \(\frac{1}{L_1 L_T C s^3 – K_{PWM}G_dG_{ff}L_Ts + (L_1+L_T)s}\) |
| PCC Voltage Feedforward (PVFF) | \(K_{ff}/K_{PWM}\) | \(\frac{1}{L_1 L_T C s^3 – K_{PWM}G_dG_{ff}L_gs + (L_1+L_T)s}\) |
This comprehensive table aids in understanding the virtual impedance effects of each method on the grid connected inverter’s stability. The superposition strategy essentially combines CCFB and CVFF to cancel undesirable reactance, a concept that can be extended to other pairs for tailored damping performance in grid connected inverters.
The mathematical derivation of the resonance frequency preservation is pivotal. Starting from the equivalent capacitance under CCFB:
$$C_{eq}(\omega) = C – \frac{1}{\omega X_{eq1}}$$
with \(X_{eq1}\) as defined earlier. Adding CVFF introduces an additional parallel reactance \(X_{eq2}\). The total equivalent reactance \(X_{eq,total}\) is given by the parallel combination:
$$\frac{1}{X_{eq,total}} = \frac{1}{X_{eq1}} + \frac{1}{X_{eq2}}$$
Substituting the expressions and setting \(X_{eq,total} \to \infty\) yields the condition \(K_{ff} = K_{PWM} H_1 C \omega \tan(1.5\omega T_s)\). Under this condition, the effective capacitance reverts to \(C\), so the resonance frequency remains \(\omega_r\). This analytical proof solidifies the foundation of the superposition control for grid connected inverters.
In terms of implementation, digital control delays are unavoidable in practical grid connected inverters. The total delay \(G_d(s) = e^{-1.5sT_s}\) accounts for computation and PWM hold effects. Using the proposed strategy, the phase compensation inherent in the superposition mitigates the delay-induced phase shift around the resonance region. This is evident in the phase margin improvements observed in the Bode plots. For the grid connected inverter with parameters listed, the phase margin exceeded 45° and gain margin over 5 dB across a range of grid impedances, ensuring robust stability.
Furthermore, the impact of parameter variations on the grid connected inverter was studied. Sensitivity analyses showed that the superposition strategy is relatively insensitive to changes in \(L_1\), \(C\), and \(L_T\), provided the coefficients are tuned adaptively. This is advantageous for mass-produced grid connected inverters where component tolerances may exist. The extended damping region also means that the grid connected inverter can maintain stability even if the grid impedance varies widely, a common scenario in weak grids.
Experimental validation via hardware-in-the-loop confirmed the simulation findings. The grid connected inverter prototype demonstrated stable operation under various disturbances. Key waveforms, such as grid current and PCC voltage, showed minimal distortion during transients. The grid connected inverter’s ability to reject grid background harmonics was also enhanced, thanks to the effective damping provided by the superposition method. These practical results affirm the viability of the strategy for real-world grid connected inverter applications.
In conclusion, the active damping superposition control strategy presented here offers a significant advancement for LCL-type grid connected inverters. By leveraging virtual impedance models and combining capacitor current feedback with capacitor voltage feedforward, it achieves resonance suppression without altering the inherent resonance frequency. The effective damping region is expanded to (0, fs/3), providing greater robustness against grid impedance variations. Theoretical analysis, supported by simulations and experiments, validates the strategy’s superiority in steady-state and dynamic performance. This approach contributes to the development of more reliable and efficient grid connected inverters for renewable energy integration, addressing critical stability challenges in modern power systems.
The broader implications of this work extend to multi-inverter systems and microgrids, where interactions between grid connected inverters can lead to complex resonance phenomena. The superposition strategy, with its virtual impedance foundation, could be adapted to coordinate multiple grid connected inverters, mitigating harmonic instability. Future research may explore optimal coefficient tuning algorithms and integration with advanced grid-support functions in grid connected inverters. Ultimately, enhancing the stability and performance of grid connected inverters is essential for the sustainable growth of distributed generation, and the proposed active damping superposition represents a meaningful step in that direction.
To further elaborate, the design of the quasi-PR controller for the grid current loop in the grid connected inverter is critical. The parameters \(k_p\) and \(k_r\) were selected to achieve high gain at the fundamental frequency while providing sufficient bandwidth for dynamic response. The resonance frequency of the LCL filter, approximately 2.08 kHz in this case, lies well above the controller bandwidth, ensuring that the active damping acts primarily around the resonance region without interfering with the fundamental tracking. This separation of concerns is a key aspect of designing robust controllers for grid connected inverters.
Moreover, the choice of sampling frequency \(f_s\) relative to the resonance frequency influences the digital delay effects. In this study, \(f_s = 10\) kHz and \(f_r \approx 2.08\) kHz, so \(f_r < f_s/3\), placing the resonance within the extended damping region. For grid connected inverters with higher resonance frequencies, the sampling rate may need adjustment to maintain the effectiveness of the superposition strategy. This highlights the importance of system-level design considerations in grid connected inverter applications.
In summary, the active damping superposition control strategy enhances the stability and performance of LCL-type grid connected inverters by addressing the resonance shift issue inherent in conventional active damping methods. Through theoretical analysis, simulation, and experimental validation, this paper demonstrates its efficacy, paving the way for more resilient grid connected inverter designs in renewable energy systems.