In the realm of renewable energy integration, solar power has emerged as a pivotal contributor to global electricity generation. As large-scale photovoltaic (PV) plants become increasingly prevalent, their interaction with the grid poses significant challenges to power system stability. Among these, the ability of solar inverters to remain connected during grid faults—known as low voltage ride-through (LVRT)—is critical. In this article, I present a comprehensive approach to achieving LVRT in high-power solar inverters through DC bus voltage control, detailing the design, implementation, and experimental validation. The focus is on maintaining grid support during voltage sags while ensuring efficient maximum power point tracking (MPPT), with emphasis on the role of solar inverters in modern power networks.
The proliferation of solar inverters in utility-scale applications necessitates adherence to stringent grid codes. These codes mandate that solar inverters must not disconnect during temporary voltage dips, as sudden loss of generation can exacerbate grid instability. Instead, solar inverters are required to provide reactive current support to aid voltage recovery. The core challenge lies in balancing the dual objectives of power extraction from PV arrays and grid compliance during faults. Traditional MPPT methods, such as perturb-and-observe, often struggle under rapid environmental changes or grid disturbances, leading to DC link voltage collapse and reduced efficiency. To address this, I propose a control strategy centered on the DC bus voltage, which not only enhances MPPT robustness but also facilitates seamless transition into and out of LVRT modes. Throughout this discussion, the term “solar inverters” will be frequently referenced to underscore their central role in this technology.
Grid codes for solar inverters, such as those outlined in the Chinese standard Q/GDW 617-2011, specify LVRT requirements similar to those for wind turbines. As shown in the curve below, solar inverters must remain connected for voltage dips above a defined threshold and are permitted to disconnect only for deeper sags. During LVRT, solar inverters must inject reactive current proportional to the voltage deviation, with a response time within 30 milliseconds. Post-fault, active power must recover at a rate of at least 30% of rated power per second. These demands place considerable stress on the control algorithms of solar inverters, necessitating precise voltage detection and current regulation.
To quantify the LVRT requirements, consider the following table summarizing key parameters for large-scale solar inverters:
| Parameter | Value | Description |
|---|---|---|
| Voltage Dip Threshold | 0.9 pu to 0.2 pu | Range where LVRT is mandatory |
| Reactive Current Support | $$I_q \geq 1.6 \times (0.9 – U_d^*/U_d) \times I_{\text{rated}}$$ | Required during dips, where \(U_d^*/U_d\) is per-unit voltage |
| Response Time | < 30 ms | Time to reach 90% of required reactive current |
| Power Recovery Rate | ≥ 30% \(P_{\text{rated}}\)/s | Minimum rate after fault clearance |
| DC Bus Voltage Range | 450 V to 820 V | Typical for high-power solar inverters |
The design of solar inverters for LVRT capability involves both hardware topology and control software. A common configuration for high-power solar inverters is a two-level voltage source inverter (VSI) with an LC filter, directly connected to the PV array without a DC-DC converter. This simplifies the structure but demands robust control to manage DC link dynamics. The inverter’s switching frequency is typically around 3 kHz, with a rated power of 500 kW or more. The control strategy employs a dual-loop proportional-integral (PI) regulator in the synchronous reference frame (dq-frame), where the outer loop regulates DC bus voltage and the inner loop controls grid current. This setup enables precise manipulation of active and reactive power outputs from solar inverters.
Mathematically, the current control loop can be expressed as follows. Let the grid voltage vector be represented in dq coordinates as \( \mathbf{V}_{gdq} = [V_{gd}, V_{gq}]^T \), and the inverter output current vector as \( \mathbf{I}_{dq} = [I_d, I_q]^T \). The reference values for current are derived from power commands: \( I_d^{\text{ref}} \) for active power and \( I_q^{\text{ref}} \) for reactive power. The PI controllers generate voltage references \( V_d^{\text{ref}} \) and \( V_q^{\text{ref}} \), which after decoupling feed the PWM modulator. The decoupling terms account for cross-coupling due to grid inductance and filter capacitance. The equations are:
$$ V_d^{\text{ref}} = \left( K_p + \frac{K_i}{s} \right) (I_d^{\text{ref}} – I_d) – \omega L I_q + \frac{1}{\omega C} I_q $$
$$ V_q^{\text{ref}} = \left( K_p + \frac{K_i}{s} \right) (I_q^{\text{ref}} – I_q) + \omega L I_d – \frac{1}{\omega C} I_d $$
where \( \omega \) is the grid angular frequency, \( L \) is filter inductance, \( C \) is filter capacitance, and \( K_p \), \( K_i \) are PI gains. This control framework is standard in solar inverters, but its adaptation for LVRT requires modifications in the reference generation.
Maximum power point tracking is fundamental to the efficiency of solar inverters. Conventional MPPT methods perturb the current reference \( I_d^{\text{ref}} \) directly, but this can lead to instability during fast irradiation changes. Instead, I utilize a DC bus voltage-based MPPT strategy. The PV array characteristic exhibits a non-linear power-voltage (P-V) curve with a distinct maximum power point (MPP). By perturbing the DC voltage reference \( V_{dc}^{\text{ref}} \) and observing the resulting power change, the algorithm locates the MPP. The power from the PV array is given by:
$$ P_{pv} = V_{dc} \cdot I_{pv} $$
where \( I_{pv} \) is the array current. The relationship between voltage and power near the MPP can be approximated by a quadratic function, allowing for gradient-based adjustment. The update rule for \( V_{dc}^{\text{ref}} \) is:
$$ V_{dc}^{\text{ref}}(k+1) = V_{dc}^{\text{ref}}(k) + \Delta V \cdot \text{sign}\left( \frac{P_{pv}(k) – P_{pv}(k-1)}{V_{dc}^{\text{ref}}(k) – V_{dc}^{\text{ref}}(k-1)} \right) $$
Here, \( \Delta V \) is a small perturbation step. This method inherently prevents DC link overvoltage during grid faults, as the voltage reference is bounded, and it ensures smooth power transitions—a key advantage for solar inverters in LVRT scenarios.
When a voltage sag occurs, solar inverters must rapidly detect the severity and type of dip. For balanced sags, the positive-sequence voltage component \( V_{gd}^+ \) suffices to determine depth. However, unbalanced sags—common in distribution networks—require per-phase analysis. I employ a virtual three-phase construction method for single-phase voltages. For instance, if phase A voltage \( v_a(t) \) is sampled, the instantaneous amplitude can be estimated by creating virtual phases B and C using time-shifted values:
$$ v_b^{\text{virt}}(t) = v_a\left(t – \frac{T}{6}\right) $$
$$ v_c^{\text{virt}}(t) = -v_a(t) – v_b^{\text{virt}}(t) $$
where \( T \) is the grid period. From these, the positive-sequence amplitude \( \hat{V}_a \) is computed via Clarke and Park transformations, enabling fast detection within one-sixth of a cycle. This technique is crucial for solar inverters to respond accurately to asymmetrical faults.
During LVRT, solar inverters prioritize reactive current injection. The required reactive current \( I_q^{\text{ref}} \) is calculated based on the voltage dip depth \( \Delta V = 0.9 – V_{gd}^+ / V_{\text{base}} \), as per the formula in the table. Simultaneously, the active current \( I_d^{\text{ref}} \) is reduced to limit total inverter current within safe bounds, typically 1.2 pu of rated current. The constraint is:
$$ I_d^{\text{ref}} \leq \sqrt{ I_{\text{max}}^2 – (I_q^{\text{ref}})^2 } $$
where \( I_{\text{max}} \) is the maximum allowable current for the semiconductor devices. This current management protects the solar inverters from thermal overload while fulfilling grid support duties.
Post-fault power recovery is another critical aspect for solar inverters. After voltage recovery, the DC bus voltage may be near the open-circuit value, and MPPT must resume quickly to meet the 30%/s recovery rate. To accelerate this, I implement a two-step process: first, a fixed DC voltage reference set at 85% of the open-circuit voltage rapidly brings the operating point close to the MPP; second, the voltage-perturbation MPPT refines the point. The recovery power trajectory can be modeled as:
$$ P_{\text{rec}}(t) = P_{\text{pre-fault}} \left(1 – e^{-t/\tau}\right) $$
with time constant \( \tau \) adjusted via PI gains in the voltage outer loop. This ensures compliance even under light-load conditions, where natural recovery might be sluggish.
To validate the proposed control strategy, I conducted experiments on a 500 kW solar inverter testbed. The setup includes a PV simulator to emulate array behavior and a grid emulator to create voltage sags. Key parameters of the solar inverter are listed below:
| Component | Specification |
|---|---|
| Rated Power | 500 kW |
| DC Voltage Range | 450–820 V |
| Grid Voltage (line-line) | 270 V |
| Switching Frequency | 3 kHz |
| Filter Inductance (L) | 0.5 mH |
| Filter Capacitance (C) | 50 μF |
| Control Sampling Rate | 10 kHz |
The MPPT performance was evaluated under steady and dynamic irradiation. The static MPPT efficiency exceeded 99.3%, and under rapid changes from 30% to 100% of rated power in 10 seconds, the dynamic efficiency remained above 98.8%. This demonstrates the resilience of the DC voltage-based approach in solar inverters. For LVRT tests, balanced dips to 50% voltage and unbalanced dips (single-phase to 80%) were applied. The solar inverters responded within 20 ms, meeting reactive current requirements. The power recovery rate was measured at over 90%/s, surpassing the standard. Waveforms captured during these tests show stable operation without disconnection.
The image above illustrates a modern solar inverter system with battery storage, highlighting the integration capabilities in real-world installations. Such systems benefit from advanced LVRT controls to ensure grid stability.
In terms of mathematical analysis, the stability of the control loop during transitions can be assessed using small-signal modeling. The transfer function from DC voltage reference to grid current involves the PV array impedance, which is non-linear. Linearizing around an operating point, the open-loop transfer function \( G(s) \) for the voltage controller is:
$$ G(s) = \frac{K_{pv} + K_{iv}/s}{1 + s T_{pv}} \cdot \frac{1}{C_{dc} s} \cdot \frac{1}{1 + s T_{inv}} $$
where \( K_{pv}, K_{iv} \) are PI gains for voltage, \( C_{dc} \) is DC link capacitance, \( T_{pv} \) is PV time constant, and \( T_{inv} \) is inverter delay. The phase margin should exceed 45° for robustness. For the current loop, bandwidth is set above 500 Hz to ensure fast tracking. These design considerations are essential for reliable performance of solar inverters under disturbances.
Furthermore, the impact of grid impedance on solar inverters during LVRT cannot be overlooked. The voltage at the point of common coupling (PCC) may differ from the ideal grid due to line impedances. The control algorithm includes a feedforward term based on estimated grid impedance \( Z_g = R_g + jX_g \) to compensate for this. The modified voltage reference becomes:
$$ V_d^{\text{ref}}’ = V_d^{\text{ref}} + I_d R_g – I_q X_g $$
$$ V_q^{\text{ref}}’ = V_q^{\text{ref}} + I_q R_g + I_d X_g $$
This enhancement improves the accuracy of reactive current injection, making solar inverters more effective in weak grid conditions.
To summarize the LVRT implementation, the following flowchart outlines the decision process in solar inverters:
| Step | Action | Control Outcome |
|---|---|---|
| 1 | Measure grid voltages (per phase) | Detect sag depth and type |
| 2 | If sag > threshold, compute \( I_q^{\text{ref}} \) | Reactive current command generated |
| 3 | Limit \( I_d^{\text{ref}} \) based on \( I_{\text{max}} \) | Active current reduced for safety |
| 4 | Switch to LVRT control mode | DC voltage reference held constant |
| 5 | Upon voltage recovery, ramp \( V_{dc}^{\text{ref}} \) | Power restored at ≥ 30%/s rate |
| 6 | Resume voltage-perturbation MPPT | Efficient operation regained |
In conclusion, the DC bus voltage control method for solar inverters offers a robust solution for low voltage ride-through. By integrating MPPT and LVRT functions through voltage reference manipulation, solar inverters can maintain grid connection during faults, provide requisite reactive support, and recover power rapidly afterwards. The use of virtual phase construction enables precise handling of unbalanced sags, a common grid anomaly. Experimental results on a 500 kW platform confirm the effectiveness, with compliance to standards and high efficiency. As solar penetration grows, such advanced controls will be indispensable for the reliability of power systems. Future work may explore adaptation to higher power levels or integration with energy storage, further enhancing the versatility of solar inverters.
The overarching theme is that solar inverters are not merely power converters but active grid participants. Their design must evolve to meet dynamic grid demands, and the proposed strategy represents a step forward in this direction. Through continuous innovation, solar inverters will play a pivotal role in the sustainable energy landscape, enabling greater renewable integration without compromising stability.