In the context of rapid development and widespread application of new energy sources, an increasing number of power electronic devices are being integrated into the grid. However, this trend introduces new challenges, one of which is the presence of high-order harmonics in the output voltage of grid connected inverters, adversely affecting the quality of grid currents. The key to mitigating this issue lies in selecting appropriate filters to attenuate these harmonics. Compared to traditional single-inductor L-type filters, the inductor-capacitor-inductor (LCL) type filter, due to the introduction of a capacitor, provides an additional path for high-frequency harmonics, resulting in significantly reduced volume and weight under equivalent high-frequency filtering performance. Nonetheless, the parameter calculation for LCL filters becomes more complex as the number of parameters increases from one to three. In engineering design, traditional LCL filter parameter calculation must consider not only factors such as ripple current magnitude on the filter, grid current quality, and power factor but also balance cost, size, and weight. Therefore, this article aims to propose a reasonable and rapid calculation method for LCL filter parameters under basic constraints, simplifying the practical engineering calculation process.
Currently, numerous studies have investigated LCL parameter calculation. For instance, some literature analyzes mathematical formulas for harmonic content in inverter output voltages under various topologies and modulation methods, focusing on deriving formulas for inverter-side inductance. Others explore NPC-type three-level inverters but find that different modulation methods lead to variations in calculated parameters, particularly affecting inverter-side parameter calculation. Some optimize based on reactive power concentrated on capacitance, but single-parameter optimization inevitably leads to other parameters becoming larger or less ideal. Multi-boundary constraint design methods provide calculation ranges and derivation processes but are overly complex, potentially requiring multiple adjustments to output target parameters. Some focus primarily on grid-side inductance calculation and provide basic control parameters for closed-loop design without optimizing other LCL parameters. Graph-based methods, while intuitive, complicate the design process and often overlook inverter-side inductor current ripple. Approaches combining control loops and delay issues impose additional restrictions on filter parameters, designing comprehensive methods including current controller and filter parameters. Weak grid condition-based methods, incorporating delay factors and undamped control, propose resonance frequency stability domains, allowing direct determination of resonance frequency and thus LCL parameters without verification but neglecting attenuation characteristics. Intelligent optimization algorithms, such as multi-objective optimization functions, optimize filter parameters by considering multiple constraints like component cost and grid current total harmonic distortion (THD) within traditional parameter calculation ranges. However, these methods essentially use optimization algorithms to replace manual search, failing to simplify the parameter calculation process in engineering. Moreover, without appropriate software, these methods may become relatively difficult to compute.
In summary, methods like trial-and-error, graphing, and intelligent optimization still lack simplicity and speed when calculating LCL filter parameters. To address this, this article proposes an approximate calculation method for LCL filter parameters. This method prioritizes high-frequency filtering of the LCL filter as the primary constraint, making parameter design more purposeful. After adjusting parameter calculation formulas, parameters can be calculated based solely on the resonance ratio, effectively reducing parameter design difficulty. Additionally, it is found that the error in approximate calculation is only related to the square of the resonance ratio, allowing parameter adjustment based on error magnitude.
The structure of a three-level LCL grid connected inverter is considered. The circuit topology includes a DC bus voltage, DC-side support capacitors, switching transistors forming a three-phase T-type inverter, and LCL filter sections composed of inverter-side inductors L1, filter capacitor C, and grid-side inductors L2. The grid voltages are denoted as Vga, Vgb, and Vgc. The inverter-side inductor currents are i1a, i1b, i1c, and the grid-side inductor currents are iga, igb, igc. The harmonic equivalent model of the LCL grid connected inverter is analyzed, deriving transfer functions from the inverter output voltage Vinv to the inverter-side current i1 and to the output grid current ig. The transfer functions are given by:
$$G_{i1}(s) = \frac{i_1(s)}{V_{inv}(s)} = \frac{s^2 L_2 C + 1}{s^3 L_1 L_2 C + s L_T} = \frac{s^2 L_2 C + 1}{s^2 L_2 C + 1 + m} \cdot \frac{1}{s L_1}$$
and
$$G_{ig}(s) = \frac{i_g(s)}{V_{inv}(s)} = \frac{1}{s^3 L_1 L_2 C + s L_T} = \frac{1}{s L_T} \cdot \frac{\omega_{res}^2}{s^2 + \omega_{res}^2} = \frac{1}{s L_1 L_2 C} \cdot \frac{1}{s^2 + \omega_{res}^2}$$
where m = L2/L1 is the inductance ratio, ω_res is the resonance angular frequency, and L_T = L1 + L2 is the total inductance of the inverter-side and grid-side inductors. The resonance angular frequency is defined as:
$$\omega_{res} = \sqrt{\frac{L_T}{L_1 L_2 C}}$$
The frequency characteristics of these transfer functions show that in the low-frequency range near the fundamental frequency, they approximate 1/(s L_T), equivalent to an L-type filter with inductance L_T. In the high-frequency range near the switching frequency, the inverter-side current transfer function approximates 1/(s L1), while the grid-side current transfer function has a slope of -60 dB/dec. Since the LCL structure introduces resonance, damping is necessary. Active damping via state variable feedback is commonly used to simulate a virtual resistor, avoiding additional losses. The transfer function with virtual damping is:
$$G_{ig}(s) = \frac{i_g(s)}{V_{inv}(s)} = \frac{1}{s^3 L_1 L_2 C + s^2 L_1 L_2 / R + s L_T}$$
With appropriate damping, the frequency response can be segmented: the low-frequency segment dominated by L_T and the high-frequency segment approximated by 1/(s^3 L_1 L_2 C). This segmentation forms the basis for the approximate calculation method.
To design the LCL filter parameters, constraints from standards such as IEEE Std. 1547-2003 must be considered, which limit individual harmonic components and total harmonic distortion (THD) in grid currents. The priority of constraints is rearranged: first, ensure that harmonic components near the switching frequency in the grid current are below maximum limits; second, satisfy constraints on total inductance L_T, inverter-side inductance L1, capacitor C, and resonance frequency range; finally, additional constraints like cost and volume can refine parameter design.
Based on the segmentation approach, when suitable damping is applied, the high-frequency attenuation at the switching frequency ω_sw is critical. From the high-frequency approximation, the gain at ω_sw is:
$$G_{gh}(j\omega_{sw}) = \frac{i_g(j\omega_{sw})}{V_{inv}(j\omega_{sw})} = \frac{\lambda_h I_g}{V_{inv}(j\omega_{sw})} = \frac{\omega_{res}^2}{\omega_{sw}^3 L_T} = \frac{k^2}{\omega_{sw} L_T}$$
where λ_h is the ratio of the target harmonic current to the rated grid current, I_g is the maximum harmonic current, and k = ω_res / ω_sw is the resonance ratio. This equation shows that for fixed G_gh(jω_sw) and ω_sw, L_T depends only on k. Comparing LCL and L filters under the same switching frequency attenuation, the LCL filter allows for smaller total inductance. The capacitance C is related to the inductance split ratio r = L1 / L_T:
$$\lambda_{gh}(j\omega_{sw}) = \frac{1}{\omega_{sw}^3 L_1 L_2 C} = \frac{1}{\omega_{sw}^3 r (1 – r) L_T^2 C}$$
When L_T is sufficiently large and exceeds twice the minimum value of L1, the required C is smaller as r approaches 0.5.
Parameter limits must be calculated. The maximum total inductance L_T_max is constrained by power factor and DC bus voltage prevention of over-modulation:
$$L_{T\_max} = \frac{0.25 U_{dc}^2 – E_m^2}{\omega_0 I_m}$$
where U_dc is the DC bus voltage, E_m is the peak grid voltage, ω_0 is the grid fundamental angular frequency, and I_m is the peak grid current. The minimum inverter-side inductance L1_min is limited by ripple current to reduce switch current stress and losses:
$$L_{1\_min} = \frac{U_{dc} M_r}{4 \sqrt{3} \Delta I_{1\_max} f_{sw}}$$
where ΔI1_max is the maximum ripple current amplitude, typically 20% of the peak current, M_r is the modulation index, and f_sw is the switching frequency. The maximum filter capacitor C_max is constrained by reactive power, usually limited to 5% of the system active power:
$$C_{max} = \frac{\lambda_c P_0}{\omega_0 V_g^2}$$
where λ_c is the ratio of reactive power to active power, P_0 is the rated power, and V_g is the grid voltage.
Analyzing these limits, the minimum value of k is derived by combining the high-frequency gain equation and L1_min:
$$k > \frac{\pi \lambda_h}{2 \sqrt{6}} \cdot \frac{U_{dc} M_r}{\delta V_{inv}(j\omega_{sw})}$$
where δ is the ripple coefficient. This shows that k depends mainly on the modulation method and DC bus voltage. The minimum switching frequency is also derived:
$$f_{sw} > \frac{2 k^2 f_0}{\lambda_h} \cdot \frac{V_{inv}(j\omega_{sw})}{\sqrt{0.25 U_{dc}^2 – E_m^2}}$$
Both k_min and f_sw_min rely on the accuracy of L1_min and L_T_max calculations, which vary with topology and modulation methods.
Error analysis compares the exact transfer function G_g(s) with the high-frequency approximation G_gh(s). The error ratio is:
$$E(j\omega) = \frac{G_{gh}(j\omega)}{G_g(j\omega)} = 1 – \frac{\omega_{res}^2}{\omega^2}$$
At the switching frequency, the error is:
$$E(j\omega_{sw}) = 1 – k^2$$
Thus, the error decreases as k decreases. For small k, the approximate method yields parameters close to the exact ones. In practice, parameters may be slightly increased to compensate for damping imperfections.
The design flowchart for LCL filter parameters is as follows:
- Determine basic design requirements: rated power P, DC bus voltage U_dc, grid voltage V_g, switching frequency f_sw, and modulation method.
- Simulate to obtain V_inv(jω_sw) harmonic content under design conditions. Select λ_h based on harmonic standards like IEEE Std. 1547-2003.
- Calculate constraint limits: L_T_max, L1_min, C_max using the formulas above.
- Compute k_min and choose k between k_min and 0.5.
- Calculate L_T using the high-frequency gain equation: L_T = k^2 / (ω_sw G_gh(jω_sw)).
- If L_T < 2 L1_min, set L1 = L1_min and L2 = L_T – L1_min. Otherwise, set L1 = L2 = 0.5 L_T.
- Calculate C from the resonance frequency formula: C = L_T / (ω_res^2 L1 L2). If C ≥ C_max, adjust k or increase f_sw. If C < C_max, output the LCL parameters.
This method simplifies the design by focusing on high-frequency attenuation and using the resonance ratio as a key parameter. It reduces trial-and-error iterations and enhances engineering applicability.
To validate the method, simulation and experimental analysis are conducted. For a T-type three-level grid connected inverter with U_dc = 800 V, grid line voltage 380 V, frequency 50 Hz, switching frequency 50 kHz, and SPWM modulation, parameters are designed using the flowchart. With λ_h = 0.22%, k = 0.32, ripple coefficient δ = 22%, and modulation index M_r = 0.7, the calculated parameters are L_T ≈ 356 μH, L1_min ≈ 340 μH, and C ≈ 6.5 μF. Simulation in PLECS software shows that the inverter-side inductor current has a ripple amplitude of about 2.7 A, which is 12.5% of the rated current peak, below the 20% limit. The grid-side current has a target harmonic amplitude of about 37 mA, which is 0.24% of the grid current RMS value, slightly above the designed 0.22% but within the 0.3% limit. The THD is about 0.7%, satisfying grid standards. Variations in inductance split ratio and capacitance confirm that the high-frequency attenuation remains consistent, demonstrating the method’s robustness.
Experimental tests on a prototype yield similar results. The inverter-side inductor current has a maximum ripple of 3 A, and the grid-side current has a harmonic amplitude of about 50 mA at 50 kHz, which is 0.32% of the grid current, with a THD of 2.5%. Minor discrepancies from simulation are attributed to parameter tolerances and noise, but the overall performance meets grid connection requirements. The results validate the approximate calculation method for LCL filter parameters in grid connected inverters.
In conclusion, this article addresses the complexity of LCL filter parameter design by proposing an approximate calculation method based on active damping. Prioritizing high-frequency attenuation as the primary constraint, the method simplifies the design process through resonance ratio adjustment. Formulas for parameter limits and error analysis are provided, and a flowchart guides the calculation. Simulation and experimental results confirm the method’s effectiveness in meeting grid standards while reducing design effort. This approach is particularly valuable for engineers designing grid connected inverters in renewable energy systems, offering a balance between performance and simplicity. Future work could extend the method to other topologies and modulation schemes, further enhancing its applicability in diverse grid connected inverter applications.
The integration of grid connected inverters into modern power systems is crucial for harnessing renewable energy. However, harmonic distortion remains a significant challenge. The LCL filter, with its superior high-frequency attenuation, is a preferred solution, but its parameter design often requires iterative processes. The approximate method presented here streamlines this by leveraging frequency-domain segmentation and active damping principles. By focusing on the resonance ratio, designers can quickly determine parameters that satisfy harmonic limits, reactive power constraints, and component sizing. This method not only accelerates the design cycle but also ensures reliable operation of grid connected inverters in compliance with international standards. As the demand for clean energy grows, such simplified design techniques will become increasingly important for deploying efficient and cost-effective grid connected inverter systems worldwide.
To further illustrate the parameter relationships, key formulas are summarized in the following table:
| Parameter | Formula | Description |
|---|---|---|
| Total Inductance L_T | $$L_T = \frac{k^2}{\omega_{sw} G_{gh}(j\omega_{sw})}$$ | Calculated from resonance ratio and high-frequency gain |
| Resonance Frequency ω_res | $$\omega_{res} = \sqrt{\frac{L_T}{L_1 L_2 C}}$$ | Defines the LCL filter resonance point |
| Inverter-side Inductance L1_min | $$L_{1\_min} = \frac{U_{dc} M_r}{4 \sqrt{3} \Delta I_{1\_max} f_{sw}}$$ | Minimum value based on ripple current limits |
| Grid-side Inductance L2 | $$L_2 = L_T – L_1$$ | Determined from total inductance and split ratio |
| Capacitance C | $$C = \frac{L_T}{\omega_{res}^2 L_1 L_2}$$ | Calculated to achieve desired resonance frequency |
| Resonance Ratio k | $$k = \frac{\omega_{res}}{\omega_{sw}}$$ | Key design parameter between 0.1 and 0.5 |
| Error at Switching Frequency | $$E(j\omega_{sw}) = 1 – k^2$$ | Approximation error relative to exact model |
This table encapsulates the core equations used in the approximate calculation method. Designers can refer to it for quick parameter estimation. Additionally, the method’s reliance on simulation for harmonic content of V_inv underscores the importance of accurate modeling in grid connected inverter design. By combining simulation tools with analytical formulas, the method bridges the gap between theoretical analysis and practical implementation.
In practice, grid connected inverters must operate under varying grid conditions, including impedance changes and voltage fluctuations. The active damping approach incorporated in this method enhances robustness against such variations. By feeding back capacitor current or other state variables, the virtual resistor effectively stabilizes the system without adding physical losses. This aligns with the trend towards more efficient and adaptive grid connected inverter controls. Furthermore, the approximate calculation method can be extended to multi-objective optimization, where parameters are tuned to minimize cost, size, or THD simultaneously. However, for initial design, the simplicity of the resonance ratio-based approach offers a significant advantage.
Another aspect to consider is the impact of switching frequency on filter design. Higher switching frequencies allow for smaller filter components but increase switching losses. The method here accommodates this by linking f_sw to k and L_T. Designers can trade off switching frequency and filter size based on application requirements. For instance, in high-power grid connected inverters, lower switching frequencies may be preferred to reduce losses, necessitating larger filters. The approximate method provides a framework to evaluate these trade-offs quickly.
The experimental validation involved a prototype grid connected inverter with the designed LCL filter. Measurements confirmed that harmonic emissions were within specified limits, demonstrating the method’s practical efficacy. While minor adjustments were needed due to component tolerances, the overall design process was straightforward. This highlights the method’s suitability for real-world engineering, where time and resource constraints are common. As grid codes become stricter, such efficient design methods will be essential for compliance.
In summary, the approximate calculation method for LCL filter parameters in grid connected inverters presented here offers a balanced approach to filter design. By prioritizing high-frequency attenuation and using the resonance ratio as a central parameter, it simplifies the complex task of LCL parameter selection. The method is supported by mathematical derivations, error analysis, and practical validation, making it a valuable tool for engineers. As the adoption of grid connected inverters continues to rise, such methods will play a key role in ensuring power quality and system stability. Future research could explore integration with digital control algorithms or application to emerging topologies, further advancing the field of grid connected inverter technology.