Impact of Power Plant Controller on Voltage Oscillations in RE complex: Insights from Hardware-in-the-Loop Simulation

The Government of India aims to establish 500 GW of non-fossil fuel generation capacity by 2030 and achieve net zero emissions by 2070[1]. However, as renewable energy (RE) generation increases, ensuring grid reliability poses several challenges. Renewable generation, primarily from wind and solar resources, is getting connected to pooling stations in a concentrated manner in India, i.e., in the size of around 3-5 GW, generation is pooled at one particular substation. Many of these pooling stations do not have synchronous machine connectivity, resulting in a weak grid. In those regions with high renewable penetration, such as the Rajasthan RE complex in India (Bhadla-Fatehpur-Bikaner areas), there have been frequent grid disturbances, including multiple instances of voltage oscillations [2].

From mid-December 2022 through November 2024, several voltage oscillations were observed in these areas. These oscillations, mainly in voltage and reactive power, sometimes reached magnitudes of 0.1 p.u., causing voltage fluctuations ranging from 380 kV to 420 kV in the 400 kV grid. Figure 1 shows an example of oscillation observed at 400kV Fatehgarh-2 pooling station(India) on 30th September 2024. These oscillations were primarily observed in voltage and reactive power. Some of these oscillations have resulted in generation loss (even to the order of 7GW) and line tripping as their intensity increases during transient disturbances in the grid [2]. In all these events, there were no triggering events in the grid, such as reactor switching, line charging, etc., that could be considered as the cause of oscillations.

Identifying the causes of these oscillations is critical to maintaining grid stability and preventing generation loss. The PMU plot of a PV plant during such oscillation is shown in Figure 2. From the PMU data analysis, it is observed that some of the PV plants inject reactive power with considerable delay, which results in the injection of reactive power in phase with the voltage oscillations thereby amplifying them. The oscillations subsided when some plants were kept in fixed reactive power control mode. The preliminary analysis shows that one of the reasons for these voltage oscillations is the delays associated with PPCs, as the real and reactive power command from the PPC reaches the inverter in some of the RE plants after considerable time delays associated with the Power Plant Controller. These time delays amplify some of the oscillation modes. In this paper, a detailed analysis of oscillation due to PPC delay is done through hardware in-loop simulation of PPC hardware.

The simplified block diagram of the PPC is shown in Figure 3. The PQ meter provides active power, reactive power, frequency and voltage measurements (P, Q, f, V) at the plant terminal to the PPC. The PPC calculates the required real and reactive power that needs to be injected into the POI based on the operational mode. The operational mode, droop settings (dead band, droop gain, offset) and the corresponding reference values (Pref, Qref, Vref, fref) are provided to the PPC from the SCADA HMI. In the power factor control mode, the PPC generates a reactive power reference based on the p.f. reference and active power reference value. The voltage magnitude signal error is given to the PI controller in voltage control mode to get the reactive power reference.

The reactive power output of the plant in Q-V droop control mode is decided by the droop gain, dead band and offset of the plant as illustrated in Figure 3. For example, a plant with 4% droop and 1% dead band (with zero offsets) may vary its reactive power to 100 per cent of plant capacity for a change of 4% of its POI voltage from its reference value. However, the reactive power of the RE plants is usually limited to 33% of the plant capacity, and the support from the individual inverters is usually limited to 60% of the inverter capacity.

The reactive power injection by each RE plant is controlled by its PPC configuration. The RE plant often has multiple plants commissioned at various times, each with different PPCs, PQ meters, and communication protocols. The PPCs manage the power output of individual inverters in an RE plant, and any delay in their response to grid changes can induce oscillations as well as amplify the existing oscillations in the grid. These delays may stem from the PPC design itself or from limitations in the communication network, which takes the signals from the PQ meter to the PPC and sends the control commands from the PPC to the inverters. Low sampling rates in PQ meters and reduced polling rates for inverters also contribute to these delays. PQ meter has different registers for processing fast and slow data. Fast data channels can transmit data within 10-20 ms, whereas slower channels can take between 250 ms and 1 second. Many older plants have PQ meters that operate only on slow channels, resulting in expected delays of at least 250ms before the PPC receives data from the PQ meter, Figure 4. After receiving this data, the PPC requires an additional 10-20ms to process it and provide the setpoints to the inverters. The PPC sends active (P_iref ) and reactive (Q_iref ) power commands through the communication network, with delays influenced by the protocol used, such as MODBUS or IEC 61850. The MODBUS communication protocol is slower than IEC 61850 and thus increases communication delay. Due to polling rate limitations, inverters cannot continuously receive updated commands from the PPC. Typical polling rates range from 200 ms to 1 second, restricting the frequency of PPC commands to the inverters. This variability in response times in different plants results in each plant responding differently to voltage changes, which can exacerbate oscillations in the grid and impact grid stability.

The RE-penetrated grid generally has low short circuit strength and thus forms a weak grid. According to CEA standards, RE plants must provide 0.33 p.u. of reactive power while delivering 1 p.u. of active power [2]. In a weak grid, the voltage is very sensitive to reactive power injection. The PPC primarily controls the reactive power injection in an RE plant. However, it is important that the power plant accurately injects the required reactive power in the correct phase. The reactive power injection out of phase will dampen the voltage oscillation, and in phase will aggravate the voltage oscillation The injection of reactive power from the RE plant depends on the PPC settings. The PPC commands the individual inverters to generate /absorb reactive power based on the grid voltage. However, it is often observed that the reactive power injection from the inverter experiences significant delays after the PPC detects a change in voltage, as discussed in section (IV).

As illustrated in Figure 5, the reactive power reference signal command should be out of phase with the voltage oscillations to damp out the voltage oscillations. During the low voltage period in the oscillation, the plant should inject the reactive power, and for the higher voltage period in the oscillation, the plant should absorb the reactive power, as illustrated in Figure 5(b). However, due to the delays associated with PPC, the reference command signal for reactive power comes in phase with the voltage oscillations as illustrated in Figure 5(c). This results in the injection of reactive power in phase with voltage oscillation, which will amplify the voltage oscillations at POI.

The experimental setup of the hardware in the loop simulation of PPC is shown in Figure 6. The Real-Time Digital Simulator facility in the lab is used to integrate the PV plant model of 1800 MW and the hardware of the PPC. The interfacing is done through GTAI/GTAO cards, as shown in Figure 7. The details of the modelled plant, as well as the delays associated with PPC, are tabulated in Table 1. The delays and sampling rates modeled in the study are selected to represent realistic values observed in renewable energy plants.

A low droop setting is recommended for RE plants with shorter PPC delays, while a high droop setting is more suitable for those with longer PPC delays. For plants experiencing significant reactive power fluctuations, increasing the dead band and droop may help stabilize operation. However, the plant with higher PPC delays may be operated in a fixed power factor mode or in fixed q mode to reduce the amplification of oscillation. However, this may cause the inverter to enter ride-through mode due to excessively high or low terminal voltage during peak instances of voltage oscillation. If the plant operates in leading power factor (p.f.) mode, it may inject capacitive reactive power during high voltage conditions, further increasing the inverter terminal voltage. Thus, it is recommended to raise the ride-through settings of the inverter before operating the plant in fixed Q or fixed p. f mode. This prevents the inverter from entering ride-through mode during the voltage oscillations in the grid.

A detailed analysis of voltage oscillations under different operating modes of PPC is done though hardware in loop simulation of PPC hardware. An actual PPC hardware, which is intended for installation in a renewable energy plant is used for this study. The delays and sampling rates modelled in the study are chosen to reflect realistic values observed in renewable energy plants. It is observed that the PPC delays can result in voltage oscillation and can amplify some of the oscillation modes. The outcome of the study provides recommendations for operating plants, taking into account the delays related to their PPC hardware. It also emphasizes that the choice of droop settings and deadband in renewable energy plants should be considering the PPC delays. Additionally, the significance of maintaining appropriate ride-through settings in each operating mode is discussed.

The PPC delays typically arise from factors such as communication delays and inverter polling rates. As these delays associated with each plant are fixed, they can be mitigated by adopting suitable delay compensation algorithms in PPC. This will enable PPC to dampen oscillations in any mode without amplifying them. While several delay compensation techniques have been documented in the literature, the research gap exists in identifying and implementing a delay compensation scheme tailored to PPC architectures in renewable energy (RE) plants. Recognizing these challenges, ongoing research is focused on developing and implementing a novel delay compensation algorithm within the PPC. This approach aims to enhance the PPC's capability to damp voltage oscillations, thereby improving grid stability in systems with high RE penetration.