In recent years, with the increasing imbalance between supply and demand of traditional energy sources and the severe situation of rising global carbon emissions, countries worldwide are concurrently undergoing an energy revolution. New energy sources, represented by photovoltaic and wind power, are developing rapidly. However, as wind power generation, photovoltaic generation, and other new energy technologies mature and are applied on a large scale, the characteristics of strong randomness and large volatility displayed during grid integration have become increasingly apparent, greatly limiting the absorption capacity of new energy and leading to serious issues such as wind and solar curtailment. Large-scale energy storage technology is a key technology for the widespread application of new energy. Due to the volatility of power generated by new energy sources like wind and photovoltaic, which have poor adjustability, configuring energy storage devices can effectively regulate the energy balance between power supply and demand, overcome the intermittency and instability of new energy output, improve the operational efficiency of power equipment, ensure safe and reliable system operation, and serve as a critical supporting technology for new energy systems. Currently, over 20 countries worldwide are constructing or have put into operation more than a hundred megawatt-level energy storage projects involving the four main power segments: generation, transmission, distribution, and consumption.
Lithium iron phosphate (LiFePO4, LFP) batteries occupy an important position in energy storage fields such as microgrid systems containing renewable energy, electric vehicles, and communication base stations due to their excellent charge and discharge performance, good safety, and cycle life. In the application of LiFePO4 batteries, a corresponding battery management system must be equipped. Among them, real-time and accurate estimation of the battery state of charge (SOC) is a key and difficult point in the industrial application of battery management systems, and it is also an important parameter for the scientific and rational use of battery energy. However, China has an extremely vast geographical range, and climates vary greatly across regions. In northwest and northeast China, due to long sunlight hours in summer and short daylight hours in winter, as well as obvious heat island effects, the annual temperature difference and day-night temperature difference are both very large. For example, in Urumqi, Xinjiang, the extreme high temperature is 47.8°C, and the extreme low temperature is -41.5°C. Such large temperature changes can significantly affect the electrical performance of lithium-ion batteries. Temperature is an important influencing factor in the application of LiFePO4 batteries, directly affecting the capacity and material activity of lithium-ion batteries under operating conditions. This is because the electrochemical reaction at the electrode/electrolyte interface is related to ambient temperature. The electrode/electrolyte interface is considered the heart of lithium batteries. If the temperature drops, the reaction rate of the electrode also decreases: assuming the voltage of the lithium battery remains constant, the discharge current decreases, and the power output of the lithium battery also decreases; if the temperature rises, the opposite occurs. That is, the output power of the lithium battery increases. Temperature also affects the transport speed of the electrolyte: it accelerates when the temperature rises and slows down when the temperature drops, affecting the charge and discharge performance of the lithium battery. However, if the temperature is too high, it can disrupt the chemical balance inside the lithium battery. For lithium batteries, there is currently no clear theoretical support in the industry for the inevitable relationship between internal resistance, discharge platform, lifespan, capacity, and other performance parameters under various temperature conditions. Related calculation formulas and mathematical models are still in the exploratory stage.
In energy storage systems participating in grid frequency control, physical models are based on conventional model data, and control schemes are generated based on inaccurate simulation data and models. These lead to differences between the control effect of the strategy and the actual state of the real grid. Traditional mechanism model analysis and optimal control methods are difficult to meet the requirements of monitoring analysis, operational optimization, and stability control of new power systems. In this paper, when energy storage participates in grid frequency control, ambient temperature factors are considered, and a real-time correction method for model parameters based on environmental temperature correlation is proposed. Combined with actual test data provided by equipment manufacturers, the accuracy of model parameters in the power system modeling process is improved. On this basis, during the process of energy storage systems participating in grid frequency control, a droop control method based on real-time corrected SOC is adopted to achieve real-time mapping of actual grid frequency regulation conditions, improving frequency control efficiency.
To better analyze the methods of energy storage systems participating in grid regulation, it is first necessary to construct a physical model that reflects the characteristics of energy storage. The energy storage system is composed of battery equipment, a power conversion system (PCS), and filtering links. The connection of energy storage devices to the microgrid is achieved through inverters. A widely used control method for inverters is the dual-loop controller. The outer loop controller reflects different control objectives and generates reference signals for the inner loop, while the inner loop controller is used for fine adjustment. The typical structure of the voltage source inverter control system used in this paper is shown in the figure. In the figure, \(U_{abc}\) is the AC grid-side voltage of the three-phase inverter, \(C_f\) and \(L_f\) are the VSC filter capacitor and inductor, respectively, \(L_c\) is the coupling inductor, \(f_s\) and \(\theta_s\) are the frequency and phase of the A-phase voltage at the connected AC bus, respectively, \(U_{sdq}\) and \(I_{sdq}\) are the dq-axis components of the AC-side voltage and current after Park transformation, \(P_{out}\) and \(Q_{out}\) are the calculated instantaneous output power of the inverter, and \(I_{sdq0}\) are the dq-axis components of the inverter input current after Park transformation.
The power calculation block is executed based on Park’s dq-axis theory. In the figure, \(V_{d0}\), \(V_{q0}\), \(I_{d0}\), and \(I_{q0}\) are the dq-axis components of the AC-side voltage and current after Park transformation, and \(\omega_c\) is the low-pass filter parameter. The calculations are as follows:
$$p = v_{d0} i_{d0} + v_{q0} i_{q0}$$
$$q = v_{d0} i_{q0} – v_{q0} i_{d0}$$
To obtain the corresponding fundamental \(P_{out}\) and \(Q_{out}\) components, the instantaneous active power \(p\) and reactive power \(q\) are passed through a low-pass filter to the outer loop controller. The outer loop controller includes droop control and power control. The energy storage inverter quickly responds to frequency and voltage deviations in the outer loop by controlling the dq-axis current components, adjusting the reference values of active and reactive power of the energy storage system. In the figure, \(K_f\) and \(K_u\) are droop coefficient parameters, \(K_{Pp}\), \(K_{Ip}\), \(K_{Pq}\), and \(K_{Iq}\) are PI control parameters for power control, \(f_s\) and \(U_s\) are the system frequency and AC bus voltage amplitude, respectively, and \(f_0\), \(U_0\), \(P_0\), and \(Q_0\) are the reference values for system frequency, AC bus voltage, active power, and reactive power, respectively. \(I_{dref}\) and \(I_{qref}\) are the reference values for the inner loop dq-axis currents.
The state of charge (SOC) of an energy storage battery refers to the ratio of the remaining battery capacity to the rated capacity at a certain discharge rate. It is usually estimated using the ampere-hour integral method, which calculates the change in battery charge and discharge capacity by integrating the battery charge and discharge current over time. The formula is:
$$SOC = SOC_0 – \frac{\int i_b(t) dt}{C_N} \eta$$
where \(SOC\) is the state of charge of the battery, \(SOC_0\) is the initial state of charge of the battery, \(C_N\) is the rated capacity of the battery, \(i_b\) is the instantaneous current value, and \(\eta\) is the battery charge and discharge efficiency.
$$\eta = \begin{cases} \eta_c & \text{for charging} \\ \eta_d & \text{for discharging} \end{cases}$$
where \(\eta_c\) is the charging efficiency and \(\eta_d\) is the discharging efficiency. SOC, as a state quantity representing the remaining capacity of the battery, its accurate estimation is one of the core technologies in battery management systems and the basis for controlling the energy balance of battery energy storage systems. Accurate estimation of SOC can not only effectively prevent overcharging and over-discharging but also serve as a basis for rational use and effective maintenance of batteries.
Based on the SOC state of the energy storage battery, the operating state of the battery is divided into normal operating state and alert state. When the energy storage battery operates in low-alert and high-alert states for a long time, the battery life is greatly shortened. In this paper, when the SOC is in the normal state, power charging and discharging can be performed; in the high-alert state, only discharging is allowed; and in the low-alert state, only charging is allowed, to avoid affecting battery life due to overcharging or over-discharging.
Temperature has a significant impact on the performance parameters of LiFePO4 batteries. The actual capacity \(C_{N,T}\) of the battery is one of the important battery characteristic parameters, and ambient temperature, the number of battery cycles, and the average discharge current are the three main factors affecting the actual capacity of the battery. In lithium batteries, operating temperature has the most significant impact on the capacity of LFP batteries. Numerous studies have shown that the actual available capacity of LiFePO4 batteries increases with rising ambient temperature. The main reason for this phenomenon is that the electrochemical reaction at the electrode/electrolyte interface is related to ambient temperature. The electrode/electrolyte interface is considered the heart of lithium batteries. When the ambient temperature increases, the viscosity of the electrolyte decreases, activity improves, the diffusion ability of ions in the electrolyte enhances, the utilization rate of active materials increases, and the actual battery capacity increases. When the ambient temperature decreases, the opposite occurs, and the actual battery capacity decreases. However, if the temperature is too high, it can disrupt the chemical balance inside the battery, leading to side reactions, thereby affecting the charge and discharge performance of the battery. The performance of battery materials degrades, and the battery cycle life is greatly shortened. This damage is also irreversible. Excessively high temperatures can cause the battery outer packaging to bulge and rupture, and the chemicals in the battery can explode when exposed to air.
The operating temperature limit for LiFePO4 batteries is generally -40°C to 60°C. When the temperature is below 0°C, the performance of the battery seriously declines, and activity severely decreases, so below 0°C is considered the low-temperature stage. When the temperature is above 40°C, the SOC of the battery may reach 1.1, so above 40°C is considered the high-temperature stage. The fully functional operating temperature for lithium-ion batteries is generally 0°C to 40°C.
Charge and discharge efficiency \(\eta\) is an important parameter for energy storage participating in grid operation. Charge and discharge efficiency, also known as Coulombic efficiency, is defined by the United States Advanced Battery Consortium (USABC) as the percentage of the capacity discharged to the capacity charged before discharge. Traditional charge and discharge efficiency does not consider the effects of charge and discharge rates and operating temperature. Literature has conducted high and low-temperature experiments on 100 Ah, soft-packaged LFP batteries. The experimental results show that as the ambient temperature increases, the charge and discharge efficiency of the battery also improves; at low temperatures, the charge and discharge capacity and efficiency of the battery decrease significantly; at high temperatures, the charge and discharge capacity and efficiency of the battery increase somewhat. Within the operating range of lithium batteries, the charge and discharge efficiency of the battery is on a relatively high platform. The discharge efficiency of nickel-cadmium, nickel-metal hydride, and lithium batteries decreases significantly at low temperatures (e.g., below -15°C), and at -20°C, the electrolyte reaches its freezing point, and the battery charging speed also greatly decreases. Charging at low temperatures below 0°C increases internal battery pressure and may cause the safety valve to open.
In this paper, the CATL 280 Ah LiFePO4 battery is taken as the research object. Based on actual manufacturer test result data, the approximate relationship of the rated capacity of the battery at different temperatures is obtained as shown in the following formula. This allows the battery rated capacity to be corrected through temperature correlation coefficients.
$$C_{t2} = C_{t1} [1 + \alpha (t_2 – t_1)]$$
where \(C_{t1}\) is the capacity at temperature \(t_1\)°C, \(C_{t2}\) is the capacity at temperature \(t_2\)°C, and \(\alpha\) is the temperature correlation coefficient for capacity change with temperature, obtained from experimental measurements. Different temperatures have different \(\alpha\). Table 1 shows the energy efficiency of the complete discharge process of the CATL 280 Ah LiFePO4 cell at different temperatures provided by the battery supplier Ningde Times. The energy efficiency is the ratio of the energy released during the complete discharge process under specified conditions to the rated charging energy of the battery. The rated charging energy is the charging energy obtained by charging the cell from an empty state to a full voltage of 3.65 V at 1P power under rated conditions (25°C constant temperature environment, pressure 101.3 kPa, RH ≤ 85%). Test conditions: charging condition—25°C, 1P CP to 3.65V; discharging condition—25, 45, 5, 0, -10, -20°C, 1P DP to 2.0V. Due to the cost of testing in practical applications, engineering prefers to monitor the discharge performance of batteries at representative temperature points, so the temperature granularity is large. Based on measured data, the calculated \(\alpha\) is as shown in the figure.
| Temperature (°C) | Energy Efficiency (%) | Actual Available Capacity (Ah) |
|---|---|---|
| -20 | 66.9 | 187.32 |
| -10 | 78.8 | 220.64 |
| 0 | 77.6 | 217.28 |
| 5 | 82.7 | 231.56 |
| 25 | 100 | 280 |
| 45 | 101.4 | 283.92 |
In addition to the direct impact of temperature on rated capacity, battery aging and self-discharge also affect the rated capacity of the battery: LiFePO4 batteries have good charge retention, meaning that under long-term static conditions, charge loss is small, and battery capacity changes little; after multiple charge and discharge cycles, battery capacity decreases. Many scholars have considered the number of cycles in rated capacity estimation and added aging coefficients to the calculation formula, but have not separately listed the impact parameters of temperature on capacity reduction due to battery aging. Therefore, in this paper, battery self-discharge and aging coefficients are not considered in SOC estimation.
Since most batteries have internal resistance, there are significant losses during charge and discharge processes, so energy storage systems must consider charge and discharge efficiency, i.e., Coulombic efficiency. Temperature also affects the Coulombic efficiency of the battery, as shown in the following formula. The charge and discharge efficiency can be corrected through the Coulomb correlation coefficient.
$$\eta_E = \eta_e K_E$$
where \(\eta_E\) is the equivalent charge and discharge efficiency considering temperature, \(K_E\) is the Coulomb correlation coefficient considering temperature, and \(\eta_e\) is the equivalent charge and discharge coefficient without considering temperature (based on the battery charge and discharge efficiency at an operating temperature of 25°C as the reference efficiency). The \(K_E\) calculated based on manufacturer measured data is as shown in the figure.
In this paper, it is considered that during the participation of the energy storage system in grid operation, all parameters of the voltage source inverter control are not affected by environmental temperature, etc., and only SOC is affected by temperature. The battery SOC with added ambient temperature parameters is:
$$SOC(T) = SOC_0(T) – \frac{\int i_b(T) dt}{C_N(T)} \eta(T)$$
In actual operation, the SOC considering environmental factors will be significantly different from the SOC without consideration under extreme environments, which will also affect the arrival time of the high-alert and low-alert positions in the battery operating state, thereby affecting the actual output of the battery during operation.
When applying energy storage systems to grid frequency regulation, fixed droop coefficient control is usually adopted, but this may lead to overcharging or over-discharging. Therefore, the SOC state must be considered in the energy storage frequency regulation strategy. In this paper, the corrected SOC parameter based on measured temperature is incorporated into frequency control, and a droop control method based on real-time corrected SOC is proposed. In the stage where the energy storage system participates in grid regulation, the temperature of the energy storage system is collected in real-time, and the SOC state is continuously corrected. When the system generation power suddenly increases, the energy storage system charges, as shown by the red dashed line. If \(SOC(T) < 0.2\), the energy storage can charge at the maximum charging power to reduce the system imbalance power. During the charging process, as SOC gradually increases, the droop coefficient decreases in an exponential form, and it still participates in frequency regulation. When \(SOC > 0.8\), to protect the energy storage from overcharging and reduced lifespan, charging stops. In this way, the energy storage can real-time correct the energy storage charge and discharge droop coefficient based on the corrected SOC state. The droop coefficients are as follows:
$$K_{f_d} = \begin{cases} K_{\text{max}} & \text{if } SOC < 0.2 \\ K_{\text{max}} \left( \frac{SOC(T) – 0.2}{0.6} \right)^m & \text{if } 0.2 \leq SOC \leq 0.8 \\ 0 & \text{if } SOC > 0.8 \end{cases}$$
$$K_{f_c} = \begin{cases} 0 & \text{if } SOC < 0.2 \\ K_{\text{max}} \left( \frac{0.8 – SOC(T)}{0.6} \right)^m & \text{if } 0.2 \leq SOC \leq 0.8 \\ K_{\text{max}} & \text{if } SOC > 0.8 \end{cases}$$
where \(m\) is the exponential change rate, \(K_{\text{max}}\) is the maximum droop coefficient, \(K_{f_d}\) is the discharge droop coefficient, and \(K_{f_c}\) is the charge droop coefficient.
To verify the effectiveness of the proposed method, the modified IEEE 13-node system is used as an islanded microgrid. The grid consists of a 3.125 MVA diesel generator (with synchronous generator, governor, and exciter), a photovoltaic system with an installed capacity of 2 MW, a 1.5 MVA LiFePO4 energy storage system (battery cell is CATL 280 Ah), and loads. By simulating this isolated grid system in Simulink, the effectiveness of considering environmental factors in frequency control is verified through the process of energy storage system droop control based on SOC participating in grid frequency regulation.
The initial state of the energy storage battery is set to 0.5. The charge and discharge efficiency without considering environmental factors is 0.9, and the maximum charge and discharge power is 50 kW. The battery control parameters are shown in Table 2. Three temperature scenarios are set in this paper: ① extremely cold temperature condition of -10°C in winter in a certain area of Inner Mongolia; ② extremely hot temperature condition of 45°C in summer in a certain area of Inner Mongolia; ③ standard temperature condition of 25°C in a certain area of Inner Mongolia.
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| \(L_f\) (mH) | 3 | \(K_{Pp}\) | 0.05 |
| \(L_c\) (mH) | 0.1 | \(K_{Ip}\) | 100 |
| \(C_f\) (μF) | 100 | \(K_{Aq}\) | 0.42 |
| \(\omega_c\) (rad/s) | 30 | \(K_{Iq}\) | 6 |
| \(K_{\text{max}}\) | 30 | \(m\) | 0.3 |
When the grid load suddenly increases by 0.6 MW, under this condition, the adjustable margin of the diesel generator is insufficient, and the energy storage discharges to participate in system frequency control. The frequency response curves and energy storage SOC change curves without considering environmental factors and considering environmental factors are shown in the figure.
It can be seen from the figure that without considering environmental factors (T=25°C), the SOC of the energy storage is always in the normal range within 1000 s in this scenario. The energy storage always participates in grid regulation, and \(K_{f_d}\) is in a higher position due to the higher SOC, so the frequency drops slowly, and the frequency regulation effect is good. However, in actual operation, high temperature (T=45°C) will cause the SOC to drop faster, and the droop coefficient \(K_{f_d}\) decreases rapidly, reducing the frequency regulation effect and increasing the frequency deviation. Additionally, at high temperatures, the battery enters the low-alert state (SOC=0.2) around 855 s. To ensure safe and reliable operation of the battery, under the control requirements of this paper, the energy storage can only charge at this time and cannot discharge for frequency control. That is, high temperature causes the energy storage to end frequency regulation sooner. At this time, the only frequency regulation resource left is the diesel generator, and the frequency will enter a lower steady-state value, still meeting the frequency deviation limit range of the isolated grid. In the low-temperature scenario (T=-10°C), the SOC drop rate slows, the frequency regulation effect is slightly better than that in the normal temperature operating environment (T=25°C), and it will discharge for a longer time to participate in frequency regulation.
To verify the effectiveness of the proposed method for continuous photovoltaic fluctuations, this example uses a more extreme historical scenario of photovoltaic fluctuations. The energy storage system participating in grid frequency regulation has fast response characteristics and can cooperate with the diesel generator to smooth out photovoltaic power fluctuations. The frequency and SOC change curves of energy storage under different test conditions under continuous disturbance are shown in the figure.
The SOC change curves after energy storage participates in frequency regulation under different temperature environments still show significant differences. Based on the control in this paper, the droop coefficient \(K_{f_c}\) is affected by SOC, so the frequency regulation effects of the three ambient temperatures in the early stage of disturbance are not exactly the same. Among them, under high-temperature conditions (T=45°C), since the SOC is slightly higher than the other two conditions, the droop coefficient \(K_{f_c}\) will be smaller, so the frequency regulation effect is worse, and the frequency offset is larger. In addition, in the later stage of photovoltaic power fluctuation, the high-temperature environmental condition enters the high-alert position at 1738 s, after which the energy storage system can only discharge and cannot continue charging, thus exiting frequency control. That is, high temperature prompts the energy storage to exit frequency control as soon as possible, and subsequently, the grid can only be supported by the diesel generator for frequency. The energy storage system under normal environmental temperature (T=25°C) will exit frequency control at 2118 s, and the low-temperature environment (T=-10°C) can participate in frequency control within the entire 2750 s.
Based on the above analysis, the proposed energy storage system frequency control method considering environmental temperature factors and correcting SOC parameters can effectively estimate the SOC of the battery more accurately for extreme weather conditions, making the frequency control effect more suitable for actual operating scenarios and improving the efficiency and reliability of energy storage participation in grid frequency control. The method in this paper can estimate the SOC of the energy storage system more accurately and improve the efficiency of frequency control compared to methods that do not consider environmental temperature factors. Future research directions include coupling other environmental factors such as humidity and salinity to further improve the accuracy of model parameters for energy storage devices participating in grid regulation.