In the pursuit of the dual-carbon goals, photovoltaic power generation has experienced rapid growth as a leading renewable energy source. However, the inherent intermittency, randomness, and volatility of photovoltaic output present significant challenges to the stable operation of power systems. Solar power output is heavily influenced by weather conditions, diurnal cycles, and other factors, often resulting in a mismatch between the generation curve and the load demand curve, which can cause supply-demand imbalances. Battery energy storage systems, leveraging their technical advantages such as fast response times, high energy density, and long cycle life, have emerged as a critical technological pathway for mitigating the grid-integration challenges of photovoltaic power generation.
Existing research has focused on grid integration and the synergistic operation of energy storage, advancing from hierarchical control and Energy Management System (EMS) optimization to capacity-power co-configuration and life-cycle/economic joint optimization. For instance, some studies have examined the economics of battery energy storage systems in peak-load shaving applications for integrated photovoltaic-storage microgrids. Others have developed optimal storage configuration models solved by improved particle swarm optimization algorithms, using load rate as an evaluation index to assess peak-shaving effectiveness and determine the optimal storage capacity by balancing comprehensive cost and load standard deviation. However, current research often lacks validation using high-resolution fluctuation data, has incomplete coupling between capacity-power and life-cycle economics, and faces challenges in replicability due to data sources.
In my project, which focuses on an industrial park scenario integrating photovoltaic generation with lithium battery storage, I propose a comprehensive method that includes capacity-power co-configuration and a multi-objective EMS optimization strategy. This approach effectively enables the temporal and spatial transfer of electrical energy. The system stores energy when photovoltaic output is abundant and releases it when output is insufficient, thereby smoothing generation fluctuations and significantly enhancing the stability, reliability, and economic benefits of the photovoltaic power generation system.
Project Overview
My project is a “photovoltaic + storage” integrated clean energy demonstration initiative built within an industrial park. The park covers approximately 150 acres and features a 10 MW distributed photovoltaic power generation system, complemented by a 5 MW/10 MWh lithium battery energy storage system. The park houses 15 enterprises with a total daily electricity consumption of approximately 150 MWh. The peak electricity demand primarily occurs during working hours from 9:00 to 17:00 and 19:00 to 22:00, creating a significant peak-valley electricity price differential of 0.6 CNY/kWh.
The project site boasts excellent solar resources, with an average annual sunshine duration of 2,280 hours and an annual total solar radiation of about 1,350 kWh/m², providing a robust natural energy base for the photovoltaic system. Under the traditional power supply model, the park relied heavily on the external grid, leading to high electricity costs and dual pressures related to power supply stability and carbon emission control. To align with national energy policies and address the challenges of rising fossil fuel costs and power supply constraints, the park urgently needed to optimize its energy structure. The construction of a battery energy storage system was identified as the key to capturing, storing, and managing the energy generated by the photovoltaic system, enabling a transition towards an optimized and sustainable energy structure.
Application of Battery Energy Storage Systems in Photovoltaic Power Generation
Capacity Configuration Design
Based on the solar resource conditions at the project site, the photovoltaic system’s average daily power generation is approximately 32.88 MWh, with about 60% of this energy concentrated between 10:00 and 16:00. The park’s load demand exhibits a bimodal pattern, with morning peaks from 9:00 to 11:00 and afternoon peaks from 14:00 to 17:00, while the nighttime base load is about 40% of the peak load. This load profile aligns with the photovoltaic output characteristics, indicating a power surplus during high-irradiance daylight hours and a power deficit at night. Therefore, the design of the energy storage system’s capacity must thoroughly consider the diurnal fluctuation in energy and the matching requirements between supply and demand. Based on the analysis of the load and photovoltaic output curves, I derived the following formula for the energy storage capacity:
$$E_{storage} = \alpha E_{remaining} + \beta E_{deficit}$$
where:
- \( E_{storage} \) is the total capacity of the battery energy storage system.
- \( E_{remaining} \) is the surplus electricity generated by the photovoltaic system.
- \( E_{deficit} \) is the deficit electricity required to meet the load demand.
- \( \alpha \) is the proportion of surplus photovoltaic power that can be efficiently stored, taken as 0.8.
- \( \beta \) is the effective compensation ratio of the battery energy storage system for the load deficit, taken as 0.6.
Considering factors such as battery charging/discharging efficiency, converter efficiency, usable State of Charge (SOC) window, and temperature and aging-related derating to prevent over-discharge and extend battery life, I calculated the optimal energy storage capacity to be 10 MWh. This capacity allows the system to store approximately 70% of the surplus photovoltaic energy and supplement about 6 hours of base electricity demand during periods without sunlight.
For the power configuration, the primary considerations were smoothing photovoltaic power fluctuations and meeting peak load regulation requirements. During meteorological events like cloud cover, the photovoltaic system’s power change rate can reach up to 2 MW/min. The battery energy storage system must possess fast response capabilities to suppress these fluctuations. To match the demand during peak electricity consumption periods, I configured the storage power at 5 MW, thereby simultaneously satisfying both the peak load regulation and photovoltaic power smoothing requirements.
Charge and Discharge Control Design
Taking into account the multi-faceted factors of photovoltaic output forecasting, load forecasting, electricity price information, battery health status, and system economic benefits, I implemented a charge and discharge control strategy for the battery energy storage system based on a multi-objective optimization algorithm, aligning with the configured storage capacity. The system employs a time-differentiated control strategy, implementing different charge and discharge protocols across various time periods to maximize system benefits.
There are two primary charging modes: photovoltaic charging and grid charging. Photovoltaic charging is given the highest priority. When the photovoltaic power generation exceeds the park’s load demand and the battery SOC is below 90%, the battery energy storage system automatically enters the charging state. The power control formula for photovoltaic charging is:
$$P_{charging} = \min \left( P_{pv\_remaining}, P_{charging\_rated}, \frac{(SOC_{max} – SOC) \times C_{rated}}{\eta_{charging}} \right)$$
where:
- \( P_{charging} \) is the charging power of the battery energy storage system.
- \( P_{pv\_remaining} \) is the remaining power from photovoltaic generation.
- \( P_{charging\_rated} \) is the rated charging power of the battery energy storage system (5 MW).
- \( C_{rated} \) is the rated capacity of the battery (10 MWh).
- \( \eta_{charging} \) is the charging efficiency (95%).
- \( SOC \) and \( SOC_{max} \) are the current and maximum battery State of Charge, respectively.
Grid charging mode is initiated during valley electricity price periods (23:00 to 07:00 the next day) when the electricity price is below 0.4 CNY/kWh and the battery SOC is below 20%. In this mode, the system charges from the grid at a constant power rate to balance energy requirements and capitalize on the price differential, thereby optimizing economic performance.
From the perspective of discharging strategies, I designed three modes tailored to different application scenarios: load-tracking discharge, peak-load shaving discharge, and emergency backup discharge. When photovoltaic generation is insufficient, the load-tracking discharge mode is activated, dynamically adjusting the discharge power based on the load deficit to ensure sufficient power supply during low irradiance or peak demand hours. During peak electricity price periods (08:00-11:00 and 18:00-21:00), the peak-load shaving discharge mode is executed. When the electricity price exceeds 0.8 CNY/kWh and the battery SOC is above 20%, the battery energy storage system supplies power to the park, thereby reducing the park’s electricity procurement costs. To extend battery lifespan, I implemented a SOC tiered management strategy. During normal operation, the SOC is maintained within the 20% to 90% range, preventing deep discharge and overcharging, which helps ensure a battery cycle life exceeding 6,000 cycles. Under emergency conditions, the system allows the SOC to drop to 10% to guarantee sustained power supply in extreme situations.
In a 1-minute resolution evaluation, the introduction of the battery energy storage system reduced the maximum power ramp rate from 2 MW/min to 0.6 MW/min. The standard deviation of power was reduced by 45%, and the 95th percentile ramp rate was lowered by 65%. This data indicates that the proposed method has a significant smoothing effect on short-term power fluctuations. The following table shows the charge and discharge regulation data for photovoltaic generation and load demand over a 24-hour period, reflecting the energy balance characteristics but not intended for dynamic performance validation.
| Hour (h) | Solar Output (MWh) | Load Demand (MWh) | Battery Discharge (MWh) | Battery Capacity (MWh) |
|---|---|---|---|---|
| 1 | 27.96 | 150.00 | 0 | 0 |
| 3 | 29.01 | 139.21 | 0 | 0 |
| 5 | 36.81 | 122.06 | 0 | 0 |
| 7 | 34.10 | 142.73 | 0 | 0 |
| 9 | 31.53 | 125.21 | 0 | 0 |
| 11 | 34.08 | 120.84 | 0 | 0 |
| 13 | 36.65 | 112.61 | 0 | 0 |
| 15 | 33.71 | 124.18 | 0 | 0 |
| 17 | 35.01 | 120.37 | 0 | 0 |
| 19 | 32.75 | 137.79 | 4.33 | 0 |
| 21 | 30.16 | 132.07 | 4.45 | 0 |
| 23 | 29.48 | 138.92 | 5.44 | 0 |
Grid-Connected Operation Design
The project utilizes advanced power conversion and grid synchronization technologies to control the grid-connected operation of the battery energy storage system, ensuring stable and efficient interaction between the system and the power grid.
Grid synchronization control is a core technology for grid-connected operation. The system employs a Phase-Locked Loop (PLL) to track the phase of the grid voltage in real time, ensuring that the output voltage of the battery energy storage system is synchronized with the grid voltage in both frequency and phase. By using a second-order generalized integrator-based PLL control algorithm, the system achieves a phase tracking accuracy of ±0.1° and a frequency tracking accuracy of ±0.01 Hz, enabling precise grid synchronization. At the point of common coupling, a constant voltage control strategy is adopted, utilizing a PI regulator to adjust the output voltage amplitude of the storage converter. This keeps the voltage at the point of common coupling stable within ±5% of the nominal value. This precise voltage regulation strategy facilitates a seamless connection between the battery energy storage system and the grid, minimizing the impact of voltage fluctuations on grid stability.
The power control system design enables independent regulation of active and reactive power, ensuring the system can flexibly respond to grid demands. The active power response time is controlled within 100 ms, with a regulation accuracy of ±1% of the rated power. The reactive power control range is -2.5 to +2.5 Mvar, providing effective voltage support and reactive power compensation to the grid, thereby improving overall grid performance. The battery energy storage system is also equipped with Low Voltage Ride Through (LVRT) and High Voltage Ride Through (HVRT) capabilities, ensuring it can remain connected during grid voltage dips down to 0.2 per unit for up to 0.15 seconds, in compliance with national grid technical standards, thus safeguarding system and grid stability.
Real-Time Operation Design
Real-time battery monitoring and an EMS are critical for ensuring the safe and efficient operation of the battery energy storage system. My system utilizes a multi-layered monitoring architecture and intelligent management algorithms to achieve comprehensive situational awareness and precise control. It employs a cloud-edge collaborative monitoring architecture, where a local EMS handles real-time control and data acquisition tasks, while a cloud platform provides big data analysis and remote operation and maintenance services.
For real-time battery monitoring, the Battery Management System (BMS) provides meticulous supervision through a distributed sensor network that collects key parameters such as voltage, current, and temperature for each battery module. The system uses an Extended Kalman Filter algorithm for SOC estimation, integrating the Coulomb counting method with the open-circuit voltage method through fusion calculation to achieve an SOC estimation accuracy of ±3%. By employing a dual-indicator model of capacity fade and internal resistance growth, the system evaluates the State of Health (SOH) of the battery. Machine learning algorithms are used to analyze historical operational data to predict the remaining useful life and performance degradation trajectory of the batteries. Furthermore, the system features sophisticated fault diagnosis and alarm capabilities, using multivariate statistical analysis and an expert system to identify abnormal conditions, enabling early fault warning and localization. A three-level alarm system was established: yellow warnings for parameter deviations, orange warnings for abnormal states, and red alarms for emergency faults. All monitoring data is transmitted to the cloud platform in real time via 4G/fiber optic networks, allowing for remote monitoring from PCs and mobile devices, resulting in intelligent, unattended operation.
The EMS integrates intelligent algorithm modules for photovoltaic generation forecasting, load forecasting, and electricity price forecasting. The photovoltaic generation forecast uses a hybrid model combining numerical weather prediction and artificial neural networks, achieving a day-ahead prediction accuracy of over 85%. The load forecast is based on historical load data and influencing factor analysis, utilizing a Support Vector Machine regression algorithm with an accuracy of over 90%. Integrating forecasts for photovoltaic generation and load demand, the EMS executes a rolling optimization control strategy. Through the following optimization objective, the system maximizes energy utilization efficiency and minimizes operational costs:
$$Optimization = \min (C_{grid} P_{grid} + C_{battery} P_{battery})$$
where:
- \( Optimization \) is the system optimization objective.
- \( C_{grid} \) is the grid electricity price.
- \( P_{grid} \) is the power exchanged with the grid.
- \( C_{battery} \) is the battery aging cost.
- \( P_{battery} \) is the battery charge/discharge power.
Through integrated intelligent algorithms and real-time optimization control, the EMS can minimize system operating costs while ensuring power supply stability, thereby improving the energy self-sufficiency rate and economic benefits of the park.
Implementation Methodology
Key Equipment Selection
The equipment selection for my project’s photovoltaic system was based on the local resource conditions of 2,280 annual sunshine hours and 1,350 kWh/m² of total solar radiation. I selected 540 W monocrystalline PERC photovoltaic modules with an efficiency of over 21.2% and a temperature coefficient of -0.35%/°C, ensuring high performance even in hot environments. To meet the 10 MW installed capacity, a total of 18,520 photovoltaic modules were configured in an electrical arrangement of 22 strings by 841 parallel connections, ensuring an annual power generation exceeding 12 million kWh. For the grid-connected inverter, I selected five 2 MW centralized three-phase inverters with a maximum efficiency of 98.8% and a Maximum Power Point Tracking efficiency of over 99%. These inverters offer LVRT and frequency response capabilities to meet grid adaptability requirements. For the battery energy storage system, I chose a 5 MW/10 MWh lithium iron phosphate battery system. The single cell capacity is 280 Ah, and the cycle life exceeds 6,000 cycles, ensuring efficient energy storage and long-term stability. The Power Conversion System (PCS) is a bidirectional energy storage converter with four-quadrant operation capability and a power conversion efficiency exceeding 97%. It supports grid voltage support and frequency regulation and includes islanding detection, under-frequency load shedding, and black start functions, providing an uninterruptible power supply for the park.
Engineering Implementation Process
The project implementation followed a strategy of overall planning and phased construction, encompassing stages such as design, civil construction, equipment installation and commissioning, and grid-connected acceptance. Before project initiation, preliminary work including feasibility studies, environmental impact assessments, and power access scheme approval was completed to ensure compliance. During the design phase, the spacing and tilt angle of the photovoltaic module layout were optimized to ensure array-to-array shading losses were less than 3%. The layout of the battery energy storage system considered safety distances for fire protection and ease of maintenance, with a minimum spacing of 6 meters between storage containers. In the civil construction phase, concrete strip foundations were used for the photovoltaic support structures, buried to a depth of 1.2 meters with a rebar cover thickness of 50 mm. The foundations for the battery energy storage system containers measured 3.0 m x 12.5 m x 0.8 m, with a surface flatness tolerance of less than 5 mm. During the equipment installation and commissioning phase, the photovoltaic system and battery energy storage system were installed strictly according to technical requirements, with installation accuracy controlled within ±5 mm. Electrical connections were carried out strictly per installation specifications, and the system was energized for testing only after the insulation resistance test passed. The system integration phase involved individual equipment commissioning to verify the technical indicators of the photovoltaic inverters, storage converters, and the EMS.
Performance Analysis
After the system renovation, the energy utilization efficiency of the photovoltaic power generation system was significantly improved. The synergistic operation between the battery energy storage system and the generation components effectively resolved the original problems of energy waste and fluctuation. Comparing data before and after project implementation, the average annual power generation of the photovoltaic system increased from 32,880 MWh to 35,400 MWh, representing an increase of about 7.6%. The introduction of the battery energy storage system allowed surplus photovoltaic energy to be stored in a timely manner and released during periods of low irradiance or peak demand, greatly alleviating the pressure of power fluctuations on the grid. The battery energy storage system can achieve a maximum discharge power of 5 MW and sustain discharge for over 6 hours, effectively mitigating power shortages during peak consumption periods.
Based on real-time 1-minute resolution measurements from the BMS/EMS and a before-and-after comparison, the optimization of the EMS dispatch and PCS parameters resulted in an increase in the overall system energy utilization efficiency from approximately 86% to about 90%. The park’s self-sufficiency rate (self-consumption + storage feed-in relative to total consumption) rose from 60% to 80%, with a significant reduction in net purchased electricity during peak periods. Combined with peak shifting and fluctuation smoothing, the annual utilization rate of photovoltaic generation increased by about 7.6%, and the system’s reliance on the external power grid was effectively reduced.
Conclusion
Through the implementation of this project, the application of lithium battery energy storage technology has effectively smoothed the fluctuations of photovoltaic power generation and aligned the output with the park’s electricity demand, enhancing the stability and economic viability of the photovoltaic power generation system. By precisely configuring the storage capacity and optimizing the charge-discharge control strategy, the project significantly improved energy utilization rates, reduced dependence on the external power grid, and consequently lowered electricity procurement costs. As technology continues to advance and its adoption expands, the role of energy storage in promoting the transition to clean energy, refining the energy structure, and achieving green and low-carbon development goals becomes increasingly critical. This project provides a replicable technical solution for realizing the dual-carbon objectives, thereby supporting the implementation of sustainable development strategies.