Due to the large number of photovoltaic power plant modules and strings, inefficient power generation is inevitable in each string inverter or combiner box unit during actual operation due to factors such as module quality issues, harsh environmental conditions, and design and construction defects. Identifying, analyzing, and resolving inefficient power generation units is crucial for improving the power plant's output. To encourage photovoltaic power plant operation and maintenance personnel to pay attention to inefficient units, this paper focuses on the O2O (Online to Offline) operation and maintenance concept, guiding them to effectively utilize centralized monitoring systems for online analysis and offline diagnosis, enabling them to identify, analyze, and address inefficient units.
Characteristics and troubleshooting strategies for inefficient power generation units
■ Inefficient power generation unit: This is determined by the fixed properties of the components in a certain string. That is, if a string has inefficient components or low-power mixed components, the power generation or power generation hours of the inverter corresponding to that string will be consistently lower than those of a normal string for several consecutive days. Generally, there will be no phenomenon of fluctuating between high and low.
■ Inefficient string operation: environmental factors, construction factors, design factors, orientation issues, power grid curtailment, and others.
Before locating inefficient strings, maintenance personnel must be thoroughly familiar with the basic information of the power station, such as the installed capacity, inverter model, number of connected strings, number of modules in each string, and module specifications. For power stations using string inverters, all inverters must be analyzed first to identify inefficient inverter generating units. To quickly pinpoint inefficient units, string inverters or DC combiner boxes are used as initial investigation units. Through on-site verification or other diagnostic methods, the problem can be narrowed down to a specific string branch or module. This approach also applies to centralized power stations using combiner boxes. The following methods and approaches do not necessarily need to be strictly followed; on-site inspections can be based on experience. While the search process varies from person to person, obtaining the same result through different methods is acceptable.
A simple approach to analyzing inefficient inverter power generation through data analysis and on-site investigation:
1) First step: Use the backend to analyze whether the capacity of each inverter is consistent with the actual capacity. If there is a discrepancy, it is necessary to reconfigure. Export the daily effective power generation hours data of each inverter for about one month (it needs to be calculated based on the actual inverter array to prevent incorrect results from leading to misjudgment. This is very important. Each power station should compile a string capacity information table, i.e., string inverter number, number of connected strings, number of modules in series, module power level, etc., which must be verified with the site one by one). Determine if there are inefficient inverters. The original data needs to be saved.
2) Second: Identify inefficient strings or components: ① Conduct online diagnostics through monitoring systems, dispersion rate analysis, current benchmarking analysis, etc. ② Alternatively, you can skip online dispersion rate analysis and directly analyze the causes of inefficient inverter power generation through on-site inspections, following your own approach.
3) Third: Based on the on-site diagnosis of the surrounding environment in the second step, conduct voltage and current tests if necessary, analyze the causes (starting from internal and external factors), and make relevant records (take photos).
4) Fourth: Propose solutions and assess their feasibility. Based on feasible measures, rectify inefficient series and keep corresponding records.
For power plants with centralized inverters, the following methods can be used:
1) Using centralized inverters as the analysis unit, the analysis of inverter power generation hours over a continuous month reveals inefficient inverter units.
2) For inefficient inverter units, export the historical 5-minute branch current data and use string current dispersion analysis. Focus on strings with large dispersion. If the dispersion is normal, but the overall branch current of the combiner box is low (using the current benchmarking method below), the specific cause needs to be verified on-site, and the branch components are likely inefficient.
For western power plants experiencing power curtailment, data analysis is not recommended when photovoltaic (PV) output exceeds the planned output period issued by the provincial dispatch center. At this time, it is difficult to rule out the factors causing the curtailment. Therefore, analysis should be conducted only when irradiance is relatively high and PV output is less than the AGC (Automatic Generation Control) planned output value, or by querying historical power generation data and selecting a period with a very low overall curtailment rate for analysis.
Inefficient string or component location locking methods
1. Online diagnostic methods
The purpose of online diagnostics is to conduct preliminary analysis and screening of inefficient power generation units (modules or strings) in the power plant, especially through methods such as comparing power generation hours. Methods like string current dispersion analysis and current benchmarking require maintenance personnel to possess certain big data analysis capabilities. Maintenance personnel can make judgments through offline diagnostics, which may require the use of certain testing methods (such as multimeters, clamp meters, etc.) or by checking the surrounding environment of the strings for issues like shadows or obstructions, to confirm and analyze the causes of inefficiency.
1.1 Instantaneous Analysis of Monitoring System
Figure 1 shows a sub-array inverter. As can be seen, branches 5 and 6 have low operating voltages, only around 290V, while the normal branch voltage is around 587V. However, the current is basically the same as the other branches. In this situation, maintenance personnel need to bring multimeters, clamp meters, and other tools to the site to check for problems in these two branches.
1.2 Dispersion Analysis
Using the backend management system, compare the current and voltage data of each branch of the inefficient inverter with those of the normal inverter to identify strings with lower current or voltage. Taking a power plant as an example, if a certain array inverter has the lowest generating hours, it's necessary to analyze the string current dispersion rate. The analysis of current dispersion rate is best done on a sunny day with good irradiance. On cloudy or rainy days, the string current is lower, making it difficult to reflect differences between strings. For example, by observing the current data of each branch of the inverter, if the current of the second branch is found to be lower between 15:00 and 18:00, the problem can be pinpointed to the second branch. The string number of this inverter should be noted for future on-site verification. Based on experience, if the string current is normal during the midday period with good irradiance but lower in the afternoon, it may be due to shading. (Ideally, the problem can be identified through real-time data from the monitoring system, allowing for a more targeted on-site investigation. If the real-time monitoring data cannot determine the issue, on-site investigation is necessary.)
1.3 Current Benchmarking Analysis
If the dispersion rate analysis fails to pinpoint a specific string, we need to further analyze whether the low power generation hours are caused by an overall low string current. If the overall string current corresponding to the inverter is low, then the string current dispersion rate may be normal. Judging solely from the dispersion rate is inaccurate; that is, a normal dispersion rate does not necessarily mean there is no problem, while an abnormal dispersion rate definitely indicates a problem. This is something that maintenance personnel need to pay attention to.
The purpose of current benchmarking is to identify strings with real problems based on their actual characteristics. Generally, we need to find dates or periods with relatively stable irradiance. Unstable irradiance, fluctuating wildly, will affect our analysis, primarily because rapid irradiance changes can lead to differences in the actual tracking response accuracy of string inverters. The obtained output current may not accurately reflect the actual output current under irradiance, thus impacting our judgment. Current benchmarking requires that the component models between strings be consistent. High-power components generally have higher output currents than low-power components, but their different ratings make direct comparison impossible. In actual analysis, we compare string inverters with lower average currents to those with higher average currents. If a significant deviation is found, we need to retrieve the string capacity information table and check the component power used in that string. If the low current is due to the use of low-power components, this should be noted. If the component model used is the same as that of the high-current component, then the low-current string has a problem and requires on-site verification.
1.4 Summary of Online Diagnosis
1) Verify if there is a problem with the inverter capacity in the system backend. If it is correct, use the power generation hours automatically calculated by the backend. If the system backend capacity is incorrect, the power generation hours need to be calculated based on the actual inverter capacity.
2) Export the power generation hours of each inverter for the past month from the system backend (if the capacity is incorrect, calculate according to the actual capacity), rank them from smallest to largest, and filter the inverter with the highest number of hours.
3) Preliminary online diagnosis: On a day with good sunshine, export 5 minutes of historical data for the inefficient inverter (mainly the current of each branch in the string) to obtain the current dispersion rate of each branch. If the dispersion rate is normal, it indicates that the overall current of the inverter string branches is too low, which can be confirmed using the current benchmarking method described above.
2. Offline diagnostic methods
1.1 Check the surrounding environment
The main focus is on checking whether the strings corresponding to the inverter are affected by external environmental factors, such as shade, dust, weeds, towers, or utility poles. For mountain photovoltaic power stations, check whether the string is located in a valley, whether it faces due south, and whether its installation tilt angle is consistent with other strings, and take photos for documentation.
1.2 Test strings or components
For strings with low current detected by the online backend system, the current value measured with a clamp meter may differ from the backend reading. Check for damage to the photovoltaic modules, such as broken glass surfaces, internal cells, junction boxes, and photovoltaic cable insulation. Check for incorrect string wiring, as incorrect wiring can easily lead to low voltage in a particular string. Use a clamp meter to test the operating current of each branch of the corresponding string in the inverter, and simultaneously test a high-performing inverter as a benchmark, recording the string number and test data for subsequent analysis. If an infrared thermal imager is available, it can detect the presence of hot spots on the modules, whether the open-circuit voltage is normal, and whether the bypass diodes in the module junction box are overheating.
3 Inefficient handling measures
■ String MPPT Wiring Optimization: Currently, string optimizers typically use two strings per MPPT. Check if there is sufficient slack in the photovoltaic cables at the inverter input. If so, inefficient strings can be placed in the same MPPT. This is especially important for inverters with partially connected strings, making full use of available terminals. If necessary, use photovoltaic cables and MC4 connectors to connect the strings.
■Weed Obstruction Issues: In southern regions, during the rainy season, vegetation flourishes, leading to large areas of weeds around the array. This requires close monitoring, adjusting our inspection focus accordingly, and developing a suitable weeding plan. If any components are found to be obstructed by weeds during routine inspections, they must be removed immediately. Additionally, weeding tools should be readily available and replaced promptly when worn. Personal safety should be prioritized during weeding.
■ Change the wiring method: For vertical double-row installation brackets, the traditional U-shaped wiring has the drawback that the photovoltaic modules of the entire bracket will be affected by the shadow of the front row array, especially in winter, when the impact of shading is greater. Therefore, the wiring method can be changed to a straight line, that is, the upper row modules of adjacent brackets are connected in series to form a string, and the lower row modules are connected in a string. Sufficient cables and MC4 connectors need to be prepared.
■ Installing power optimizers: Each power station needs to count the strings or module units that can be installed with power optimizers. This mainly includes tall iron towers, utility poles, modules that are blocked from the left, right, front, and back, parapet walls (for distributed power stations), buildings that block some modules, and trees (where trees cannot be cut down), and count the number of modules that can be affected by the blockage.
4. Summary
The analysis and handling of inefficient units in photovoltaic power plants is a long-term task, falling under the category of refined management and operation and maintenance. Traditional extensive operation and maintenance management models pose a significant challenge to the analysis and handling of inefficiencies. Therefore, the analysis and handling of inefficient units cannot be separated from intelligent monitoring systems and data analysis and processing platforms. This paper elaborates on the definition of inefficiency, its types, and how to identify and analyze inefficient units in different types of power plants, such as string inverters and centralized inverters. Finally, it presents an economic benefit assessment through a typical case study, hoping to provide a starting point for the professional operation and maintenance of power plants in China.
