This is the second piece in a three-part series exploring predictive maintenance of grid-scale operational battery energy storage systems for improved safety and operation. The first article can be found here.
Battery energy storage systems (BESS) are an increasingly popular grid resource, which could provide more resilient, reliable energy systems. While relatively new technology, BESS systems are expected to continue to grow in scale and volume as society continues its global clean energy transition.
One of the most significant challenges facing BESS systems is how to ensure high reliability and operational safety. Having accurate, robust data is paramount to improving standard practice for BESS operation and maintenance. Sufficient BESS data may offer the ability for not only increased system-wide performance but may also aide in predictive efforts for safety, operations, and preventative maintenance. While storage performance standards are still evolving, uniformity in what data available, collected, and used may help ensure BESS meet the ever-growing storage needs of the utility market.
Currently available energy storage systems offer a wide range of data accessibility. For example, some systems only provide inverter AC-level information to the end user. These sparse datasets can allow for integration of BESS but also pose challenges when it comes to longer-term reliability and emerging operation and maintenance practices. Alternatively, some systems present robust and granular datasets, offering insight into system behavior which can alert operators to potential hazards and facilitate targeted service or investigation. The research team at Electric Power Research Institute (EPRI) has analyzed data from both robust and sparse data fielded systems, working with numerous energy companies including Arizona company Salt River Project (SRP) as part of the Energy Storage Performance and Reliability Foresight project.
In a statement, SRP explained, “SRPs generation fleet includes one 25MW/100MWHr battery and two 10MW/40MWHr batteries. Peak shaving and energy arbitrage are the core functions of these batteries in our system. The advanced controls and data collection offered from the integration of the 25MW/100MWHr battery has given SRP the invaluable opportunity to analyze utility to battery control interactions, as well as battery performance. The Energy Storage Performance and Reliability Foresight project has allowed SRP to enhance its portfolio of battery analytics using site level and module level data, which allows for greater insight into battery performance.”
Systems offering more robust data present greater opportunities to analyze performance and derive best operational practices. In 2017, a two enclosure, 1 MW, 2MWh BESS site was deployed, testing a variety of applications. In this robust data system, visibility was available down to the DC cell-level, with 2380 cells per enclosure and refresh rates of 6 seconds, allowing application of sophisticated analysis techniques. Access to this robust data led to more in-depth findings.
A core metric which arose from this data analysis was the measured standby loss — the percentage of state-of-charge lost in a given period without any power flow in or out of the battery system. A portion of these losses can be attributed to parasitic loads, such as the thermal management system (for a BESS, not using auxiliary supplies), control power, and cell balancing. At this site, standby losses were measured from 2017 to 2021, as shown in Figure 1. Evidently, the distribution over average daily standby loss was bimodal, with “low,” and relatively “high” loss days.
With the robust dataset available, it was possible to retrospectively evaluate information at the cell level in each storage system container once the system operator realized these losses may bean indication on abnormal behavior. Figures 2 and 3 show the standby losses correlated with cell balancing on “low” and “high” loss days, respectively. On days with high losses, rack level balancing was observed and strongly linked to periods of major State of Charge (SOC) decline without any power flow in or out of the BESS.
Low loss days exhibited no notable cell balancing, highlighting that the parasitic loads responsible for high standby losses were attributed to internal balancing; high energy cells discharged into the low energy cells to narrow the SOC distribution on the cell level. Uniformly balanced cells are critical to ensuring maximum system performance with minimal degradation.
Frequent, and power-intensive, cell balancing could be indicative of cell or module level defects, imbalances, potential safety hazards in an early stage, or otherwise dysfunctional equipment. The high standby loss metric provides evidentiary value to a system owner to trigger maintenance and investigation before more serious symptoms or issues develop. Without the data to signal maintenance, or without sufficient granularity, issues like this may fly under the radar until they give off serious alarm at the system level — and when preventative maintenance and intervention is no longer feasible.
In addition to a robust electrical data for this system, module temperatures were available for analysis, as seen in Figure 4. Temperature deviations within individual containers of identical design varied and indicated problems with the thermal management system in Container 1. With the ability to track module temperatures across the enclosure, the HVAC system was able to be effectively modified and augmented to increase temperature uniformity. Without access to this level of individual module temperature, the lack of uniformity would not provide insight for needed temperature distribution improvement. Access to this level of data also provides opportunities for long-term health tracking, preventative maintenance and upgrades, and prolonged life.
As BESS installations grow and serve an increasingly critical role in utility operations, access to appropriate levels of BESS data will be needed to ensure performance expectations. The intricacies of BESS equipment present a challenge not only in terms of allowing independent performance analysis but also in terms of defining best operational practices. Traditional utility-based, routine, walk-through maintenance schedules need to be tuned to these new BESS assets. Increasing owner and operator data visibility can allow for a targeted approach for large scale O&M and efficient performance, as well as insight to degradation and problems that need to be addressed before they hinder operation.
EPRI’s Energy Storage Integration Council has generated numerous tools to aid understanding storage specifications, data guides, as well as operational reporting, including: Electrical Energy Storage Data Submission Guidelines, Version 2, Energy Storage Operations and Maintenance Tracker, Summary of Energy Storage Control Performance Metrics, and ESIC Energy Storage Technical Specification Template, version 3.0.
