What Standard BMS Data Still Misses in Modern Battery Storage Systems

A battery management system is not a complete picture of a battery energy storage system's health. It is a picture of the electrochemical layer — cell voltage, pack current, temperature at specific sensor points, and the derived states calculated from these inputs. For a decade, the industry has treated that picture as largely sufficient. The growth in grid-scale BESS deployments, the accumulation of failure incident data, and the rising pressure of insurance and regulatory requirements are collectively making visible how much that picture leaves out.

Understanding what standard BMS data misses is not an academic exercise. EPRI's analysis of documented BESS failure incidents found that only 11 percent of failures originate in cell defects. The remainder trace to balance-of-system components, cooling infrastructure, integration quality, and operational factors — systems that standard BMS architecture does not monitor at all. The data gap is not primarily about sensor accuracy or algorithm sophistication within the BMS. It is about the fundamental scope boundary of what a BMS is designed to measure, and what lies outside that boundary at an architectural level.

What a Standard BMS Actually Measures

To understand the gaps, the starting point is clarity on what a standard BMS does measure. A production-grade BMS in a utility-scale BESS provides continuous monitoring across several data categories:

• Cell-level measurements: individual cell voltage, module-level temperature from thermistors at selected points, string current from shunt or Hall-effect sensors
• Derived state estimates: state of charge (SoC) calculated from coulomb counting combined with voltage-based correction, state of health (SoH) estimated from capacity fade and internal resistance changes observed over operating cycles
• Protection functions: overvoltage, undervoltage, overcurrent, and overtemperature thresholds that trigger contactor disconnection when limits are exceeded
• Cell balancing: passive or active equalization circuits that reduce voltage divergence across series-connected cells
• Communication: data output to SCADA, EMS, or cloud analytics platforms via CAN, Modbus, or proprietary protocols

These are the core functions and data streams that every BMS specification covers. The IEEE published its recommended practice for BMS in stationary energy storage applications in February 2025 under 2686-2024, which formalizes these functions along with interoperability, communication, and cybersecurity guidance. The standard is a significant step toward harmonization — and its scope accurately reflects what the industry considers to be the BMS domain.

What the scope definition also does is formalize what is not in the BMS domain: the thermal management system, the container environment, off-gas conditions, and balance-of-system health. These are treated as external systems that the BMS interacts with but does not own.

The SOH Estimation Problem — Accurate on Average, Blind to Divergence

State of health is arguably the most consequential data point a BESS operator needs, and it is also the one where standard BMS data is most systematically misleading. The gap operates at two levels: estimation methodology and spatial resolution.

At the methodology level, SOH is not a directly measurable quantity. It is inferred. The most common approach combines coulomb counting — integrating current over charge and discharge cycles — with periodic measurements of internal resistance and capacity using available operating data. These methods produce a pack-level or string-level SOH estimate that represents an average across all cells in that assembly. The estimate is useful for gross tracking of capacity fade over the asset's life, but it does not capture the distribution of health states within the pack.

Real battery packs are not homogeneous. Even cells from the same manufacturing batch exhibit production-related differences in capacity and impedance that are small at the start of life but compound over thousands of cycles. A pack with a mean SOH of 85 percent may contain individual cells at 70 percent and others at 95 percent. The BMS SoH number reports the former but not the latter. The cells at 70 percent are the ones most likely to reach voltage limits first during cycling, to generate excess heat under load, and to initiate the divergence that eventually forces the operator to derate or replace the entire string.

At the spatial resolution level, temperature sensing in standard BMS hardware is a sampling problem. A typical module may have one to four thermistors placed at selected locations — usually near the center of the module or at the terminal connections. The actual temperature distribution across a cell stack under load is non-uniform, with localized hot spots that can significantly exceed the average temperature measured at the sensor points. The Arrhenius relationship between temperature and degradation rate is well established: sustained operation 10°C above optimal approximately doubles the aging rate. A cell running 8°C hotter than its nearest thermistor, consistently, over three years, ages substantially faster than the BMS data suggests it should. The BMS reports a healthy average and misses the local condition driving accelerated degradation.

Internal State Invisibility — What Voltage and Current Cannot Tell You

The BMS measures terminal voltage and current — the external observables of what is happening electrochemically inside the cell. The internal states that actually determine how a cell is aging, whether it is approaching unsafe conditions, and how much usable capacity remains are not directly accessible through standard instrumentation. Several critical internal parameters fall into this category:

• Lithium plating: at low temperatures or high charge rates, lithium deposits on the anode surface instead of intercalating into the graphite structure. This produces metallic lithium deposits that reduce capacity, increase internal resistance, and can create internal short circuit conditions if the plating is severe enough to form dendrites. Standard BMS data — voltage, current, temperature — cannot reliably detect early-stage lithium plating. The voltage signature during charging can provide indirect signals, but these require specialized analysis not present in standard BMS firmware.
• State of lithiation gradients: in large-format cells in stacks, the local state of charge within different regions of the electrode is not uniform, particularly at high charge and discharge rates. The BMS measures terminal voltage, which reflects a mixed average of electrode states. Cells that appear balanced by voltage can have significant internal gradients that accelerate stress and local degradation.
• Electrolyte degradation: the electrolyte progressively degrades over the life of a lithium-ion cell through oxidation, reduction, and the formation of solid electrolyte interphase layers on electrode surfaces. Electrolyte degradation affects impedance, capacity, and eventually off-gas generation. Standard BMS instrumentation has no direct measurement pathway for electrolyte condition. The effect is observable only after it has already caused measurable capacity loss or impedance increase — by which point significant degradation has already occurred.

Electrochemical impedance spectroscopy offers a pathway to access some of these internal states non-invasively by probing the cell's frequency-domain response. EIS measurements yield parameters including ohmic resistance, charge transfer resistance, and double-layer capacitance that correlate with specific degradation mechanisms. EIS as a diagnostic technique is mature in laboratory settings, and research efforts are underway to implement it in operational BMS hardware without requiring system downtime. As of 2025, EIS-capable BMS hardware is available from specialized vendors but is not yet standard in utility-scale BESS deployments. The data gap it would address is real and documented.

The Cell-to-Pack Gap in Practice

A distinct category of BMS limitation is the aggregation problem — the translation from cell-level data to pack-level and system-level status. Most utility-scale BESS architectures monitor cell voltage at the module level but aggregate temperature and current data at the string or rack level. The result is a monitoring hierarchy where the most granular data — individual cell voltage — reaches the BMS, but complementary cell-level data — local temperature, local impedance — does not.

The consequence is that certain failure modes propagate silently through the hierarchy. A cell with elevated internal resistance generates more heat per unit of current than neighboring cells. If the temperature sensor is positioned at the module center rather than adjacent to the high-resistance cell, the localized thermal signature is averaged into the surrounding cells' readings. The BMS does not see a temperature anomaly. It sees a string within thermal limits. The cell continues degrading faster than its neighbors, the capacity divergence grows, and the string eventually hits a voltage limit that either triggers a protective disconnect or requires derating — at which point the cause has been present for months.

The table below summarizes the categories of internal and cell-level parameters that influence BESS degradation but are outside the measurement scope of standard BMS architecture:

Environmental and Balance-of-System Data — Entirely Outside BMS Scope

The most operationally significant data gap in standard BESS monitoring is not a limitation of BMS sensor technology or algorithm sophistication. It is the complete absence of monitoring for the systems adjacent to the battery stack that drive the majority of documented failure incidents.

Container humidity and dew point are not BMS data. The BMS has no humidity sensors. HVAC cycling inside a BESS container creates temperature differentials that produce condensation spikes — periods when relative humidity inside the container exceeds 75 percent — that occur dozens of times per day and are never captured in the BMS data stream. DNV's analysis of documented BESS fires identified condensation on electrical components from faulty humidity control as a direct fire cause. Korean ESS fire investigations found condensation combined with dust contamination caused insulation breakdown. Neither failure mechanism produces any signal in standard BMS data.

HVAC system health is not BMS data. The cooling infrastructure — compressors, heat exchangers, refrigerant circuits, fans — is the single most consequential determinant of how fast the batteries age under load. A cooling system operating at half its commissioning-level efficiency produces cells that run hotter on every cycle, accelerating degradation silently. The BMS sees the consequence — slightly elevated cell temperatures — but cannot distinguish between a thermal management system that is working normally in high ambient conditions and one that is degraded and delivering inadequate cooling at standard ambient. The cause and the symptom look identical in BMS data.

Off-gas concentration patterns are not BMS data. When cells approach thermal runaway, they emit volatile organic compounds, hydrogen, and CO that appear 5 to 20 minutes before the runaway event — the only reliable pre-runaway detection window. Standard BESS installations include gas detectors as required by NFPA 855, but these devices output binary threshold alarms that are not analyzed in the BMS context. The gas sensor and the BMS are separate systems with no data integration. The BMS can report all cells within voltage and temperature limits at the same moment a gas sensor is detecting early off-gas signatures from a developing thermal event.

This is the data architecture that underlies most documented BESS safety incidents: the BMS is functional, reporting normal cell-layer parameters, while the actual failure mechanism is developing in a system the BMS does not see.

What Complete BESS Monitoring Requires Beyond the BMS

The IEEE 2686-2024 recommended practice positions the BMS as a functionally distinct component of the BESS — a correct characterization that implicitly acknowledges the BMS is not the BESS monitoring layer, only one part of it. Complete monitoring coverage requires data from several additional systems:

• Environmental monitoring: continuous measurement of humidity, dew point, and condensation probability inside the container at multiple heights and locations; correlation with HVAC cycling events to detect humidity spikes during HVAC transitions
• HVAC health monitoring: compressor current signature analysis and vibration trending to detect bearing wear and refrigerant loss weeks before failure; continuous coefficient of performance tracking normalized to ambient temperature to detect efficiency degradation over time; HVAC power consumption trending relative to cooling output
• Off-gas analytics: pattern-based analysis across multiple gas species simultaneously — VOC, hydrogen, CO, CO₂ — to distinguish genuine pre-runaway signatures from environmental interference and HVAC cycling artifacts
• Thermal gradient monitoring: spatial temperature arrays across racks and enclosures to detect hotspot formation and uneven cooling distribution that threshold-based BMS temperature sensors miss

None of these monitoring functions require BMS integration, modification of the SCADA system, or changes to the battery OEM's equipment. They operate as an additional monitoring layer alongside existing BMS and SCADA infrastructure, processing their own sensor data at the edge and generating alerts based on patterns that the BMS data stream is architecturally incapable of producing.

The shift from viewing BMS data as a sufficient monitoring foundation to viewing it as one layer of a more complete monitoring architecture is the practical engineering response to what EPRI's failure data and field incident records consistently show. The BMS monitors the 11 percent of failure causes that originate in cell defects. The remaining failure causes require monitoring infrastructure that starts where the BMS scope ends.

Quick Overview

A standard BMS monitors cell voltage, current, temperature at selected sensor points, and derived state estimates including SoC and SoH. These data streams cover the electrochemical layer of battery performance and provide the protection functions that prevent cell-level threshold violations. EPRI's analysis of documented BESS failures found that only 11 percent of incidents originate in cell defects — the majority involve HVAC and cooling systems, container environmental conditions, balance-of-system components, and integration factors that standard BMS architecture does not monitor. Complete BESS monitoring requires environmental sensing, HVAC health tracking, off-gas analytics, and spatial thermal monitoring as separate layers operating alongside the BMS.

Key Applications

Grid-scale and C&I BESS assets where failure incident history shows balance-of-system failures exceed cell-level failures, stationary storage installations subject to insurance requirements mandating demonstrated hazard mitigation, BESS facilities in the first two years of operation where failure rates are documented to be highest, and multi-vendor BESS fleets where OEM-agnostic monitoring is needed without BMS or SCADA integration.

Benefits

Environmental monitoring captures humidity spikes and condensation probability — documented fire causes invisible to standard BMS. HVAC health scoring and COP trending detects cooling degradation months before it affects cell temperatures or triggers BMS alarms. Pattern-based off-gas analytics reduces alarm fatigue by distinguishing genuine pre-runaway signatures from environmental interference. Spatial temperature arrays detect hotspot formation that point-sensor BMS temperature monitoring systematically misses.

Challenges

Adding environmental and HVAC monitoring requires physical sensor deployment inside containers and HVAC enclosures. Edge processing systems need commissioning periods to establish site-specific baselines before anomaly detection becomes reliable. Integrating alert outputs from supplementary monitoring layers into existing O&M workflows requires organizational change alongside the technical deployment. SOH estimation at the cell level, rather than the string level, requires either higher-density instrumentation or physics-informed models that most deployed BMS platforms do not implement.

Outlook

IEEE 2686-2024 establishes the BMS as a defined component with explicit scope boundaries, which formally creates space for the industry to develop and specify the monitoring layers that complement it. EU battery regulation increasing transparency and data reporting requirements, combined with insurance market pressure following high-profile BESS incidents, is creating commercial incentives for operators to invest in monitoring coverage beyond the BMS. Edge AI platforms that deploy without BMS integration are the most practical near-term approach to closing the environmental and balance-of-system monitoring gap across existing fleets.

Related Terms

BMS, BESS, SoC, SoH, state of health estimation, thermal runaway, off-gas detection, HVAC monitoring, COP tracking, condensation monitoring, IEEE 2686-2024, EPRI BESS failure database, cell balancing, coulomb counting, internal resistance, electrochemical impedance spectroscopy, EIS, lithium plating, SEI layer, NFPA 855, edge AI, balance of system