2024 White Paper

Insights from EPRI’s Battery Energy Storage

Systems (BESS) Failure Incident Database
Analysis of Failure Root Cause

0
 INTRODUCTION
The global installed capacity of utility-scale battery energy storage systems (BESS)
has dramatically increased over the last five years. While recent fires afflicting
some of these BESS have garnered significant media attention, the overall rate of
incidents has sharply decreased,1 as lessons learned from early failure incidents
have been incorporated into new designs and best practices. Between 2018 and
2023, the global grid-scale BESS failure rate has dropped 97%. The battery indus-
try continues to engage in R&D activities to improve prevention and mitigation
measures, including development of a better understanding of the diverse causes
of BESS failures.
TABLE OF CONTENTS
Introduction................................. 2
Methodology............................... 3
The BESS Failure Incident
Database.................................. 3
Data Collection......................... 4
Classification of Failure
Incidents................................... 4

Results......................................... 5
Results Overview...................... 5
Root Causes of Incidents.......... 7
Failed Element.......................... 9
Biaxial Analysis........................10
Mitigations and
Recommendations...................12
Figure 1. Global Grid-Scale BESS Deployment and Failure Statistics
Looking Ahead.........................13
Several entities compile information on battery fires that have occurred in vari-
Conclusion..................................14
ous products (e.g., mobile, stationary, consumer product) categorized by differing
battery technologies (e.g., lead acid, lithium ion). EPRI has produced the most
comprehensive compilation of stationary BESS incidents, called the EPRI BESS
Incident Database,2 based on publicly accessible underlying data. Other notable
databases include UL’s Lithium-Ion Battery Incident Reporting3 and EV FireSafe.4

1 Technology Innovation Spotlight: Lithium Ion Battery Fires in the News. EPRI, Palo Alto, CA: 2023.
3002028411.
2 BESS Failure Incident Database. This was formerly known as the BESS Failure Event Database.
It has been renamed to the BESS Failure Incident Database to align with language used by the
emergency response community. An ‘incident’ according to the Federal Emergency Manage-
ment Agency (FEMA) is an occurrence, natural or man-made, that requires an emergency
response to protect life or property, while an ‘event’ is a planned, non-emergency activity. The
use of incident is prevalent, for example, in referring to the Incident Command, or Incident Com-
mand System used by public and private agencies to coordinate incident management opera-
tions, [Link]
3 Lithium-ion Battery Incident Reporting. UL Solutions. [Link]
battery-incident-reporting.
4 EV FireSafe Database. [Link]

0
2 | EPRI White Paper May 2024
The UL Lithium-Ion Battery Incident Reporting encompasses METHODOLOGY
incidents caused by utility-scale, C&I, and residential BESS, This report relies on data from EPRI’s BESS Failure Incident
as well as EVs, e-mobility, and consumer products. This Database along with findings from incident reports and
database focuses exclusively on lithium ion technologies. EV root case analyses and expert interviews conducted by the
FireSafe tracks EV and electric micro-mobility fires involv- authors to build robust descriptions of each event. Each
ing (though not necessarily caused by) the traction battery, incident from the database is categorized through a biaxial
and categorizes incidents by cause. Both the UL Lithium-Ion framework to allow for analysis of two distinct failure
Battery Incident Reporting and EV FireSafe provide statistics facets. BESS failures were classified by a) the root cause
and figures, but do not disclose details of individual failures of failure (design; manufacturing; integration, assembly &
or sources. construction; or operation); and b) by the element of the
There is currently no public resource that categorizes BESS BESS that experienced the failure (cell/module, controls, or
incidents by cause of failure. This information would pro- balance of system). The study examines the proportion of
vide industry-level insights on common and uncommon fail- failures sharing a root cause or responsible element, the re-
ure modes, and would help to prioritize needed mitigation lationship between root cause and the element experienc-
technology R&D. This knowledge is particularly important ing failure, and the trends in failure type and rate over time.
because individual incident details and root cause informa- Results from this analysis will inform the industry’s efforts
tion are not always easily accessible, but are crucial to im- to optimize safety research and product development.
prove safety and understand risk. Failure classification can
help determine the role of different components of a BESS, The BESS Failure Incident Database
from controls to battery cell/module, in contributing to EPRI’s BESS Failure Incident Database is the main source
an incident and in preventing future incidents. No current of data for this report. The database was initiated in 2021
federal, state, or local jurisdiction requires incident report- following the series of lithium ion BESS fires in South Korea
ing. Even in cases where detailed root cause investigations and the Surprise, AZ, incident in the US. The database
are conducted, legal barriers often prevent the results from gathers information on stationary BESS failure events for
being shared publicly. New York state encouraged Original commercial and industrial (C&I) and utility-scale BESS. This
Equipment Manufacturers (OEMs) to disclose root cause database defines utility-scale BESS as a system that is inter-
analyses (RCAs) after failure incidents, but stopped short connected to the grid, with no capacity limitations, while
of including a requirement for disclosure in their pending C&I systems could include behind-the-meter installations.
update5 to the fire code. Residential energy storage system failures are not tracked
by this database and were not considered in this report.
This report is intended to address the failure mode analysis
gap by developing a classification system that is practical for It contains incidents as far back as 2011 and continues to
both technical and non-technical stakeholders. Once cat- be updated with new incidents as they occur. The focus
egorized in a standardized manner, the aggregated failure of the database is on occurrences that had a wider public
data was analyzed to better understand trends in how, why, health and safety risk or impact, rather than on operational
and how infrequently BESS fail, and to provide recommen- failures where no additional risk to personnel or equipment
dations for future safety improvements. was present or likely. EPRI defines failure incident as an oc-
currence which resulted in increased safety risk, caused by
5 New York State Inter-Agency Fire Safety Working Group: Fire Code
a BESS system or component failure rather than an exog-
Recommendations. NYSERDA. Feb 6, 2024. [Link] enous cause of failure (e.g., wildfire impacting the BESS).
[Link]/-/media/Project/Nyserda/Files/Programs/Energy-Storage/
Draft-New-York-State-Inter-Agency-Fire-Safety-Working-Group-Fire-
[Link].

0
3 | EPRI White Paper May 2024
The database captures incidents occurring globally and Classification of Failure Incidents
cites information from publicly available sources, including Incidents can result from a variety of causes, such as water
media reports, published root cause analyses (RCA), and intrusion, retrofitting errors, operating conditions, cool-
corporate press releases. Source documents are identified ant leaks, temperature stress, quality control, component
by active searching of global English-language media, and manufacturing defects and other factors. For meaningful
passive collection of reports through keyword flagging on analysis, these causes were grouped into classifications.
internet websites and RSS feeds. Crowdsourced information Each failure incident with sufficient information was clas-
that can be verified through publicly available documenta- sified by root cause and by failed element. Definitions for
tion is also incorporated. EPRI has used academic publica- each classification are provided below:
tions, and collaborated with other organizations tracking
failures, to ensure all publicly known stationary BESS events Root Cause:
are captured. However, many incidents are not reported • Design
in news media, especially before 2018-19 when there was A failure due to planned architecture, layout, or func-
a renewed industry focus on safety. There is no guarantee tioning of the individual components or the energy
that the database captures every relevant BESS failure storage system as a whole. Design failures include
incident, nor that all project data related to an incident those due to a fundamental product flaw or lack of
is captured. Despite these caveats, this remains the most safeguards against reasonably foreseen misuse.
comprehensive stationary BESS failure database available. • Manufacturing
A failure due to a defect in an element of an energy
Data Collection storage system introduced in the manufacturing pro-
At the time of writing, the database contained 81 incidents. cess, including but not limited to, the introduction of
Of these, 26 incidents had sufficient information to assign foreign material into cells, forming to incorrect physical
tolerances, or missing or misassembled parts.
a root cause and to identify the element that experienced
failure. Certain incidents had published root cause analy- • Integration, Assembly & Construction
sis reports that explicitly noted the cause of failure. The A failure due to poor integration, component incompat-
ibility, incorrect installation of elements of an energy
remaining incidents were classified based on engineering
storage system or due to inadequate commissioning
judgement by subject matter experts at EPRI, TWAICE, and
procedures.
Pacific Northwest National Laboratory (PNNL). The authors
• Operation
reviewed publicly available technical details and inter-
A failure due to the charge, discharge, and rest behav-
viewed other industry experts involved in failure incident
ior of the energy storage system exceeding the design
analysis. No proprietary information was discussed in these
tolerances of an element of an energy storage system
interviews nor used in the classification of the incidents.
or the system as a whole. Operational failures include,
Transparency on the cause of BESS failures continues to be but are not limited to, incorrect sensing of voltage,
limited. Battery OEMs and BESS integrators are often reluc- current, temperature, and other set point values, or
tant to disclose the cause of failure, and many investigation operation above designed temperature, C-rate, state of
charge, or voltage limits of the energy storage system.
reports are not released to the public. In several instances,
legal complications prevent site owners or manufacturers
Failed Element:
from divulging information about the nature of the fail-
• Cell/Module
ure. Aggregation and anonymization by a third-party can
A failure originating in the lithium ion cell or battery
encourage disclosure of such information to support safety module, the basic functional unit of the energy stor-
research advancement. age system. It consists of an assembly of electrodes,
electrolyte, casing, terminal, and usually separators.6

6 IEC Glossary. [Link]

orm&ievref=482-01-01.

0
4 | EPRI White Paper May 2024
 Cell failures usually begin with short circuits within the through the container caused electrical arcing within the
cell leading to eventual thermal runaway. They can system, leading to thermal runaway within one BESS unit
originate from poor cell design, manufacturing defects, on site. A water ingress point in the enclosure had been
incorrect installation, or cell abuse. created when an umbrella valve had been dislodged during
• Controls the improper installation of a vent shield. As a confounding
A failure in the sensing, logic circuits, and communica- factor, insulation loss alarms were not properly escalated
tion systems. Control systems coordinate the operation to the operator. Two days after the initial insulation alarms
of the ESS, including the battery management system were recorded, smoke and fire were reported to the fire
(BMS), energy management system (EMS), plant con-
department. Appropriate reporting of the insulation loss
trollers, and any subsystems. Controls failures include
alarms could have prevented escalation of the initial failure
those due to control system incompatibility, incorrect
into a fire that consumed the whole BESS unit. Therefore,
installation of the control system, defects leading to er-
the root cause was classified as both an integration, assem-
rors in sensors or controls, or inappropriate operation
limits. bly & construction failure in the BOS and a design failure of
the control system.
• Balance of System (BOS)
A failure in any of the elements of a BESS excluding the
cells, modules, and controls. BOS typically comprises RESULTS
of, but is not limited to: busbars, cabling, enclosures, Results Overview
power conversion systems, transformers, fire suppres- The following section contains insights from the 26 in-
sion systems, HVAC, or liquid cooling systems.
cidents that were classified. The distributions along the
An incident may have multiple failure elements or root biaxial classification system are examined in detail. As
causes; such incidents are assigned multiple classifica- described above, investigations into battery failures are
tions. The following example illustrates this classification often inconclusive, and there is a lack of transparency that
methodology. The Elkhorn battery facility located at Moss further limits the sharing of lessons learned. The industry
Landing, CA, experienced a fire on September 20, 2022. The experts who provided additional information beyond public
investigation report7 was shared publicly by Tesla (the BESS reports are based in the United States, so information on
manufacturer and integrator) and Pacific Gas & Electric (site incidents in other parts of the world is more limited in this
owner). The investigation found that rainwater intrusion report.

7 Report: Elkhorn Battery Energy Storage System Fire of September 20,

2022 - PGE Currents. [Link]
port-elkhorn-battery-energy-storage-system-fire-september-20-2022.

0
5 | EPRI White Paper May 2024
Figure 2. Fraction of BESS Failures with Identified Cause

Of the 9 incidents recorded in the BESS Failure Incident Database between 2011 and 2017, none were able to be classified,
while 36% of incidents between 2018 and the present had root causes identified. The availability of root cause information
starting in 2018 is an indication of both energy storage industry maturity as well as collective action and scrutiny on lithium
ion BESS safety.

Figure 3. BESS Failures with Identified Root Cause Over Time

0
6 | EPRI White Paper May 2024
Between 2017 and 2018, the lithium ion BESS deployments incidents, and provided summarized findings for the failures
increased by ~1 GW, more than doubling total global de- in aggregate. Subsequent academic papers provided more
ployment, and signaling the advent of the commercial BESS detailed root cause analyses for individual incidents.10
industry.8 The period between 2017-2019 also experienced
In the United States, a fire and explosion at a BESS facility
a spike in BESS failure incidents. Of the 30 incidents in the
in Surprise, AZ in 2019 injured four firefighters. Following
database between 2018 and 2019, 27 occurred in South Ko-
the incident, multiple root cause investigation reports were
rea. The Korean government had provided strong economic
released publicly, and safety became a priority issue for the
incentives for BESS, especially paired with solar PV genera-
energy storage industry in the US. In the subsequent years,
tion. The number of installed BESS in South Korea rose from
root cause investigations have occasionally been made
30 in 2013 to 947 in 2018. The rapid deployment was not
public to support industry learnings. However, the number
accompanied by robust safety standards and regulations,
of unclassified incidents in the preceding figures are a clear
which contributed to the failures.9 After the first spate of
indication of the continued challenges around failure data
fires, the South Korean government investigated the
access and transparency.

Root Causes of Incidents

Figure 4. Breakdown of BESS Failures by Root Cause

8 WoodMackenzie Energy Storage Database. Accessed Apr 17, 2024.

9 Im, D-H and J-B Chung. Social construction of fire accidents in bat-
tery energy storage systems in Korea. Journal of Energy Styorage, 10 Na, Y-U and J-W Jeon. Unraveling the Characteristics of ESS Fires in
Volume 71, 1 November 2023, 108192. [Link] South Korea: An In-Depth Analysis of ESS Fire Investigation Out-
est.2023.108192. comes, Fire, 6(10), 389, 2023. [Link]

0
7 | EPRI White Paper May 2024
Figure 4 shows the root cause classification for the 26 inci- the number of failures in the database that happen early in
dents considered in the analysis. Note that two incidents the project lifecycle. Referring back to Figure 1, deployment
were classified with dual root causes (Design as well as has increased significantly in recent years, and there are
Integration, Assembly & Construction), and the discrepancy relatively few older BESS that are operational. This may be
in total incidents is due to this double-counting. There is no why there are not many recorded failure incidents of aged
clear phase across the product lifecycle that is particularly systems so far. It remains to be seen if this trend will be
susceptible to failure, with all phases contributing to several sustained as systems being installed today age over time.
failures. EPRI has also gathered information on failure inci- Regardless, the majority (72%) of failures where the system
dents during manufacturing, transportation, and recycling age is known happen during construction, commission-
of batteries, which can be found in the ‘Other’ table in the ing, or within the first two years of operation. Integration,
database.11 These incidents were not considered for this Assembly & Construction is a critical phase in BESS risk miti-
analysis. gation. This root cause is examined further in subsequent
sections of this report.
Integration, Assembly & Construction was the most com-
mon root cause of failure in this analysis. Figure 5 highlights

Figure 5. BESS Age at Failure, where known

11 BESS Failure Event Database.

0
8 | EPRI White Paper May 2024
Manufacturing as a root cause has the fewest failures Electrical Rail Applications and Stationary Applications) and
attributed to it. This is most likely due to the difficulty in de- UL 1642 (Standard for Lithium Batteries) have improved the
finitively identifying a manufacturing defect as a root cause quality of manufactured batteries. Product certifications
with the loss of physical evidence after a fire or explosion. include quarterly and annual audits of factories to review
Earlier failures from 2018-2020 in particular may have quality control procedures, part inspection standards, and
involved cell or module manufacturing defects as a contrib- more. A recent report from Clean Energy Associates (CEA)
uting factor. Several product recalls from major EV manu- summarizes findings from BESS factory quality audits. Of
facturers during those years cited manufacturing issues by the identified issues in cell and module manufacturing, the
battery OEMs.12,13 Some residential ESS products were also majority were classified as minor issues, meaning they were
recalled during the same timeframe.14 It is important to not expected to impact safety in the short or long term.15
note that recalls do not definitively point to manufacturing
issues, but indicate the probable failure cause. In recent Failed Element
years, more robust product standards such as Underwriters The distribution of failure sources across BESS elements
Laboratory (UL) 1973 (Standard for Batteries for Use in Light (i.e. components) provides an insightful view of the vulner-
abilities within the system.

Figure 6. Breakdown of BESS Failures by Failed Element

12 Gitlin, J. Multiple recalls spark Fed investigation of LG’s electric car 15 BESS Quality Report. February 2024. Clean Energy Associates Insights.
batteries. Ars Technica, 2022. [Link]
multiple-recalls-spark-fed-investigation-of-lgs-electric-car-batteries/.
13 De Chant, T. GM recalls every Chevy Bolt ever made, blames LG
for faulty batteries. Ars Technica, 2021. [Link]
cars/2021/08/gm-recalls-every-chevy-bolt-ever-made-blames-lg-for-
faulty-batteries/.
14 United States Consumer Product Safety Commission. LG Energy
Solution Michigan Recalls Home Energy Storage Batteries Due to Fire
Hazard.

0
9 | EPRI White Paper May 2024
The BOS and controls account for the vast majority of failed were classified as controls failure rather than cell/module
components. The prevalence of BOS failures is corroborated since the failures could have been prevented if more limited
by the recent CEA report cited above, which found that operational windows were maintained.
nearly 50% of quality assurance items were in the BOS.
Only 3 incidents, or 11% of classified incidents, are attrib- Biaxial Analysis
uted directly to the cells. However, it should be noted that The following analysis looks at the combination of root
many of the failures classified as controls were related to cause and failed elements across the 26 incidents consid-
operational issues aimed at restricting cell state of charge ered.
(SOC), voltage and current, due to cell limitations. These

Figure 7. Relationship between Root Cause and Failed Element

1. Integration, Assembly & Construction and BOS failure due to assembly quality issues. During commis-
Integration is the most common root cause of BESS fail- sioning, a leak in the coolant system led to a fire that
ures, and the vast majority of incidents with this clas- spread across two BESS units.16
sification involved BOS components. These components 2. Operation and Controls
included DC and AC wiring, HVAC subsystems, and Operation is the second most common root cause,
safety elements such as the fire suppression system. and in all cases, the operation failure occurred in the
Lithium ion BESS contain components from multiple controls system. Seven of these incidents occurred in
suppliers, which are not necessarily designed to work 2018-2019 in South Korea, reflecting the early chal-
together. Integration is a critical part of the deployment lenges in determining appropriate BESS operation limits
and installation process to ensure all interfaces are for parameters such as voltage and SOC.
compatible and functional. A 2021 incident in Australia
at the Victoria Big Battery facility is an example of BOS
16 Lessons Learned from Past Failures Around the World, Session 6:
Responding to a Safety Event. EPRI, Palo Alto, CA: 2023. [Link]
[Link]/app/uploads/sites/163/2023/06/2023ESSRF_Session6.2_
Srinivasan_Lakshmi.pdf.

0
10 | EPRI White Paper May 2024
Figure 8. Root Cause and Failed Element Trends Over Time

Considering root cause trends over time, the bulk of inspected BESS units had defects in the fire suppression sys-
operational failures occurred in 2018-2019 when a signifi- tem, while 18% had thermal management system defects.18
cant number of BESS installed in South Korea experienced Both subsystems are critical for BESS safety. It is important
fires. Many of these were classified as operational failures to note that some of these failures occurred during the
since the SOC just before incidents was higher than recom- commissioning phase, when monitoring and communi-
mended limits. Investigation of the failures revealed that cations were not online, thus allowing leaks or isolation
a significant fraction of those failures occurred when the failures to cascade into large-scale fires. Site-specific hazard
SOC was above 90%.17 It is possible that these failures could assessments, monitoring, and procedures during commis-
also be attributed to manufacturing or design issues with sioning are recommended to avoid failures. EPRI published
the cell, but there was not sufficient evidence to make that an updated commissioning guide19 in 2023 through the
determination with confidence. Energy Storage Integration Council (ESIC) that captures rec-
ommendations and lessons learned to improve safety.
Integration-related failures have become more common.
The vast majority of these failures are related to poor build While the core battery technology has been in commer-
quality in the BOS, whether it is AC or DC wiring, coolant cial development since the 1990s, fully integrated BESS
systems, or safety systems such as water suppression pip- products arrived much later to market. BOS subsystems
ing. The CEA report corroborates these findings: 26% of like cooling, and especially safety components are not yet

17 Na, Y-U and J-W Jeon. Unraveling the Characteristics of ESS Fires in 18 BESS Quality Report. February 2024. Clean Energy Associates Insights.
South Korea: An In-Depth Analysis of ESS Fire Investigation Out- 19 ESIC Energy Storage Commissioning Guide. EPRI, Palo Alto, CA: 2023.
comes, Fire, 6(10), 389, 2023. [Link] 3002013972.

0
11 | EPRI White Paper May 2024
mature. BESS products have rapidly evolved from walk-in systems are needed to prevent integration, assembly, and
containers assembled on-site to module, pre-integrated sys- construction failures going forward.
tems. There is a diversity of products, architectures, ther-
mal management approaches etc., leading to integration Mitigations and Recommendations
challenges and the potential for incompatible interfaces or Reducing the risks associated with lithium ion BESS is a
unexpected interactions between components. complex task. Safety must be embedded at every scale of
a project, from material selection at the cell level to public
As deployment increases, many more individuals and
health impacts at the community level. As illustrated by this
organizations are working on BESS for the first time. New
analysis, safety must also to be considered at every phase
products without long operational histories are entering
of the project lifecycle, from design to operation to decom-
the market. A lack of experience and training in integration
missioning. For an overview of related lithium ion BESS
and assembly could have contributed to the assembly and
safety resources, including state-of-the-science documen-
construction-related failures in the recent years. Designs
tation of safety technology and hazard assessments, visit
may have flaws, or may not account for all operating and
EPRI’s Storage Wiki Safety Page.20
ambient conditions. For example, three of the four design-
related failures in 2023 occurred due to same BOS design The recommendations in this section focus on addressing
flaw in a BESS product. The enclosure design for systems in the gaps identified in this report. These are not intended to
New York and Idaho allowed water intrusion into the bat- be exhaustive. Preventative and mitigative measures against
tery compartment, leading to loss of isolation and thermal thermal runaway can take many forms, included compo-
runaway. Global storage deployment is expected to grow nents design/engineering, monitoring, procedural, and
exponentially, and many new entrants to the industry are site-level analyses. A comprehensive view of risk mitigation
expected. Sufficient training for manufacturers and integra- options can be found in the ESIC Energy Storage Reference
tors/developers and more extensive product quality control Hazard Mitigation Analysis.21

Table 1. Mitigations and Recommendations for Each Root Cause

ROOT CAUSE FAILED ELEMENT MITIGATIONS AND RECOMMENDATIONS

• Compliance with relevant codes and standards (UL, NFPA). Latest
revisions have incorporated lessons learned from past failures.
Design Controls, BOS
• Site-specific hazard assessments to consider all risks and failures.
• Robust sensing and monitoring to provide early alert for design failures.
• Workforce training and quality checks during energy storage
Integration/Assembly/ commissioning and installation.
BOS, Controls
Construction • System-level failure analysis, especially for interfaces between
components.
• Increased manufacturing quality controls.
Cell/Module, • Supplier quality verification.
Manufacturing
Controls • Robust system specifications.
• Factory acceptance testing.
• Battery monitoring and analytics to augment BMS operation, generating
Operation Controls
trends and predictive analyses to identify potential failures early.

20 Storage Safety. EPRI, Palo Alto, CA.

21 ESIC Energy Storage Reference Fire Hazard Mitigation Analysis. EPRI,
Palo Alto, CA: 2021. 3002023089.

0
12 | EPRI White Paper May 2024
Looking Ahead the EPRI database. A comparison of deployments in energy
This analysis is the first look at BESS failure root causes in capacity and reported failures in recent years by country
aggregate. For a significant fraction of the incidents, the points to a possible information gap. The number of failures
root cause was unknown, highlighting challenges in trans- is taken from the EPRI BESS Failure Incident Database, while
parency around BESS failures. Additionally, it is possible installed capacity numbers are from Rho Motion Consult-
that there are BESS failures that have not been captured in ing.22

Figure 9. Failures and Cumulative Deployment by Country

22 Rho Motion Consulting. Battery Energy Stationary Storage Monthly

Database.

0
13 | EPRI White Paper May 2024
EPRI and the other co-authors of this paper call for more CONCLUSION
transparency and data-sharing by the storage industry, Industry efforts to improve BESS safety during a period of
especially of root cause investigations. With additional rapid deployment expansion have led to a sharp decrease in
incident identification and classification, future work could the failure rate, but areas of needed improvement remain.
build on this initial report to provide deeper insights on This analysis demonstrated that all stages of the product
root causes and effectiveness of preventative measures. lifecycle contribute significantly to BESS safety and must be
EPRI continues to conduct research in BESS safety, and the rigorously engineered and diligently tested. Notably, the
current portfolio23 includes projects on thermal runaway data challenges the widespread assumption that the lithium
off-gas characterization, propagation mitigation technolo- ion battery cell is the primary cause of failure. The BOS and
gies, characterizing risks of siting BESS near critical infra- controls were the leading causes of failure, with the cell
structure, first responder training, and more. These activi- having a relatively small number of failures attributed to it.
ties are done in collaboration with a variety of industry Finally, this analysis is limited by the data that is publicly
stakeholders including electric power companies, OEMs, available. Of the known incidents, less than a third were
fire departments, and other research organizations. Ongo- assigned a cause of failure due to lack of sufficient informa-
ing regulatory development, voluntary industry efforts, tion. Industry transparency on details of BESS failures will
and focused research initiatives will continue to support be essential to more comprehensive analysis, to ongoing
increased BESS safety. safety research, and to future development that will ensure
the continued safe operation of BESS facilities.

23 Battery Energy Storage Fire Prevention and Mitigation Phase III. EPRI,
Palo Alto, CA: 2023. 3002028531.

0
14 | EPRI White Paper May 2024
0
DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES About EPRI
THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED Founded in 1972, EPRI is the world’s preeminent independent, non-
BELOW AS AN ACCOUNT OF WORK SPONSORED OR COSPONSORED BY profit energy research and development organization, with offices
THE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, around the world. EPRI’s trusted experts collaborate with more than
ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S) 450 companies in 45 countries, driving innovation to ensure the public
BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM: has clean, safe, reliable, affordable, and equitable access to electricity
across the globe. Together, we are shaping the future of energy.
(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER,
EXPRESS OR IMPLIED, (I) WITH RESPECT TO THE USE OF ANY INFORMA- About TWAICE
TION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN TWAICE provides predictive analytics software for companies working
THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS FOR A with batteries addressing key challenges throughout the entire lifecycle.
PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON Customers using TWAICE de-risk their battery business and outperform
OR INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PAR- their peers by increasing battery performance and lifetime. Uniquely
TY’S INTELLECTUAL PROPERTY, OR (III) THAT THIS DOCUMENT IS SUIT- combining deep battery knowledge and artificial intelligence on a scal-
ABLE TO ANY PARTICULAR USER’S CIRCUMSTANCE; OR able analytics platform, TWAICE is committed to increasing the lifetime,
(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY efficiency, safety, and sustainability of the products that power the
WHATSOEVER (INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF economy of tomorrow.
EPRI OR ANY EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE POSSI- About PNNL
BILITY OF SUCH DAMAGES) RESULTING FROM YOUR SELECTION OR USE
Mission statement: We transform the world through courageous discov-
OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD,
ery and innovation.
PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT.
PNNL takes on some of the world’s toughest research challenges and
REFERENCE HEREIN TO ANY SPECIFIC COMMERCIAL PRODUCT, PRO- biggest scientific questions in chemistry, Earth sciences, biology, and
CESS, OR SERVICE BY ITS TRADE NAME, TRADEMARK, MANUFACTURER, data science. We do the difficult, demanding, and exciting fundamental
OR OTHERWISE, DOES NOT NECESSARILY CONSTITUTE OR IMPLY ITS research to discover new knowledge and apply it to innovations in tech-
ENDORSEMENT, RECOMMENDATION, OR FAVORING BY EPRI. nology for our energy future and national security. ​​​​​​​
This report was co-authored by EPRI, TWAICE, and PNNL. We are a U.S. Department of Energy national lab with Pacific Northwest
• EPRI: Lakshmi Srinivasan, Eula Billaut, and Stephanie Shaw roots and global reach. Whether we are unlocking the mysteries of

• TWAICE: Ryan Franks, Sebastian Becker, Jordi Mueller

Earth’s climate, helping modernize the U.S. electric power grid, or safe-
guarding ports around the world from nuclear smuggling, we accept
• PNNL: Matthew Paiss
great challenges for one purpose: to create a world that is safer, cleaner,
more prosperous, and more secure.

EPRI CONTACTS
LAKSHMI SRINIVASAN , Principal Team Lead
202.293.7512, lsrinivasan@[Link]
STEPHANIE SHAW , Technical Executive
650.855.2353, sshaw@[Link]
EULA BILLAUT , Project Engineer
650.855.1019, ebillaut@[Link]

EPRI Customer Assistance Center

For more information, contact: 800.313.3774 • askepri@[Link]

3002030360 May 2024

EPRI
3420 Hillview Avenue, Palo Alto, California 94304-1338 USA • 650.855.2121 • [Link]
© 2024 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research Institute, EPRI, and TOGETHER…SHAPING THE FUTURE OF ENERGY are
registered marks of the Electric Power Research
0 Institute, Inc. in the U.S. and worldwide.