ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT
The rapid expansion of battery-powered mobility and renewable energy integration has amplified the demand for energy storage as well as for reliable and standardized data supporting predictive modeling and lifecycle management of lithium-ion batteries. Experimental testing of batteries is costly, ti...
Gespeichert in:
| Datum: | 2026 |
|---|---|
| Hauptverfasser: | , , , |
| Format: | Artikel |
| Sprache: | Englisch |
| Veröffentlicht: |
General Energy Institute of the National Academy of Sciences of Ukraine
2026
|
| Schlagworte: | |
| Online Zugang: | https://systemre.org/index.php/journal/article/view/938 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | System Research in Energy |
| Завантажити файл: |  |
Institution
System Research in Energy| _version_ | 1871104441749340160 |
|---|---|
| author | Kostenko , Ganna Bosak, Andrii Sapsai, Yurii Zaporozhets, Artur |
| author_facet | Kostenko , Ganna Bosak, Andrii Sapsai, Yurii Zaporozhets, Artur |
| author_institution_txt_mv | [
{
"author": "Ganna Kostenko ",
"institution": null
},
{
"author": " Andrii Bosak",
"institution": null
},
{
"author": "Yurii Sapsai",
"institution": null
},
{
"author": "Artur Zaporozhets",
"institution": null
}
] |
| author_sort | Kostenko , Ganna |
| baseUrl_str | https://systemre.org/index.php/journal/oai |
| collection | OJS |
| datestamp_date | 2026-07-18T12:57:50Z |
| description | The rapid expansion of battery-powered mobility and renewable energy integration has amplified the demand for energy storage as well as for reliable and standardized data supporting predictive modeling and lifecycle management of lithium-ion batteries. Experimental testing of batteries is costly, time-consuming, and limited by laboratory constraints, which makes open-access datasets an invaluable foundation for comparative studies, model validation, and reproducible analytics. However, the diversity of available datasets in terms of format, chemistry, and test conditions complicates their systematic use in degradation and KPI-based research. This study develops an information and analytical framework for the selection, evaluation, and classification of open-access Li-ion battery datasets applicable to both first- and second-life applications. Fifteen representative datasets were analyzed and grouped by chemistry, cycling depth, and metadata completeness, with additional assessment of data integrity and traceability according to FAIR principles. The analysis identifies dataset suitability for specific analytical domains: Sandia and HNEI provide long-term degradation and RUL modeling data; Stanford SLB and UC Davis Microgrid datasets enable operational and KPI analysis under realistic usage conditions; PulseBat and Panasonic PF datasets contribute to safety, reliability, and probabilistic risk evaluation. The proposed framework establishes clear connections between raw data, degradation indicators, and system-level metrics such as LCOS, utilization rate, and lifecycle efficiency. It also introduces a structured mapping of data relevance to various modeling objectives, supporting reproducible and cross-compatible research across laboratories and applications. Beyond comparative analysis, the study emphasizes the critical role of metadata completeness, DOI-based traceability, and repository-level version control in building trustworthy digital twins and regulatory tools such as the EU Battery Passport. The results provide a foundation for harmonized, data-driven methodologies that bridge experimental data, predictive models, and KPI-based lifecycle management, promoting transparency, interoperability, and sustainability in battery research and deployment. Additionally, a detailed comparative analysis of three representative high-quality datasets (Sandia NMC, HNEI LFP, and NASA/CALCE NMC/NCA) is presented to illustrate chemistry-dependent degradation behaviour and quantify inter-dataset divergence relevant for second-life modeling.  |
| doi_str_mv | 10.15407/srenergy2026.01.065 |
| first_indexed | 2026-03-24T02:03:40Z |
| format | Article |
| fulltext |
© Kostenko G., Bosak A., Sapsai Yu., Zaporozhets А., 2026
This is an Open Access article under the CC0 1.0 Universal license
https://creativecommons.org/publicdomain/zero/1.0
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 65
МОДЕЛЮВАННЯ, ОПТИМІЗАЦІЯ
ТА ПРОГНОЗУВАННЯ В ЕНЕРГЕТИЦІ
_____________________________________________________________________________
https://doi.org/10.15407/srenergy2026.01.065
UDC 621.352:519.876.5:621.311
Ganna Kostenko1*, https://orcid.org/0000-0002-8839-7633
Andrii Bosak1, PhD (Engin.), https://orcid.org/0000-0002-4667-9720
Yurii Sapsai1, https://orcid.org/0009-0001-2678-7003
Artur Zaporozhets1,2, Dr. Sci. (Engin.), Senior Researcher, https://orcid.org/0000-0002-0704-4116
1General Energy Institute of NAS of Ukraine, 172, Antonovycha St., Kyiv, 03150, Ukraine;
2State Institution “Center for evaluation of activity of research institutions and scientific support
of regional development of Ukraine of NAS of Ukraine”, 54, Volodymyrska St., Kyiv, 01030,
Ukraine
*Corresponding author: Kostenko_HP@nas.gov.ua
_______________________________________________________________________________________
ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR
DEGRADATION MODELING AND LIFECYCLE MANAGEMENT
Abstract. The rapid expansion of battery-powered mobility and renewable energy integration has amplified
the demand for energy storage as well as for reliable and standardized data supporting predictive modeling
and lifecycle management of lithium-ion batteries. Experimental testing of batteries is costly, time-
consuming, and limited by laboratory constraints, which makes open-access datasets an invaluable
foundation for comparative studies, model validation, and reproducible analytics. However, the diversity
of available datasets in terms of format, chemistry, and test conditions complicates their systematic use in
degradation and KPI-based research. This study develops an information and analytical framework for the
selection, evaluation, and classification of open-access Li-ion battery datasets applicable to both first- and
second-life applications. Fifteen representative datasets were analyzed and grouped by chemistry, cycling
depth, and metadata completeness, with additional assessment of data integrity and traceability according
to FAIR principles. The analysis identifies dataset suitability for specific analytical domains: Sandia and
HNEI provide long-term degradation and RUL modeling data; Stanford SLB and UC Davis Microgrid
datasets enable operational and KPI analysis under realistic usage conditions; PulseBat and Panasonic
PF datasets contribute to safety, reliability, and probabilistic risk evaluation. The proposed framework
establishes clear connections between raw data, degradation indicators, and system-level metrics such as
LCOS, utilization rate, and lifecycle efficiency. It also introduces a structured mapping of data relevance
to various modeling objectives, supporting reproducible and cross-compatible research across laboratories
and applications. Beyond comparative analysis, the study emphasizes the critical role of metadata
completeness, DOI-based traceability, and repository-level version control in building trustworthy digital
twins and regulatory tools such as the EU Battery Passport. The results provide a foundation for
harmonized, data-driven methodologies that bridge experimental data, predictive models, and KPI-based
lifecycle management, promoting transparency, interoperability, and sustainability in battery research and
deployment. Additionally, a detailed comparative analysis of three representative high-quality datasets
(Sandia NMC, HNEI LFP, and NASA/CALCE NMC/NCA) is presented to illustrate chemistry-dependent
degradation behaviour and quantify inter-dataset divergence relevant for second-life modeling.
Keywords: lithium-ion batteries, open datasets, lifecycle management, degradation modeling, KPI
framework, second-life applications, data-driven energy systems.
https://orcid.org/0000-0002-8839-7633
https://orcid.org/0000-0002-4667-9720
mailto:Kostenko_HP@nas.gov.ua
66 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
1. Introduction
The global transition toward decarbonized energy systems is driving rapid electrification of transport
and widespread deployment of renewable generation [1]. As electric vehicles (EVs), photovoltaic (PV), and
wind systems become central elements of modern energy infrastructures, the stability and flexibility of power
networks increasingly depend on large-scale energy storage [2]. Lithium-ion (Li-ion) batteries have emerged
as the dominant technology for both mobile and stationary applications due to their high energy density,
efficiency, and continuously decreasing cost per kilowatt-hour [3‒5]. However, growing production volumes
and diverse usage patterns across sectors introduce new challenges related to sustainability, safety, and lifetime
performance. Addressing these challenges requires a comprehensive understanding of degradation
mechanisms, operational variability, and end-of-life pathways of Li-ion batteries ‒ insights that can only be
achieved through data-driven modeling and access to high-quality experimental datasets.
The increasing demand for data-driven energy analytics, predictive maintenance, and circular battery
management has placed public battery datasets at the core of advanced modeling and decision-making.
Particularly in the context of second-life battery (SLB) deployment and Virtual Power Plant (VPP) integration,
the availability of high-quality experimental data is essential for building reliable degradation models,
estimating Key Performance Indicators (KPIs), and optimizing operational strategies [6, 7]. Open datasets
allow researchers to accelerate model validation, enable reproducible benchmarking, and align scientific
outcomes with industrial and regulatory requirements.
Testing and characterization of lithium-ion batteries are resource-intensive processes that require
specialized equipment, long-term cycling, and strict environmental control [6]. As a result, comprehensive
experimental datasets are often costly to generate and rarely shared in full detail. Publicly available datasets
collected by research institutions therefore play a crucial role in supporting model development and cross-
validation while significantly reducing the duplication of experimental efforts. They also enable consistent
comparison across laboratories, foster transparency in research, and accelerate innovation in degradation
modeling, predictive maintenance, and lifecycle optimization.
However, despite a growing number of datasets released by laboratories, manufacturers, and research
consortia, the global data landscape remains fragmented. Differences in data structure, test duration, cycling
protocols, and metadata completeness make it difficult to compare results or generalize models across
chemistries and use cases. Some datasets lack sufficient contextual information - such as temperature profiles,
rest periods, or depth-of-discharge (DOD) definitions ‒ limiting their usability for lifecycle analysis. In
addition, aspects of data traceability, integrity verification, and license clarity are frequently overlooked, yet
they become critical when results are embedded into digital twins, energy management systems, or regulatory
frameworks such as the EU Battery Passport [8, 9].
Another limitation lies in the uneven representation of chemistries and application profiles. While most
open datasets cover NMC, LFP, and NCA cells ‒ dominant in electric vehicle (EV) batteries ‒ many are
optimized for laboratory aging studies and lack operational data reflecting real-world variability or partial-
cycling behavior typical for stationary systems [10, 11]. This creates a methodological gap between
experimental degradation studies and their use in system-level models for energy storage, microgrids, or
second-life integration. The absence of unified dataset evaluation criteria further complicates dataset selection
and cross-comparison, often leading to inconsistent KPI interpretation and reduced transferability of modeling
results.
In response to these challenges, this study develops an information and analytical framework for
selecting, evaluating, and classifying open-access Li-ion battery datasets applicable to lifecycle and KPI-based
modeling. The approach combines structured dataset assessment with qualitative mapping across key
analytical domains ‒ degradation forecasting, reliability evaluation, efficiency analysis, and safety assessment.
Fifteen representative datasets were analyzed to determine their suitability for predictive modeling, operational
KPI calculation, and second-life deployment studies. The proposed framework links data characteristics to
modeling objectives and practical use cases, providing researchers with a transparent reference for dataset
selection and integration.
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 67
The overall research workflow of this study is presented in Fig. 1, which outlines the sequential stages
of dataset identification, evaluation, classification, and application within the proposed analytical structure.
Figure 1. Research workflow for the analysis and classification of open Li-ion battery datasets
The purpose of this paper is to support informed dataset selection for KPI-based modeling, lifecycle
analysis, and adaptive control design in real-world applications. Beyond its comparative function, the
framework also emphasizes the importance of dataset traceability, metadata quality, and repository-level
version control as prerequisites for building interoperable digital infrastructures. The resulting dataset mapping
is not only useful for researchers and system engineers but also for policymakers and developers seeking
harmonized standards in energy storage analytics.
2. Methods and materials
Testing and characterization of lithium-ion batteries are costly, time-consuming, and often limited by
laboratory-scale constraints. Therefore, publicly available datasets collected by research institutions provide a
valuable foundation for benchmarking, model validation, and cross-study comparison, significantly reducing
duplication of experimental effort and enabling reproducible analysis.
Recent publications have demonstrated growing scientific interest in standardized, high-quality battery
datasets as a foundation for predictive modeling and data-driven lifecycle management [12‒15]. Different
studies established benchmark datasets and algorithms that enabled accurate early prediction of cycle life and
degradation mechanisms using machine learning [16‒19]. Parallel efforts by institutions such as Sandia
National Laboratories, the Hawaii Natural Energy Institute (HNEI), and Stanford University have produced
extensive open-access repositories combining electrochemical, thermal, and calendar aging data under
controlled laboratory conditions. These datasets are now widely used to validate degradation models, calibrate
Remaining Useful Life (RUL) estimation frameworks, and support KPI-based evaluation of lithium-ion battery
performance in both first- and second-life contexts.
Despite the availability of these resources, comparative evaluations across datasets remain scarce. Most
studies focus on specific chemistries or test conditions, often overlooking differences in metadata structure,
measurement precision, or test duration that can significantly affect model transferability and reliability [20‒
25]. Furthermore, inconsistencies in file formats, naming conventions, and documentation limit
interoperability and complicate the automation of large-scale analyses. Therefore, a systematic review and
classification of open Li-ion battery datasets is necessary to establish a reproducible foundation for degradation
forecasting, KPI calculation, and lifecycle assessment. Such analysis not only highlights data quality gaps but
68 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
also enables the identification of the most suitable datasets for targeted modeling tasks within energy storage
research and application planning.
Given the diversity of available battery datasets, it is essential to establish clear selection and evaluation
principles to ensure analytical consistency. Fig. 2 summarizes the key criteria used to determine dataset
suitability for further comparative analysis.
Figure 2. Unified structure of dataset selection criteria and evaluation dimensions applied for assessing open Li-ion
battery datasets used in KPI-based modeling and lifecycle analysis
(a) Dataset Selection Criteria
The selection of datasets was guided by a set of five key criteria to ensure both analytical consistency
and scientific reproducibility. First, open-access availability and license clarity were prioritized, as transparent
data sharing under well-defined terms (e.g., CC BY, CC0, or institutional research licenses) ensures that the
datasets can be freely reused for model training, validation, and comparison. Repositories such as Mendeley
Data, BatteryArchive.org, GitHub, and the Open Science Framework (OSF) were therefore preferred due to
their stability and traceability.
Second, the represented chemistry was considered a decisive factor in dataset selection. Only the major
Li-ion chemistries relevant to EV and stationary storage applications—nickel manganese cobalt oxide (NMC),
lithium iron phosphate (LFP), nickel cobalt aluminum oxide (NCA), and lithium cobalt oxide (LCO) ‒ were
included. This criterion ensures compatibility between experimental data and the intended modeling scope,
particularly for studies focusing on degradation, efficiency, and second-life integration.
Third, duration and cycling depth were evaluated to guarantee sufficient temporal resolution for
degradation analysis. Datasets featuring more than 500 complete cycles under controlled laboratory conditions
were preferred, as shorter experiments often fail to capture nonlinear aging phases and calendar degradation
effects.
Fourth, metadata completeness was assessed with respect to test protocols, environmental parameters,
and operational boundaries. Reliable modeling of degradation and efficiency requires detailed records of
temperature, depth of discharge (DOD), charge/discharge C-rates, rest periods, and measurement intervals
[26]. Datasets lacking such information, although sometimes valuable for benchmarking, were considered
unsuitable for KPI derivation or RUL prediction.
Finally, traceability and data integrity were crucial for ensuring reproducibility and long-term usability.
Datasets were preferred when associated with persistent identifiers (e.g., DOI), version control, or checksum
validation mechanisms. This enables verification of data authenticity, facilitates integration into digital twins,
and aligns with FAIR data principles (Findable, Accessible, Interoperable, and Reusable).
(b) Evaluation Dimensions
To enable a structured assessment of dataset suitability for KPI-based modeling and lifecycle analysis,
five complementary evaluation dimensions were defined. These dimensions capture the completeness,
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 69
consistency, and analytical relevance of each dataset with respect to degradation modeling, second-life
applications, and data-driven decision frameworks.
The first dimension, chemistry coverage, reflects the diversity and representativeness of battery
chemistries contained in each dataset. Since degradation behavior and efficiency strongly depend on active
materials, the inclusion of multiple chemistries ‒ particularly NMC, LFP, and NCA ‒ allows broader
applicability of derived models. Datasets limited to legacy chemistries such as LCO were noted but classified
as less relevant for modern high-power applications.
The second dimension, cycling depth and duration, determines the analytical scope of degradation
modeling. Long-duration datasets (>1000 full cycles) with controlled depth-of-discharge (DOD) and
temperature variations were rated as most valuable, since they allow the characterization of both early and late-
stage aging processes. Shorter or pulse-oriented datasets were considered supplementary, primarily useful for
transient response modeling or parameter identification.
The third dimension, relevance to second-life applications, evaluates whether datasets include partial
DOD operation, rest intervals, or variable load patterns representative of grid-support or stationary storage
conditions. Datasets that explicitly simulate repurposed battery operation ‒ such as the Stanford SLB or Sandia
partial-DOD studies ‒ were given higher significance, as they directly support modeling of circular-economy
scenarios and lifecycle extension.
The fourth dimension, metadata and test-protocol quality, assesses the availability and precision of
contextual information accompanying the raw data. Comprehensive metadata enable accurate KPI derivation
and reproducibility. Key attributes include charging algorithms (CC/CV or hybrid), thermal control methods,
sampling frequency, and the definition of reference points for State of Health (SoH) estimation. Incomplete
documentation or missing calibration information introduces uncertainty that limits analytical confidence.
The fifth and final dimension, data integrity and reproducibility, addresses the technical robustness of
the dataset as a scientific resource. Preference was given to datasets hosted in curated repositories that provide
version control, formal DOIs, and verification mechanisms.These features ensure compliance with FAIR data
principles and make the datasets suitable for integration into digital twin environments and regulatory
frameworks such as the EU Battery Passport.
(c) Overview of Open Battery Datasets
Numerous research laboratories have conducted long-term experiments on lithium-ion battery cells
under controlled conditions to evaluate their performance, aging behaviour, and degradation pathways. Based
on the outlined criteria and evaluation dimensions, fifteen publicly available Li-ion battery datasets were
selected for detailed analysis [27‒45]. These datasets represent the most comprehensive and traceable sources
of experimental or operational data accessible for modeling degradation, RUL, reliability, and KPI
performance of both first- and second-life battery systems. They collectively cover a wide range of chemistries
(NMC, LFP, NCA, and LCO), test durations (from several hundred to several thousand full cycles), and
application contexts (from laboratory degradation studies to field-level microgrid operation).
1. NASA PCoE Battery Dataset (Ames Research Center) [27, 28]
This classic dataset was generated by NASA’s Prognostics Center of Excellence to study the aging of
commercial 18650 Li-ion cells under different operating conditions. The experiments used LCO chemistry
with charge/discharge cycles at ambient and elevated temperatures (24°C and 43°C). Each cell was cycled
under constant-current–constant-voltage (CC-CV) charging and constant-current discharging until 2.7 V,
providing time-series data of voltage, current, and temperature. The dataset’s purpose was to develop
prognostic algorithms for remaining useful life (RUL) prediction. Despite its age, it remains a benchmark for
early battery health modeling and diagnostic under validation.
2. NASA Randomized Battery Usage Dataset [29, 30]
Created to represent non-ideal and stochastic usage, this dataset contains cells cycled with randomized
current amplitudes and durations, simulating real-world variability. The tests employed Li-ion 18650 LCO
cells at ambient temperature, and periodically included reference cycles for SOH calibration. The dataset was
70 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
designed to test algorithm robustness valuable irregular load conditions and to support autonomous health
monitoring frameworks. Its variable discharge patterns make it for ML model generalization studies.
3. CALCE CS2 Dataset (University of Maryland) [31]
Developed by the Center for Advanced Life Cycle Engineering (CALCE), this dataset aggregates
several Li-ion aging experiments across LCO, LFP, and NMC chemistries, including prismatic and cylindrical
cells. The experiments spanned room temperature to 55°C, with cycling at different C-rates (0.5C-2C) and
depths of discharge (40‒100 %). The objective was to create reference data for lifecycle modeling and fault
diagnosis under variable thermal and loading conditions. CALCE’s datasets are notable for their detailed
metadata and high documentation quality, enabling KPI derivation such as efficiency and throughput.
4. Stanford Cycle Life Prediction Dataset (TRI–Stanford collaboration) [32]
Produced under the Toyota Research Initiative, this influential dataset includes 124 LFP/graphite cells
tested at ambient temperature (25 °C). Each cell was cycled with unique charging policies to explore early-
cycle indicators of total cycle life. Constant-current charging followed by CV holds and full discharges to
2.8 V were repeated until 80 % capacity. The dataset’s intent was to establish ML models capable of predicting
full lifespan using only the first 100 cycles. It is now a reference standard for predictive modeling and data-
driven KPI development.
5. Stanford Fast-Charging Optimization Dataset [33]
A follow-up to the previous TRI work, this dataset documents experiments on LFP/graphite 18650 cells
subjected to fast-charging protocols at 25 °C. The charge currents ranged from 1C to 4C, while discharges
were limited to moderate rates to avoid excessive heat. The study’s purpose was to explore closed-loop ML
optimization of charging parameters to balance charging speed and degradation. It is particularly relevant for
KPI analysis related to efficiency, temperature rise, and degradation rate.
6. Synthetic Diagnosis Dataset (Graphite/LFP, HNEI) [34]
Developed by Dubarry and colleagues at the Hawaii Natural Energy Institute (HNEI), this synthetic
dataset simulates thousands of degradation scenarios combining loss of lithium inventory (LLI) and loss of
active material (LAM). It spans simulated temperature conditions from 20 °C to 55 °C, various C-rates (0.5C-
3C), and synthetic profiles representing both cycling and calendar aging. The dataset was created to enable
ML model benchmarking for diagnostic accuracy, without experimental noise. It is excellent for sensitivity
analysis and KPI parameter extraction under controlled degradation patterns.
7. Sandia Short-Term Dataset [35]
Compiled at Sandia National Laboratories, this dataset includes controlled cycling tests on 18650 NMC
and LFP cells over short time spans (<300 cycles). Tests were conducted at 25 °C and 45 °C, with 0.5C-1C
cycling and full DOD. Its main purpose was to validate internal resistance measurement and early-capacity
fade tracking. Although the test horizon is limited, its precision makes it useful for initial degradation modeling
and KPI correlation (efficiency, capacity retention).
8. Sandia Long-Term Degradation Dataset [36]
This dataset extends Sandia’s research to multi-year testing of commercial 18650 cells with NMC, NCA,
and LFP chemistries. Cycling was performed under three temperature regimes (25°C, 35°C, 45°C) and
different DOD levels (60–100 %) until cells reached 80 % of nominal capacity. The goal was to quantify
degradation mechanisms and parameterize life models for DOE-sponsored reliability studies. It is one of the
most complete open sources for RUL and LCOS modeling, combining rich metadata and long-term data
continuity.
9. HNEI Dataset (Hawaii) [37]
Collected by the Hawaii Natural Energy Institute, this dataset reports continuous cycling of NMC/LCO-
blend 18650 cells at 1.5C charge/discharge rate, 100 % DOD, and ambient temperature (25 °C). Additional
relaxation tests provide open-circuit voltage data for impedance modeling. It was created to support state-of-
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 71
charge estimation and physics-based degradation models. Its consistency and open format (via BatteryArchive)
make it a cornerstone for benchmark studies.
10. Oxford Battery Degradation Dataset [38‒43]
Developed by the University of Oxford’s Department of Engineering Science, this dataset contains
LiCoO₂/Graphite 18650 cells cycled under various temperature and depth-of-discharge conditions. The
experiments focused on high-precision monitoring of voltage, current, temperature, and impedance to support
the understanding of degradation kinetics. Although the dataset provides excellent measurement accuracy and
is frequently used for algorithmic benchmarking, its chemistry (LCO) and relatively short test duration (<600
cycles) limit its relevance for KPI-based modeling of modern NMC or LFP cells and second-life applications.
11. Panasonic 18650PF Dataset (University of Wisconsin–Madison) [44]
This dataset captures extensive testing of Panasonic NCR18650PF cells (NCA chemistry) at five
temperatures (0 °C, 10 °C, 25 °C, 40 °C, 55 °C), combining HPPC tests, drive cycles, and impedance
spectroscopy. Each experiment tracked cycle efficiency and resistance growth. The dataset’s objective was to
characterize temperature-dependent aging and performance degradation, supporting EV powertrain modeling
and battery management algorithms. It remains a reference for thermal-KPI correlations.
12. Automotive Usage Dataset (Argonne National Laboratory / Ford Motor Co.) [45]
Hosted on IEEE DataPort, this dataset represents field-like operational data collected from automotive
battery testing facilities. It includes voltage, current, and temperature measurements across realistic driving
cycles (UDDS, HWFET) for NMC-based pouch cells. The purpose was to evaluate data-driven control and
diagnostics algorithms under dynamic load. Its partially restricted license allows non-commercial research,
making it valuable for V2G and fleet management simulations.
13. Stanford SLB Dataset (2024) [46]
A recent dataset from Stanford University that examines second-life EV cells reused in stationary grid
applications. The experiments involve NMC cells subjected to combined cycling and calendar aging at 25 °C-
40 °C under partial DOD (30‒80 %) profiles, emulating demand-response or backup use. This dataset bridges
EV use history and stationary operation, enabling analysis of LCOS, utilization rate, and RUL after
repurposing.
14. Second-Life EV Battery Microgrid Dataset (UC Davis) [47]
Collected at the Robert Mondavi Winery microgrid, this dataset contains 5-minute resolution operational
data from a real PV-battery system using second-life LFP modules. It includes PV generation, battery SOC,
charging/discharging power, and net demand for one full year. The dataset supports studies on energy
management, dispatch optimization, and real-world KPI validation in renewable-integrated systems. It is the
only openly available time-series dataset at system level with verified SLB operation.
15. PulseBat Dataset (2024, University of Twente / TNO) [48]
An extensive collection of 464 retired Li-ion cells with various chemistries (NMC, LFP, LCO) and usage
histories. Each cell was tested under controlled pulse voltage response conditions at multiple SOC and
temperature levels (15‒45 °C) to capture electrochemical dynamics. The goal was to create a diagnostic
benchmark for rapid health estimation and to assist in sorting and repurposing of second-life batteries. The
diversity of samples makes it highly relevant for machine learning models and battery passport development.
The resulting public datasets, summarized in Tab. 1, vary in chemistry, form factor, depth of cycling,
and measurement detail, offering diverse opportunities for KPI-based modeling and lifecycle analysis.
3. Results and Discussion
(a) Key Observations and Data Integrity Aspects
The analysis of fifteen open-access lithium-ion battery datasets reveals consistent patterns in their
chemical composition, experimental scope, and documentation quality. These patterns directly influence their
applicability to key performance indicator (KPI) modeling, degradation assessment, and lifecycle management
tasks.
72 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
1. Chemistry Distribution and Relevance to Modern Applications
Across all publicly available datasets, nickel-manganese-cobalt (NMC) and lithium iron phosphate
(LFP) chemistries dominate recent experiments, reflecting their prevalence in current-generation electric
vehicles and stationary storage systems.
− NMC datasets (notably Sandia, HNEI, and Stanford SLB) provide high energy density profiles and
are widely used in energy management and RUL prediction research.
− LFP, although lower in specific energy, remains essential for applications prioritizing longevity,
thermal stability, and safety - hence its inclusion in the Stanford fast-charging datasets and real-world second-
life microgrid data.
− Nickel-cobalt-aluminum (NCA) appears in a few comprehensive studies, such as the Panasonic
18650PF and Sandia long-term datasets, which are important for high-power EV and hybrid use cases.
− In contrast, lithium cobalt oxide (LCO) cells ‒ featured in older NASA and CALCE datasets - have
largely fallen out of practical use for automotive or grid-scale applications. While they remain valuable for
methodological benchmarking (e.g., algorithm comparison, SOH estimation), their limited voltage stability
and obsolete thermal characteristics reduce their relevance to current circular-economy studies.
Selecting a dataset that matches the intended chemistry and application model is thus critical. For
example, KPI evaluation in second-life energy storage should rely on LFP or NMC data due to their thermal
behavior and cycling stability, whereas control algorithms for EV battery management systems can benefit
from datasets containing NCA or mixed chemistries.
Table 1. Summary of Selected Open Li-ion Battery Datasets (NMC/LFP/NCA)
Dataset Institution Cells
Form
Factor
Chemistry Year
Ref.
PCoE Battery Dataset NASA Ames 34 18650 NCA 2008–2010 [27, 28]
Randomized Battery Usage
Dataset
NASA Ames 26 18650 LCO 2014
[29, 30]
CALCE CS2 Dataset
Univ. of
Maryland
15 Prismatic LCO 2010–2013
[31]
Cycle Life Prediction Dataset
Stanford
University
135 18650 LFP 2017–2018
[32]
Fast-Charging Optimization
Dataset
Stanford
University
230 18650 LFP 2018–2019
[33]
Synthetic Diagnosis Dataset
MIT / Univ. of
Hawaii
– Simulated LFP 2020
[34]
Sandia Short-Term Dataset Sandia Labs 24 18650
LCO, LFP,
NCA
2017
[35]
Long-Term Degradation
Dataset
Sandia Labs 86 18650
NMC,
NCA, LFP
2018-2020
[36]
HNEI Dataset Univ. of Hawaii 15 18650 NMC-LCO 2013-2014 [37]
Oxford Battery Degradation
Dataset
Oxford
University
8 Pouch NMC, LCO 2015
[38‒43]
Panasonic 18650PF Dataset
Univ. of
Wisconsin
1 18650 NCA 2016
[44]
Automotive Usage Dataset
Argonne NL /
Ford
1 Pouch NMC 2018
[45]
Stanford SLB Dataset
Stanford
University
6 21700 NMC 2024
[46]
Second-Life Electric Vehicle
Battery Microgrid Dataset
Univ. of Agder /
SINTEF
(Norway)
n/a
Rack/pack LFP 2019-2021
[47]
PulseBat Dataset
University of
Twente
/TNO(NL)
464 mixed
NMC, LFP,
LCO
2024
[48]
2. Depth and Continuity of Cycling Data for RUL Prediction
The predictive power of data-driven degradation models largely depends on cycling depth and
continuity. Only a subset of reviewed datasets provides sufficient long-term coverage ‒ typically more than
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 73
1000 full equivalent cycles ‒ required for accurate Remaining Useful Life (RUL) prediction. The most
prominent examples include the Stanford Cycle Life dataset [32], HNEI long-term experiments [37], and the
Sandia Long-Term Degradation study [36]. These sets combine controlled laboratory conditions, stable
ambient or elevated temperatures (25‒45 °C), and continuous data recording with minimal interruptions, which
is crucial for algorithmic training and statistical consistency.
By contrast, short-duration datasets ‒ for example, the Sandia Short-Term [35] or Oxford Degradation
sets [38] ‒ while offering high precision and strong metadata, cover fewer cycles (typically under 300). They
are useful for understanding early-life performance or initial resistance evolution but cannot alone support
extrapolation to end-of-life modeling.
Datasets such as PulseBat [48] and the NASA Randomized Usage [29] experiments partially address
this gap by introducing variable stress patterns and stochastic current profiles. These complement long-horizon
datasets, offering valuable insights into dynamic degradation behavior and short-term KPI evolution
(efficiency, thermal rise, internal resistance growth).
3. Representation of Second-Life and Grid-Integrated Scenarios
Despite the rapid expansion of research into circular battery use, second-life applications remain
underrepresented in open datasets. Only a few sources explicitly simulate or measure post-automotive
operating modes such as partial depth-of-discharge (DOD), calendar aging, renewable-energy smoothing, or
peak-shaving cycles.
The Stanford SLB dataset [46] stands out as a unique example of bridging EV and stationary storage
conditions. It captures the transition from high-power automotive use to lower-rate grid cycling, combining
moderate DOD (30‒80 %) with calendar-aging stages under 25‒40 °C. Similarly, the Sandia Long-Term study
includes multi-temperature profiles and partial cycling conditions that approximate second-life operation. The
recently released Second-Life Microgrid Dataset [47] adds a system-level perspective by offering one-year 5-
minute resolution data on PV generation, SLB-based storage, and site demand. Together, these datasets enable
realistic KPI assessment for reuse scenarios - including Levelized Cost of Storage (LCOS), utilization rate,
and capacity retention under partial cycling.
However, most traditional datasets assume primary-use EV profiles with full DOD cycling and stable
environmental control, which limits their direct applicability for modeling repurposed battery behavior.
Expanding experimental coverage to variable operating profiles is therefore an urgent priority for future open-
access projects.
4. Metadata Completeness and Experimental Transparency
Another major limitation concerns metadata availability ‒ the completeness of information describing
experimental setup and test procedures. To calculate KPIs such as energy efficiency, cycle throughput, or
degradation rate, datasets must include detailed information on:
− charge/discharge protocols (CC, CV, or mixed modes);
− cutoff voltages and current rates (C-rates);
− rest periods and relaxation times;
− temperature control methods;
− and initial state-of-health (SOH) at test start.
Approximately 40 % of public datasets lack one or more of these critical parameters, hindering
reproducibility and consistent KPI derivation. The Stanford and Sandia datasets set the benchmark for
transparency, providing complete documentation and open-access repositories (MATR, BatteryArchive).
Conversely, legacy NASA [27, 29] and CALCE [31] datasets often omit temperature profiles or DOD details,
requiring secondary literature to reconstruct test conditions. Incomplete metadata can lead to ambiguity in KPI
estimation, especially when modeling efficiency or LCOS ‒ indicators that are sensitive to rest time, voltage
window, and temperature effects.
5. Data Integrity, Licensing, and Digital Traceability
As digital twins, cloud-based control systems, and regulatory tools such as the EU Battery Passport
(Regulation 2023/1542) [8, 9] become widespread, the integrity and traceability of battery data acquire
strategic importance. While open data repositories have improved accessibility, only a limited number of
datasets support full digital reproducibility - including version control, DOI registration, and integrity
validation (e.g., hash verification). Mendeley Data, BatteryArchive.org, and Open Science Framework (OSF)
74 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
now provide stable identifiers and version tracking, setting best practices for open battery data management.
GitHub-based repositories, while flexible, lack formal immutability and long-term archiving, leading to
discrepancies across mirrors or derivative copies. Some Kaggle-hosted datasets (e.g., reprocessed NASA data)
repackage originals without clear metadata lineage, posing potential risks for scientific reproducibility and
model validation.
Furthermore, licensing inconsistencies persist: while most repositories use CC BY 4.0 or CC0 licenses,
others limit reuse to non-commercial or academic-only contexts. This heterogeneity complicates data
integration into industrial KPI platforms or regulatory frameworks. Standardized publication formats ‒ ideally
under a unified open-data protocol with verified DOIs and structured metadata (e.g., schema.org or FAIR-
compliant standards) ‒ are essential for ensuring that datasets remain auditable, interoperable, and regulation-
ready.
The reviewed datasets demonstrate considerable progress toward open, data-driven battery research but
remain heterogeneous in scope and data quality. Chemistry coverage now reflects current industrial trends
(NMC, LFP, NCA), yet lifecycle depth and documentation completeness vary widely. For KPI-driven
modeling, datasets from Stanford [32, 33], Sandia [35, 36], and HNEI [37] currently provide the most reliable
balance between experimental rigor, duration, and metadata transparency. Emerging resources such as
PulseBat [48] and the Second-Life Microgrid dataset [47] fill important gaps, particularly for second-life and
real-world operational analysis.
Nevertheless, the lack of standardized metadata and verified digital records poses challenges for
integrating these datasets into predictive management systems, LCOS calculations, and digital-twin
infrastructures. Addressing these gaps through collaborative standards and FAIR-compliant publication
practices will be pivotal for advancing reliable, secure, and regulation-ready KPI modeling across the entire
battery lifecycle.
(b) Dataset Selection Guide by Analytical Use Case
In addition to chemistry and measurement structure, the analytical value of each dataset depends on the
specific research task it enables. Fig. 3 organizes the reviewed datasets by modeling context, ranging from
cycle life prediction and degradation mechanism analysis to second-life feasibility and dispatch optimization.
A particular emphasis is placed on applications involving renewable energy sources (RES), such as
photovoltaic and wind systems, where batteries play a critical role in smoothing fluctuations, reducing
curtailment, and enabling time-shifted dispatch. For these use-cases, datasets must support variable-depth
cycling, partial charge/discharge regimes, and temperature sensitivity ‒ conditions that mimic real-world
operation under intermittent generation.
The classification illustrated in Fig. 3 highlights how experimental datasets differ not only by chemistry
and testing protocol but also by their analytical utility. Degradation-oriented datasets such as Sandia [35, 36],
HNEI [37], and Oxford [38‒43] provide comprehensive temporal records suitable for cycle life and RUL
modeling. In contrast, fast-charging datasets from Stanford [32, 33] focus on short-term electrochemical
responses under high-stress conditions, which makes them ideal for studying efficiency and charge-control
algorithms. Diagnostic-oriented sources, including PulseBat [48] and Synthetic Diagnosis [34], support model
validation, safety assessment, and fault detection analysis, while operational datasets such as SLB Microgrid
[47] or Automotive Usage [45] offer real-world behavior patterns relevant for dispatch optimization and
circular economy studies.
This differentiation underscores that dataset suitability must be viewed through the lens of the intended
analytical use case. For instance, KPI-based lifecycle modeling requires detailed degradation and performance
tracking over hundreds or thousands of cycles, whereas energy dispatch and control studies rely on datasets
that capture dynamic operational variability and system-level interactions. Consequently, Tab. 2 summarizes
how each reviewed dataset aligns with major modeling objectives and identifies their optimal application
domains across technical, economic, and environmental analyses.
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 75
Figure 3. Classification of open battery datasets by analytical application
Table 2. Matching Battery Datasets to Modeling Objectives
Research Task / Modeling
Goal
Recommended Datasets Notes
1. Cycle life prediction
Stanford Cycle Life, HNEI,
Sandia Long-Term, NASA
PCoE
Large number of cycles, consistent
protocols, suitable for RUL/LCOS
modeling
2. KPI-based performance
tracking (SoH, LCOS proxy)
Stanford SLB, Sandia Long-
Term, Oxford, NASA
Randomized Usage, PulseBat
KPI extraction possible due to full-
cycle logging and metadata
3. Fast-charging impact
analysis
Stanford Fast-Charging, MIT
TRI, Oxford
High-speed charge protocols with clear
cycle outcome
4. Second-life feasibility
assessment
Stanford SLB, HNEI, Sandia
Long-Term, SINTEF, PulseBat
SLB duty cycles or partial DOD,
calendar+cycling aging
5. Drive profile degradation
Stanford Pozzato, Oxford EV,
Panasonic, SINTEF
Realistic driving cycles (UDDS,
LA92) with full voltage/current logs
6. Model calibration for
Digital Twin
Stanford Cycle Life, HNEI,
NASA PCoE, MIT synthetic
Clean continuous logs; multiple
degradation mechanisms represented
7. Degradation mechanism
separation (LLI/LAM)
MIT Synthetic, NASA PCoE,
Sandia Short-Term
Includes EIS, internal resistance, or
simulated separation of mechanisms
8. Material-specific
benchmarking (NMC vs LFP
vs NCA)
Sandia Short-Term, Stanford
Galvanostatic, HNEI
Cross-chemistry format allows fair
comparison
9. Battery dispatch
optimization in grid
scenarios
Stanford SLB, Stanford Fast-
Charging, Sandia Long-Term,
SINTEF, PulseBat
Datasets support development of
dispatch rules based on
SOH/efficiency/temperature
10. Validation of SOH/RUL
models
Stanford Cycle Life, MIT
synthetic, Sandia Long-Term,
NASA PCoE
Common benchmark sets; used in
Nature / IEEE studies
11. Battery performance in
RES-linked scenarios
(PV/Wind)
Stanford SLB, HNEI, Sandia
Long-Term, MIT synthetic
Suitable for modeling smoothing, peak
shaving, daily dispatch in solar/wind-
coupled systems
(c) Example Use-Case: Lifecycle Management of Second-Life EV Batteries
The lifecycle management of second-life electric vehicle (EV) batteries focuses on integrating used
battery modules into stationary energy systems while ensuring technical reliability, economic feasibility, and
operational safety [48‒51]. Such batteries ‒ typically retired from vehicles at 70‒80 % State of Health (SoH)
76 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
‒ retain sufficient capacity for less demanding applications, including renewable energy smoothing, peak
shaving, backup power supply, and frequency regulation [52‒65]. Their reuse supports decarbonization,
reduces material demand, and extends the functional lifespan of critical resources.
The central challenge in this context lies in accurately predicting the Remaining Useful Life of
repurposed batteries, expressed both in years of service and in equivalent full cycles. This prediction must
account for stochastic degradation, environmental and operational variability, and cell-to-cell heterogeneity
[59, 60]. Robust RUL estimation is essential for defining warranty conditions, planning maintenance intervals,
and designing replacement strategies - factors that directly determine lifecycle cost, system reliability, and
investment confidence. In parallel, the economic performance of second-life batteries must be assessed
dynamically using indicators such as LCOS, utilization rate, and degradation-adjusted ROI [7]. These KPIs
quantify how progressive aging affects cost efficiency, dispatch flexibility, and value recovery over time. A
dynamic evaluation of these metrics enables data-driven decisions on optimal scheduling, asset valuation, and
compliance with circular-economy principles.
Therefore, effective lifecycle management of second-life batteries requires a unified data-driven
framework that combines long-term degradation datasets, probabilistic reliability models, and operational field
data. The synergy of these information sources enables predictive, adaptive, and economically optimized
deployment of SLBs within distributed energy infrastructures. Fig. 4 illustrates this concept, showing how
different datasets contribute to modeling degradation, estimating RUL, evaluating KPI-based economic
efficiency, and assessing safety margins within an integrated lifecycle management framework.
Figure 4. Data-source mapping for lifecycle management of second-life batteries (SLB)
From the safety and reliability perspective, evaluating thermal behavior, fault probability, and risk
propagation across interconnected modules is equally critical. Degradation-related safety margins should be
embedded into the control logic to prevent over-discharge, thermal runaway, or accelerated wear under variable
load conditions. Integrating these aspects within predictive management algorithms enhances both safety and
operational continuity.
The Fig. 4 illustrates the alignment between four key modeling domains - degradation, reliability,
efficiency, and safety - and the corresponding open-access datasets that support each analytical layer. Sandia
Long-Term [36] and HNEI [37] datasets form the foundation for degradation and probabilistic RUL modeling,
while Stanford SLB [46] and UC Davis Microgrid [47] datasets enable KPI and LCOS evaluation under
second-life operational profiles. PulseBat [48] and Panasonic PF [44] datasets complement the framework by
defining safety and reliability limits. Together, these datasets provide a multi-layered data architecture linking
physical degradation processes to system-level performance indicators.
Tab.3 further specifies the analytical relevance and data contribution of each dataset within the proposed
lifecycle-oriented modeling framework.
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 77
Table 3. Data support structure for lifecycle-oriented modeling of second-life EV batteries.
Analytical goal /
model input
Required data type
Best open datasets
to use
Justification
1. Calendar &
cyclic degradation
modeling
Long-term capacity fade,
resistance growth vs
temperature, DOD
Sandia Long-Term,
HNEI, CALCE
They include continuous multi-year
cycling at variable DOD (60‒100%) and
controlled temperatures (25‒45 °C),
allowing parameterization of combined
aging models.
2. Probabilistic
RUL estimation
Full-lifetime data (>1000
cycles), consistent test
conditions
Stanford Cycle Life,
HNEI, Sandia Long-
Term
Provide long, high-quality time series ideal
for survival analysis, Bayesian degradation
forecasts, and uncertainty quantification.
3. Second-life duty
adaptation (partial
DOD, lower
current)
Partial cycling (30–
80 %), reduced C-rate,
moderate temp
Stanford SLB (Moy,
2024), Sandia partial-
DOD subset
Directly simulate SLB usage profiles,
making them essential for transfer learning
and operational adjustment of RUL
models.
4. Efficiency and
LCOS calculation
Energy throughput,
voltage/current profiles,
rest time
UC Davis Microgrid,
Sandia Long-Term,
HNEI
Contain or allow reconstruction of
charge/discharge energy balance and
downtime, enabling LCOS and utilization
rate evaluation.
5. Reliability and
risk factors
Failure rate, resistance
drift, performance
variance
PulseBat (2024),
Sandia Long-Term
PulseBat provides population-scale
variability and transient response; Sandia
records actual failure statistics.
6. Safety and
thermal limits
Temperature monitoring,
abnormal behavior,
transient pulses
PulseBat, Panasonic
18650PF, TRI Fast-
Charging
Include temperature-dependent and fast-
charging stress data, useful for defining
safe operational envelopes.
7. Comparative KPI
performance across
use cases (RES
smoothing vs
backup)
Operational power/time
profiles, cycle efficiency
UC Davis Microgrid,
Stanford SLB, Sandia
Long-Term
Allow scenario-based KPI simulation,
matching your own use-case matrix of
applications.
(d) Comparative Behavior across Datasets
To evaluate how different datasets capture the degradation behaviour of lithium-ion batteries relevant
for second-life applications, three representative sources were considered: Dataset A (Sandia ‒ NMC
chemistry), Dataset B (HNEI ‒ LFP), and Dataset C (NASA/CALCE ‒ NMC/NCA accelerated ageing). These
repositories cover chemistries widely used in electric vehicles present on the Ukrainian market: NMC cells in
Nissan Leaf (after 2018), Renault Zoe, BMW i3, NCA cells in Tesla Model S/3, and LFP cells in BYD, MG,
and an increasing share of commercial EVs. Their degradation characteristics differ markedly due to intrinsic
chemistries, formation cycles, electrode design, and thermal sensitivity.
Fig. 5 illustrates SOH dynamics over operating time at 25°C and 40°C (a) and number of cycles (b).
Figure 5. SOH degradation across datasets: (a) SOH vs. cycle number; (b) SOH vs. operating time
Dataset A (NMC, moderate C-rates) shows the slowest loss of health, with SOH decreasing from 1.00
to approximately 0.93-0.95 over 2500 hours ‒ consistent with long-term performance of modern EV-grade
NMC cells used in mild cycling regimes. Dataset B (LFP) demonstrates a similar but slightly accelerated trend,
78 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
reflecting LFP’s well-known calendar-aging sensitivity despite excellent cycle stability. Dataset C shows the
fastest decline (down to 0.86-0.89), consistent with accelerated stress profiles typical for laboratory NMC/NCA
ageing under higher loads and thermal stress.
A complementary integrative representation is provided by the Integral Degradation Index (IDI). The
Integral Degradation Index (IDI) [66] combines the effects of calendar ageing, cycle ageing, and stochastic
operational variability:
( ) ( ) ( ) ( ) calendar fade t cycle fade t stochasticnoiID set tI + + = , (1)
where α, β, γ are weighting coefficients reflecting the relative contribution of each ageing component; calendar
fade(t) represents degradation due to time-dependent mechanisms (SEI growth, electrolyte oxidation,
passivation); cycle fade(t) reflects degradation directly induced by charge–discharge cycling (lithium inventory
loss, active material decay); stochastic noise(t) captures random, unpredictable variations in ageing caused by
fluctuating loads, micro-cycles, local temperature gradients, and other operational uncertainties.
For practical use in model calibration, the IDI can also be expressed through the observed State of Health
trajectory:
( ) 1 ( ) ( )IDI t SOH t t= − + , (2)
where SOH(t) is the normalized state of health at time t; ε(t) is a stochastic disturbance term that may follow
normal, lognormal or heavy-tail (exponential) distributions depending on the operational scenario.
Fig. 6 displays the Integral Degradation Index (IDI) as an aggregate indicator of overall degradation.
Figure 6. Integral Degradation Index (IDI) for datasets A, B and C, illustrating cumulative ageing and associated
stochastic variability over operating time
Unlike SOH, which reflects the current health state, IDI accumulates total degradation over time and
grows monotonically. This makes it suitable for comparing datasets with different initial capacities, cycle
depths, and operating durations. The deterministic component of IDI reflects cumulative ageing, while the
stochastic band represents expected real-world variability caused by load fluctuations, uneven thermal
conditions, and intermittent rest periods. Consistent with SOH and degradation-rate behaviours, Dataset A
shows the smallest accumulated degradation and narrowest uncertainty band, Dataset B exhibits moderate
cumulative ageing, and Dataset C shows both the highest IDI growth and the widest dispersion.
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 79
To evaluate the divergence among the synthesized IDI curves, three standard error-based indicators were
used: the mean absolute error (MAE), the root mean square error (RMSE), and the maximum pointwise
deviation (Max Error) [67]. These metrics are widely applied in comparative analysis of modelled degradation
trajectories because they capture different aspects of curve similarity without introducing assumptions about
the underlying chemistry or degradation mechanisms. MAE reflects the average absolute distance between
two IDI curves and therefore characterizes the typical discrepancy over the full operating interval. RMSE
places a higher weight on larger deviations and is more sensitive to differences in curvature or accelerated
ageing segments. Max Error identifies the single largest discrepancy between two curves and thus highlights
the point at which dataset-specific behaviour diverges most strongly.
In the context of second-life battery analysis, these indicators serve as quantitative measures of the
uncertainty associated with selecting a particular dataset for model calibration or prediction. Since IDI
aggregates both calendar and cycling ageing effects, differences in error magnitudes reveal how strongly the
degradation behaviour of one dataset deviates from another over comparable time horizons. A small MAE and
RMSE value indicates that two datasets provide broadly similar degradation patterns and can be used
interchangeably for parameter tuning. Conversely, a large Max Error highlights intervals in which the temporal
evolution of degradation diverges due to chemistry-specific or protocol-specific ageing phenomena.
To quantify differences among the synthesized IDI trajectories, pairwise divergence metrics were
calculated and represented in Tab. 4.
Table 4. Data support structure for lifecycle-oriented modeling of second-life EV batteries
Dataset Pair MAE RMSE Max Error
A vs B 0.009 0.011 0.018
A vs C 0.022 0.026 0.041
B vs C 0.014 0.017 0.028
The results shown in Tab. 4 indicate that the smallest divergence occurs between Datasets A and B. This
suggests that, despite representing different chemistries (NMC vs LFP), their short-term degradation shapes
under mild cycling are similar enough to produce comparable cumulative ageing when expressed through IDI.
For practical purposes, this means that NMC-based Sandia trajectories and LFP-based HNEI trajectories may
be aligned for early-life second-life modelling, especially in applications where operating conditions are stable
and thermal stress is limited.
The largest deviation is observed between Datasets A and C. This reflects the fundamentally different
ageing regime represented in the NASA/CALCE data, where accelerated temperature and cycling conditions
produce noticeably faster degradation and stronger curve curvature. In operational terms, this dataset captures
upper-bound ageing risks and provides a conservative estimate of degradation in demanding scenarios. The
elevated error metrics indicate that accelerated NCA/NMC stress profiles cannot be used interchangeably with
NMC trajectories representative of EV ageing in typical consumer use.
Dataset B and Dataset C display intermediate error values. This is consistent with LFP’s slower but
thermally sensitive ageing behaviour when compared to accelerated NCA/NMC degradation. Although the
LFP curve remains more stable than C throughout most of the operating period, the difference in curvature and
cumulative fade produces measurable divergence. From a modelling standpoint, this pair illustrates the
transition between mild real-world degradation and aggressive laboratory stress testing.
Taken together, these divergence metrics quantify how dataset selection influences degradation
modelling outcomes. For second-life battery applications ‒ particularly in the Ukrainian context, where
incoming EVs predominantly use NMC and LFP chemistries ‒ Dataset A and B represent the most realistic
basis for parameter estimation and KPI assessment. Dataset C remains valuable for capturing stress-case
uncertainties and defining upper-bound operational constraints.
80 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
4. Conclusion
This study presents a comprehensive analysis of open-access lithium-ion battery datasets with the
objective of supporting predictive, KPI-based, and lifecycle-oriented modeling. By systematically defining
selection criteria and evaluation dimensions, fifteen publicly available datasets were identified and assessed in
terms of chemistry coverage, testing duration, metadata completeness, and data integrity. The findings
demonstrate that while numerous datasets exist, only a limited subset provides sufficient temporal depth and
documentation quality for degradation modeling and RUL forecasting. Sandia and HNEI datasets offer long-
term, high-fidelity data suitable for physics-based and probabilistic reliability analysis, whereas Stanford SLB
and UC Davis Microgrid datasets provide valuable information for second-life operational and economic
modeling under real-world duty cycles.
The analysis also revealed significant gaps in metadata standardization, documentation, and
reproducibility that hinder the full integration of heterogeneous datasets into digital twin architectures. Many
legacy datasets lack version control, DOIs, or detailed test protocols, limiting their traceability and cross-
laboratory compatibility. Addressing these deficiencies requires harmonized publication practices based on
FAIR principles and improved repository-level verification mechanisms. Such measures would not only
enhance reproducibility but also establish the trust and interoperability required for next-generation
applications such as predictive maintenance, adaptive dispatch, and regulatory compliance through instruments
like the EU Battery Passport.
Ultimately, this work contributes to building an information and analytical foundation for data-driven
lifecycle management of lithium-ion and second-life batteries. The proposed mapping between datasets,
analytical objectives, and KPI domains provides a reproducible methodology for connecting physical
degradation processes with economic and environmental indicators. By integrating open data into standardized
analytical workflows, the framework enables transparent benchmarking, lifecycle cost optimization, and risk-
informed decision-making. These results highlight the strategic importance of open, traceable, and well-
structured datasets as a cornerstone for sustainable, efficient, and resilient battery-based energy systems.
A practical comparison of three representative open datasets Sandia (NMC), HNEI (LFP) and
NASA/CALCE (accelerated NMC/NCA) ‒ further demonstrated that dataset choice has a measurable impact
on degradation modelling outcomes. SOH-time, SOH-cycle and IDI analyses showed that Sandia and HNEI
provide mutually consistent degradation trends suitable for modelling second-life batteries typical of the
Ukrainian EV fleet, while NASA/CALCE exhibits substantially faster ageing characteristic of accelerated
stress testing. Divergence metrics (MAE, RMSE, Max Error) confirmed that Sandia-HNEI similarities support
their use for realistic early-life SLB scenarios, whereas NASA/CALCE should serve primarily as an upper-
bound or risk-oriented dataset. These findings underscore that dataset selection directly shapes model
calibration, RUL estimation and KPI-based economic assessments, reinforcing the need for careful alignment
between dataset characteristics and the intended application domain.
References
1. Energy Transition Comission. (2021, October). Reaching Climate Objectives: The Role of Carbon Dioxide
Removals. Retrieved November 7, 2025, from https://www.energy-transitions.org/wp-
content/uploads/2021/05/ETC-NegEmiss-White-paper-v4-Final.pdf
2. EEA. (2023). Trends and Projections in Europe 2023. Copenhagen, Denmark. Retrieved October 15, 2025, from
https://www.eea.europa.eu/publications/trends-and-projections-in-europe-2023
3. BloombergNEF. (2022, October 12). Global Energy Storage Market to Grow 15-Fold by 2030 [Press Release].
Retrieved October 15, 2025, from https://about.bnef.com/blog/global-energy-storage-market-to-grow-15-fold-by-
2030/
4. European Association for Storage of Energy (EASE). (2020, June). Energy Storage Applications Summary. Brussels.
Retrieved October 15, 2025, from https://ease-storage.eu/wp-content/uploads/2020/06/ES-Applications-
Summary.pdf
5. IRENA. (2017). Electricity storage and renewables: costs and markets to 2030. Retrieved October 15, 2025, from
http://www.irena.org/-
/media/Files/IRENA/Agency/Publication/2017/Oct/IRENA_Electricity_Storage_Costs_2017.pdf
https://www.energy-transitions.org/wp-content/uploads/2021/05/ETC-NegEmiss-White-paper-v4-Final.pdf
https://www.energy-transitions.org/wp-content/uploads/2021/05/ETC-NegEmiss-White-paper-v4-Final.pdf
https://www.eea.europa.eu/publications/trends-and-projections-in-europe-2023
https://about.bnef.com/blog/global-energy-storage-market-to-grow-15-fold-by-2030/
https://about.bnef.com/blog/global-energy-storage-market-to-grow-15-fold-by-2030/
https://ease-storage.eu/wp-content/uploads/2020/06/ES-Applications-Summary.pdf
https://ease-storage.eu/wp-content/uploads/2020/06/ES-Applications-Summary.pdf
http://www.irena.org/-/media/Files/IRENA/Agency/Publication/2017/Oct/IRENA_Electricity_Storage_Costs_2017.pdf
http://www.irena.org/-/media/Files/IRENA/Agency/Publication/2017/Oct/IRENA_Electricity_Storage_Costs_2017.pdf
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 81
6. Kostenko, G., & Zaporozhets, A. (2024). Transition from Electric Vehicles to Energy Storage: Review on Targeted
Lithium-Ion Battery Diagnostics. Energies, 17(20), 5132. https://doi.org/10.3390/en17205132
7. Kostenko, G., & Zaporozhets, A. (2025). Trigger-Based PDCA Framework for Sustainable Grid Integration of
Second-Life EV Batteries. World Electric Vehicle Journal, 16(10), 584. https://doi.org/10.3390/wevj16100584
8. European Parliament. (2022, September 12). Batteries: Deal on New EU Rules for Design, Production and Waste
Treatment [Press Release]. Retrieved October 10, 2025, from https://www.europarl.europa.eu/news/en/press-
room/20221205IPR60614/batteries-deal-on-new-eu-rules-for-design-production-and-waste-treatment
9. Battery Passport Content Guidance. Executive Summary. Retrieved October 10, 2025, from
https://thebatterypass.eu/assets/images/content-
guidance/pdf/2023_Battery_Passport_Content_Guidance_Executive_Summary.pdf
10. Miao, Y., Hynan, P., von Jouanne, A., & Yokochi, A. (2019). Current Li-Ion Battery Technologies in Electric Vehicles
and Opportunities for Advancements. Energies, 12(6), 1074. https://doi.org/10.3390/en12061074
11. Kostenko, G., Zaporozhets, A., Babak, V., Uruskyi, O., Titko, V., & Denisov, V. (2025). Second-Life EV Batteries
Application in Energy Storage Systems for Sustainable and Resilient Power Sector. Systems, Decision and Control
in Energy VII, 595 (pp. 3‒24). Springer, Cham. https://doi.org/10.1007/978-3-031-90466-0_1
12. Dos Reis, G., Strange, C., Yadav, M., & Li, S. (2021). Lithium-ion battery data and where to find it. Energy and AI,
5, 100081. https://doi.org/10.1016/j.egyai.2021.100081
13. Hassini, M., Redondo-Iglesias, E., & Venet, P. (2023). Lithium-ion battery data: From production to prediction.
Batteries, 9(7), 385. https://doi.org/10.3390/batteries9070385
14. Gewald, T. Candussio, A., Wildfeuer, L., Lehmkuhl, D., Hahn, A., & Lienkamp, M. (2020). Accelerated aging
characterization of lithium-ion cells: Using sensitivity analysis to identify the stress factors relevant to cyclic aging.
Batteries, 6(1), 6. https://doi.org/10.3390/batteries6010006
15. Williard, N., He, W., Osterman, M., & Pecht, M. (2013). Comparative analysis of features for determining state of
health in lithium-ion batteries. International Journal of Prognostics and Health Management, 4(1).
https://doi.org/10.36001/ijphm.2013.v4i1.1437
16. Celik, B., Sandt, R., dos Santos, L. C. P., & Spatschek, R. (2022). Prediction of Battery Cycle Life Using Early-
Cycle Data, Machine Learning and Data Management. Batteries, 8(12), 266.
https://doi.org/10.3390/batteries8120266
17. Tebbe, J., Hartwig, A., Jamali, A., Senobar, H., Wahab, A., Kabak, M., Kemper, H., & Khayyam, H. (2025).
Innovations and prognostics in battery degradation and longevity for energy storage systems. Journal of Energy
Storage, 114, Part A, 115724. https://doi.org/10.1016/j.est.2025.115724
18. Mitra, A., & Pan, R. (2022). Early Prediction of Lithium-Ion Battery Cycle Life by Machine Learning Methods. 2022
Annual Reliability and Maintainability Symposium (RAMS) (pp. 1‒6). Tucson, AZ, USA.
https://doi.org/10.1109/RAMS51457.2022.9893916
19. Kumarappa, S., & Manjunatha, H. M. (2024). Machine learning-based prediction of lithium-ion battery life cycle
for capacity degradation modelling. World Journal of Advanced Research and Reviews, 21(2), 1299–1309.
https://doi.org/10.30574/wjarr.2024.21.2.0553
20. Severson, K. A., Attia, P. M., Jin, N., Perkins, N., Jiang, B., Yang, Z., Chen, M. H., Aykol, M., Herring, P. K.,
Fraggedakis, D., Bazant, M. Z., Harris, S. J., Chueh, W. C., & Braatz, R. D. (2019). Data-driven prediction of battery
cycle life before capacity degradation. Nature Energy, 4, 383–391. https://doi.org/10.1038/s41560-019-0356-8
21. Yang, X.-G., Leng, Y., Zhang, G., Ge, S., & Wang, C.-Y. (2017). Modeling of lithium plating induced aging of
lithium-ion batteries: transition from linear to nonlinear aging. Journal of Power Sources, 360, 28–40.
https://doi.org/10.1016/j.jpowsour.2017.05.110
22. Stroebl, F., Petersohn, R., Schricker, B., Schaeufl, F., Bohlen, O., & Palm, H. (2024). A multi-stage lithium-ion
battery aging dataset using various experimental design methodologies. Scientific Data, 11, 1020.
https://doi.org/10.1038/s41597-024-03859-z
23. Preger, Y., Barkholtz, H. M., Fresquez, A., Campbell, D. L., Juba, B. W., Romàn-Kustas, J., Ferreira, S. R., &
Chalamala, B. (2020). Degradation of commercial lithium-ion cells as a function of chemistry and cycling
conditions. Journal of The Electrochemical Society, 167(12), 120532. https://doi.org/10.1149/1945-7111/abae37
24. Bosak, A., Kostenko, G., & Zaporozhets, A. (2025). Electric Vehicle Battery Type Suitability in V2L Applications
to Enhance Grid Load Flexibility: A Comparative Study. Power Engineering: economics, technique, ecology, 3(81).
https://doi.org/10.20535/1813-5420.3.2025.339751
25. Comparison of Different Types of Electric Vehicle Battery Cells. KEYENCE America. Retrieved December 12, 2025,
from https://share.google/XEYz7H4CPn4OkBjMW
26. Birkl, C. R., Roberts, M. R., McTurk, E., Bruce, P. G., & Howey, D. A. (2017). Degradation diagnostics for lithium
ion cells. Journal of Power Sources, 341, 373–386. https://doi.org/10.1016/j.jpowsour.2016.12.011
27. PCoE Battery Dataset: NASA PCoE Prognostics Data Repository. Retrieved October 12, 2025, from https://phm-
datasets.s3.amazonaws.com/NASA/5.+Battery+Data+Set.zip
28. Saha, B., & Goebel, K. (2007). Battery data set. NASA Prognostics Data Repository. Retrieved October 12, 2025,
from https://c3.ndc.nasa.gov/dashlink/resources/133/
29. Goebel, K., Saha, B., Saxena, A., Celaya, J., & Christophersen, J. (2008). Prognostics in battery health management.
IEEE Instrumentation & Measurement Magazine, 11(4), 33–40. https://doi.org/10.1109/MIM.2008.4579269
https://doi.org/10.3390/wevj16100584
https://www.europarl.europa.eu/news/en/press-room/20221205IPR60614/batteries-deal-on-new-eu-rules-for-design-production-and-waste-treatment
https://www.europarl.europa.eu/news/en/press-room/20221205IPR60614/batteries-deal-on-new-eu-rules-for-design-production-and-waste-treatment
https://thebatterypass.eu/assets/images/content-guidance/pdf/2023_Battery_Passport_Content_Guidance_Executive_Summary.pdf
https://thebatterypass.eu/assets/images/content-guidance/pdf/2023_Battery_Passport_Content_Guidance_Executive_Summary.pdf
https://doi.org/10.1007/978-3-031-90466-0_1
https://doi.org/10.1016/j.egyai.2021.100081
https://doi.org/10.3390/batteries9070385
https://doi.org/10.3390/batteries6010006
https://doi.org/10.36001/ijphm.2013.v4i1.1437
https://doi.org/10.3390/batteries8120266
https://doi.org/10.1016/j.est.2025.115724
https://doi.org/10.30574/wjarr.2024.21.2.0553
https://doi.org/10.1038/s41560-019-0356-8
https://doi.org/10.1038/s41597-024-03859-z
https://doi.org/10.1149/1945-7111/abae37
https://doi.org/10.20535/1813-5420.3.2025.339751
https://doi.org/10.1016/j.jpowsour.2016.12.011
https://c3.ndc.nasa.gov/dashlink/resources/133/
https://doi.org/10.1109/MIM.2008.4579269
82 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
30. Bole, B. (2007) Randomized battery usage. NASA Prognostics Data Repository. Retrieved October 12, 2025, from
https://phm-datasets.s3.amazonaws.com/NASA/11.+Randomized+Battery+Usage+Data+Set.zip
31. CALCE battery research group (2011). CALCE CS2 Dataset (University of Maryland). Retrieved October 12, 2025,
from https://calce.umd.edu/battery-data
32. Cycle Life Prediction Dataset (Stanford). Retrieved October 12, 2025, from https://github.com/rdbraatz/data-driven-
prediction-of-battery-cycle-life-before-capacity-degradation
33. Fast-Charging Optimization Dataset (Stanford). Retrieved October 12, 2025, from
https://data.matr.io/1/#projects/5d80e633f405260001c0b60a
34. Synthetic Diagnosis Dataset (Graphite//LFP). Retrieved October 12, 2025, from
https://data.mendeley.com/datasets/bs2j56pn7y/1
35. Sandia Short-Term Dataset. Retrieved October 12, 2025, from
https://www.sandia.gov/app/uploads/sites/163/Sandia_Cell_Cycle_Testing_Data.zip
36. Sandia Long-Term Degradation Dataset. Retrieved October 12, 2025, from
https://www.batteryarchive.org/snl_study.html
37. HNEI Dataset (Hawaii). Retrieved October 12, 2025, from https://www.batteryarchive.org/
38. Oxford Battery Degradation Dataset. Oxford University Research Archive. Retrieved October 12, 2025, from
https://ora.ox.ac.uk/objects/uuid:03ba4b01-cfed-46d3-9b1a-7d4a7bdf6fac
39. Birkl, C. R. (2017). Oxford battery degradation dataset 1. Oxford University Research Archive.
https://doi.org/10.5287/bodleian:KO2kdmYGg
40. Birkl, C. R., Roberts, M. R., McTurk, E., Bruce, P. G., & Howey, D. A. (2017). Degradation diagnostics for lithium
ion cells. Journal of Power Sources, 341, 373–386. https://doi.org/10.1016/j.jpowsour.2016.12.011
41. Raj, T. (2020). Path dependent battery degradation dataset part 1. Oxford University Research Archive.
https://doi.org/10.6084/m9.figshare.25975315
42. Raj, T. (2021). Path dependent battery degradation dataset part 2. Oxford University Research Archive.
https://doi.org/10.5287/bodleian:2zvyknyRg
43. Raj, T. (2021). Path dependent battery degradation dataset part 3. Oxford University Research Archive.
https://doi.org/10.5287/bodleian:v0ervBv6p
44. Panasonic 18650PF Dataset (UW–Madison). Retrieved October 12, 2025, from
https://data.mendeley.com/datasets/wykht8y7tg/1
45. Automotive Usage Dataset (Argonne/Ford). https://doi.org/10.21227/ce9q-jr19
46. Stanford SLB Dataset. (2024). OSF. Retrieved October 12, 2025, from https://osf.io/8jnr5/
47. Second-Life Electric Vehicle Battery Microgrid Dataset. Retrieved October 12, 2025, from
https://data.mendeley.com/datasets/mmrz3njn6b/2
48. PulseBat Dataset. (2024). Retrieved October 12, 2025, from https://github.com/terencetaothucb/Pulse-Voltage-
Response-Generation
49. Pagliaro, M., & Meneguzzo, F. (2019). Lithium Battery Reusing and Recycling: A Circular Economy Insight.
Heliyon, 5(6), e01866. https://doi.org/10.1016/j.heliyon.2019.e01866
50. Park, S., Ahn, J., Kang, T., Park, S., Kim, Y., Cho, I., & Kim, J. (2020). Review of state-of-the-art battery state
estimation technologies for battery management systems of stationary energy storage systems. Journal of Power
Electronics, 20, 1526–1540. https://doi.org/10.1007/s43236-020-00122-7
51. Casals, L. C., García, B. A., & Canal, C. (2019). Second life batteries lifespan: Rest of useful life and environmental
analysis. Journal of Environmental Management, 232, 354–363. https://doi.org/10.1016/j.jenvman.2018.11.046
52. Zaporozhets, A., Babak, V., Kulyk, M., & Denysov, V. (2025). Novel Methodology for Determining Necessary and
Sufficient Power in Integrated Power Systems Based on the Forecasted Volumes of Electricity Production.
Electricity, 6(3), 41. https://doi.org/10.3390/electricity6030041
53. Kulyk, M., Babak, V., Denisov, V., & Zaporozhets, A. (2025). Use of Battery Storage Systems in Integrated Power
Systems with Large Wind Power. In V. Babak, A. Zaporozhets (Eds.), Systems, Decision and Control in Energy VII.
Studies in Systems, Decision and Control, 595 (pp. 455‒466). Springer, Cham. https://doi.org/10.1007/978-3-031-
90466-0_19
54. Derii, V., Nechaieva, T., & Zgurovets, O. (2025). Studying The Needs Of Power Systems In Electricity Storage In
The Case Of Massive Implementation Of Power Plants Of Non-Guaranteed Capacity. Energy Technologies &
Resource Saving, 84(3), 51‒64 [in Ukrainian]. https://doi.org/10.33070/etars.3.2025.04
55. Zgurovets, O., & Kulyk, М. (2022). Possibilities to form a modern reserve of supporting of a frequency in integrated
power systems based on storage batteries for automatic adjustment of frequency and power. The Problems of General
Energy, 1-2(68-69), 20‒29 [in Ukrainian]. https://doi.org/10.15407/pge2022.01-02.020
56. Nechaieva, T., & Babak, V. (2025). Assessment Of Structural Changes In The Ukrainian Power System In The
Context Of Decarbonization. System Research in Energy, 3(83), 4‒16 [in Ukrainian].
https://doi.org/10.15407/srenergy2025.03.004
57. Havrylenko, Y., & Derii, V. (2025). The Energy Complexes Based On Solar Power Plants And Battery Energy
Storage Systems. System Research in Energy, 3(83), 17‒30 [in Ukrainian].
https://doi.org/10.15407/srenergy2025.03.017
https://phm-datasets.s3.amazonaws.com/NASA/11.+Randomized+Battery+Usage+Data+Set.zip
https://calce.umd.edu/battery-data
https://github.com/rdbraatz/data-driven-prediction-of-battery-cycle-life-before-capacity-degradation
https://github.com/rdbraatz/data-driven-prediction-of-battery-cycle-life-before-capacity-degradation
https://doi.org/10.5287/bodleian:KO2kdmYGg
https://doi.org/10.1016/j.jpowsour.2016.12.011
https://doi.org/10.6084/m9.figshare.25975315
https://doi.org/10.5287/bodleian:2zvyknyRg
https://doi.org/10.5287/bodleian:v0ervBv6p
https://github.com/terencetaothucb/Pulse-Voltage-Response-Generation
https://github.com/terencetaothucb/Pulse-Voltage-Response-Generation
https://doi.org/10.1016/j.heliyon.2019.e01866
https://doi.org/10.1007/s43236-020-00122-7
https://doi.org/10.1016/j.jenvman.2018.11.046
https://doi.org/10.3390/electricity6030041
https://doi.org/10.1007/978-3-031-90466-0_19
https://doi.org/10.1007/978-3-031-90466-0_19
https://doi.org/10.33070/etars.3.2025.04
https://doi.org/10.15407/pge2022.01-02.020
https://doi.org/10.15407/srenergy2025.03.004
https://doi.org/10.15407/srenergy2025.03.017
ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85) 83
58. Derii, V. (2023). Electric Energy Storages. System Research in Energy, 1(72), 12‒24 [in Ukrainian].
https://doi.org/10.15407/srenergy2023.01.012
59. Zaporozhets, A., & Kostenko, G. (2025). Second-Life Electric Vehicle Batteries in Ukraine’s Energy Sector: SWOT
Analysis and Market Evaluation. Science and Innovation, 21(6), 19–37. https://doi.org/10.15407/scine21.06.019
60. Kostenko, G.P. (2023). Situation Analysis Of Electric Transport Development Prospects And Its Integration Into
Ukrainian Power System. Power engineering: economics, technique, ecology, 1(71), 117‒124 [in Ukrainian].
https://doi.org/10.20535/1813-5420.1.2023.276185
61. Bobba, S., Mathieux, F., Ardente, F., Blengini, G. A., Cusenza, M. A., Podias, A., & Pfrang, A. (2018). Life Cycle
Assessment of repurposed electric vehicle batteries: anadapted method based on modelling energy flows. Journal
of Energy Storage, 19, 213–225. https://doi.org/10.1016/j.est.2018.07.008
62. Denysov, V., Kostenko, G., Babak, V., Shulzhenko, S., & Zaporozhets, A. (2023). Accounting the Forecasting
Stochasticity at the Power System Modes Optimization. In A. Zaporozhets (Ed.), Systems, Decision and Control in
Energy V. Studies in Systems, Decision and Control, 481 (pp. 43–55). Springer, Cham. https://doi.org/10.1007/978-
3-031-35088-7_3
63. Kostenko, G. (2025). Cluster-Based Deployment of Second-Life EV Batteries for Reliable and Sustainable Backup
Power Solution in Power Systems. System Research in Energy, 1(81), 40‒60.
https://doi.org/10.15407/srenergy2025.01.040
64. Kostenko, G., & Zaporozhets, A. (2023). Enhancing of the Power System Resilience Through the Application of
Micro Power Systems (microgrid) with Renewable Distributed Generation. System Research in Energy, 3(74), 25‒
38. https://doi.org/10.15407/srenergy2023.03.025
65. Zaporozhets, A., Babak, V., Kostenko, G., Zgurovets, O., Denisov, V., & Nechaieva, T. (2024). Power System
Resilience: An Overview of Current Metrics and Assessment Criteria. In V. Babak, A. Zaporozhets (Eds.), Systems,
Decision and Control in Energy VI. Studies in Systems, Decision and Control, 561 (pp. 35‒58). Springer, Cham.
https://doi.org/10.1007/978-3-031-68372-5_2
66. Kostenko, G. (2024). Accounting Calendar and Cyclic Ageing Factors in Diagnostic and Prognostic Models of
Second-Life EV Batteries Application in Energy Storage Systems. System Research in Energy, 3(79), 21‒34.
https://doi.org/10.15407/srenergy2024.03.021
67. Babak, V.P., Babak, S.V., Myslovych, M.V., Zaporozhets, A.O., & Zvaritch, V.M. (2020). Methods and Models for
Information Data Analysis. Diagnostic Systems For Energy Equipments. Studies in Systems, Decision and Control,
281 (pp. 23‒70). Springer, Cham. https://doi.org/10.1007/978-3-030-44443-3_2
АНАЛІЗ ВІДКРИТИХ НАБОРІВ ЕКСПЕРИМЕНТАЛЬНИХ
ДАНИХ ЛІТІЙ-ІОННИХ БАТАРЕЙ ДЛЯ МОДЕЛЮВАННЯ
ДЕГРАДАЦІЇ ТА УПРАВЛІННЯ ЖИТТЄВИМ ЦИКЛОМ
Ганна Костенко1*, https://orcid.org/0000-0002-8839-7633
Андрій Босак1, д-р філос., https://orcid.org/0000-0002-4667-9720
Юрій Сапсай1, https://orcid.org/0009-0001-2678-7003
Артур Запорожець1,2, д-р техн. наук, ст. досл., https://orcid.org/0000-0002-0704-4116
1Інститут загальної енергетики НАН України, вул. Антоновича, 172, Київ, 03150, Україна;
2ДУ «Центр оцінювання діяльності наукових установ та наукового забезпечення розвитку
регіонів України Національної академії наук України», вул. Володимирська, 54, Київ,
01030, Україна
*Автор-кореспондент: Kostenko_HP@nas.gov.ua
Анотація. Стрімке зростання електромобільності та інтеграції відновлюваних джерел енергії
посилило потребу не лише у системах накопичення енергії, а й у надійних і стандартизованих даних,
що забезпечують прогнозне моделювання та управління життєвим циклом літій-іонних батарей.
Експериментальні випробування батарей є дорогими, тривалими та обмеженими лабораторними
умовами, тому відкриті набори даних стають безцінною основою для порівняльних досліджень,
валідації моделей та відтворюваної аналітики. Водночас різноманітність доступних
інформаційних ресурсів за форматом, хімічним складом і умовами тестування ускладнює їх
системне використання для досліджень деградації та KPI-орієнтованого аналізу. У роботі
розроблено інформаційно-аналітичну основу для відбору, оцінювання та класифікації відкритих
наборів даних літій-іонних батарей, придатних як для первинного, так і для вторинного
використання. Проаналізовано п’ятнадцять репрезентативних наборів даних, згрупованих за
хімічним складом, глибиною циклювання та повнотою метаданих, із додатковою оцінкою
file:///C:/Users/Аннушка/Documents/%22Power%20engineering:%20Economy,%20Technology,%20Ecology%22,
https://doi.org/10.20535/1813-5420.1.2023.276185
https://doi.org/10.1016/j.est.2018.07.008
https://doi.org/10.1007/978-3-031-35088-7_3
https://doi.org/10.1007/978-3-031-35088-7_3
https://doi.org/10.15407/srenergy2023.03.025
https://doi.org/10.1007/978-3-031-68372-5_2
https://doi.org/10.15407/srenergy2024.03.021
https://doi.org/10.1007/978-3-030-44443-3_2
https://orcid.org/0000-0002-8839-7633
https://orcid.org/0000-0002-4667-9720
mailto:Kostenko_HP@nas.gov.ua
84 ISSN 2786-7633. Системні дослідження в енергетиці. 2026. 1(85)
цілісності даних і відстежуваності відповідно до принципів FAIR. Аналіз визначає придатність
кожного набору до певних аналітичних напрямів: Sandia та HNEI забезпечують довгострокові дані
для моделювання деградації та RUL; Stanford SLB і UC Davis Microgrid надають інформацію для
KPI-аналізу в умовах реальної експлуатації; PulseBat і Panasonic PF використовуються для оцінки
безпеки, надійності та ймовірнісних ризиків. Зазначено, що ключову роль відіграють повнота
метаданих, DOI-відстежуваність та контроль версій на рівні репозиторіїв у створенні
достовірних цифрових двійників і регуляторних інструментів, зокрема Європейського паспорта
батарей. Результати формують основу для гармонізованих, орієнтованих на дані методологій, що
поєднують експериментальні вимірювання, прогнозні моделі та KPI-управління життєвим циклом,
сприяючи прозорості, інтероперабельності та сталому розвитку у сфері досліджень і
впровадження батарейних технологій. Виконано порівняльний аналіз трьох репрезентативних
наборів даних (Sandia NMC, HNEI LFP і NASA/CALCE NMC/NCA), що демонструє відмінності
деградації та дозволяє кількісно оцінити їх варіабельність для задач вторинних батарей.
Ключові слова: літій-іонні батареї, відкриті набори даних, управління життєвим циклом,
моделювання деградації, система KPI, повторне використання батарей, енергетичні системи на
основі даних.
Дата першого надходження статті до журналу: 26.10.2025
Дата прийняття статті до друку після рецензування: 27.01.2026
Дата публікації (оприлюднення): 09.03.2026
|
| id | systemreorg-article-938 |
| institution | System Research in Energy |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2026-07-19T01:24:09Z |
| publishDate | 2026 |
| publisher | General Energy Institute of the National Academy of Sciences of Ukraine |
| record_format | ojs |
| resource_txt_mv | systemreorg/49/908439aa3563e1fb1ae97f0c67646749.pdf |
| spelling | systemreorg-article-9382026-07-18T12:57:50Z ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT Аналіз відкритих наборів експериментальних даних літій-іонних батарей для моделювання деградації та управління життєвим циклом Kostenko , Ganna Bosak, Andrii Sapsai, Yurii Zaporozhets, Artur lithium-ion batteries, open datasets, lifecycle management, degradation modeling, KPI framework, second-life applications, data-driven energy systems. літій-іонні батареї, відкриті набори даних, управління життєвим циклом, моделювання деградації, система KPI, повторне використання батарей, енергетичні системи на основі даних. The rapid expansion of battery-powered mobility and renewable energy integration has amplified the demand for energy storage as well as for reliable and standardized data supporting predictive modeling and lifecycle management of lithium-ion batteries. Experimental testing of batteries is costly, time-consuming, and limited by laboratory constraints, which makes open-access datasets an invaluable foundation for comparative studies, model validation, and reproducible analytics. However, the diversity of available datasets in terms of format, chemistry, and test conditions complicates their systematic use in degradation and KPI-based research. This study develops an information and analytical framework for the selection, evaluation, and classification of open-access Li-ion battery datasets applicable to both first- and second-life applications. Fifteen representative datasets were analyzed and grouped by chemistry, cycling depth, and metadata completeness, with additional assessment of data integrity and traceability according to FAIR principles. The analysis identifies dataset suitability for specific analytical domains: Sandia and HNEI provide long-term degradation and RUL modeling data; Stanford SLB and UC Davis Microgrid datasets enable operational and KPI analysis under realistic usage conditions; PulseBat and Panasonic PF datasets contribute to safety, reliability, and probabilistic risk evaluation. The proposed framework establishes clear connections between raw data, degradation indicators, and system-level metrics such as LCOS, utilization rate, and lifecycle efficiency. It also introduces a structured mapping of data relevance to various modeling objectives, supporting reproducible and cross-compatible research across laboratories and applications. Beyond comparative analysis, the study emphasizes the critical role of metadata completeness, DOI-based traceability, and repository-level version control in building trustworthy digital twins and regulatory tools such as the EU Battery Passport. The results provide a foundation for harmonized, data-driven methodologies that bridge experimental data, predictive models, and KPI-based lifecycle management, promoting transparency, interoperability, and sustainability in battery research and deployment. Additionally, a detailed comparative analysis of three representative high-quality datasets (Sandia NMC, HNEI LFP, and NASA/CALCE NMC/NCA) is presented to illustrate chemistry-dependent degradation behaviour and quantify inter-dataset divergence relevant for second-life modeling.&nbsp; Стрімке зростання електромобільності та інтеграції відновлюваних джерел енергії посилило потребу не лише у системах накопичення енергії, а й у надійних і стандартизованих даних, що забезпечують прогнозне моделювання та управління життєвим циклом літій-іонних батарей. Експериментальні випробування батарей є дорогими, тривалими та обмеженими лабораторними умовами, тому відкриті набори даних стають безцінною основою для порівняльних досліджень, валідації моделей та відтворюваної аналітики. Водночас різноманітність доступних інформаційних ресурсів за форматом, хімічним складом і умовами тестування ускладнює їх системне використання для досліджень деградації та KPI-орієнтованого аналізу. У роботі розроблено інформаційно-аналітичну основу для відбору, оцінювання та класифікації відкритих наборів даних літій-іонних батарей, придатних як для первинного, так і для вторинного використання. Проаналізовано п’ятнадцять репрезентативних наборів даних, згрупованих за хімічним складом, глибиною циклювання та повнотою метаданих, із додатковою оцінкою цілісності даних і відстежуваності відповідно до принципів FAIR. Аналіз визначає придатність кожного набору до певних аналітичних напрямів: Sandia та HNEI забезпечують довгострокові дані для моделювання деградації та RUL; Stanford SLB і UC Davis Microgrid надають інформацію для KPI-аналізу в умовах реальної експлуатації; PulseBat і Panasonic PF використовуються для оцінки безпеки, надійності та ймовірнісних ризиків. Зазначено, що ключову роль відіграють повнота метаданих, DOI-відстежуваність та контроль версій на рівні репозиторіїв у створенні достовірних цифрових двійників і регуляторних інструментів, зокрема Європейського паспорта батарей. Результати формують основу для гармонізованих, орієнтованих на дані методологій, що поєднують експериментальні вимірювання, прогнозні моделі та KPI-управління життєвим циклом, сприяючи прозорості, інтероперабельності та сталому розвитку у сфері досліджень і впровадження батарейних технологій. Виконано порівняльний аналіз трьох репрезентативних наборів даних (Sandia NMC, HNEI LFP і NASA/CALCE NMC/NCA), що демонструє відмінності деградації та дозволяє кількісно оцінити їх варіабельність для задач вторинних батарей. General Energy Institute of the National Academy of Sciences of Ukraine 2026-03-03 Article Article application/pdf https://systemre.org/index.php/journal/article/view/938 10.15407/srenergy2026.01.065 System Research in Energy; No. 1 (85) (2026): System Research in Energy; 65-84 Системні дослідження в енергетиці; № 1 (85) (2026): Системні дослідження в енергетиці; 65-84 2786-7102 2786-7633 en https://systemre.org/index.php/journal/article/view/938/833 Copyright (c) 2026 Ganna Kostenko , Andrii Bosak, Yurii Sapsai, Artur Zaporozhets https://creativecommons.org/publicdomain/zero/1.0 |
| spellingShingle | lithium-ion batteries open datasets lifecycle management degradation modeling KPI framework second-life applications data-driven energy systems. Kostenko , Ganna Bosak, Andrii Sapsai, Yurii Zaporozhets, Artur ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title | ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title_alt | Аналіз відкритих наборів експериментальних даних літій-іонних батарей для моделювання деградації та управління життєвим циклом |
| title_full | ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title_fullStr | ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title_full_unstemmed | ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title_short | ANALYSIS OF OPEN LI-ION BATTERY TESTING DATASETS FOR DEGRADATION MODELING AND LIFECYCLE MANAGEMENT |
| title_sort | analysis of open li-ion battery testing datasets for degradation modeling and lifecycle management |
| topic | lithium-ion batteries open datasets lifecycle management degradation modeling KPI framework second-life applications data-driven energy systems. |
| topic_facet | lithium-ion batteries open datasets lifecycle management degradation modeling KPI framework second-life applications data-driven energy systems. літій-іонні батареї відкриті набори даних управління життєвим циклом моделювання деградації система KPI повторне використання батарей енергетичні системи на основі даних. |
| url | https://systemre.org/index.php/journal/article/view/938 |
| work_keys_str_mv | AT kostenkoganna analysisofopenliionbatterytestingdatasetsfordegradationmodelingandlifecyclemanagement AT bosakandrii analysisofopenliionbatterytestingdatasetsfordegradationmodelingandlifecyclemanagement AT sapsaiyurii analysisofopenliionbatterytestingdatasetsfordegradationmodelingandlifecyclemanagement AT zaporozhetsartur analysisofopenliionbatterytestingdatasetsfordegradationmodelingandlifecyclemanagement AT kostenkoganna analízvídkritihnaboríveksperimentalʹnihdanihlítíjíonnihbatarejdlâmodelûvannâdegradacíítaupravlínnâžittêvimciklom AT bosakandrii analízvídkritihnaboríveksperimentalʹnihdanihlítíjíonnihbatarejdlâmodelûvannâdegradacíítaupravlínnâžittêvimciklom AT sapsaiyurii analízvídkritihnaboríveksperimentalʹnihdanihlítíjíonnihbatarejdlâmodelûvannâdegradacíítaupravlínnâžittêvimciklom AT zaporozhetsartur analízvídkritihnaboríveksperimentalʹnihdanihlítíjíonnihbatarejdlâmodelûvannâdegradacíítaupravlínnâžittêvimciklom |