Source-linked AI summary
Quantifying the Gap Between Laboratory Battery Test Patterns and Field Duty Profiles
Chunyang Zhao, Chresten Træholt
TL;DR
Laboratory battery tests do not directly represent stochastic, application-shaped field duty profiles, limiting how their results transfer across operating contexts. This paper compares controlled, drive-cycle, dynamic, chemistry-aligned, charging-trace, and fleet SOH evidence, finding substantial duty-pattern differences, including DSI values from 0.630 to 2.936 and distinct ageing trajectories.
Problem
Laboratory test patterns do not directly represent stochastic field duty profiles, making results valid for protocols only partially informative for real applications.
Method
The paper compares six battery evidence sources using duty-pattern, usage, transient-response, variability, and long-term-retention measures.
Results
Single-segment DSI ranged from 0.630 for the EV-fleet source trace to 2.936 for Oxford, while ageing reached 80% retention near 351, 6292, and 1019 respective checkpoints or cycles.
Takeaways & Limitations
Application-oriented battery studies should report explicit duty-profile descriptors alongside chemistry, capacity, temperature, and ageing metrics.
Abstract
from arXiv · showhide
Laboratory battery tests provide the main empirical basis for battery performance and degradation studies, but their operating patterns do not directly represent field duty profiles. This paper quantifies the gap by comparing six accessible evidence sources covering controlled cycling, drive-cycle testing, dynamic cycling, NMC811 laboratory ageing, a real electric-vehicle charging trace, and fleet-scale electric-vehicle state-of-health (SOH) data. The analysis combines usage frequency, usage intensity, usage C-rate, and a duty-structure index (DSI) based on normalized current dispersion and ramping. The representative single-segment DSI ranges from 0.630 for the field source trace and 0.699 for NASA to 2.936 for Oxford and 2.855 for Imperial, while usage C-rate ranges from 0.14-0.40 for Imperial, NASA, Stanford, and Hyundai to 2.00 for Oxford. Long-term ageing also differs: the 80 percent retention region occurs near 351 NASA cycles, 6292 Oxford checkpoints, and 1019 Stanford cycles. In chemistry-aligned NMC/NCM evidence, Imperial retains 0.813 under standard cycling and 0.865 under drive-cycle ageing, while the field source has median SOH 0.889 with visible dispersion. Field operation further shows a median use intensity of 137.2 km/day and 56.9 percent of charges ending at or above 95 percent SOC. These results show that battery performance metrics are conditional on the duty pattern that generated them; application-oriented studies should report explicit duty-profile descriptors together with chemistry, capacity, and ageing metrics.
I. INTRODUCTION
Battery behavior depends on operating patterns as well as chemistry and cell design, yet most evidence still comes from structured laboratory tests that only partially represent stochastic field duty profiles. This paper compares evidence at the test-pattern level and proposes a structured framework for interpreting performance across the laboratory-to-field gap.
- Motivation: Operating patterns—including current transients, voltage windows, SOC limits, temperature exposure, rest periods, and supervisory control—shape battery behavior across applications.These applications include electric vehicles, stationary storage, microgrids, hybrid renewable systems, and grid-support systems.
- Evidence base: Controlled laboratory tests remain indispensable because they provide repeatability, interpretability, and a defensible basis for cross-study comparison.The evidence base includes controlled cycling, constant-current constant-voltage charging, periodic reference tests, and temperature-regulated ageing campaigns.
- Laboratory-to-field gap: Field duty profiles are stochastic, partially observed, and shaped by battery-management systems, charger behavior, route conditions, driver behavior, and system-level constraints.This mismatch creates a transfer problem: laboratory results can be valid under a protocol while remaining only partially informative for a real application.
- Research scope: The paper compares controlled, application-inspired, dynamic, chemistry-aligned, and field-derived battery evidence to assess how far each evidence type remains from actual field duty structure.The comparison is designed around two questions: which performance statements each evidence type supports and how closely its operating structure reflects field use.
- Contribution: The key contribution is a duty-pattern-based comparison framework that interprets battery performance evidence across the laboratory-to-field gap rather than introducing a new dataset or lifetime model.The paper defines the evidence base and workflow, presents comparison results, discusses evidence limits, and concludes with the study’s implications.
II. STUDY DESIGN AND EVIDENCE BASE
The study compares heterogeneous laboratory and field evidence using long-period usage metrics and a short-period duty-structure index. Because SOC is unavailable for all sources, cycle count is estimated consistently from current throughput when direct counts are unavailable.
- Evidence base: Table I spans sub-Ah laboratory cells, commercial 21700 cells, large EV cells or packs, and fleet-level source data.The evidence base identifies each source’s system level, chemistry, nominal capacity, signal range, and scientific role.
- Usage metrics: Usage frequency (UF), usage intensity (UI), and usage C-rate (UC) describe how often and how intensely a battery is used over an application period.Conventional descriptors such as chemistry, nominal capacity, voltage, current, and temperature do not fully describe application-period use.
- Cycle-count assumption: Cycle count is estimated consistently from current throughput when direct SOC-based cycle counts are unavailable.Active usage time accumulates charging and discharging time, application time includes standby, and active charging time supports the usage C-rate calculation.
- Duty-structure index: The duty-structure index (DSI) captures short-period current-shape variability after each record is resampled onto a common progress coordinate.DSI combines normalized-current dispersion and cumulative ramping; logarithmic compression reduces sensitivity to extreme transients.
III. RESULTS · A. Duty Structure and Cycle Response
The results compare laboratory cycling and dynamic-operation tests with field applications using accessible battery-testing and charging profiles. The comparison extends from duty structure to measured response through normalized progress in selected records.
- A. Duty Structure and Cycle Response: Fig. 1 contrasts cycling-test profiles with dynamic-operation profiles across laboratory testing and field applications.The cycling tests appear in subfigures a and c, while dynamic operation tests appear in b and d.
- A. Duty Structure and Cycle Response: The 2025 EV fleet panel contains one accessible pack-level charging trace in segment 1, with segments 2–5 empty.This contrasts with the Hyundai panel, which uses five repeated charger-control blocks.
- A. Duty Structure and Cycle Response: The Hyundai panel uses five repeated charger-control blocks, whereas the 2025 EV fleet panel leaves segments 2–5 empty.The fleet panel places its single accessible pack-level charging trace in segment 1.
- A. Duty Structure and Cycle Response: The NASA profile shows repeated charge-discharge laboratory cycles in the duty-structure comparison.Fig. 1 compares the battery-testing profile with field applications through cycling and dynamic-operation tests.
- A. Duty Structure and Cycle Response: Fig. 2 extends the comparison from duty structure to measured response using normalized progress through each selected response record.The figure examines diagnosis cycles in which stable charging-discharging processes are implemented.
- A. Duty Structure and Cycle Response: Reference performance tests are periodic diagnostic tests used during ageing campaigns to measure capacity retention under controlled conditions.They provide the controlled-response reference for the comparison of selected records.
- A. Duty Structure and Cycle Response: NASA is shown with a charge–discharge pair, while Oxford is shown with an application-shaped example cycle.These records are presented within the diagnosis-cycle comparison.
- A. Duty Structure and Cycle Response: Voltage is retained in physical units, current is normalized by peak magnitude, and temperature is shown if measured.These conventions define the displayed voltage, current, and temperature traces.
B. Short-Term and Long-Term insights
Long-term ageing trajectories differ across NASA, Oxford, and Stanford, while chemistry-aligned laboratory and field evidence shows distinct retention outcomes. Field records also reveal varied intensity, charging behavior, and substantial SOH dispersion.
- Long-term degradation: NASA shows the faster decline with more fluctuation, Oxford degrades more gradually, and Stanford remains the shallowest.The trajectories use normalized retention and normalized ageing progress; NASA and Stanford use cycle count, while Oxford uses checkpoint count.
- Chemistry-aligned comparison: 0.865 retention occurs for Imperial’s drive-cycle laboratory condition by RPT9, versus 0.813 for the standard-cycle control by RPT15.Both laboratory conditions use Imperial 21700 NMC811 controls at 25 ◦C.
- Field operating context: 137.2 km/day is the median field use intensity, while 56.9% of charging events end at or above 95% SOC.The 10th–90th percentile use-intensity range is 90.6 to 197.6 km/day; median charge start SOC is 38.0% and median charge end SOC is 96.0%.
- Field operating context: 11.30 h is the selected Hyundai Ioniq 5 charging-session duration, with a peak summed phase current of 48.1 A and SOC increasing by 53.0 percentage points.The session comes from the KIT record.
- Field health outcomes: The field SOH-versus-mileage panel shows an overall downward trend with substantial spread, indicating mileage alone does not explain field health outcomes.This supplements the single duty trace with broader field context from Liu et al..
IV. DISCUSSION
Battery test patterns represent distinct kinds of evidence, producing different duty structures, ageing trajectories, and field relevance. Chemistry alignment improves interpretation but does not eliminate the gap between laboratory tests and field operation.
- Evidence diversity: Laboratory, fleet, and charging records differ in temporal structure even after normalization, making battery tests different kinds of evidence.The fleet source combines a low-DSI single trace with broad operating and SOH dispersion, while the Hyundai session shows repeated charger-control structure.
- Ageing trajectories: Near cycle 351 for NASA, checkpoint 6292 for Oxford, and cycle 1019 for Stanford, the selected laboratory records reached the 80% retention region.These are different ageing trajectories shaped by different test logic.
- Chemistry-aligned evidence: 0.813 under standard cycling and 0.865 under drive-cycle ageing were retained by the Imperial NMC811 control.Chemistry-aware alignment improves interpretation but does not eliminate the field gap.
- Field operation: 137.2 km/day was the median use intensity, while 56.9% of charging events ended at or above 95% SOC.The median charging window extended from 38.0% to 96.0% SOC.
- Application relevance: Controlled cycling supports mechanism isolation and stable benchmarking, whereas drive-cycle and dynamic protocols better suit application-oriented claims.The relevant question is which application claim a given test can defend.
V. CONCLUSION
The paper quantifies a laboratory-to-field duty-profile gap and shows that selected records are not interchangeable benchmarks. Duty structure, usage metrics, and ageing outcomes vary substantially across protocols and field operation.
- Duty-profile comparison: DSI ranges from 0.630 for the EV-fleet source trace to 2.936 for Oxford, demonstrating distinct temporal current structures across records.Other reported values are 0.699 for NASA, 0.639 for Stanford, 2.855 for Imperial, and 2.268 for the Hyundai charger-control block.
- Duty-profile comparison: UI = 0.817 and UC = 2.00 for Oxford, versus UI = 0.082 and UC = 0.40 for Stanford, showing differences in usage scale and temporal structure.The selected cell records therefore differ not only in scale, but also in temporal structure.
- Degradation comparison: 80% retention appears near 351 NASA cycles, 6292 Oxford checkpoints, and 1019 Stanford cycles, while Imperial retains 0.813 under standard cycling and 0.865 under drive-cycle ageing.The field source reports median SOH of 0.889 with an interquartile spread.
- Practical implications: Application-oriented studies should report chemistry, capacity, temperature, cycle count, cycling windows, temporal variability, rest structure, usage frequency, intensity, C-rate, and application analogy.The conclusion states that laboratory testing is necessary but not sufficient for application-oriented claims.