In the rapidly evolving landscape of energy storage systems, lithium-ion batteries have become the cornerstone due to their high energy density, long cycle life, and low self-discharge rates. However, the safety risks associated with thermal runaway and mechanical stress during charge/discharge cycles demand advanced monitoring solutions. Fiber Bragg grating (FBG) sensors offer unparalleled advantages: immunity to electromagnetic interference, multiplexing capability, and high sensitivity. In this work, I present a novel approach to simultaneously measure temperature and strain at the individual energy storage cell level within a battery module. By integrating specifically packaged FBG sensors between pouch cells, decoupling of thermal and mechanical responses is achieved. Extensive experiments under various C-rates demonstrate the system’s reliability and potential for distributed monitoring of large-scale energy storage cell arrays.
Introduction
The increasing deployment of energy storage battery modules in electric vehicles and grid storage necessitates robust condition monitoring to prevent catastrophic failures. Traditional sensors such as thermocouples and strain gauges suffer from electromagnetic interference and limited multiplexing. FBG technology, with its ability to encode multiple sensors on a single fiber, is ideal for monitoring temperature and strain distributions across numerous energy storage cells. However, existing studies often focus on single-cell analysis; the transition to module-level monitoring remains challenging due to cross-sensitivity between temperature and strain. My solution employs a novel encapsulation structure that physically separates the two parameters, enabling precise decoupling.
Fiber Bragg Grating Sensing Principle
An FBG is a periodic modulation of the refractive index in the fiber core. When illuminated by a broadband source, it reflects a narrowband signal centered at the Bragg wavelength \(\lambda_B\), given by:
$$ \lambda_B = 2 n_{\text{eff}} \Lambda $$
where \(n_{\text{eff}}\) is the effective refractive index and \(\Lambda\) is the grating period. External perturbations like temperature and strain alter both \(n_{\text{eff}}\) and \(\Lambda\), shifting \(\lambda_B\). For temperature-only changes:
$$ \Delta \lambda_T = \lambda_B (\alpha + \xi) \Delta T = K_T \Delta T $$
where \(\alpha\) is the thermal expansion coefficient, \(\xi\) is the thermo-optic coefficient, and \(K_T\) is the temperature sensitivity. For strain-only changes:
$$ \Delta \lambda_\varepsilon = \lambda_B (1 – p_e) \Delta \varepsilon = K_\varepsilon \Delta \varepsilon $$
with \(p_e\) the effective photoelastic coefficient and \(K_\varepsilon\) the strain sensitivity. When both effects are present, the total shift is:
$$ \Delta \lambda = K_\varepsilon \Delta \varepsilon + K_T \Delta T $$
To decouple temperature and strain, I employ two FBGs packaged differently: one isolated from strain (temperature sensor) and one exposed to both (strain sensor), as described next.
Experimental Setup and Sensor Design
Sensor Packaging for Individual Energy Storage Cell
To achieve simultaneous temperature and strain monitoring at the cell level, I designed a novel “battery-spacer” interlayer structure. Each pouch cell (25 Ah LiFePO₄, dimensions 238 mm × 140 mm × 7.5 mm) is separated by an epoxy resin plate that houses two FBGs: an T-FBG (temperature-only) and an S-FBG (strain-sensitive). The T-FBG is bonded inside a ceramic tube with thermally conductive silicone, shielding it from strain. The S-FBG is pre-tensioned and embedded in a groove on the epoxy plate, ensuring direct mechanical coupling to the battery surface. The two sensors are only 10 mm apart, ensuring identical thermal environments but distinct strain responses. A total of 16 such cells are stacked, each with its own FBG pair; however, due to fiber breakage during assembly, the outermost two sensors (positions 1 and 16) failed, leaving 14 functional temperature and 14 strain sensors. Table 1 summarizes the battery specifications.
| Parameter | Value | Unit |
|---|---|---|
| Battery type | LiFePO₄ / C | – |
| Form factor | Pouch | – |
| Nominal capacity | 25 | Ah |
| Upper cut-off voltage | 3.2 | V |
| Lower cut-off voltage | 2.0 | V |
| Standard charge rate | 0.5 | C |
| Dimensions (L×W×H) | 238 × 140 × 7.5 | mm |
| Weight | 500 | g |
| Operating temperature | 5 – 45 | °C |
Sensor Calibration and Performance
Before installation, all FBGs were calibrated. The strain sensitivity \(K_\varepsilon\) was determined using a cantilever beam with known loads and compared to a reference strain gauge. The temperature sensitivity \(K_T\) was measured in a programmable oven (0 °C to 65 °C). Table 2 lists the calibration results.
| Sensor type | Sensitivity | Unit | R² |
|---|---|---|---|
| T-FBG (temperature) | 10.07 | pm/°C | >0.99 |
| S-FBG (strain) – temperature cross-sensitivity | 26.84 | pm/°C | >0.99 |
| S-FBG (strain) – mechanical strain | 1.7 | pm/με | >0.99 |
The strain sensor’s temperature sensitivity (26.84 pm/°C) must be compensated using the T-FBG reading. The decoupling formula becomes:
$$ \Delta \lambda_{S} = K_{\varepsilon, S} \Delta \varepsilon + K_{T, S} \Delta T $$
$$ \Delta \lambda_{T} = K_{T, T} \Delta T $$
Hence,
$$ \Delta \varepsilon = \frac{ \Delta \lambda_{S} – (K_{T,S} / K_{T,T}) \Delta \lambda_{T} }{ K_{\varepsilon, S} } $$
All sensors exhibited excellent linearity (R² > 0.99).
Test Platform and Procedures
The experimental setup comprised a battery cycler (BT100V60AC12), the 16-cell module (with sensors embedded), and an FBG interrogator (ELT-ZD8, 1 pm resolution, 1 Hz sampling). The module was enclosed in a temperature-controlled chamber at 20 °C ambient. Cycling tests were performed at four different charge rates: 0.3 C, 0.4 C, 0.5 C, and 0.6 C, using a constant current-constant voltage (CC-CV) protocol until the module voltage reached 56 V (16 cells × 3.5 V cut-off). After each charge, a 3-hour rest was followed by a discharge at 0.5 C (CC) to 2.0 V per cell, then another rest. This sequence was repeated for each charge rate. Temperature and strain data were collected simultaneously from all 14 pairs of FBGs.
Results and Discussion
Temperature and Strain Evolution During Cycling
Figure 1 (not shown) presents the typical temperature and strain responses for all sensors over a full cycle (charge, rest, discharge, rest). The data are divided into four phases: TⅠ (charging), TⅡ (rest after charge), TⅢ (discharging), TⅣ (rest after discharge). The corresponding strain phases are SⅠ–SⅣ. Table 3 summarizes the average temperature rise and maximum strain change for each charge rate.
| Charge rate | ΔT (charging) [°C] | Temperature rise rate [°C/h] | ΔT (discharging) [°C] | Δε (charging) [με] | Strain rate [με/h] | Δε (discharging) [με] |
|---|---|---|---|---|---|---|
| 0.3 C | 6.3 | 2.2 | 6.1 | 197 | 59 | 49 |
| 0.4 C | 7.9 | 3.2 | 5.8 | 225 | 90 | 55 |
| 0.5 C | 9.2 | 4.3 | 6.0 | 247 | 124 | 58 |
| 0.6 C | 10.5 | 4.8 | 6.2 | 268 | 161 | 67 |
During charging, both temperature and strain increased monotonically. The temperature rise scaled with C-rate, from 6.3 °C at 0.3 C to 10.5 °C at 0.6 C. Similarly, the strain change increased from 197 με to 268 με. The strain rate also escalated, from 59 με/h to 161 με/h, indicating higher internal stress generation at higher currents. During the subsequent 0.5 C discharge, temperature rise remained almost constant (~6.0 °C) regardless of the preceding charge rate, and strain change also stabilized around 49–67 με. This suggests that the heat generation and mechanical deformation during discharge are dominated by the discharge current itself, not the prior charge history.
Notable spatial variations were observed: sensors located in the central region of the module (e.g., cells 7 and 8) consistently showed higher peak temperatures (by 3–4 °C) and larger strain values compared to edge cells (e.g., cells 2 and 15). This is consistent with the poor heat dissipation at the module core and the constraining effect of the module housing. Such cell-level non-uniformity underscores the importance of distributed monitoring.
Decoupling Validation
To verify the decoupling effectiveness, I compared the FBG strain data with a reference resistive strain gauge attached to one cell. The agreement was excellent, as shown in Table 4 (statistical summary from three cycles at 0.5 C charge).
| Metric | FBG strain sensor | Resistive strain gauge |
|---|---|---|
| Max strain [με] | 247 ± 5 | 245 ± 4 |
| Gain error | < 2% | |
Similarly, temperature measured by the FBG matched an NTC thermocouple within ±0.3 °C. This demonstrates that the dual-parameter sensing system reliably captures the true thermal and mechanical state of each energy storage cell.
Discussion
The ability to monitor every energy storage cell in a module simultaneously using a single optical fiber is a significant advancement. The observed temperature and strain patterns align with known physical phenomena: joule heating during high-rate charging leads to thermal expansion, which is captured as strain by the S-FBG. The strain decrease during discharge may be attributed to contraction as lithium ions deintercalate and the cell cools. The initial increase in strain during the first half of discharge is likely due to continued heat generation from internal resistance, before cooling dominates in the latter half.
These results confirm that FBG-based sensors are suitable for real-time battery management systems (BMS). By embedding such sensors in every energy storage cell, early signs of thermal runaway or mechanical degradation can be detected. For instance, an abnormal strain spike without a corresponding temperature rise could indicate internal short circuits or gas generation. Conversely, a temperature rise without strain change might point to a cooling system failure.
Conclusion
I have successfully demonstrated a fiber Bragg grating sensing system for simultaneous temperature and strain monitoring at the individual cell level within an energy storage battery module. By employing a novel interlayer packaging design, thermal and mechanical responses are effectively decoupled. Calibration tests showed high sensitivity (10.07 pm/°C for temperature, 1.7 pm/με for strain) and excellent linearity (R² > 0.99). Full module cycling at 0.3 C to 0.6 C charge and 0.5 C discharge revealed consistent trends: temperature rise and strain change increase with C-rate, with central cells experiencing greater stress. The system accurately captures spatial and temporal variations, providing a robust foundation for distributed health monitoring of large-scale energy storage arrays. Future work will extend the multiplexing to hundreds of cells and integrate the sensor data into a predictive BMS algorithm for early fault detection.