The rapid global integration of renewable energy sources like solar and wind has created an unprecedented demand for large-scale, cost-effective, and safe energy storage solutions. Among the various technologies, sodium-ion battery systems have emerged as a formidable candidate, primarily due to the abundant geographical distribution and lower cost of sodium compared to lithium. However, the operational safety and longevity of any battery system are intrinsically tied to its thermal behavior. During charge and discharge cycles, sodium-ion battery cells generate heat, which, if not managed efficiently, can lead to accelerated degradation, reduced performance, and in severe cases, thermal runaway. Therefore, developing a profound understanding of the heat-generation characteristics specific to sodium-ion battery technology and formulating optimized thermal management strategies is not just beneficial but critical for their successful deployment in grid-scale storage.
My research focuses on bridging this gap by employing a combined experimental and numerical simulation approach. The core objective is to decipher the asymmetric heat-generation profile of a commercial 210 Ah sodium-ion battery and leverage this understanding to design a liquid-cooling thermal management system that is both effective and energy-efficient. Traditional cooling systems often operate at constant parameters, which can be wasteful. Our work proposes a paradigm shift towards an asymmetric, multi-stage variable flow strategy that dynamically responds to the real-time thermal demands of the sodium-ion battery module.
Experimental Investigation of Heat Generation
The foundation of any effective thermal management design is accurate data on the heat source itself. To this end, we conducted rigorous experiments using an Accelerating Rate Calorimeter (ARC) under adiabatic conditions. This setup allows for the precise measurement of the heat generation power (Q_gen) of a single sodium-ion battery cell during operation, isolated from environmental heat exchange. We tested the cell under two distinct regimes, 1P (555W) and 0.5P (255W), reflecting different intensities of charge and discharge cycles relevant to real-world storage applications.
The results were revealing and formed the cornerstone of our optimization strategy. A stark asymmetry was observed between the charge and discharge processes. For instance, under the 1P regime, the discharge process generated a peak heat power of approximately 70 W, whereas during charging, the peak barely reached 25 W. More importantly, the total cumulative heat generated during a full discharge was found to be roughly three times greater than that during a full charge. This fundamental asymmetry immediately suggests that a one-size-fits-all cooling strategy is inherently inefficient.
Furthermore, the temporal evolution of heat generation exhibited distinct stages, particularly during discharge. The profile was not a simple curve but a sequence of phases: an initial ramp-up, a sustained high-power plateau, a peak surge, and a final rapid decay. This staged characteristic, quantified in our experiments, provides the critical temporal map needed for implementing a predictive, variable-control cooling strategy. The heat generation power Q_gen(t) extracted from these experiments serves as the direct input for our numerical model.
Numerical Modeling and Simulation Framework
To translate cell-level findings to a practical pack-level system, we developed a comprehensive 3D computational fluid dynamics (CFD) model. The model represents a module configured in a 1P52S arrangement, comprising 52 prismatic cells, inter-cell insulating layers, and an aluminum cold plate with integrated liquid channels. The governing equations for fluid flow and heat transfer are solved within this domain.
The energy conservation for the solid regions (battery cells, cold plate, insulation) is governed by:
$$ \frac{\partial}{\partial t} (\rho_s c_{p,s} T_s) = \nabla \cdot (k_s \nabla T_s) + Q_{gen} $$
where $\rho_s$ is density, $c_{p,s}$ is specific heat capacity, $k_s$ is thermal conductivity, $T_s$ is temperature, and $Q_{gen}$ is the volumetric heat generation rate sourced from our ARC experiments.
For the fluid domain (coolant), we solve the continuum equations:
$$ \nabla \cdot (\rho_f \vec{u}_f) = 0 $$
$$ \frac{\partial}{\partial t} (\rho_f \vec{u}_f) + \nabla \cdot (\rho_f \vec{u}_f \vec{u}_f) = -\nabla p_f + \nabla \cdot (\mu_f \nabla \vec{u}_f) $$
$$ \frac{\partial}{\partial t} (\rho_f c_{p,f} T_f) + \nabla \cdot (\rho_f c_{p,f} \vec{u}_f T_f) = \nabla \cdot (k_f \nabla T_f) $$
where the subscript $f$ denotes fluid properties, $\vec{u}$ is velocity, and $p$ is pressure.
The thermal properties of all materials used in the simulation are summarized in the table below. Notably, the sodium-ion battery is modeled with anisotropic thermal conductivity, acknowledging its layered internal structure.
| Component | Density, $\rho$ (kg/m³) | Thermal Conductivity, $k$ (W/m·K) | Specific Heat, $c_p$ (J/kg·K) |
|---|---|---|---|
| Coolant (Water) | 998.2 | 0.60 | 4182 |
| Cold Plate (Aluminum) | 2719 | 202.4 | 871 |
| Insulation Material | 50 | 0.02 | 1000 |
| Sodium-ion Battery | 2719 | $k_x=k_y=20$, $k_z=3$ | 871 |
The boundary conditions simulate a realistic environment with convective heat loss from module surfaces. A grid independence study was rigorously performed to ensure the accuracy of the numerical results was not compromised by discretization choice. The validated model allows us to virtually test various cooling strategies rapidly and at low cost.
Analysis of Baseline Thermal Performance
Simulating the module with a constant coolant flow rate provided a baseline understanding of its thermal behavior and confirmed the implications of our experimental data. Under a 1P discharge, the module’s maximum temperature rose significantly, following the staged heat generation input. The temperature distribution was non-uniform, with hotter spots concentrated in the core regions of the module, away from the cooling plate. In contrast, during a 1P charge cycle with the same cooling flow, the maximum temperature remained much lower, consistently staying close to the coolant inlet temperature. This baseline analysis visually and quantitatively confirmed the primary thesis: the thermal management demand is profoundly different between charge and discharge for a sodium-ion battery system.
We then parametrically analyzed the effect of coolant flow rate on two key metrics: the peak module temperature ($T_{max}$) and the system pressure drop ($\Delta P$), which is directly proportional to the pumping power consumption. The relationship can be expressed as:
$$ \Delta P \propto f_{d} \cdot \frac{L}{D_h} \cdot \frac{\rho_f u_f^2}{2} $$
where $f_d$ is the Darcy friction factor, $L$ is flow length, $D_h$ is hydraulic diameter, and $u_f$ is flow velocity. As expected, increasing the flow rate improved cooling (lower $T_{max}$) but at the expense of a sharply increasing $\Delta P$. The law of diminishing returns was clearly observed; beyond a certain flow rate, further increases yielded negligible temperature reduction but caused a quadratic rise in pumping power. This trade-off curve is the central challenge in optimizing thermal management system (TMS) efficiency.
Proposed Asymmetric Multi-Stage Optimization Strategy
Informed by the experimental and numerical findings, we propose a two-tiered optimization strategy for the sodium-ion battery thermal management system.
Tier 1: Process-Asymmetric Flow. This is the macro-level strategy. Given the order-of-magnitude difference in heat generation:
– Discharge Process: Activate the liquid cooling system with an optimized base flow rate.
– Charge Process: Deactivate the liquid cooling pump entirely (0 kg/s flow). Our simulations confirmed that for the studied sodium-ion battery module, the natural convection and thermal mass are sufficient to keep temperatures within a safe limit during the mild charging process. This simple asymmetry can eliminate the TMS energy consumption for nearly half of the operating cycle.
Tier 2: Discharge Phase-Dependent Variable Flow. This is the micro-level, intra-discharge optimization. We dissect the discharge process into phases based on the measured heat generation power $Q_{gen}(t)$:
– Phase I (Initial/Low Heat): $t_0$ to $t_1$. $Q_{gen}$ is low and building. Apply a low flow rate ($\dot{m}_1$).
– Phase II (High/Sustained Heat): $t_1$ to $t_2$. $Q_{gen}$ is at its sustained peak. Apply a high flow rate ($\dot{m}_2$).
– Phase III (Final/Decaying Heat): $t_2$ to $t_{end}$. $Q_{gen}$ decays rapidly. Apply a very low or minimal flow rate ($\dot{m}_3$).
The core optimization problem is to find the set of flow rates $\{ \dot{m}_1, \dot{m}_2, \dot{m}_3 \}$ and their switch times $\{t_1, t_2\}$ that minimize total pumping energy $E_{pump}$ over the discharge cycle:
$$ E_{pump} = \int_{t_0}^{t_1} \frac{\Delta P(\dot{m}_1) \cdot \dot{V}_1}{\eta} \, dt + \int_{t_1}^{t_2} \frac{\Delta P(\dot{m}_2) \cdot \dot{V}_2}{\eta} \, dt + \int_{t_2}^{t_{end}} \frac{\Delta P(\dot{m}_3) \cdot \dot{V}_3}{\eta} \, dt $$
subject to the constraint that the maximum cell temperature never exceeds a safety threshold $T_{safe}$:
$$ T_{max}(t, \dot{m}_1, \dot{m}_2, \dot{m}_3) \leq T_{safe} \quad \forall \, t \in [t_0, t_{end}] $$
Here, $\dot{V}$ is volumetric flow rate and $\eta$ is pump efficiency.
For our specific module, an example implementation derived from simulation is:
| Discharge Phase | Time Period | Coolant Mass Flow Rate | Rationale |
|---|---|---|---|
| Phase I | 0 – 2025 s | 0.2 kg/s | Sufficient for low initial heat load. |
| Phase II | 2025 – 2850 s | 0.6 kg/s | High flow to tackle peak heat generation. |
| Phase III | 2850 – 4000 s | 0.1 kg/s | Minimal flow as heat generation plummets. |
Results and Discussion of the Optimized Strategy
Simulating the proposed multi-stage variable flow strategy and comparing it against a conventional constant high-flow strategy yielded significant insights. The primary finding is that both strategies achieved nearly identical thermal performance, maintaining the peak sodium-ion battery temperature below the safe limit. The temperature evolution curves were virtually superimposed, confirming that the variable strategy does not compromise cooling efficacy.
The dramatic improvement was in system efficiency. The pumping power, proportional to the pressure drop $\Delta P$, showed vast differences. During Phase I, the variable strategy operated at 0.2 kg/s, resulting in a $\Delta P$ roughly 64% lower than the constant 0.6 kg/s strategy. During the critical Phase II, the flow rates were equal, so the power draw was similar. However, in Phase III, reducing the flow to 0.1 kg/s caused the $\Delta P$ to plummet, achieving a reduction of approximately 91% compared to the constant high-flow scenario.
Integrating this power saving over the entire discharge duration leads to a substantial reduction in the total energy consumed by the thermal management system for the cycle. This efficiency gain directly translates to higher overall system efficiency for the sodium-ion battery energy storage unit, a crucial metric for grid operators. Furthermore, by reducing the pump’s operational time at high load, the proposed strategy may also contribute to improved pump reliability and longevity.
The implementation of this strategy requires a control system equipped with logic tied to the state of charge (SOC) or, more directly, to real-time temperature feedback or pre-programmed timers based on the characterized $Q_{gen}(t)$ profile. For a sodium-ion battery system, whose heat signature is repeatable, such a predictive feedforward control is highly feasible.
Conclusion and Perspective
This investigation underscores that the thermal behavior of sodium-ion battery systems is not merely a lesser version of lithium-ion systems but has its own distinct signature, characterized by strong asymmetry between charge and discharge and a staged heat generation profile during discharge. Ignoring this specificity leads to thermally safe but energetically wasteful thermal management systems.
Our work demonstrates that through detailed experimental characterization and physics-based modeling, significant optimization potential can be unlocked. The proposed asymmetric, multi-stage variable flow liquid cooling strategy is a direct outcome of understanding the unique “thermal personality” of the sodium-ion battery. It proves that one can achieve the required thermal control—maintaining safe operating temperatures for the sodium-ion battery module—while simultaneously minimizing the parasitic energy draw of the cooling system. This is a vital step towards making large-scale sodium-ion battery energy storage not only technologically viable but also economically and operationally superior.
Future work will involve experimental validation of this strategy on a full module, exploring the use of advanced coolants, and integrating this thermal management approach with broader battery management system (BMS) functions for holistic optimization of the sodium-ion battery energy storage system’s performance, lifespan, and safety.