A Correlational Study on Architectural Design and Thermal Distribution Patterns Using a Novel Multi-Terminal Approach in Cylindrical Li-Ion Cell-Integrated Battery Packs

Abstract

A novel architectural design is proposed to mitigate uneven thermal distribution, peak temperature, and heat spot generation, which are common issues that are observed in conventional battery packs. This approach features a multi-terminal configuration, incorporating a modified battery pack structure along with a multi-terminal switching algorithm that identifies the optimal terminal for current flow to the load. In the proposed design, the first and second terminals are placed at the first and fourth series string while the battery pack is divided into four regions, each corresponding to one series string. Additionally, terminal points represent the four thermal zones at the pack level. Experiments were conducted to evaluate the performance of the dual-terminal switching mechanism in three configurations—1S, 2S, and 3S. The 1S setup outperformed the single-terminal design, achieving a 6.23% improvement in reducing the zone temperature difference (ΔP_z). The 2S configuration demonstrated an 11.11% improvement, while the 3S setup achieved an improvement in peak region difference (ΔP_r) of >50%, without a cooling system. Finally, while forced air cooling effectively lowers peak temperature, it is insufficient in addressing thermal distribution and heat spot formation. However, integrating the proposed multi-terminal approach enables the effective control and management of all three critical thermal parameters—peak temperature, thermal distribution, and heat spot generation.

1. Introduction

Thermal management plays a critical role in cylindrical lithium-ion battery systems due to its significant impact on performance, lifespan, and safety [1]. As the demand for high-performing battery systems increases in electric vehicle (EV) applications, under-standing thermal issues and developing effective mitigation strategies have become increasingly important [2]. The architectural design of battery packs and modules greatly influences the rate of heat generation and its distribution within the pack [3,4,5]. The architecture of battery packs built for cylindrical lithium-ion cells presents unique thermal challenges due to its design and configuration [6,7]. Key factors influencing thermal behavior include the type of cells used, cooling system design, cell connection methods (series and parallel combinations), terminal placements, inlet and outlet positions, and control strategies such as charging and discharging cycles [8,9]. Thermal zones and heat spot areas within the module or on individual cells exhibiting temperature variations compared to the overall region are primarily the result of these architectural choices [10].

Numerous studies are being conducted to develop sustainable thermal environments for the effective operation of lithium-ion battery (LiB) packs, particularly for large-scale applications where a narrow operating temperature range is critical. Several studies have focused on understanding thermal behavior in terms of temperature gradients, such as in the work of S. C. Chen et al. [11], who developed a 3D thermal model highlighting natural convection. Their findings highlight that radiation contributes approximately 43–63% of natural convection. For cylindrical cells, Christophe Forgez et al. [12] report a temperature difference of up to 10 °C between the cell interior and surface, which is crucial for improving the accuracy of dynamic thermal models in battery management systems (BMSs).

Cell geometry and design features also critically impact thermal performance. Alastair et al. [13,14] examine LiB pouch cells, comparing two cell models and analyzing tab and surface temperatures, concluding that terminal points illustrate higher values compared to those of the body. Yan Zhao et al. [15] further demonstrated that increasing the tab thickness in pouch cells enhances heat removal, contributing to improved thermal uniformity.

To address external cooling challenges, various pack-level designs are explored. Shahid et al. [16] examine a cylindrical cell battery pack model with four different configurations. Their results showed that incorporating an inlet plenum, jet inlets, and multiple vortex generators led to a 21.5% reduction in overall battery pack temperature and a 16% improvement in thermal uniformity. Rui Huang et al. [17] investigated a battery pack containing 25 parallel 18,650 cells combined with a phase-change material (PCM), which highlights enhanced thermal conductivity and latent heat, contributing to improved thermal regulation.

Cooling method comparisons have also been investigated. Dafen Chen et al. [18] studied various cooling methods—fin-based, air, direct, and indirect liquid cooling—and their effects on battery performance. They found that air cooling consumed 2–3 times more energy, fins added 40% weight, and among liquid methods, indirect cooling offered a better practicality and thermal performance. Similarly, Liu et al. [19] examined the impact of a reciprocating airflow system on Sony US-18,650 cylindrical cells. Their findings indicated that an airflow velocity of 6 m/s and an inlet temperature of 283.15 K were optimal for cooling performance, although these parameters increased both the system’s power requirements and design complexity.

The thermal response under high rates has also been investigated. Kazuo Onda et al. [20] focused on the thermal behavior of lithium cells under high C rates, confirming that radial temperature gradients were minimal. Meanwhile, Yang E. et al. [21] investigated the relationship between uniform cooling and cell alignment, staggering patterns, and inter-cell distances in battery packs using LiFePO₄ cylindrical cells. The study concluded that a staggered configuration with 34 mm longitudinal and 32 mm transverse spacing provided optimal thermal uniformity across the pack. In the broader context, Todd M. Bandhauer et al. [22] reviewed safety, performance, and cost-related thermal issues, highlighting the need for more precise heat generation measurements and the lack of standardized strategies in battery thermal management systems.

To address these challenges, recent research has shifted toward design-based thermal control, which involves modifying the internal architecture and power delivery paths of battery packs. Concepts such as reconfigurable multi-terminal architectures and support for multiple loads within a single battery pack are being explored [23]. Some studies focus on developing adaptive reconfigurable battery packs that can dynamically adjust their configuration based on real-time load demands through cell-level connections [24,25]. In other cases, reconfigurable voltage and load handling at the pack level are introduced to enhance fault tolerance, scalability, cost-effectiveness, and the reuse of secondary packs using load-aware scheduling mechanisms. Building upon this direction, the present study introduces a multi-terminal battery module architecture, in which the battery pack is equipped with multiple paired positive and negative terminals. A key innovation lies in the implementation of a software-controlled, toggle-based switching mechanism that enables the electrical load to dynamically connect to different terminal pairs. This architecture allows for the redistribution of current paths during discharge, thereby facilitating thermal energy dissipation across different zones of the battery pack. The goal is to achieve a more uniform temperature profile while maintaining the same power demand.

In this paper, an experimental setup is presented to manage heat generation and distribution, as well as to reduce peak temperatures in battery cells and modules. The proposed battery module design features multiple paired (positive and negative) terminals that are capable of delivering the same power output. A software-controlled, toggle-based switching mechanism enables the load to draw power from any selected terminal pair. This design incorporates three novel techniques across different levels of conventional architecture. First, at the module level, a multi-terminal configuration is introduced in place of the traditional single-terminal approach, allowing for toggling between terminals. Second, at the pack level, a new hardware architecture is implemented to support the multi-terminal design. Finally, a control algorithm is developed to manage the terminal switching process. The primary objective of this study is to develop an effective thermal management strategy that achieves uniform thermal distribution across the battery pack without relying on external cooling systems, while still meeting the required power demand.

2. Proposed Battery Pack Technology and Experiments

The proposed battery pack technology addresses two key aspects—the architecture of the battery module and the overall system-level design. The architectural design typically includes the type of cells used, the configuration of the pack, the type of coolant (along with its inlet/outlet positions and flow direction), terminal placement, and supporting components such as bus bars and cell holders. A major limitation observed in conventional designs is the tendency to develop hotspots, uneven thermal distribution, and localized high-temperature zones. These issues can negatively impact charge/discharge rates, battery lifespan, and overall safety. To mitigate these problems, the proposed design focuses on several critical areas, i.e., identifying sources of heat generation and analyzing heat dissipation patterns in lumped cell configurations to assess the role and relevance of coolant-assisted approaches (managing coolant flow and its influence on heat removal, as well as optimizing the direction and positioning of coolant channels) for effective thermal regulation. At the system level, the emphasis is on integrating the proposed module design with other sub-components to form a cohesive battery pack that enhances performance, reliability, and safety.

2.1. Proposed Novel Architecture of Battery Module and Battery Pack

The proposed battery module (BM) introduces a multi-terminal (dual-terminal) approach in contrast to the conventional single-terminal design. The two-terminal BM is constructed using LiFePO₄ 18,650 cells arranged in a 4s4p configuration, with overall dimensions of 200 × 200 × 90 mm, as shown in Figure 1a. The first terminal (T₁) has its positive and negative connections located at the 1s1p and 4s1p positions, respectively. The second terminal (T₂) is positioned at the 1s4p (positive) and 4s4p (negative) locations of the BM. Each terminal delivers a nominal voltage of 12.8 V and a current of 6 Ah. Both terminals are connected to separate battery management systems (BMSs) to regulate uniform charging and discharging.

Upon evaluating various cell-to-cell spacing values, a spacing of 22 mm (S) was adopted, providing adequate space to observe heat dissipation and evaluate the thermal influence of individual cells on the overall battery module. Figure 1b illustrates the schematic of the proposed battery pack (BP) architecture developed to support the novel BM design. In traditional BPs, the series terminals of the BM are connected to a single battery management system, which then supplies power to the actuators. In contrast, the proposed BP architecture accommodates the dual-terminal BM by integrating two BMS units. Each BMS is connected to one end of the BM’s series strings—BMS_1 is connected to the 1s series and BMS_2 is connected to the 4s series.

Both BMS units are linked to relays, through which power exchange is managed. Figure 1c details the components within the BP, while Figure 1d shows the physical implementation of the dual-terminal BM tested in this study. Specifications related to typical EV pack requirements, the selected single cell used in the BM, and the tested lumped BM configuration are summarized in

Table 1. Specifications of the EV drive pack requirement, test cell, and lumped cell pack model.

	EV Drive	Test Cell	Lumped Cell
Chemical Composition	LiFePO4	LiFePO4	LiFePO4
Configuration	Depends on load requirement	1S1P	4S4P
Nominal discharge capacity (mAh)	20,000–100,000+	1500	6000
Nominal voltage (V)	Typically 3.2–3.3 V per cell in series	3.2	12.8
Standard charge (A)	0.5–1C	0.75	3.0
Maximum continuous discharge (A)	2–10C	4.5	18
Discharge cut-off voltage (V)	Typically 2.5–2.8 V per cell	2.5–2.8	10–11.2
Cell/Pack weight (g)	20,000–100,000	39	~800
Cell/Pack height (mm)	Typically, 70–200	65	95
Cell diameter/pack dimensions (mm)	Varies by configuration	18	200 × 200 × 90
C-rate (C)	0.5–10	0.5–3	0.5–3

. It is important to note that although both terminals are integrated within the same BM, only one terminal is active at any given time for charging or discharging. The relay switching controller determines the active terminal, enabling precise control over heat generation and distribution across different regions of the BM.

2.2. Experimental Setup and Procedure

Numerical simulations were conducted based on the proposed design outlined in Section 2.1. The test cases and configurations, as summarized in

Table 2. Specifications of the test case: features and states.

Battery State	C-Rate	Con-V (V)	Capacity (Ah)	Limited-V (V)	Limited- C (A)	Convection	Terminal Count	Terminal Switch Count
Con-C and V Charge	1	14.8	6	--	0.6	Natural, Forced	1, 2	0, 1, 2, 3
Con-C Discharge	1	--	6	11.2	--	Natural, Forced	1, 2	0, 1, 2, 3

, were developed to evaluate the effectiveness and efficiency of the components involved in achieving the desired outcomes. However, simulations alone require validation through real-time experimental models. Therefore, corresponding experimental setups were developed, as illustrated below.

2.2.1. Experimental Setup

The schematic and actual experimental hardware setup are shown in Figure 2. A LiFePO₄ lithium-ion cell (Orange A-grade IFR18650) with a 3C discharge rate was used in the proposed lumped-cell battery module, which was placed inside a custom battery pack casing. The casing, constructed from acrylic material, features integrated coolant inlet/outlet channels and cable channels carved into the side surfaces. The dimensions of the casing are 200 × 200 × 90 mm, with three coolant inlet channel orifices (diameter: 25 mm).

A cooling system was incorporated into the experimental setup to enable comparative analysis between the proposed design and a conventional forced-air cooling approach. To monitor thermal behavior within the battery pack and its surroundings, a thermal data acquisition module was developed. This system includes a microcontroller, an InfiRay P2 Pro thermal camera (256 × 192 IR resolution), a 1112 °F high-temperature testing camera, and LM35 temperature sensors with an operating range of −55 °C to 150 °C. Ambient room temperature was maintained at 25 ± 2 °C during testing by conditioning the entire room using an air conditioning (AC) system before the experiments commenced.

A DALY LiFePO₄ 8s 24 V BMS, with a 40 A discharge and a 20 A charge current, was integrated to enable controlled and balanced charge/discharge cycles. The Semco S1 BCDS 99 V 20 A 1CH battery charge/discharger and analyzer was used as both the charger and load for the battery pack, operated via the Semco Lithium Battery Charging System_4.3 software installed on a computer. A relay module was employed to manage terminal switching for power delivery. For forced convection cooling, a PGSA2Z axial fan was used, operating on a 4–12 V variable speed range (1600–3900 rpm) with a power rating of 78 W and a flow rate of 28 cubic feet per minute (CFM).

2.2.2. Experimental Procedure

The test cases were designed to examine two key aspects—the heat generation rate of the BM across different regions and zones, and the thermal distribution patterns in both single-terminal and two-terminal configurations. An infrared thermal camera was positioned 150 mm above the top surface of the BM to capture a top-down thermal view. The BM was divided into four regions, corresponding to the four rows of series cell connections, and four thermal zones, each aligned with a terminal point, as illustrated in Figure 3a. Figure 3b illustrates the schematic of the BM. In this experiment, charging and discharging cycles were performed to analyze heat generation and distribution in both the proposed and conventional BMs.

The goal is to evaluate the thermal characteristics and determine the impact of the modified design. It is noted that the charging cycle maintains relatively static and constant conditions; therefore, test cases were focused on the 1C discharge cycle and its influence on thermal behavior. Terminal switching was controlled using logic based on the remaining Depth of Discharge (DOD), as outlined in

Table 3. Selected terminal switching condition.

Terminal Switch Count	Time (min)	DOD Condition (%)
0	--	--
1	30	50
2	20, 40	33, 66
3	15, 30, 45	25, 50, 75

. This table maps time and DOD values to the number of terminal switches expected during a standard 1C discharge cycle. Based on the states in the feature, eight test cases are considered to cover all combinations of conditions given in Table 2. (The test case naming convention is as follows: Test_case_1—18_D_1C_E_N_1T_0S—this represents an 18 mm diameter Li-ion cell undergoing a discharge cycle at a 1C rate, in an enclosed environment with natural convection. A single-terminal model is considered and the terminals are alternated zero times).

2.3. Numerical Simulations

The velocity and mass flow rate of air for the fan are determined using the RPM and diameter of the fan, based on Equations (1)–(8) [26]. The circumferential velocity (tip speed, V_tip) is calculated using Equations (1) and (2), which relate the velocity of air (V_air), fan RPM, and fan diameter (D). Since the actual V_air is unknown in our case, we utilize the volumetric flow rate (VFS) and cubic feet per minute (CFM) values from Equations (5) and (6) to derive Equation (3). CFM is calculated using the cross-sectional area (A) of the fan, as defined in Equation (4). The air flow rate from the coolant chamber to the battery module is then obtained using Equation (7), considering the density of air ($\rho$) at 25 °C. Finally, the power consumed by the fan (P) is calculated using Equation (8).

$${\mathrm{V}}_{\mathrm{t}\mathrm{i}\mathrm{p}}={\displaystyle \frac{{\mathrm{V}}_{\mathrm{a}\mathrm{i}\mathrm{r}}}{\mathsf{\eta }}}$$

(1)

$${\mathrm{V}}_{\mathrm{t}\mathrm{i}\mathrm{p}}={\displaystyle \frac{\mathsf{\pi }\ast \mathrm{D}\times \mathrm{R}\mathrm{P}\mathrm{M}}{60}}$$

(2)

$${\mathrm{V}}_{\mathrm{a}\mathrm{i}\mathrm{r}}={\displaystyle \frac{\mathrm{V}\mathrm{F}\mathrm{S}}{\mathrm{A}}}$$

(3)

$$\mathrm{A}=\mathsf{\pi }{\left({\displaystyle \frac{\mathrm{D}}{2}}\right)}^2$$

(4)

$$\mathrm{C}\mathrm{F}\mathrm{M}={\mathrm{V}}_{\mathrm{a}\mathrm{i}\mathrm{r}}\times \mathrm{A}\times 60$$

(5)

$$\mathrm{V}\mathrm{F}\mathrm{S}\left({\displaystyle \frac{{\mathrm{m}}^3}{\mathrm{s}}}\right)={\displaystyle \frac{\mathrm{C}\mathrm{F}\mathrm{M}\times 0.0283168}{60}}$$

(6)

$$\mathrm{M}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{F}\mathrm{l}\mathrm{o}\mathrm{w}\mathrm{R}\mathrm{a}\mathrm{t}\mathrm{e}\left({\displaystyle \frac{\mathrm{g}}{\mathrm{s}}}\right)=\mathrm{V}\mathrm{F}\mathrm{S}\times \rho$$

(7)

$$\mathrm{P}=\mathrm{V}\times \mathrm{I}$$

(8)

The experimental setup with respect to coolant parameters is summarized in

Table 4. Coolant data setup for natural and forced convection.

Parameter	Natural Convection (Enclosed Chamber)	Natural Convection (Open Chamber)	Average Speed
Velocity (m/s)	<0.5	<1	3.354
RPM (rpm)	NA	NA	1600
Mass Flow Rate (g/s)	NA	<1	19.982
Power Consumed (W)	0	0	1.044
Reynolds Number	<2000	<2000	6808

. The heat generation and dissipation rates in the cylindrical cells and modules are set and measured using Equations (9)–(14) [27]. Volumetric heat generation inside the cell is calculated using Equation (9) using the internal resistance, radius, and height of the cylindrical cell and current, which defines the average volumetric heat generation rate under the assumption of uniform heat generation across the entire cylindrical cell. Given that the cell is modeled as being thermally homogeneous, this volumetric rate is multiplied by the cell volume and time interval defining total heat generated, which is used in the overall energy balance in Equation (10) and cell temperature in Equation (11). Equation (13) derives the air temperature rise in the pack chamber using Equations (12) and (14), where Equation (12) derives air mass flow (m_air) using air density ($\rho$) and air volume flow rate (V_air). Meanwhile, Equation (14) derives the heat transferred to air, which is also known as the convective heat transfer rate (Q_t), where (h) represents the convective heat transfer co-efficient and (A) stands for the surface area of the cell exposed to convection. These equations account for the heat generation rate, cell temperature variation over time, heat transfer, and the temperature change in the air.

$${\mathrm{Q}}_{\mathrm{g}\mathrm{e}\mathrm{n}\_\mathrm{p}\mathrm{e}\mathrm{r}\_\mathrm{v}\mathrm{o}\mathrm{l}}={\displaystyle \frac{{\mathrm{I}}^2\times {\mathrm{R}}_{\mathrm{i}}}{\mathsf{\pi }{\mathrm{r}}^2\mathrm{h}}}$$

(9)

$${\mathrm{Q}}_{\mathrm{g}\mathrm{e}\mathrm{n}}=\mathrm{m}\mathrm{c}∆\mathrm{T}+\mathrm{h}\mathrm{A}({\mathrm{T}}_{\mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}}-{\mathrm{T}}_{\mathrm{a}\mathrm{i}\mathrm{r}})$$

(10)

$${\mathrm{T}}_{\mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}}={\displaystyle \frac{{\mathrm{Q}}_{\mathrm{g}\mathrm{e}\mathrm{n}}+\mathrm{h}\mathrm{A}{\mathrm{T}}_{\mathrm{a}\mathrm{m}\mathrm{b}}}{{\mathrm{m}}_{\mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}}\mathrm{C}+\mathrm{h}\mathrm{A}}}$$

(11)

$${\mathrm{m}}_{\mathrm{a}\mathrm{i}\mathrm{r}}={\rho }_{\mathrm{a}\mathrm{i}\mathrm{r}}\times {\mathrm{V}}_{\mathrm{a}\mathrm{i}\mathrm{r}}$$

(12)

$$∆{\mathrm{T}}_{\mathrm{a}\mathrm{i}\mathrm{r}}={\displaystyle \frac{{\mathrm{Q}}_{\mathrm{t}}}{{\mathrm{m}}_{\mathrm{a}\mathrm{i}\mathrm{r}}\times {\mathrm{C}}_{\mathrm{a}\mathrm{i}\mathrm{r}}}}$$

(13)

$${\mathrm{Q}}_{\mathrm{t}}=\mathrm{h}\mathrm{A}({\mathrm{T}}_{\mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}}-{\mathrm{T}}_{\mathrm{a}\mathrm{m}\mathrm{b}})$$

(14)

Numerical simulations are designed using mathematical models derived and are shown in Figure 4a,b. Figure 4a showcases the average temperature of the simulated battery pack model vs. the average temperature of the experimental results. The simulated and experimental average temperature results are aligned with each other, and the error rate is observed to be <1 °C.

Figure 4b illustrates the peak temperature of the battery pack, which clearly indicates a bigger difference between the simulated and experimental values. As the simulation model assumes a uniform and steady heat generation and thermal distribution for simplification and computational efficiency, it does not account for the non-uniformities observed in the experimental setup. These arise due to factors such as cell-to-cell variation, thermal contact resistance, localized airflow, and fixed current flow paths, all of which contribute to uneven heat buildup—clearly evident in the experimental results.

3. Results

The experimental setup described in Section 2 is tested under various conditions of switch count and cooling, as presented in Section 3.1, Section 3.2 and Section 3.3. The test cases focus on the discharge cycle, as the load requirement is dynamic in nature based on the input used. In contrast, the charging process is considered static and constant. Therefore, the primary focus is on the discharge cycle.

3.1. Thermal Characteristics of Conventional Battery Packs Under Natural and Forced Convection for Single-Terminal Conditions (STNC/STFC)

As discussed in Equation (9) in Section 2.3, heat generation within the cell during the charging and discharging process is the primary source of thermal behavior in battery packs. Numerous analytical studies have concluded that cells within a pack exhibit non-uniform thermal characteristics depending on their position, which significantly affects heat generation patterns [28]. Therefore, understanding the spatial distribution of heat generation and identifying regions with higher-than-expected heat density are crucial steps in designing an effective cooling strategy for battery packs. This highlights the importance of investigating heat generation patterns across different regions, particularly when evaluating the feasibility and stability of a multi-terminal approach under varying operating conditions.

It can be observed that heat generation and thermal distribution are non-uniform in conventional battery pack designs, as illustrated in

Table 5. Thermal images of single-terminal discharge cycle.

Test Cases	00:00	15:00	30:00	45:00	60:00
(a) 18_D_E_N_1T
(b) 18_D_E_F_1T_M

. The table presents thermal images of the battery pack at five different time points during a discharge cycle (1C-rate), corresponding to varying depths of discharge (DODs), as follows: 0 min = 0% DOD, 15 min = 25% DOD, 30 min = 50% DOD, 45 min = 75% DOD, and 60 min = 100% DOD. The data are organized into two rows, showing results from setups with natural convection and forced convection, respectively.

Test case A (18_D_1C_N_1T_0S) illustrates the flow of heat generation and dissipation from the terminal position located in Region 1 (R1) to Region 4 (R4). The ideal thermal operating conditions for Li-ion cells are estimated to be within the range of 15 °C to 35 °C [29,30]. However, studies show that in a battery pack, the temperature difference between cells should not exceed 5 °C [31]. Therefore, in the experimental setup, an upper temperature threshold (T_h) is considered to be 5 °C above the initial temperature, which serves as a practical benchmark to assess how quickly and severely thermal gradients develop during testing, ensuring that the temperature difference between regions and zones does not exceed this limit. At 0 min, the battery pack has an approximate initial temperature of 25.0 °C, which is mapped in the graphs shown in Figure 5a and Figure 6a. These figures depict the thermal characterization of the complete discharge cycle under natural convection with respect to the four regions and four zones in the pack, respectively. As the discharge progresses, the 15 min thermal image shows initial heat densities emerging in Zone 1 (Z1) and Zone 2 (Z2), with a temperature rise from 25.8 °C to 29.9 °C in Z1, and from 26.8 °C to 32.9 °C in Z2. The temperature in Z2 crosses the threshold at 350 s (Tz2), followed by Z1 at 1050 s (T_z1), and continues to rise throughout the discharge. Z2 reaches a peak temperature (P_T1+) of 36.9 °C, while Zone 3 (Z3) peaks at 28.6 °C (P_T2−), resulting in a maximum zone temperature difference (ΔP_Z) of 9.0 °C. This clearly indicates that the terminal zones are of particular interest, especially when a single-terminal approach is considered in the design. Regionally, R1 shows a continuous increase in temperature throughout the discharge cycle, reaching a peak (P_r1) of 31.4 °C, followed by R2 at 30.2 °C, R3 at 29.3 °C, and R4 at 28.3 °C. The peak regional temperature difference (ΔP_r) is 3.4 °C.

In Test Case B (18_D_1C_F_1T_0S), the results of the forced-convection-based setup are illustrated, where air is used as the cooling medium and a fan is employed to regulate the airflow velocity. The inlet is positioned at R1, and the airflow is maintained at a constant speed of 3.354 m/s. At 0 min, the thermal image records the average temperature of the regions and zones at approximately 25.0 °C, which is reflected in the graphs shown in Figure 5b and Figure 6b.

As the discharge progresses to 25% DOD, the variation from the initial temperature remains below 2.0 °C across the regions and remains below 3.0 °C across the zones of the battery pack. As the discharge cycle progresses to 100% DOD, Regions R1, R2, and R4 maintain relatively constant temperatures. This is because R1 and R2 are directly aligned with the inlet channels, resulting in effective cooling and temperature reduction. R4, being the farthest from the load terminal position and located at the end of the pack, experiences a slower temperature rise compared to the other regions. In contrast, Region 3 (R3) exhibits higher temperatures due to the reduced influence of air cooling and its closer proximity to the load terminals, which contributes to a higher heat generation rate.

Z2 reaches a peak temperature (P_T1+) of 29.2 °C, which is significantly lower than the P_T1+ value observed in Test Case 1. Nevertheless, the cooling system has demonstrated an effective performance by maintaining the peak region temperature difference (ΔP_r) below 1.2 °C and the peak zone temperature difference (ΔP_z) below 2.3 °C, while also keeping all temperature values under the defined threshold (T_h). However, this performance comes at the cost of 19.5 Wh of power consumption due to the additional cooling system integrated into the battery pack. Despite the improvements, the results clearly indicate that complete thermal uniformity has not yet been achieved.

3.2. Thermal Characteristics of Battery Packs Under Natural Convection for Multi-Terminal Conditions (MTNC)

Upon analyzing the drawbacks of the single-terminal approach and the inefficiency of external cooling for managing pack temperature, as discussed in Section 3.1, a multi-terminal-based experimental setup was tested. The resulting thermal images are presented in Table 5, where the three rows correspond to the results under the 1S, 2S, and 3S conditions indicating the number of terminal switches occurring within a single discharge cycle. The specific terminal switching times are listed in

Table 6. Thermal images of multi-terminal discharge cycles under natural convection.

	Time (Mins)
Test Cases	00:00	15:00	30:00	45:00	60:00
(a) 18_D_E_N_2T_1S
(b) 18_D_E_N_2T_2S
(c) 18_D_E_N_2T_3S

and are visualized in the sub-parts (a, b, and c) of Figure 7 and Figure 8. (Figures contain colored dotted lines representing: terminal switching point using black color, threshold value using red color, and region and zone wise color combination to represent threshold crossover.)

Test Case A (18_D_1C_N_2T_1S) reflects the results obtained through a two-terminal switching algorithm, where a single switch occurs at 50% DOD. This means that for the first 30 min, current is drawn from the terminal points located in Z3 and Zone 4 (Z4) of R4, where Z3 represents T₂₋ and Z4 represents T₂₊. The initial temperature across the pack is maintained at approximately 25.0 °C.

In Figure 7 sub-part a (1S) and Figure 8 sub-part a (1S), the discharge cycle under the 1s condition is presented, where the process begins with Terminal 2, located in R4. The terminal points are connected to Zone 3 (T₂₋) and Zone 4 (T₂₊). As a result, the temperatures in Regions R3 and R4 increase at a faster rate compared to R1 and R2, due to their proximity to the active terminal. These temperature trajectories continue steadily for the first 30 min of the discharge cycle, rising from an initial value of 26.5 °C to 29.8 °C in R3, and from 26.5 °C to 29.4 °C in R4, respectively. These temperature values are comparable to those observed in the single-terminal Test Case A at 50% DOD. At the 30 min mark, the terminal switching algorithm triggers the relay to switch from Terminal 2 to Terminal 1, which is located in R1. Terminal 1 connects to Z1 (T₁₋) and Z2 (T₁₊), effectively shifting the current draw to the opposite end of the pack.

As noted in prior studies [32], the positive terminal side tends to reflect higher temperature values compared to the negative side during discharge; this pattern also emerged in the experimental results. Following the terminal switch, the temperatures in Regions R3 and R4 begin to decrease, while the temperatures in Regions R1 and R2 increase sharply. During the first half of the discharge cycle, the average heat generation rates in R1, R2, R3, and R4 were recorded as 0.09 °C/min, 0.087 °C/min, 0.11 °C/min, and 0.097 °C/min, respectively. After the terminal switch, these rates shifted to 0.126 °C/min (R1), 0.096 °C/min (R2), 0.056 °C/min (R3), and 0.022 °C/min (R4).

This transition leads to peak region temperatures of P_r1 = 31.8 °C, P_r2 = 30.9 °C, P_r3 = 31.0 °C, and P_r4 = 29.6 °C, resulting in a regional peak temperature difference (ΔP_r) of <2.3 °C. On the other hand, Zone 3 (T_z3) exceeds the threshold at 1080 s, which is 750 s later than the earliest threshold crossing in the single-terminal natural convection (STNC) setup. A better thermal distribution is clearly observed, with all four zones—Z3, Z4, Z2, and Z1—crossing the threshold at 1080 s, 1320 s, 1920 s, and 2160 s, respectively. This results in a zone peak temperature difference (ΔP_z) of <4.90 °C, showing an improvement of 1.1 °C in ΔP_r and of 4.1 °C in ΔP_z compared to the STNC setup. These improvements translate to a 32.35% reduction in regional peak temperature and a 45.56% reduction in zone peak temperature, clearly indicating enhanced thermal uniformity and a lower overall peak temperature at the pack level.

Test Case B (18_D_1C_N_2T_2S) presents the results of a two-terminal switching algorithm where switching occurs twice—once at 33.33% DOD and again at 66.66% DOD. This means that the relay is triggered every 20 min during a 1C-rate discharge cycle, switching terminals in the following sequence: Terminal 2 → Terminal 1 → Terminal 2 (i.e., from R4 to R1 to R4). Figure 7 sub-part b (2S) and Figure 8 sub-part b (2S) depict the progression of Test Case B. As outlined in

Table 7. The 2S condition switch-count-based discharge cycle.

Switch Count (2S)	R1	R2	R3	R4	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{r}}$	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{z}}$	Pr	Pz
	(°C/min)	(°C/min)	(°C/min)	(°C/min)	(°C)	(°C)	(°C)	(°C)
0	0.026	0.032	0.063	0.095	1.6	3.5	29.3	31.1
1	0.116	0.068	0.021	−0.005	1.3	4.7	29.7	32.8
2	0.107	0.095	0.107	0.108	1	2.7	30.9	32.5

, the discharge cycle is divided into three parts based on the number of terminal switches. Upon reviewing the tabulated values, it is evident that during the initial phase (0 switches), the highest heat generation rate occurs in Region R4, as it serves as the initial load point. Consequently, the highest zone peak temperature (P_z) is observed at P_z2+, reaching 31.1 °C.

At the first switch, the terminal transitions to Region R1, making it the new load discharge point. This change results in a descending pattern of heat generation rates from R1 to R4. Notably, the average heat generation rate for R4 becomes negative, indicating a cooling phase, with a rate of −0.005 °C/min, representing active thermal reduction.

Following the second switch, the heat generation pattern shifts back, increasing from R4 to R1 as the discharge point returns to R4. During this phase, the highest peak temperature in the pack is recorded at P_T2+, reaching 32.8 °C—a value that still remains lower than the peak temperature observed in the STNC setup.

Figure 7 sub-part c (3S) and Figure 8 sub-part c (3S) illustrate the progression of Test Case C (18_D_1C_N_2T_3S), where the terminals are switched three times, at 15 min intervals during a 1C-rate discharge cycle. The discharge cycle for this 3S switching condition is divided into four segments, based on switch count, as shown in

Table 8. The 3S condition switch-count-based discharge cycle.

Switch Count (3S)	R1	R2	R3	R4	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{r}}$	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{z}}$	Pr	Pz
	(°C/min)	(°C/min)	(°C/min)	(°C/min)	(°C)	(°C)	(°C)	(°C)
0	0.073	0.087	0.1	0.107	1.4	2.5	28.1	29.4
1	0.107	0.067	0.027	−0.013	1.1	3.9	29.4	32.1
2	0.014	0.05	0.071	0.093	0.8	1.5	29.6	31.1
3	0.15	0.125	0.108	0.067	1.2	4.2	31.1	34.2

.

During the 0th switch (i.e., the first 15 min), the heat generation rate follows a descending order from R4 to R1, as the load is initially drawn from R4 region terminals (T₂₊ and T₂₋). Compared to the 2S condition, the 3S configuration exhibits reduced ΔP_r and ΔP_Z values, primarily due to the shorter switching interval (15 min vs. 20 min), which allows for better thermal regulation and distribution. Following the first switch, the terminal connection shifts from Zones Z3 and Z4 to Zones Z1 and Z2, reversing the heat generation pattern observed in the 0th switch. Among all segments of the discharge cycle, the highest ΔP_r of 1.4 °C occurs during the 0th switch, while the highest ΔP_z of 4.2 °C is observed during the 3rd switch. The peak region temperature (P_r1) crosses the T_h at 3000 s into the discharge; compared to the 2S case, the second region to cross the threshold is delayed by 100 s. In terms of zone behavior, the P_z > T_h event experiences a more than 600 s delay compared to the 2S setup, demonstrating the effectiveness of the 3S strategy in prolonging the threshold crossing time and allowing for a greater temperature uniformity across regions. When compared to the STNC setup, the 3S configuration delays the zone threshold crossing (P_z > T_h) by approximately 950 s and achieves a 2 °C reduction in ΔP_r, clearly highlighting its superior thermal management performance.

3.3. Thermal Characteristics of Battery Packs Under Forced Air-Based Convection for Multi-Terminal Conditions (MTFC)

In

Table 9. Thermal images of multi-terminal discharge cycles under forced convection.

	Time (Mins)
Test Cases	00:00	15:00	30:00	45:00	60:00
(a) 18_D_E_F_M_2T_1S
(b) 18_D_E_F_M_2T _2S
(c) 18_D_E_F_M_2T_3S

, similar to Section 3.2, three test cases are considered with respect to the switching mechanism, along with the addition of an integrated external coolant system using air as the cooling agent. In Test Case a (18_D_1C_F_2T_1S), one switching mechanism is showcased and the overall thermal characteristics with respect to regions and zones are reflected in Figure 9a and Figure 10a. Forced convection-based air cooling is achieved by introducing a fan into the enclosed chamber of the battery pack, where three equidistant valves are positioned as inlets and face R1. As a result, the temperature in R1 remains mild, even when load is applied from the T₁₊ and T₂₊ terminals, while T₂₊ and T₂₋ are the initial load points for discharge. At the 0th switch (i.e., the first half of the discharge), it is observed that the R4 and R3 regions of the pack have a higher ΔP_r, as shown in

Table 10. The forced convection condition-based switch-count-based discharge cycle.

Test Case	Switch Count	R1	R2	R3	R4	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{r}}$	$\mathbf{\Delta }{\mathbf{P}}_{\mathbf{z}}$	Pr	Pz
		(°C/min)	(°C/min)	(°C/min)	(°C/min)	(°C)	(°C)	(°C)	(°C)
1S	0	0.048	0.045	0.055	0.055	1	2.2	28.1	29.1
	1	0.026	0.026	0.019	0.019	0.7	2.3	28.8	30.8
2S	0	0.06	0.065	0.07	0.075	1	2.1	28.1	29.1
	1	0	−0.005	−0.02	−0.035	0.7	1.9	28.1	29.8
	2	0	0.024	0.041	0.041	1	2.1	28.7	29.6
3S	0	0.073	0.067	0.093	0.093	1	1.6	27.6	28.3
	1	0.05	0.038	0.031	0.006	1.2	1.9	27.9	29.2
	2	0.027	0.067	0.053	0.067	1.1	2	28.2	28.9
	3	0.118	0.109	0.1	0.082	0.9	1.8	28.6	29.7

. After the switch, in the remaining half of the discharge, a drop of 0.3 °C is observed in ΔP_r.

In the regions, the impact of air cooling is clearly visible, where the Pr for the 1S case with natural convection was 31.8 °C, which is 3.3 °C higher than the 1S case with the forced convection setup. However, even with forced convection, the temperature difference between the zones can still be observed. Terminal points T₂₊ and T₁₊ have higher temperatures compared to the other negative terminal points. A distinct change observed is that in the R1 region, T₁₋ even under load does not exhibit the same proportional temperature increase as the positive side of the terminal, as observed in previous setups. This can be explained by the continuous airflow in R1 and its lower heat generation rate compared to the positive side, which leads to a reduced temperature pattern for T₁₋. Upon reviewing all the cases, it is noted that in the 1S case, the zone temperature exceeds Th at the end of the 0th switch.

It can be concluded that the addition of the external cooling system was beneficial only in reducing the peak values of the regions and zones. Another observation is that the region facing the inlet (R1 and R2) is the only one where zone and region uniformity can be observed, as shown in Figure 9c and Figure 10c. However, the issue of non-uniformity could not be fully addressed using the cooling system alone. With the addition of the switching mechanism, thermal distribution and uniformity are significantly improved. Therefore, it can be concluded that the switching mechanism plays a pivotal role in managing thermal uniformity, distribution, and temperature control in battery packs.

3.4. Proof of System Stability Using Voltage, Current, and Capacity

In this section, the system’s stability is presented using the voltage, current, and capacity of the discharge cycles for the conventional single-terminal approach and for the 1S, 2S, and 3S conditions applied in the multi-terminal approach. Figure 11 illustrates the discharge cycle at a 1C rate for the test cases considered in the experimental setup. The pattern of the standard voltage discharge curve can be seen in Figure 11a, where the maximum charge voltage is set at 13.6 V, the nominal voltage is 12.4 V, and the cut-off voltage limit is 11.2 V. A current of 6 A is continuously drawn from the pack, and the rate of discharge is shown using capacity in Ah. The readings from the complete cycle show no anomalies and have been set as the standard benchmark pattern for comparison with the proposed model.

The graph in Figure 11b illustrates the multi-terminal 1S condition, where terminals are switched at 50% DoD and mapped on the voltage line at the 1800 s timeline. Current and capacity show no fluctuation in their trajectory. The voltage line shows only a very small drop, ΔV, during the transition, which is negligible compared to the benchmark voltage trajectory. Hence, it can be concluded that incorporating a multi-terminal design in the battery pack architecture and using a software-controlled terminal switching mechanism does not impact system performance.

Upon analyzing Figure 11c,d, it is clear that the performance of current and capacity is independent of the switching mechanism and switch count, while voltage shows minute signs of fluctuation. Every time a switching event occurs, a voltage fluctuation is observed, alternating between a drop and a rise. The first switching instance results in a voltage drop, followed by a voltage rise during the second instance, and this alternating pattern continues throughout the operation.

4. Discussion

In this study, a novel architectural design and well-structured testing conditions are presented and explored to investigate the thermal characteristics of a Li-ion battery pack module built using a multi-terminal approach, controlled via a software-driven switching mechanism under two cooling conditions. The major findings are summarized below.

• Peak value in the battery pack module: Under the 0S switch condition, i.e., the conventional single-terminal model, a peak temperature of 36.9 °C was observed. In comparison, the multi-terminal 2S setup showed an improvement of 11.11%, while the 1S and 3S setups demonstrated improvements of 6.23% under natural cooling conditions.
• Set threshold temperature crossover duration: In the 0S condition, the module crosses the T_h at 400 s into the discharge cycle, leading to the formation of a heat spot in the pack and resulting in approximately 88.89% of the remaining discharge duration being above the desired temperature range. In contrast, under the 1S, 2S, and 3S conditions, the pack crosses T_h at 1110, 700, and 1300 s, respectively. This results in approximately 69.17%, 80.56%, and 63.89% of the remaining discharge cycle duration being above the desired range, indicating a significantly better thermal performance.
• Thermal distribution in the battery pack module: Thermal distribution was evaluated by measuring the difference between the maximum and minimum temperature values across the regions at any given time. As shown in Figure 12, the 0S_R (single-terminal approach) shows a steadily increasing temperature difference, reaching 3 °C at 50% DoD and persisting for the rest of the discharge cycle. The 1S setup demonstrates improvement, maintaining a temperature difference of less than 2 °C, which only crosses this value around 80% DoD. In the 2S and 3S setups, the temperature difference remains below 1.6 °C and 1.4 °C, respectively, indicating a superior thermal uniformity.
• Heat Spot Generation and Management: The heat spots were identified and categorized as zones. Figure 13 presents the temperature difference between these zones with respect to the DoD percentage. A smaller difference value indicates a better thermal distribution and a lower heat spot generation rate. In the 0S_z case, representing the natural convection-based single-terminal setup, the temperature difference between zones was as high as 9 °C, indicating significant temperature concentration in specific zones, which leads to heat spot formation. Moreover, for more than 50% DoD, the temperature difference remained above 8 °C. In contrast, for the 1S, 2S, and 3S setups, the temperature difference did not exceed 5 °C throughout the discharge cycle, resulting in lower peak temperatures in heat-prone zones and improved thermal uniformity across zones.
• Influence of External Air-Based Forced Convection on the Multi-Terminal Approach: In the forced air-cooling setup, for the single-terminal test case, the peak temperature in R1 was reduced from 31.4 °C (observed in the natural convection case) to 27.2 °C. The cooling system successfully maintained thermal uniformity in R1, R2, and R4. However, it failed to achieve similar cooling in R3, since R1 and R2 were positioned near the coolant inlets; R4, which was at the end of the pack, experienced a naturally lower heat accumulation. R3, being neither close to the air inlet nor the terminal region, retained the heat generated during discharge, leading to a consistently higher temperature compared to the other regions.

When the multi-terminal approach was tested under forced cooling conditions, both peak values in the regions and zones showed a significant reduction, and thermal distribution across the pack was better managed.

5. Conclusions

In summary, the multi-terminal approach proves to be stable and effective in maintaining and controlling thermal distribution and heat generation in the battery pack. Among the test cases, the 2S setup performed best in reducing regional peak temperatures, while the 3S setup showed a superior performance in maintaining thermal uniformity throughout the pack. This study focused on exploring the thermal characteristics and stability of a multi-terminal switching strategy for BMs under controlled laboratory conditions. These conditions allowed us to isolate and understand the behavior of the proposed architecture. However, we recognize that real-world EV operations involve more complex and dynamic discharge patterns, which were beyond the scope of this work. Another important aspect that was not covered in this study is the potential for long-term cell aging due to frequent switching and switching current paths—factors like impedance growth or lithium plating could play a role over time. In particular, future studies will explore how the controller responds under varying C-rates and dynamic load profiles, as well as how the system performs over the long term. This could help us in understanding the full potential and limitations of the proposed design in real-world conditions.