In my research on thermal management for electric vehicles, I have focused on addressing the critical challenge of maintaining optimal operating temperatures for lithium-ion batteries. As the backbone of modern EV powertrains, lithium-ion batteries exhibit high sensitivity to temperature fluctuations, with performance, safety, and longevity heavily dependent on keeping cell temperatures within a narrow window of 20°C to 40°C and module temperature differences below 5°C. While various cooling methods exist, my work explores a dripping contact liquid cooling approach, which offers direct coolant-cell contact for high heat transfer efficiency while promoting excellent temperature uniformity. To enhance this system’s adaptability and energy efficiency, I designed and implemented an active flow control system that dynamically adjusts coolant flow based on real-time battery temperature feedback. This article details the complete design process, from theoretical foundation and hardware selection to control strategy development and experimental validation, all from my firsthand perspective as the system architect.
The core principle of the dripping contact cooling method I developed involves a dielectric coolant, specifically transformer oil, being dripped uniformly onto the top surfaces of battery cells from a multi-channel reservoir. The coolant then spreads and flows down the cell surfaces along guided paths, absorbing heat before draining away. This method ensures that each cell is contacted by coolant at approximately the same initial temperature, which is paramount for minimizing temperature gradients within the module. The system’s schematic, which I conceptualized, consists of a battery module, a coolant tank, a drive pump, a control module, and a collection sump, forming a closed loop. The innovation lies in integrating a feedback-based flow control system into this loop. The control module, which I programmed, receives temperature signals from sensors attached to the lithium-ion battery cells and regulates the pump speed to modulate flow rate, thereby creating an intelligent thermal management system.
My design targets a module of 18650-type lithium-ion batteries arranged in a 7-series, 4-parallel (7S4P) configuration. A key mechanical design task was ensuring uniform coolant distribution. I engineered a reservoir divided into four flow channels, each with seven drip outlets aligned above individual cells. The uniformity of flow from these multiple outlets is critical. The flow rate through each orifice can be affected by pressure variations along the manifold. To assess this, I performed a calibration test, measuring the outflow from each orifice under a constant pump drive. The results confirmed satisfactory uniformity. The flow uniformity coefficient $\eta$ is defined as:
$$ \eta = \frac{q_{\text{min}}}{q_{\text{max}}} $$
where $q_{\text{min}}$ and $q_{\text{max}}$ are the minimum and maximum measured flow rates from the orifices, respectively. The tested system achieved a $\eta$ value exceeding 0.95, validating the manifold design for the dripping contact cooling of the lithium-ion battery pack.
Hardware Selection for the Flow Control System
The effectiveness of the flow control system hinges on precise component selection. I based my choices on criteria of accuracy, reliability, and suitability for the lithium-ion battery thermal management application.
Microcontroller Unit (MCU): The MCU is the brain of the system. I selected the STM32F103C8T6, a 32-bit ARM Cortex-M3 based microcontroller. Its key parameters, which aligned perfectly with my needs, are summarized below.
| Parameter | Specification |
|---|---|
| Core | ARM Cortex-M3 |
| RAM | 20 KB (SRAM) |
| Flash Memory | 64 KB |
| Operating Voltage | 2.0 – 3.6 V (Standard 3.3V) |
| Package | LQFP48 |
This MCU handles critical tasks: acquiring analog temperature signals from sensors, processing the data to determine the state of the lithium-ion battery pack, executing the predefined flow control algorithm, and generating a Pulse-Width Modulation (PWM) signal to drive the pump.
Temperature Sensor: For accurate temperature monitoring of the lithium-ion battery surfaces, I chose K-type adhesive thermocouples. Their specifications are as follows.
| Parameter | Specification |
|---|---|
| Lead Material | Teflon-insulated |
| Thermocouple Material | Nickel-Chromium / Nickel-Silicon |
| Measurement Range | -40°C to 200°C |
| Probe Dimensions (l×w×δ) | 20 mm × 12 mm × 0.5 mm |
These sensors provide a voltage output linearly proportional to temperature, which is read by the MCU’s Analog-to-Digital Converter (ADC).
Coolant Pump: The pump must provide a controllable flow rate within the required range. Prior simulation work indicated that a flow rate of approximately 20 mL/min per drip outlet provides effective cooling for a single lithium-ion battery cell under typical loads. For the full module (28 cells), the theoretical maximum flow demand $Q_{\text{max}}$ is:
$$ Q_{\text{max}} = 20 \times L \times P $$
$$ Q_{\text{max}} = 20 \times 4 \times 7 = 560 \, \text{mL/min} $$
where $L$ is the number of parallel strings (4) and $P$ is the number of series cells (7). I selected a TOPSFLO FG-A12-1006 brushless DC pump, which offers a maximum flow of 2 L/min, comfortably exceeding the requirement. Crucially, in the range of 30 to 800 mL/min, its flow rate exhibits a linear relationship with the input PWM duty cycle, enabling precise control.
System Software Design and Control Strategy
I developed all firmware for the MCU to implement the feedback control loop. The software architecture centers on two main functions: temperature data acquisition and PWM generation for pump speed control.
PWM Driver Program: I configured one of the MCU’s general-purpose timers (TIM) in output compare mode to generate a PWM signal. The frequency ($Freq$), duty cycle ($Duty$), and resolution ($Reso$) of the PWM wave are determined by timer register values:
$$ Freq = \frac{Ck\_PSC}{(PSC + 1) \times (ARR + 1)} $$
$$ Duty = \frac{CCR}{ARR + 1} $$
$$ Reso = \frac{1}{ARR + 1} $$
where $Ck\_PSC$ is the timer clock source frequency, $PSC$ is the prescaler value, $ARR$ is the auto-reload value, and $CCR$ is the capture/compare register value. I set $ARR = 999$ and $PSC = 71, resulting in a PWM frequency of 1 kHz and a resolution of 1/1000 (0.1%). The variable $CCR$, and thus the $Duty$, is dynamically adjusted by the control algorithm.
Temperature Acquisition: The thermocouple’s analog voltage is read via the MCU’s 12-bit ADC. The conversion from the ADC digital value ($ADValue$, range 0-4095) to temperature ($Temp$ in °C) involves two steps. First, the voltage is calculated:
$$ Voltage = \frac{ADValue}{4095} \times 3.3 \, \text{V} $$
Then, using the linear characteristic of the K-type thermocouple (0V @ -40°C, 5V @ 200°C), the temperature is derived:
$$ Temp = \frac{Voltage \times 240}{5} – 40 $$
$$ Temp = (Voltage \times 48) – 40 $$
This calculation is performed in real-time for multiple sensor inputs to monitor the lithium-ion battery module’s thermal state.
Calibration and Control Strategy Formulation: Before defining the control logic, I established the empirical relationship between the PWM duty cycle and the resulting coolant flow rate per orifice. I conducted a calibration experiment, measuring the flow rate at various fixed duty cycles. The results are tabulated below.
| Pump Drive Duty Cycle, $Duty$ (%) | Single-Orifice Flow Rate, $Q$ (mL/min) |
|---|---|
| 23.8 | 8.5 |
| 23.9 | 12.0 |
| 24.0 | 12.8 |
| 25.0 | 13.8 |
| 26.0 | 14.8 |
| 27.0 | 15.8 |
| 28.0 | 16.8 |
| 29.0 | 17.8 |
| 30.0 | 18.8 |
For $Duty \geq 24.0\%$, the relationship is effectively linear. I performed a linear regression to obtain the control equation for the flow controller:
$$ Duty = \frac{Q + 11.2}{1.00} \quad \text{or simply} \quad Duty(\%) = Q + 11.2 $$
where $Q$ is the desired per-orifice flow in mL/min. However, the control strategy is temperature-driven. Based on simulation data and the goal of maintaining the lithium-ion battery below 40°C, I devised a tiered strategy. The system remains off until the maximum monitored cell temperature ($T_{max}$) reaches 30°C. Between 30°C and 35°C, duty cycle varies linearly with temperature. Above 35°C, a stepwise strategy is employed to provide robust cooling power while avoiding excessive pump cycling. The complete control law I implemented is as follows.
| Battery Temperature Condition, $T_{max}$ | Pump Drive Duty Cycle, $Duty$ (%) | Remarks |
|---|---|---|
| $T_{max} < 30^\circ C$ | 0 | Cooling system inactive |
| $30^\circ C \leq T_{max} \leq 35^\circ C$ | $(2 \times T_{max} + 170) / 10$ | Linear control region |
| $35^\circ C < T_{max} < 36^\circ C$ | 24.0 | First cooling step |
| $36^\circ C \leq T_{max} < 37^\circ C$ | 25.0 | Second cooling step |
| $37^\circ C \leq T_{max} < 38^\circ C$ | 26.0 | Third cooling step |
| $38^\circ C \leq T_{max} < 39^\circ C$ | 27.0 | Fourth cooling step |
| $39^\circ C \leq T_{max} < 40^\circ C$ | 28.0 | Fifth cooling step |
| $T_{max} \geq 40^\circ C$ | 30.0 | Maximum cooling (safety mode) |
This strategy ensures that the flow rate increases progressively with the thermal load of the lithium-ion battery, promoting energy efficiency at lower temperatures and guaranteeing safety at high temperatures.
Experimental Validation and Results Analysis
To validate the entire dripping contact cooling system with integrated flow control, I constructed a comprehensive test platform. It included the 7S4P lithium-ion battery module, the custom-designed dripping contact cooling assembly with its reservoir and drip outlets, the brushless DC pump, the STM32-based control board, K-type thermocouples, a programmable battery cycler for applying discharge loads, and a data acquisition system for independent temperature logging.
I instrumented three representative lithium-ion battery cells within the module (positions corresponding to cells 17, 24, and 31 in the layout) with thermocouples at their bottom surfaces to capture the worst-case temperatures. The module was subjected to a continuous discharge at a 1.5C rate, a demanding condition that generates significant heat. The flow control system was active, following the strategy I defined. The temperature profiles of the three monitored cells over time are presented in the table below, which was compiled from the data acquisition system’s records at key time intervals.
| Time Elapsed (minutes) | Temperature at Cell 17 (°C) | Temperature at Cell 24 (°C) | Temperature at Cell 31 (°C) | System State / Inferred $Duty$ |
|---|---|---|---|---|
| 0 | 25.0 | 25.0 | 25.0 | Off (0%) |
| 5 | 29.5 | 29.8 | 29.3 | Approaching threshold |
| 10 | 32.1 | 32.3 | 31.9 | Linear control (~23.4%) |
| 15 | 34.7 | 34.9 | 34.5 | Linear control (~23.9%) |
| 20 | 36.0 | 36.2 | 35.8 | Step 1 (24.0%) |
| 25 | 37.1 | 37.3 | 36.9 | Step 2 (25.0%) |
| 30 | 38.0 | 38.3 | 37.8 | Step 3 (26.0%) |
| 35 | 38.7 | 39.0 | 38.5 | Step 4 (27.0%) |
| 40 | 39.2 | 39.5 | 39.0 | Step 5 (28.0%) |
| 45 | 39.5 | 39.8 | 39.4 | Step 5 (28.0%) |
| 50 | 39.7 | 40.0 | 39.6 | Near Max (28.0%) |
| 55 | 39.6 | 39.9 | 39.5 | Step 5 (28.0%) |
The results demonstrate the efficacy of the flow control system. The temperatures of all three monitored lithium-ion battery cells followed a similar trajectory: a rapid initial rise, followed by a gradual deceleration as the cooling system increased its flow rate, eventually oscillating near a steady-state value. The maximum recorded temperature for any cell was 40.0°C, successfully contained at the safety threshold. The temperature difference between the hottest and coldest monitored points remained within 1°C during the high-load phase, underscoring the inherent temperature uniformity advantage of the dripping contact method for lithium-ion battery cooling.
Furthermore, the system behavior matched the designed control strategy. The pump activated as temperatures crossed 30°C, and the duty cycle increased stepwise in response to the rising temperature of the lithium-ion battery pack. The system reached a thermal equilibrium where the heat generated by the lithium-ion battery during 1.5C discharge was effectively balanced by the heat removed by the controlled coolant flow, preventing thermal runaway.
Conclusion
In this project, I successfully designed, built, and tested an active flow control system for a dripping contact liquid cooling device aimed at thermal management of lithium-ion battery packs. The system integrates precise hardware components—an ARM Cortex-M3 microcontroller, K-type thermocouples, and a linear-response brushless DC pump—with custom firmware implementing a temperature-feedback control algorithm. The developed tiered control strategy dynamically adjusts coolant flow based on the real-time temperature of the lithium-ion battery, achieving the principle of “higher temperature, higher flow.”
Experimental validation under a strenuous 1.5C continuous discharge condition confirmed that the flow control system responds promptly and accurately to temperature changes. It successfully maintained the maximum temperature of the lithium-ion battery module below the critical 40°C limit while preserving excellent temperature uniformity. This work demonstrates that adding intelligent flow control to a dripping contact cooling system not only enhances its safety performance under high load but also improves its overall energy efficiency by minimizing unnecessary pump operation during mild conditions. The methodologies and results presented here contribute to the advancing field of adaptive thermal management systems for high-performance lithium-ion batteries in electric vehicles and energy storage applications.