Optimization of An Air-Based Heat Management System for Dusty Particulate Matter-Covered Lithium-Ion Battery Packs

This study aim to solve the problem of the cell temperature rise and the performance decline caused by the dusty particulate matter covering the surface of the cell through allocation of airflow velocities at the base of battery cooling box under the goal of low energy consumption. An optimization method simultaneously considering energy consumption and thermal performance of battery management system has been established. And it can be widely used to improve the lifecycle of the battery pack battery with minimal operating cost.

To begin the fluid flow analysis, drag fluid flow from the toolbox and analysis systems into the project schematic zone. Then hold airFEM mesh, batteryFEM mesh and dpmFEM mesh fluid flow with the left mouse button and move them to fluid flow setup. Right click on fluid flow, then set up, and select update to enter the setup window.

Confirm the validity of the FEM model and check if the mesh has a negative volume. Now enter the setting interface of the viscous model and the radiation model and select the k-epsilon model and the discreet ordinance model respectively. Change the fluid type of the numbered battery domains to the solid type.

Then on the solid window, double click each battery domain to change the DPM material to the battery material. Subsequently, choose the source terms item and check the highlighted source terms to add an energy source by assigning the number in the number of energy sources and selecting constant type to input the source value. Change the fluid type of numbered DPM domains to solid type.

Next, convert the type of all renamed surfaces, including the inner surfaces of the air domain, all sides of the battery domains and DPM domains from the default wall to the interface. To generate mesh interfaces, click the mesh interfaces and enter the create and edit mesh interfaces window. Match the cavity surfaces to all sides except the upper sides of the battery domains and the lower sides of the DPM domains.

Name and number them as interface 1 to interface 11 respectively. Then match the upper sides of the battery domains and the lower sides of the DPM domains. Name and number them as interface 12 to interface 22 respectively.

To assign the outer border surface as the wall thermal boundary, set the heat transfer coefficient as 5 in the mixed thermal condition. Then change the material from default aluminum to the previously self-defined battery box material. In the velocity inlet window, set airflow velocities of all inlets to 5 meters per second.

Then set the gauge pressure of the outlet to zero in the pressure outlet window. Next, set the state of the computing domain with an initial temperature of 300 kelvin and solution initialization type best standard initialization. Set the number of iterations to 2, 000 and click calculate to begin the simulation.

To enter the CFD post window, double click fluid flow followed by results. Then from the toolbox, double click the icon of contour. In the location selector, choose all sides of the batteries and switch from pressure to temperature.

Then click apply to generate the temperature contour of the batteries. Click file, then export to select the temperature of the selected variables, then click the dropdown button of the locations to select the battery domains. Click okay, followed by the save button to quit.

Battery pack temperature variation at different inlet airflow velocities demonstrated that the maximum battery pack temperature decreases with the increase of inlet flow velocity. The comparison of the battery pack temperature distribution and the second battery temperature distribution in different environments showed that the temperature of the battery increases under dusty conditions due to the low thermal conductivity of DPM. Begin by opening an empty spreadsheet to create a table with rows named inlet1, inlet2, and inlet3 in the first column.

Save the file as sample.xlsx. Run the optimization software and drag the spreadsheet icon onto the single arrow of task one. Then double click the spreadsheet icon to open the component editor Excel window.

Click the browse button to import sample.xlsx. Then click add this mapping to map inlet1, inlet2, and inlet3 to A1, A2 and A3 as parameters. Click okay to return to the initial window, drag the DOE icon into task one and double click on it to open the component editor DOE window.

Select Optimal Latin Hypercube. And in the general window, set the number of points to 15. Navigate to the factors window and set 5.5 as the upper limit and 5 as the lower limit for A1, A2 and A3.Switch to the design matrix window and click generate to produce random sampling points for different inlet velocities.

Close the optimization software. Combine the predictor variables x1, x2, and x3 of the velocity arrays and y of the temperature arrays to form a new table, save the table as sample. txt and import it to fit a response surface model.

Rerun the optimization software and drag the approximation icon onto the single arrow of task one. Double click the task one icon to pop up the component editor approximation window to select the response surface model. Navigate to the data file window and import the sample.

txt file containing the prediction variables. Switch to the parameters window and click scan to open the parameters in the data file window where the predictor variables of x1, x2 and x3 are defined as input and y as output. Next, go to the technique options window and select quadratic in polynomial order.

Switch to the error analysis options window and select cross validation in the error analysis method. Then switch to the view data window and click initialize now to obtain the coefficients of the quadratic linear regression equation. Click the error analysis button to open the approximation error analysis window.

Check if the errors meet the acceptable standards for each error type. Close the approximation component window. Drag the optimization icon into task one and double click it to open the component editor optimization window.

Then select adaptive, simulated and kneeling in the optimization technique. Navigate to the variables window and set 5.5 as the upper limit and 5 as the lower limit. Switch to the objectives window and select the Y parameter before closing the component editor optimization window.

Finally click the run optimization button and wait for the optimization results. Our square analysis showed that the second order polynomial response surface approximation model had a good fitting accuracy. The maximum temperature obtained through optimization was 309.39 kelvin with specific airflow velocities at the inlets.

The optimized airflow velocities led to a lower maximum temperature of 309.39 kelvin as compared to the initial case, the sum of airflow velocities of the optimized case is less than the non-optimized cases. However, the maximum temperature does not increase with decreasing airflow velocity. Further, the flow line distribution after optimization becomes wider.

Factor x1 has the greatest influence on temperature, while factors x2 and x3 have similar effects.