Parametric Battery Pack Modeling in GT-SUITE for All Existing Cooling Concepts
How would it benefit your workflow if there was a multiphysics CAE tool that performs the following tasks within a single software?
- 1D system analysis of battery and vehicle cooling including all the detailed components, DOE matrices and optimization.
- Fully parametric 3D finite element (FE) analysis of battery packs including all the cooling concepts, namely direct (air, dielectric) or indirect (plates, fins).
- Both 1D as well as 3D multidomain (thermal, electrical, chemical) modeling of battery cells for all shapes and chemistries including all the cell level subcomponents.
- 3D computational flow dynamics and 3D thermal conjugate heat transfer analyses for complex flow.
- Flexible CAD based custom 3D finite element analysis of battery packs for complex cooling concept.
- Fast running 1D-3D flow thermal models for transient drive cycle simulations and complex phenomenon such as thermal runaway.
- Switch between a detailed 3D finite element (both parametric and CAD shapes) and a 1D lumped mass thermal domain in a single model for detailed component and real time system analyses, respectively.
After a lot of innovative work done at Gamma Technologies in recent years, the latest version 2021 of GT-SUITE does all the above-mentioned tasks, as shown in Figure 1.
Here in this blog I would like to give you a sneak peek into some of the above features to be released in the coming version. The topics I chose for this blog are following:
- Fully parametric 3D finite element (FE) analyses of battery packs including all the cooling concepts, namely, direct (air, dielectric) or indirect (plates, fins)
- Switch between a detailed 3D finite element and a 1D lumped mass domain in a single model for component and system analysis, respectively
A further demonstration of these features will be available during the “Thermal Management Systems and Component Modeling” seminar at the Global GT Virtual Conference 2020.
At early stages of development cycle engineers need to define the battery packs in terms of shape, size, and cooling concept. The design goals include:
- Uniform temperature distribution at cell, module, and pack level
- Effective temperature control during charging/discharging and cold start
- Containing uncontrolled battery failures such as thermal runaway
Because multiple cooling concepts could meet the above goals, easy transition from one cooling concept to another cooling concept will enable a full evaluation of the options. Using a parametric 3D finite element approach enables a quick comparison between different cooling concepts to decide which one roughly meets the system requirements. Through model scaling they can also test the compactness and overall pack sizes while staying within the safe working temperatures.
GT-SUITE offers a fully parametric 3D finite element meshing capability to cover a wide range of geometrical shapes. For this blog I took the example of a battery module equipped with prismatic cells and I tested four cooling concepts (see Figure 2):
- Direct dielectric immersion cooling
- Direct active air flow cooling
- Indirect vertically embedded plate cooling
- Indirect horizontal plate with extended fin cooling
For each cooling concept, I identified a repetitive building block that I parametrically modeled in GT-ISE (the modeling environment of GT-SUITE) and built multiple copies to create the battery module. Within the repetitive building block, parametric 2D FE meshes were extruded to create parametric 3D FE meshes with all the parameters available in the case-setup of the model for further DOE analyses.
I ran a brief analysis of the above mentioned four cooling concepts for two given sizes of cells. In the analysis, all four battery modules were discharged at a constant C-rate of 2C. At this discharge rate, active air cooling is not suitable in keeping the module under safe temperature limit (see Figure 3). The fin extended cooling approach is effective at least for smaller cells, but it reacts slowly to both heating and cooling. But for larger cells, fin extended cooling is not suitable at least for this extreme case of 2C-rate at the given pump flow rate. For both embedded plates and immersion cooling, the cooling concepts are equally effective for the smaller and larger cell sizes.
After this initial analysis one can ask the question if fin extended cooling is good enough in terms of cost and performance or the embedded plates and immersion cooling options should be pursued. To help answer this question, the temperature distribution between cells and modules can be evaluated. For this metric the immersion cooling concept performs better than the other three cooling concepts with lowest maximum module temperature and highly uniform temperature distribution (see Figure 4).
A very compact case-setup of two cases is presented here, but for project work I would recommend to run multiple case DOE analysis to optimize the cell dimensions and extract maximum performance from the module within the safe limit of maximum allowed temperatures.
For a chosen cooling concept, cell shape, and size, as we move forward in the system development cycle the goal is to achieve an accurate system model to design and test the subsystem and system controls. Later in the development stages such analyses are done either virtually (via software in the loop) or using an ECU test bench (Hardware in the loop). But the parametric approach presented here gives the possibility to move towards the right side of the V-cycle quite early in the development stage without having to create the first CAD models and prototypes.
To integrate the controls model or hardware it is required to have real-time capable GT-SUITE models. In general practice this is achieved by simplifying the model i.e. replacing the detailed finite element structures with lumped masses. A well calibrated lumped mass model characterized with correct mass, heat transfer areas and material thermal properties can provide the accurate component thermal inertia and temperatures required for the system transient analyses.
In GT-SUITE v2021, it is possible to maintain two different levels of details i.e. 3D finite element and 1D lumped mass domain within one model. With this capability, a single model can be switched between the two domains based on the desired component or system analysis. Unlike in the past, the advantage here is that one does not have to maintain two different models and the model building time is significantly reduced (see below the new workflow in Figure 5).
In the new workflow the solid shapes are converted into 3D mesh in GEM3D and the geometrical information (distance to mass center and heat transfer areas) needed to characterize a lumped thermal mass is also extracted. As a result, the exported part in GT-ISE can be switched between the 1D thermal mass and 3D FE solution. For components that use a 3D FE mesh generated parametrically directly in GT-ISE or imported into GT-ISE from a third party tool, the solver is programmed to calculate the geometrical parameters needed to characterize the lumped thermal mass from the 3D FE mesh. These capabilities enable a real-time capable model to be developed from the detailed finite element model regardless of the source of the mesh.
Because the fidelity of the model has been reduced, the 1D thermal model requires calibration so that it can produce the same results as the 3D counterpart. For this task “Distance Multipliers” can be used to scale the “Distance to Mass Center” or in other words the distance of heat transfer in Fourier’s law. In a future version of GT-SUITE these values will be calibrated automatically from the 3D FE model to reduce the building efforts needed for 1D thermal models.
For the case presented here, the temperature different between the 3D FE and 1D thermal mass models was negligible due to the symmetrical nature of the geometry. As a result I decided not to calibrate the modules and simply add them to create a 1D battery pack model consisting of 15 battery modules (see Figure 6). This pack was integrated into a 1D electric vehicle (EV) cooling system model. Subsequently, I ran some drive-cycle simulations (The EV cooling system model used is from the example library of GT-SUITE under Cooling_Vehicle_Thermal_Management).
Below in Figure 7 are some results of the drive cycle analysis for different cooling concepts. Although the cumulative energy production over the 30 minutes of WLTC drive cycle was not as high as it was in the case of a 2C-rate constant discharge, the observation about the suitability of different cooling concepts holds true in Figure 7. The air-cooling approach is not suitable for this pack design at cold or warm conditions.
The main takeaway from the above example is that with this integrated model you have the possibility to design system controls, not just for some average lumped battery pack temperature but for specific cells that are prone to maximum temperature due to their location in the pack. Moreover, the well calibrated 1D pack model gives accurate heat rejection to the cooling system for the sizing of heat exchangers. It also accounts for accurate pressure drop in the cooling system for the sizing of coolant pumps. And not to forget that this 1D pack model was not built separately, originally it was the detailed 3D FE model build for the component analysis and simply switched to be used as a 1D model.
With this introduction I would like to invite you to join us in our “Thermal Management Systems and Component Modeling” seminar at this year’s Global GT conference to learn more about the workflow improvements and new features available in GT-SUITE v2021.
Written by Dr. Dig Vijay