Abstract

Carbon fiber organosheets are increasingly considered for electric vehicle battery enclosures due to their high stiffness-to-weight ratio and corrosion resistance. However, their crash performance depends not only on laminate design but also on manufacturing-induced fiber distortion during thermoforming. This paper presents an integrated engineering framework linking thermoforming simulation, explicit crash analysis, and data-driven modeling to evaluate crashworthiness trends in organosheet-based battery enclosures. A simplified enclosure geometry is used to generate a high-throughput dataset by varying laminate thickness, layer count, fiber orientation, and thermoforming process parameters. Post-forming fiber orientations are mapped directly into side-pole impact simulations to quantify crush load efficiency, absorbed energy, intrusion, and deceleration. Machine learning models are then trained to identify dominant design drivers and accelerate down-selection. Results show that laminate areal stiffness, governed primarily by layer count and ply thickness, dominates first-order crash response, while thermoforming temperatures and punch velocity influence local failure behavior through fiber distortion. The study demonstrates that crash simulation without forming history is incomplete and that data-driven tools are best used for screening rather than certification. The framework provides a practical pathway toward manufacturable, lightweight composite battery enclosures.

Keywords

carbon fiber organosheet; battery enclosure crashworthiness; thermoforming simulation; side-pole impact; machine learning in composites; process–structure–property relationship

1. Introduction

Battery enclosures are no longer passive housings. In modern electric vehicles, they are structural subsystems responsible for mechanical protection, energy absorption, vibration control, and system mass efficiency. As battery packs account for a significant fraction of vehicle weight, enclosure lightweighting directly affects driving range, payload capacity, and lifecycle efficiency.

Aluminum alloys dominate current enclosure designs because of their isotropic behavior, established crash performance, and thermal conductivity. However, aluminum structures are mass-intensive and offer limited tailoring of stiffness or deformation modes. Carbon fiber reinforced polymer (CFRP) organosheets provide an alternative, combining high specific stiffness with compatibility for high-rate thermoforming processes. Their adoption, however, is constrained by uncertainty in crash behavior, particularly when manufacturing-induced fiber distortion is ignored.

Many published studies on composite battery enclosures focus on material substitution or geometry optimization. Fewer consider how thermoforming alters fiber orientation and how those changes propagate into crash performance. This paper addresses that gap by presenting an integrated process–structure–crash framework that explicitly links forming simulation, impact analysis, and data-driven interpretation.

2. Background

2.1 Organosheets and Thermoforming Effects

Organosheets consist of continuous fiber reinforcements pre-impregnated with thermoplastic or thermoset matrices. During thermoforming, the sheet is heated and deformed over a tool, inducing fiber shear, rotation, and local thinning. These effects alter in-plane stiffness, bending resistance, and damage initiation paths.

Ignoring forming history in structural analysis implicitly assumes ideal fiber alignment, which can overestimate stiffness and energy absorption in crash scenarios.

2.2 Crashworthiness Metrics for Battery Enclosures

Crash performance of battery enclosures is commonly evaluated using metrics such as:

• Crush Load Efficiency (CLE): ratio of mean to peak force, indicating load stability
• Energy Absorption (EA): work done during deformation
• Intrusion: maximum inward displacement threatening battery modules
• Deceleration: peak inertial loading transmitted to the pack

These metrics collectively describe stability, safety, and energy management.

2.3 Role of Data-Driven Design

The design space of composite enclosures spans laminate architecture, thickness, orientation, and process conditions. Exhaustive exploration via explicit finite element analysis is computationally prohibitive. Machine learning provides a means to screen large parameter spaces and identify dominant trends, provided its role is limited to guidance rather than certification.

3. Methodology

3.1 Enclosure Representation

A simplified battery enclosure geometry is adopted to enable high-throughput simulation while preserving global stiffness and deformation modes. The enclosure is modeled as a shell-based composite structure with internal mass representation corresponding to the battery module.

3.2 Thermoforming Simulation

Thermoforming simulations are conducted over a wide process window, varying:

• number of laminate layers
• ply thickness
• initial fiber orientation sets
• organosheet temperature
• tool temperature
• punch velocity
• ambient air temperature

Fiber orientation after forming is extracted and mapped to the structural model.

3.3 Crash Simulation

Explicit side-pole impact simulations are performed at a representative impact velocity. The formed fiber orientations are incorporated into the composite damage model. Output metrics include CLE, EA, intrusion, and deceleration.

3.4 Data Generation and Screening

Hundreds of thermoforming simulations are attempted. A significant fraction fails due to wrinkling or non-formability, defining the manufacturable process window. Only successfully formed cases are advanced to crash analysis.

3.5 Machine Learning Analysis

Tree-based regression models are trained on the crash dataset to predict performance metrics and extract feature importance. Symbolic regression is used as a supplementary interpretability tool to identify functional relationships within the dataset.

4. Results

4.1 Manufacturability Trends

Approximately one-third of simulated process combinations result in forming failure, highlighting the narrow manufacturability window of organosheet thermoforming. This reinforces that crash performance must be evaluated only within feasible process domains.

4.2 Crash Response Characteristics

Across all evaluated designs, the dominant drivers of crash response are:

1. Number of laminate layers
2. Individual ply thickness

These parameters govern bending stiffness and load-bearing capacity, directly influencing CLE, EA, and intrusion.

4.3 Influence of Thermoforming Parameters

Thermoforming temperatures and punch velocity have secondary but non-negligible effects. Elevated temperatures reduce forming stress but increase fiber distortion, altering local stiffness and damage initiation. Air temperature influences deceleration by modifying global stiffness distribution.

4.4 Data-Driven Model Performance

Machine learning models achieve high predictive accuracy for all crash metrics within the explored design space. However, performance deteriorates outside the trained domain, underscoring that these models capture correlations rather than universal physical laws.

5. Discussion

5.1 Why Forming History Matters

Results confirm that crash simulations assuming ideal fiber alignment systematically misrepresent local stiffness and failure behavior. Fiber rotation during forming redistributes load paths, particularly in curved regions, directly affecting intrusion and energy absorption.

5.2 Structural Dominance over Process Noise

While process parameters influence local response, first-order crash behavior is governed by laminate areal stiffness. This suggests a practical design hierarchy: select laminate thickness and layer count first, then tune process parameters to maintain formability and control fiber distortion.

5.3 Role of Machine Learning in Engineering Practice

Machine learning is effective for rapid screening and trend identification but cannot replace high-fidelity simulation or testing. Its value lies in reducing the number of candidate designs before expensive validation stages.

6. Limitations

The study uses a simplified enclosure geometry without mounting interfaces, sealing features, or detailed battery modules. Only side-pole impact is considered. Thermal runaway, debris impact, and joint failure are not addressed. Material models rely on assumed damage parameters without experimental calibration.

7. Conclusions

An integrated thermoforming–crash–data framework provides deeper insight into the crashworthiness of carbon fiber organosheet battery enclosures than material substitution studies alone. Manufacturing-induced fiber distortion significantly influences crash response and must be included in structural evaluation. Laminate thickness and layer count dominate first-order crash performance, while process parameters shape local behavior and manufacturability. Data-driven models are valuable for screening but must be anchored to physics-based simulation and testing. The framework outlined here offers a practical pathway toward lightweight, manufacturable composite battery enclosures.

References

1. Jones, R.M., Mechanics of Composite Materials, Taylor & Francis.
2. Mallick, P.K., Fiber-Reinforced Composites, CRC Press.
3. Dhoke, A., Dalavi, A., Lightweight design of EV battery enclosures, Int. J. Sustainable Transportation Technology.
4. Shaikh, S.A. et al., Finite element and ML-guided design of CFRP organosheet battery enclosures, arXiv:2309.00637.