TL;DR
This paper investigates the accuracy of different simulation methods for predicting the behavior of Sheet Molding Compound (SMC) during compression molding in complex geometries. They found that while macroscopic models are good for overall flow, directly simulating fiber interactions is crucial for accurately predicting defects like fiber matrix separation in confined regions.
What Problem Does This Paper Address?
The automotive industry is increasingly using lightweight composite materials like SMC for structural components, especially with the shift towards electric vehicles. SMC compression molding is an economical way to produce these parts, but complex shapes like ribs and bosses can lead to molding defects. These defects, such as fiber matrix separation (FMS) and fiber re-orientation, significantly impact the final mechanical properties of the part. Accurately predicting these phenomena is vital for designing reliable and safe components. However, existing simulation tools struggle to capture these localized effects, especially when the fiber length is comparable to the feature size of the mold. This paper aims to compare different simulation approaches to understand their predictive capabilities for these complex scenarios.
Methods
The authors compared four simulation approaches:
- Macroscopic Models:
- MOLDEX3D 2020 & MOLDFLOW 2021: These are commercial software packages that use tensor-based models to predict fiber orientation and flow. They treat the composite as a continuous medium and assume fibers and matrix move at the same velocity, neglecting fiber-fiber interactions and detailed FMS.
- Direct Simulation Models:
- 3D TIMON (Direct Fiber Simulation - DFS): This method simulates individual fiber movements but treats them as non-interacting rods. It's a one-way coupled approach where fiber motion is dictated by the flow, but fibers don't influence the flow.
- Direct Bundle Simulation (DBS): This approach models fibers as bundles, accounting for fiber-fiber interactions and the effect of fibers on the flow. It's a more computationally intensive but potentially more accurate method for capturing complex interactions.
The simulations were validated against experimental data obtained through micro-computed tomography (µCT) for detailed structural analysis and thermal gravimetric analysis (TGA) for fiber volume content.
Datasets
The primary dataset used in this study consists of experimental results from compression molding trials of SMC in a honeycomb structure. This includes:
- µCT scans: Used to visualize the 3D fiber architecture and identify defects like fiber matrix separation.
- TGA measurements: Used to quantify the fiber volume content (FVC) in different regions of the molded parts.
- Press data: Recorded during the molding process to measure compression forces and velocities.
- Short shots: Incomplete mold fillings used to analyze the flow front progression.
The authors also utilized existing material characterization data for the SMC matrix rheology and friction properties.
Key Insights and Ideas
- Scale Separation is Crucial: The paper strongly emphasizes that when fiber lengths are comparable to the mold's geometric features (lack of scale separation), macroscopic tensor-based models struggle to predict FMS and fiber orientation accurately, especially in the thickness direction.
- Fiber-Fiber Interactions Matter: The Direct Bundle Simulation, which accounts for fiber-fiber interactions, showed better agreement with experimental results for FMS and FVC distribution compared to models that neglect these interactions.
- FMS is Inherently Random: The experimental results (µCT and TGA) revealed significant scatter in FMS and FVC, even under identical molding conditions. This is attributed to the heterogeneous nature of SMC and random events during flow, making precise prediction challenging for any model.
- Direct Simulation for Complex Geometries: For confined regions with complex geometries like ribs and bosses, directly modeling fiber behavior (like in DBS) is essential for capturing localized phenomena and achieving higher predictive accuracy.
- Macroscopic Models for Initial Flow: While not perfect for defects, macroscopic models like MOLDEX3D and MOLDFLOW can still provide acceptable predictions of the overall mold filling and compression profile, making them useful for initial design phases.
Key Results
- Compression Profile: MOLDEX3D and MOLDFLOW provided reasonable predictions of the compression profile, though MOLDFLOW slightly underestimated forces. 3D TIMON significantly overestimated compression forces due to its no-slip boundary condition assumption. DBS showed good agreement with experimental press data.
- Fiber Volume Content (FVC): MOLDEX3D predicted a uniform FVC, failing to capture FMS. 3D TIMON predicted extreme FMS but with unrealistic FVC values. DBS showed a more realistic distribution of FVC and FMS, though it sometimes overestimated the effect of fiber jamming.
- Fiber Orientation: Macroscopic models predicted larger out-of-plane (normal) fiber orientation components than observed experimentally, indicating their limitations in confined regions. Direct simulation models (DFS and DBS) showed better agreement with the experimentally observed small normal component.
- Short Shots: DBS provided the most accurate prediction of the flow front progression, especially in capturing a specific defect related to fiber bundle accumulation.
Limitations and Future Work
- Material Characterization: The paper highlights that the rheological characterization of SMC, particularly separating viscous and friction contributions, is critical for accurate simulation. The current characterization might lead to higher predicted viscosity and thus longer processing times.
- Computational Cost: Directly modeled process simulations (DFS and DBS) are computationally more expensive and time-consuming than macroscopic models.
- FMS Prediction Accuracy: While DBS showed improvement, precisely predicting the severity and exact location of FMS remains a challenge for all investigated models. The paper notes that no simulation can precisely predict the severity of FMS in all cases.
- Future Work: The authors suggest incorporating constraints into tensor-based models or improving the modeling of fiber-fiber and fiber-mold interactions in direct simulation models. Further research into more accurate material characterization and advanced modeling of FMS is also recommended.
Summary updated