Computer Aided Engineering Simulation Research Articles

Fusion of Multiple Data Sources for Vehicle Crashworthiness Prediction Using CycleGAN and Temporal Convolutional Networks

Abstract Computer-aided engineering (CAE) models play a pivotal role in predicting crashworthiness of vehicle designs. While CAE models continue to advance in fidelity, an inherent discrepancy between CAE model predictions and the responses of physical tests remains inevitable. Machine learning (ML) models have shown promising potential in improving the prediction accuracy of CAE models. Nevertheless, the scarcity of vehicle crash data poses a significant challenge for the training of such models. This paper overcomes these challenges by fusing multiple data sources from two different types of vehicles. More specifically, the Cycle-Consistent Generative Adversarial Neural networks (CycleGAN) are first employed to translate features of time-series test data from one domain to another using cycle consistency loss. Such a translation allows for the generation of synthetic crash test data for the second vehicle type by leveraging existing tests from both vehicle types. In parallel, an initial temporal convolutional network (TCN) model is trained using CAE simulation data and physical test data of the first vehicle type. This pre-trained TCN model is then fine-tuned using three sources of data from the second vehicle type, namely the CAE data, test data, and the augmented virtual test data generated using CycleGAN. Through these data fusion, the crashworthiness prediction accuracy of the second vehicle type can be improved. The proposed method is demonstrated by a real-world case study with a small-size SUV and a medium-size SUV. Results show enhancements in the predictive performance of the medium-size SUV model.

A Review of Generative Models for 3D Vehicle Wheel Generation and Synthesis

Integrating deep learning methodologies is pivotal in shaping the continuous evolution of computer-aided design (CAD) and computer-aided engineering (CAE) systems. This review explores the integration of deep learning in CAD and CAE, particularly focusing on generative models for simulating 3D vehicle wheels. It highlights the challenges of traditional CAD/CAE, such as manual design and simulation limitations, and proposes deep learning, especially generative models, as a solution. The study aims to automate and enhance 3D vehicle wheel design, improve CAE simulations, predict mechanical characteristics, and optimize performance metrics. It employs deep learning architectures like variational autoencoders (VAEs), convolutional neural networks (CNNs), and generative adversarial networks (GANs) to learn from diverse 3D wheel designs and generate optimized solutions. The anticipated outcomes include more efficient design processes, improved simulation accuracy, and adaptable design solutions, facilitating the integration of deep learning models into existing CAD/CAE systems. This integration is expected to transform design and engineering practices by offering insights into the potential of these technologies.

A Deep Learning-Based CAE Approach for Simulating 3D Vehicle Wheels Under Real-World Conditions

The implementation of deep learning (DL) in computer-aided engineering (CAE) can significantly improve the accuracy and efficiency of simulating 3D vehicle wheels under real-world conditions. While traditional CAE methods can be time-consuming and computationally expensive, DL can reduce simulation time and development cycles across all industries. This work explores the role of DL and AI in virtual manufacturing and CAE and investigates how they can be used to improve the accuracy and efficiency of simulations for 3D vehicle wheels. Deep learning models can learn the complex relationships between different wheel design parameters, such as tire load distribution, stress distribution, and fatigue life. Once trained, these models can be embedded into CAE software, allowing for faster and more accurate simulations of wheel performance. This interdisciplinary study uses various deep learning techniques, including convolutional neural networks (CNNs), generative adversarial networks (GANs), and recurrent neural networks (RNNs), to create a more efficient and accurate relationship between CAD modeling and CAE simulation. The research aims to leverage the potential of deep learning models to automate 3D CAD design, accurately predict CAE results, and provide in-depth explanations and verifications. The benefits of this research are expected to extend to the automotive industry's pursuit of more robust and resilient wheel designs. By streamlining the product development process from conceptual design to engineering performance evaluation, this study has the potential to revolutionize the automotive industry's product development cycle.

Prediction of Short-Shot Defects in Injection Molding by Transfer Learning

For a long time, the traditional injection molding industry has faced challenges in improving production efficiency and product quality. With advancements in Computer-Aided Engineering (CAE) technology, many factors that could lead to product defects have been eliminated, reducing the costs associated with trial runs during the manufacturing process. However, despite the progress made in CAE simulation results, there still exists a slight deviation from actual conditions. Therefore, relying solely on CAE simulations cannot entirely prevent product defects, and businesses still need to implement real-time quality checks during the production process. In this study, we developed a Back Propagation Neural Network (BPNN) model to predict the occurrence of short-shots defects in the injection molding process using various process states as inputs. We developed a Back Propagation Neural Network (BPNN) model that takes injection molding process states as input to predict the occurrence of short-shot defects during the injection molding process. Additionally, we investigated the effectiveness of two different transfer learning methods. The first method involved training the neural network model using CAE simulation data for products with length–thickness ratios (LT) of 60 and then applying transfer learning with real process data. The second method trained the neural network model using real process data for products with LT60 and then applied transfer learning with real process data from products with LT100. From the results, we have inferred that transfer learning, as compared to conventional neural network training methods, can prevent overfitting with the same amount of training data. The short-shot prediction models trained using transfer learning achieved accuracies of 90.2% and 94.4% on the validation datasets of products with LT60 and LT100, respectively. Through integration with the injection molding machine, this enables production personnel to determine whether a product will experience a short-shot before the mold is opened, thereby increasing troubleshooting time.

Application of mold flow analysis to the study of plastic gear rack injection molding warpage

This study employed the Taguchi method, analysis of variance, and response surface methodology for plastic gear rack injection molding parameters followed by a factorial quality validation. This study was expected to reduce the time cost of mold design and injection molding by making different combinations of the molding parameters, designing an experimental method, and performing the data simulation experiment by computer-aided engineering (CAE). With the research tool of polymer (polyacetal) for plastic material, computer-aided design mold design, and CAE mold flow analysis software, a numerical analysis of plastic molding flow was conducted. Taguchi L16 (45) orthogonal array designed 16 experimental combinations including injection molding conditions of filling time, holding pressure, holding time, plastic temperature, and mold temperature. The experimental results of molding analysis of software (Moldex3D) determined the optimum molding essentials of plastic injection: filling time 0.2 s, holding pressure 98 MPa, plastic temperature 195 °C, and mold temperature 65 °C. In this study, the parameters of the response surface method were used for the actual injection verification. The CAE simulation software can greatly improve the mold design and injection molding parameter testing time to enhance the overall working efficiency and cost control.

The Evolution of Microstructure for Carburizing and Quenching 17CrNiMo6 Steel: Forecasting and Experimentation

This paper investigates the evolution of the microstructure of 17CrNiMo6 steel produced by carburizing and quenching through computer-aided engineering (CAE) and experimental study. The chemical composition, microstructure, and properties vary from surface to the core during the carburizing and quenching, which makes the CAE simulation of temperature field and microstructure evolution more complex. The performance–temperature and field performance–microstructure iterations using different simulation methods are applied. The results showed that the CAE forecast microstructure evolution is consistent with the experiment. The error between the predicted and experimental values from the surface to 2000 μm is 5%–9%, and the predicted results are consistent with the experiment at the depth of 2000 μm.

Characterization of Epoxy-Based Rapid Mold with Profiled Conformal Cooling Channel.

Based on the experience of the foundry industry, reducing the demolding time is the key for mass production of wax patterns with sophisticated geometries. Integration of numerical simulation and rapid tooling technology for decreasing the time to market is essential in advanced manufacturing technology. However, characterization of epoxy-based rapid molds with a profiled conformal cooling channel (PCCC) using computer-aided engineering simulation of the epoxy-based rapid mold with PCCC was not found in the literature. In this study, epoxy-based rapid molds with PCCC were characterized numerically and experimentally. The cooling performance of wax injection molds with two different kinds of cross-sections of the cooling channel was investigated. Four pairs of injection molds with PCCC were implemented using four different kinds of material formulations. It was found that the cooling performance of the PCCC was better than a circular conformal cooling channel (CCCC) since the PCCC maintained a more uniform and steady cooling performance of injection-molded product than CCCC. Epoxy resin added with 41 vol.% Cu powder seems to be a cost-effective empirical material formulation in terms of cooling time and material costs. This empirical material formulation provided an injection mold with low material cost and good cooling performance simultaneously compared to an injection mold fabricated with commercial material. The cooling performance could reach 88% of that of the injection mold fabricated with commercial material. The material cost of making the injection mold was only about 60% of that of the injection mold fabricated with commercial material. The coolant flow rate had no significant effect on the cooling time, whereas the cooling time of the wax pattern was affected by coolant temperature significantly.

Dynamic analysis of the sinusoidal actuation of a flexible fin for paired fin propulsion

In this research, the sinusoidal actuation of a flexible fin for median and/or paired fin propulsion is studied. A prototype of an underwater robot is built, and the computer-aided engineering simulation method is used for the transmission system of the flexible fin. The fin ray at the end of the trajectory and the crankshaft drive the parameters as a function of the fin ray output parameters. Given the fin and the fin surface motion for the coupled simulation analysis, the flexible fin surface and the fin sinusoidal oscillation are studied. The deformation of the fin surface is established outside the edge wave trajectory equation of motion. The equations of motion are verified by simulation. The single fin wave model is solved with the computational fluid dynamics method, and the hydrodynamic performance of the fin surface is obtained. Using the dynamic mesh technique, the complex large fluctuation motion of the fin surface is solved. With different amplitudes and periods of the fin wave as the input parameters, a simulation of the fin wave propulsion is conducted, and the relationship between the fin wave propulsion force and the fin wave resistance is calculated. The correctness of the simulation results of the sinusoidal fixed wave of the fin surface is verified by a fin wave ink experiment.

Layout Design and Die Casting Using CAE Simulation for Household Appliances

Due to the development and industrialization of science and technology, aluminum alloys have been developed in various fields. Recently, the government has been pursuing ways to decrease the weight and increase the recyclability of various components in order to conserve resources, energy, and the environmental. In keeping with this trend, cast iron products are being replaced by aluminum products in the foundry industry by using high-pressure die casting (HPDC). Casting layout design, relies on the experience and knowledge of mold designers in the casting industry, which proves insufficient to respond to the rapidly changing needs of the era and to increasing production costs. Designing and producing casting layouts using CAD/CAM/CAE technology has become a critical issue. Computer-Aided Engineering (CAE) technology is rapidly increasing with the development of computer software and hardware. CAE technology not only predicts defects in mass production but also performs filling or solidification analysis during the mold design stage before production, enabling optimal mold design methods. New technologies that combine the emerging casting processes of filling and solidification analysis using computer simulation with existing technology and practical experience in the field are rapidly increasing in the foundry industry. Based on empirical knowledge, the layout and design of casting products has traditionally progressed through trial and error. The solutions achieved through scientific calculation and analysis using CAE technology can save a great deal of money and time in the building of die-casting molds and in their design and fabrication. In this study, numerical analysis of household appliances (cooking grills) quickly and accurately predicts problems arising from the filling and solidification of the melted metal in the casting process, thereby ensuring the quality of the final cast product. These results can be used to quickly establish a sound casting layout with reduced production costs.

Process design and finite element analysis of multi-station two-roller cassette hot rolling for aluminum alloy 6061 wire

In this paper, a round-oval-round roller cassette wire rolling process of the aluminum alloy 6061 was studied for the process and the roller design evaluation. The elliptic groove of roller was designed for the stand of oval cross-sectional profiles. The billet diameter for this 10- stand rolling process simulation is 20mm. The diameters for the round cross-section profile stands are 16mm,12.8mm,10.2mm,8.2mm, and 6mm, respectively. The forming conditions for the CAE simulation is 400 °C for billet temperature, 0.45 for the constant shear friction factor, 20mm/s for the initial rolling speed. The CAE (Computer Aided Engineering, CAE) results had shown the maximum dimension deviation of the billet is less than 0.25 mm at the location of the roller gap and almost perfect match at the other locations. The maximum principal stress distribution on the cross section had shown a two third area was deformed under the rolling process. The CAE results had demonstrated the proposed process and die design was able to make the smaller wire with good geometrical accuracy without breaking the wire during the rolling process because of the more uniform stress distribution at the cross section of wire.

CAE Analysis and Auralization of Driveline Clunk NVH

<div class="section abstract"><div class="htmlview paragraph">Noise and vibration evaluation of driveline clunk can be challenging as it is the result of driver input conditions and is transient in nature. As with many noise and vibration challenges, the use of computer-aided engineering (CAE) simulation is useful as it allows for detailed study of the phenomenon and prediction of potential improvements. A hybrid approach of physical test-based measurements and CAE analysis can be used to leverage the advantages of CAE in a comprehensive evaluation including the total vehicle noise, vibration, and harshness (NVH) performance. In this paper we present work performed to facilitate engineering evaluations of driveline clunk using both measured test and CAE simulation data. We explain how we used measured test data to inform the CAE analysis, how the simulation approached modelling of the transient clunk event, and how the measured data was used to provide contextual sound for realistic evaluation of the CAE output as heard by the passengers. The ability to listen to the simulated clunk in the context of the remainder of the vehicle NVH provides a direct link to evaluation of correlation between CAE and test results, as well as judgement of subjective performance and gap-to-target levels.</div></div>

Parameter values for lot size and quality level via CAE simulation, statistical method, and mathematical programming

Because of the stochastic nature of production systems, it is necessary to first build an uncertainty model for subsequent real applications. Moreover, process parameter planning, quality design, and production inventory management are interdependent elements. In this research, a computer simulation model via computer-aided engineering (CAE) was developed to determine the optimal process parameters, lot size, and back order intervals for an integrated process design and inventory management system with simultaneous quality and cost considerations. Based on the estimated process time and costs obtained using CAE, the derived production rate and unit cost were then used for production inventory applications. In consideration of the uncertainty factor, the response surface method (RSM) was employed to analyze the output, namely the total costs incurred in employing the proposed approach, as well as the inputs, which include the cutting parameters, production quantity, and back order intervals. After the RSM was used to obtain the response functions, which represent the output of the collective interests, the mathematical programming (MP) was formulated based on the response functions to determine the optimal process parameters, process quality levels, production order quantities, and back order intervals. The total cost per set time unit was minimized by determining the required quality level, process parameter values, Economic Production Quantity (EPQ), and back order intervals. A cutting example was chosen to demonstrate the proposed approach. Two cases were used for comparison: the Integrated Case (the proposed approach herein) and the Disintegrated Case.

A machine learning framework for accelerating the design process using CAE simulations: An application to finite element analysis in structural crashworthiness

This paper presents a novel framework for predicting computer-aided engineering (CAE) simulation results using machine learning (ML). The framework is applied to finite element (FE) simulations of dynamic axial crushing of rectangular crush tubes that are typically used in vehicle crashworthiness applications. A virtual design of experiments that varies the size and wall thickness of the FE model is performed to generate the necessary training data. This process generates designs with varying numbers of nodes and elements that are handled by the ML system. However, the explicit design parameters and meshing techniques that were used to generate the training data remain unknown to the ML system. Instead, 3D convolutional neural networks (CNN) autoencoders are used to process the initial FE model data (i.e., nodes, elements, thickness, etc.) to automatically determine these features in an unsupervised manner. A voxelization strategy that operates on the mass of individual nodes is proposed to handle the unstructured nature of the nodes and elements while capturing variations in the wall thickness of the FE models. The flattened latent space generated by the 3D-CNN-autoencoder is then used as input into long-short term memory neural networks (LSTM-NN) to predict the force–displacement response as well as the deformation of the mesh. The training process of both the 3D-CNN-autoencoders and LSTM-NN is systematically studied to highlight the robustness of the framework. The proposed ML system utilizes only 16% of the simulations generated in the virtual design of experiments to achieve good predictive capability. Once trained, the proposed framework can predict the deformation of the mesh and resulting force–displacement response of a new design up to ∼330 and ∼2,960,000 times faster, respectively, than the conventional FE approach with good accuracy. This computational speed up offers design engineers and scientists a potential tool for accelerating the design exploration process with CAE simulation tools, such as FE analysis.

Optimization of Injection Molding Process of Transparent Complex Multi-Cavity Parts Based on Kriging Model and Various Optimization Techniques

The injection molding process of a mold with multiple cavities can reduce the mold space, but the process is complicated and the product is prone to warpage. In order to reduce the transparent parts warpage of such a mold, by optimizing injection molding process parameters, an optimization technique combining orthogonal experiment, Kriging model and an optimization algorithm is proposed. First, the orthogonal experiment is designed, based on Computer Aided Engineering (CAE) simulation results, while the significant factors affecting the warpage are selected. On this basis, the Kriging model is established to map the nonlinear function relationship between the significant factors and the warpage. Specifically in this paper, four types of correlation functions, affecting the model precision accuracy, are used to establish the Kriging model and the high-precision accuracy model is selected. A numerical optimization algorithm, direct search approach and global exploration method are used to optimize the model, so as to obtain the optimal injection molding parameters producing the minimum warpage. Finally, CAE simulation and experimental validation prove the effectiveness and reliability of the proposed optimization method.

The study on a new method of preparing PMMA forming composite bipolar plate

The recent oil resource shortage has prompted the development of the proton exchange membrane fuel cell (PEMFC) system. PEMFC is a possible source of power that can be used in aircraft, household electricity, agriculture, fishing, motor vehicles, ships, submarines, bicycles, and other portable power systems in the future. This paper emphasizes the production of lightweight bipolar plates to solve several existing problems in the PEMFC system, including weight, cost, and integration. Conventional bipolar plates account for approximately 90% of the weight of battery packs. Therefore, an injection molded flow-field plate constructed from polymethylmethacrylate (PMMA) is developed herein to reduce the weight of the PEMFC system. Computer-aided engineering (CAE) mold flow analysis is then used to simulate the experimental design based on the finished products. Experimental analysis is also performed on the adhesion results of the plates. The results indicate that the establishment of the injection mold using CAE simulation improves mold development and reduces cost. Mechanical coarsening on the surface of the PMMA results in improved adhesion (> 50 N) at temperatures higher than 80 °C. Thus, mechanical coarsening is suitable for the PEMFC system. The problem of conventional weight is solved by reducing the weight by 70%.

Building a Scalable and Interpretable Bayesian Deep Learning Framework for Quality Control of Free Form Surfaces

Deep learning has demonstrated high accuracy for 3D object shape error modeling necessary to estimate dimensional and geometric quality defects in multi-station assembly systems (MAS). Increasingly, deep learning-driven Root Cause Analysis (RCA) is used for decision-making when planning corrective action of quality defects. However, given the current absence of scalability enabling models, training deep learning models for each individual MAS is exceedingly time-consuming as it requires large amounts of labelled data and multiple computational cycles. Additionally, understanding and interpreting how deep learning produces final predictions while quantifying various uncertainties also remains a fundamental challenge. In an effort to address these gaps, a novel closed-loop in-process (CLIP) diagnostic framework underpinned algorithm portfolio is proposed which simultaneously enhances scalability and interpretability of the current Bayesian deep learning approach, Object Shape Error Response (OSER), to isolate root cause(s) of quality defects in MAS. The OSER-MAS leverages a Bayesian 3D U-Net architecture integrated with Computer-Aided Engineering simulations to estimate root causes. The CLIP diagnostic framework shortens OSER-MAS model training time by developing: (i) closed-loop training to enable faster convergence for a single MAS by leveraging uncertainty estimates of the Bayesian 3D U-net model; and, (ii) transfer/continual learning-based scalability model to transmit meta-knowledge from the trained model to a new MAS resulting in convergence using comparatively less training samples. Additionally, CLIP increases the transparency for quality-related root cause predictions by developing interpretability model which is based on 3D Gradient-based Class Activation Maps (3D Grad-CAMs) and entails: (a) linking elements of MAS model with functional elements of the U-Net architecture; and, (b) relating features extracted by the architecture with elements of the MAS model and further with the object shape error patterns for root cause(s) that occur in MAS. Benchmarking studies are conducted using six automotive-MAS with varying complexities. Results highlight a reduction in training samples of up to 56% with a loss in performance of up to 2.1%.

Object Shape Error Response Using Bayesian 3-D Convolutional Neural Networks for Assembly Systems With Compliant Parts

This article proposes a novel object shape error response (OSER) approach to estimate the dimensional and geometric variation of assembled products and then, relate these to process parameters, which can be interpreted as root causes (RC) of the object shape defects. The OSER approach leverages Bayesian 3-D convolutional neural networks integrated with computer-aided engineering simulations for RC isolation. Compared with the existing methods, the proposed approach: 1) addresses a novel problem of applying deep learning for object shape error identification instead of object detection; 2) overcomes fundamental performance limitations of current linear approaches for RC analysis (RCA) of assembly systems that cannot be used on point cloud data; and 3) provides capabilities for unsolved challenges such as ill-conditioning, fault-multiplicity, RC prediction with uncertainty quantification, and learning at design phase when no measurement data are available. Comprehensive benchmarking with existing machine learning models demonstrates superior performance with R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.98 and MAE = 0.05 mm, thus improving RCA capabilities by 29%.

Virtual Sensor Development for Continuous Microfluidic Processes

In continuous microfluidic processes, such as factory-on-chips, most valuable variables to be sensed for implementing an optimal control are usually not accessible. This is, for instance, the case of the shape and position tracking of the flow interfaces in a flow-focusing microdevice. In this context, virtual sensors are promising tools to improve the current control systems aiming at a zero-defect strategy. This article presents a methodology for building an accurate virtual sensor, based on computer-aided engineering simulations, both analytical and numerical, and model order reduction techniques. The methodology is applied to a given flow-focusing microchip. The outcome is a real-time model (virtual sensor) able to predict the shape and location of the multiphase fluid interfaces from the volumetric flow rate measured in the system. Results are successfully validated against the experimental data. The main challenge of this approach is to minimize uncertainties associated with the microfluidics setup.

Prediction of Polymer Flow Length by Coupling Finite Element Simulation with Artificial Neural Network

In this study, computer-aided engineering (CAE) simulation software and the design of experiments (DOE) method were used to simulate the injection molding process in terms of the melt flow length, using a spiral part. Process parameters such as melt temperature, mold temperature, injection pressure and mold cavity thickness were considered as injection molding variables. A predictive model for the flow length was created using a three-layer artificial neural network (ANN). The ANN model was trained with both simulation and experimental data, and the predictive performances were compared in terms of correlation coefficient, root mean square error and mean relative error. The cavity thickness and melt temperature were found to be the most significant factors for both the simulation and the experiment, while the injection pressure and the mold temperature had little effect on the flow length. The ANN model trained with Moldex3D data shows a significantly higher prediction capacity than the ANN model trained with experimental data. However, the melt flow lengths predicted by the ANN model for both Moldex3D and Moldflow simulation data are statistically significant, indicating that the proposed prediction methodology, which combines the ANN model, DOE method and the CAE simulation technology, can effectively predict the flow length of injection molded parts, with a small number of data.

DESIGN, CHARACTERIZATION AND CAE SIMULATION FOR DEVICES APPLIED IN THE CONTROL AND SEMOVIENT IDENTIFICATION

The current research project presents a methodology product design, characterization for different thermoplastic elastomeric mixtures with mechanicals properties evaluation and computer aided engineering simulations (CAE) for ear tags devices applied in the control and semovient identification. From product design and the characterization of reference product through Fourier Transformed Infrared Spectroscopy (FTIR) is achieved new alternative mixtures to the reference, after the exposition to UV radiation for 320 hours, are evaluated through different mechanicals and chemical properties as hardness and FTIR. With the characteristics and 3 geometry designs, a design was selected regarding to the customer requirements from the Product Design Methodology - PDS [1]-[3]. This design and mixtures evaluation obtained were the initial conditions for the analysis CAE in Software Sigmasoft® specialized in mold injection simulation for thermoplastic elastomeric materials. The simulations reveal important results in temperature, packing pressure and other variables to know in the adequate mixture processing with the best performance properties, with the goal to determine de best processing conditions, and different requirements to know in the mold manufacturing achieving an optimum final product processing.