Agenda

Additively Manufactured & AI Optimized Millimetre-Wave Waveguide Filter Design for Space Applications - The Journey from Design to Manufacturing

Presented By
Laila Salman (Ansys Canada Ltd.)

Chaired By
Chair Satheesh Kandasamy

Authored By
Laila Salman (Ansys Canada Ltd.)
Sunil Acharya (Ansys Inc)
Koen Huybrechts
Loren Vancleef (3D Systems)
Diamond Liu (SynMatrix Technologies Inc.)
Gada Saad (Scientific Microwave Corp.)

High power light weight RF passive components are critical for satellite communications payloads and are required to handle increasingly larger peak/average powers over the frequency range in the transmit path of the communication link. Powder-bed fusion metal Additive Manufacturing (AM) process has matured as a breakthrough technology for the development of arbitrary shape RF and microwave components where conventional machining can be complex or impossible to use. Thus, allows the exploration of unique and optimized parts and assembly consolidation. When it comes to space applications, industries rely on testing to mitigate system failure. Reproducing space conditions for validation in laboratory setting is costly, time-consuming, and ultimately unreliable. Hence, leveraging the use of Artificial Intelligence (AI) optimization techniques in the design process of RF components is impactful. With next generation of satellites, built from fifth generation (5G) architecture, communication signals will beam down from space supporting terrestrial 5G infrastructure on Earth. The presented work demonstrates the use of physics-based simulation technologies from modelling/synthesize to manufacturing process of an optimized millimetre-wave band pass filter that can be safely launched to space for satellite communication. The proposed design manufacturing workflow can be generalized to a wide variety of radio frequency applications such as phased array antennas with complex feeding networks or multi-beam active antenna array with polarizers, couplers, and power dividers. Advanced AI optimization techniques are applied to optimize the proposed 28GHz waveguide band pass filter and maintain standard design specifications. The process simulation methodologies for the selective laser sintering allows for reducing waste and build time while maintaining adequate support structure during the manufacturing process of assembly. Internal mechanical support requirements can generate additional postprocessing challenges. This is where CAE based support generation tools can be well utilized to find the optimum build time while minimizing the distortion tendencies and the support volume. Once the optimum supports structural topology and build orientations are obtained, a coupled thermal-structural FEA simulation can be used to simulate the AM build process and estimate the residual stresses in the build for a specific build orientation and support strategy. This is important in satellite communications technology where any reduction in payload weight is considered advantageous. Measurements were obtained for the additively manufactured parts with different conducting material properties and compared to simulated results.

Acceleration of Stamping Simulation using an Advanced Artificial Intelligence Framework

Authored & Presented By
Dylan Budnick (Xccelerate AI)

Chaired By
Chair Satheesh Kandasamy

Complex simulations in engineering traditionally demands extensive computational power and time, sometimes taking days to weeks to complete, having a negative impact on project schedules and budgets. Xccelerate AI has developed an AI-based framework to swiftly predict simulation results, dramatically speeding up complex engineering simulations. The framework operates on a straightforward concept: it learns from existing simulation data to predict results for new designs. Many organizations already possess the requisite simulation data for model training, enabling fast implementation of a bespoke surrogate model tailored to an application. We will present an application of this framework related to automotive sheet metal stamping. In modern automotive engineering, vehicles are primarily designed in the virtual space to enable a rapid vehicle design process. This can lead to rapid improvements in vehicle design as more design cycles may be performed. However, this process is heavily constrained by the time and computational requirements necessary to generate the vast number of simulations conducted for vehicle design. Fortunately, modern artificial intelligence (AI) techniques may be used to drastically accelerate the generation of new simulation results. To demonstrate the approach, a cross shaped punch/die geometry is created including the punch, binder, die, and blank. A non-linear explicit dynamic finite element package is used to generate 2500 simulations with 13 parameterized input variables. These serve as the data set to train our machine learning model. In a commercial setting, existing prior simulation datasets would instead be used. This work presents for the first time an advanced recurrent deep learning framework capable of predicting the entire simulated stamping response for new geometries, materials, and process parameters. The predicted response is compared to simulated finite element results demonstrating a prediction time reduction of 740,000 times while maintaining high accuracy in key performance indicators of the stamping process. The AI-based framework thereby offers a significant opportunity to save development time and cost, for example, by enabling a more rapid evaluation of different conceptual designs. Finally, this work also highlights how this framework may be leveraged to further improve metal forming processes.

AI-Surrogates and Concept Engineering: Navigating Challenges and Solutions in Accelerated Product Development

Presented By
Ashish Aggarwal (DS Americas Corp.)

Chaired By
Chair Satheesh Kandasamy

Authored By
Ashish Aggarwal (DS Americas Corp.)
Anna Liang (Dassault Systemes)

Accelerated product development cycles present both opportunities and challenges in today's dynamic business landscape. This technical presentation explores the intricate challenges associated with compressing product development timelines and emphasizes the pivotal role of concept engineering and surrogate modeling in addressing these issues. In the fast-paced world of product development, businesses are under constant pressure to bring innovative solutions to market swiftly. However, this urgency often leads to challenges such as increased risk of design flaws, compromised quality, and higher costs. The concept engineering phase emerges as a critical component in mitigating these challenges. By focusing on ideation, feasibility analysis, and rapid prototyping, concept engineering lays the foundation for a robust and well-informed product development process. One of the primary challenges in accelerated product development is the need for efficient decision-making. The concept engineering phase acts as a decision-support mechanism by enabling teams to explore multiple design alternatives rapidly. This iterative process allows for early identification of potential pitfalls, ensuring that only the most viable concepts proceed to the subsequent stages. Another critical challenge in accelerated product development is the increasing complexity of designs. As products become more intricate, traditional methods struggle to keep pace with the demands for precision and efficiency. Concept engineering, coupled with surrogate modeling, provides a systematic approach to manage this complexity. Through the exploration of simplified prototypes and efficient modeling techniques, teams can navigate intricate design spaces with greater ease. Physics-based simulations play a crucial role in science-based design processes by providing high-accuracy simulation results. However, the high execution times often limit the design exploration capabilities. To retain comprehensive 3D simulation results while reducing executing time, we present an approach utilizing neural networks for 3D interactive design exploration. Fast executing neural networks are trained as 3D surrogates using parametric design data generated from physics simulations, and then deployed in a collaborative design environment for design exploration, allowing designers to evaluate a high number of parameters within the design space. By delivering high-fidelity results that covers both transient physical responses and 3D fields, this proposed method can accelerate the design iteration process, and enhance the efficiency and effectiveness of science-based product design approaches. In conclusion, this technical presentation delves into the challenges associated with accelerated product development cycles and underscores the significance of concept engineering and surrogate modeling in addressing these challenges. By embracing a systematic and collaborative approach in the concept-engineering phase, coupled with the efficiency of AI-empowered surrogate modeling, businesses can navigate the complexities of accelerated timelines while ensuring the delivery of high quality, innovative products to the market.

A Seamless Bridge: Integrating Cameo Systems Modeler into PLM

Authored & Presented By
Brian Schouten (Prostep Inc.)

Chaired By
Chair Max Monn

Complex systems design demand innovate solutions. Harmonizing Cameo Systems Modeler with your PLM is no easy task, but when effectively done can unlock a wealth of benefits. Fusing these two systems lays the foundation of data interoperbility, fostering a more collaborative and agile environment. Additionally, benefits of enhanced traceability, reduced redundancy, and improved descision making result. This session will explore three core strategies for integrating Cameo with PLM: cloud PLM integration, Cameo client plugin, and Cameo datahub OSLC integration. We'll examine the unique advantages of each approach and discuss which use cases are best suited for each method. Key Takeaways: 1. An understanding of the 3 strategies for integrating Cameo into a PLM platform 2. An introduction to the flexibility and compresensive tool, OpenPDM which facilitates each strategy for integration 3. Insights into best practice and methodologies for integrating Cameo into your PLM

Aligning Data Flow Between MBSE and PLM Systems at a Leading American Aerospace Company

Authored & Presented By
Brian Schouten (Prostep Inc.)

Chaired By
Chair Max Monn

Although integrating MBSE tools with PLM and other systems often pose a significant challenge, the benefits of doing so make it worthwhile. Experts at PROSTEP embark on a solution for a leading American aerospace manufacturer enabling automated, bi-directional data flow between MBSE and PLM platforms with PROSTEP's integration platform, OpenPDM. This presentation highlights the integration of 3DEXPERIENCE and SIEMENS Teamcenter with synchronization of BOM, simulation data and parameters, and documents. Additionally, the solution will provide traceability of up to 10 object types across tools use OSLC links and showcase event triggers to push and pull data from Teamcenter to 3DEXPERIENCE. Key Takeaways: 1. Discover the tool for automated synchronization between 3DEXPERIENCE and Siemens Teamcenter. 2. See how OSLC links can provide traceability between object types 3. Explore the strategies, methodologies, and best practices employed to successfully execute MBSE systems integration"

A Methodology for Configurable System Simulation Models: Integration of SysML, SSP, and FMI

Presented By
Juho Könnö (University of Oulu)

Chaired By
Chair Max Monn

Authored By
Juho Kö
nnö
(University of Oulu)
Isa Banagar
Aapo Hö
lsä
(University of Oulu)
Tero Frondelius (Wä
rtsilä
Finland Oy)

This paper introduces a comprehensive methodology for the rapid generation of configurable system simulation models by seamlessly integrating the Systems Modeling Language (SysML), the System Structure and Parameterization (SSP) standard, and the Functional Mock-up Interface (FMI). Our approach leverages SysML as the foundational framework for system architecture, employing carefully defined diagrams that facilitate the automated creation of simulation models. We target the powertrain field as this involves the integration of several disciplines, such as electrical, mechanical, and thermal domain. Hence, the need for an easy and consistent way of defining system simulation models across domains is important for virtual validation of the whole powertrain. The proposed methodology initiates the simulation model creation process using specific SysML diagrams chosen for this purpose. These diagrams capture the structural and behavioral aspects of the system, providing a clear and visual representation of the underlying architecture. To streamline the generation process, we employ the SSP standard, translating SysML constructs into simulation-ready models. The SSP standard serves as a bridge between the SysML system architecture and the simulation environment, offering a standardized framework system simulation templates. The SysML model defines the boundaries between sub-systems of the system. Based on those boundaries, required simulations and their interfaces are defined in the SysML model. After definition of required interfaces and the scope of each simulation, they are prepared using functional mock-up units (FMUs). This unique integration ensures a seamless transition from SysML-based system architecture to a simulation-ready format. Furthermore, a direct link to the underlying system architecture allows linking of product requirements and the simulation validating them. Another key advantage of our methodology lies in its configurability. By populating the FMUs with suitable models, we can tailor the system simulation to different product configurations or desired levels of simulation accuracy. This adaptability enhances the versatility and applicability of the generated simulation models, facilitating rapid prototyping, design exploration, and performance analysis. Through a practical application of the technology in a vehicle setting, we demonstrate the efficacy of our methodology in expediting the creation of configurable system simulation models. The seamless integration of SysML, SSP, and FMI empowers engineers to efficiently navigate from system architecture to simulation-ready models.

The Development of a Virtual Prototyping Review Process Using a Collaborative Immersive Approach to Enhance Processes, Awareness, and Delivery

Authored & Presented By
Alanzo Granville (SAIC)

Chaired By
Chair Todd Depauw

Immersive technology has stimulated and incentivized innovation throughout multiple industries as the technology continues to improve and open new doors to discovery. Unfortunately, the attraction of immersive development including augmented, virtual, and mixed realities gain attention, but if they are not effectively used or understood they may generate a lackluster experience within projects and leave the users or leadership thinking is this really useful. Possibly even deterring the future use or consideration of its use on projects. This leads to the big question of how do we properly use an immersive technology approach to promote progression and generate positive benefits and outcomes for our daily work. The truth is that the use of immersive technology has shown benefits, but its use and results can vary significantly among different applications. This document will serve as a guide to developing a virtual prototype review process using immersive technology and digital engineering. Providing a way to facilitate an initial connection between work and immersive technology use and enhancement capabilities. This will demonstrate a proven process utilized in previous and on current projects that has assisted in outlining how to establish a meaningful use case for immersive technology in the construction of facilities, vehicles, and other hardware to support beneficial outcomes that can boost productivity and performance through purpose and goal identification, collaboration, awareness of users capabilities and resources, and developing a virtual prototype review process that supports quick iterations and turnaround time. Both specific and generalized examples and guidance will be used to support the integration of an immersive virtual prototype review process into various applications and projects to help understand the capabilities of using immersive technology as an enhancement tool. The process will help to identify the requirements needed to solidify innovation and future use cases supporting exploration within a virtual medium for concepts but not limited to digital twin, environment design and creation, manufacturing, and more to support critical decision-making efforts utilizing a custom fast feedback loop for many projects.

Managing Key Parameters from Simulation Inputs and Results

Presented By
Leonel Garategaray (INENSIA LLC)

Chaired By
Chair Todd Depauw

Authored By
Leonel Garategaray (INENSIA LLC)
Leonel Garategaray (Inensia)

In the realm of modern engineering practices, the effective management of parameters within the Simulation Process and Data Management (SPDM) framework stands as a critical challenge. Today, organizations are faced with different and often disconnected parameter management approaches, hindering the flow of information in the simulation process. In this context, this paper delineates a strategic initiative to redefine parameter management within the SPDM paradigm. Presenting a comprehensive methodology, we aim to close existing gaps and improve parameter management in simulation activities. We believe this is essential to achieve a more agile, efficient, and integrated approach to recurring engineering and optimization techniques.. An integral aspect is how SPDM facilitates parameter management within the context of Model-Based Systems Engineering (MBSE), and its architectural model, where variables are commonly parameterized. Integrating these parameterized variables into SPDM introduces flexibility and adaptability, allowing for adjustments to analyze their holistic impact on the system, such as modifying attributes like size, weight, or performance. This capability enables the evaluation of diverse system scenarios or configurations without the need to overhaul the entire model. By adjusting these parameters, the specific effects of changes on system performance or functionalities come to light, empowering engineers with informed decision-making throughout the project's lifecycle. When we talk about parameters, we refer to input parameters, necessary to perform any simulation, and output parameters, obtained from simulation results which could be compared with product requirements. These parameters could be extracted from file content, consolidating them in a singular repository for traceability and offering a holistic view to assess design efficacy. In a typical Development Lifecycle, design process begins with an envelope and requirements, and then models that fulfill the needs. Then the simulation tth CAD designs, accompanied by design parameters, boundary conditions, and materials properties, all interpreted as Input parameters. Afterwards, the simulation process, which consists of interlinked simulation activities with various simulation branches, yield results or reports containing key information like performance values, all these are considered as Output parameters. The challenge lies in this flow, since the information is usually disconnected, due to its nature of being documentation-based, which prevents a more pragmatic use to interpret the quality of the design. The objective of this paper will be to: Demonstrate how an SPDM platform streamlines parameter extraction and programmatic management from various files. Emphasize the dual centralization of both simulation data and parameters, ensuring traceability and change control. Illustrate how this approach accelerates analysts' understanding of designs and results, fostering a more efficient workflow. Exemplify, through a practical use case, the creation of parameter repositories from an organized files repository. Highlight how this initiative ensures a Single Source of Truth for simulation data, promoting clarity and reliability. We conclude that harmonizing simulation inputs and outputs through streamlined parameter management in the SPDM platform, enhance the efficiency and reliability of decision-making.

Start Fast, Start Smart: How to Build a Solid Foundation for your SPDM Journey

Presented By
Leonel Garategaray (INENSIA LLC)

Chaired By
Chair Todd Depauw

Authored By
Leonel Garategaray (INENSIA LLC)
Thomas Lehnhaeuser (ANSYS)

In today's dynamic and competitive landscape, the need for efficient product development is more crucial than ever. Having an efficient simulation process that support lean product development, is paramount in reducing time to market of new products. That's where Simulation Process and Data Management (SPDM) solutions come in, playing a vital role in achieving reusability, agility, accuracy, and efficiency in the development cycle. As we shared in previous NAFEMS presentations, the implementation of an SPDM solution is a multi-year journey, where one of the most challenging phase is commonly the first one: What is the added value of SPDM within the organization, and how to reach the objective? To simplify and expedite the roadmap, Ansys and Inensia partnered to create a combination of tools, methodology and training, that accelerates the process. The approach we will present facilitates the initial steps, empowers key user groups, and lays the groundwork for an effective SPDM implementation plan, while also delivering a working SDM solution within two months. This comprehensive approach combines training, methodology, and workshops to expedite the deployment of a Pilot System of Record (SoR) and create a personalized SPDM Phased Roadmap within a condensed timeframe of just 8 weeks. A key advantage of this strategy lies in the rapid availability of the system, enabling users in addition to grasping theoretical concepts, tangibly experiencing its benefits.. Through this hands-on engagement, participants were empowered to contribute actively in crafting a tailored implementation roadmap, aligning with their unique organizational needs and goals. In our presentation, we'll dive into the method we developed and showcase examples from clients where we've successfully implemented it. Based on our experience, we'll cover: - Building Effective Teams: How to create productive and efficient work teams and maximizing their productivity within the stipulated timeline. - Simplified Input Gathering: Templates designed to simplify and expedite the gathering of inputs from users, ensuring a smoother transition into the SPDM ecosystem. - Project Monitoring for Speedy Execution: The importance of keeping a close eye on the project for a swift and successful implementation of the SPDM initiative. - Effective Reporting: Developing user-friendly dashboards to guide and motivate users in workshops and help them express the company's needs that could benefit from SPDM implementation. In addition to explaining the theory; we'll also share real-world experiences from our clients, enriching the audience with actionable insights gained through practical application In conclusion, the approach of streamlining the establishment of a Pilot System of Record (SoR) and developing customized SPDM plans within an 8-week timeframe reflects a commitment to industry best practices. This methodology is designed to facilitate a quick and practical learning experiences for users, empowering them to actively define a roadmap that aligns with their unique organizational objectives. Through sharing these insights, we aim to contribute valuable strategies that organizations can consider in their pursuit of an effective SPDM journey.

Multiphysics time-dependent electro-mechanical simulations for dielectric elastomers using Abaqus User Elements (UEL) & FEniCSx

Presented By
Kamalendu Ghosh (KLA)

Chaired By
Chair Tim Morris

Authored By
Kamalendu Ghosh (KLA)
Bhavesh Shrimali (Kimberly Clark)

There has been increasing interest in the constitutive modeling of dielectric elastomers due to their potential in enabling new technologies such as soft robotics, actuators, and haptic devices. Under realistic time dependent loadings, dielectric elastomers are inherently dissipative. They dissipate energy both through viscous deformation and through friction in their electric polarization process. However, majority of constitutive models and the Finite Element implementation for such models considers only mechanical dissipation. The motivation for this bias is that the mechanical relaxation time of dielectric elastomers is much larger that their electric relaxation time. Yet, Accounting for electric dissipation, in addition to mechanical dissipation is crucial when dealing with applied alternating electric fields. A fully coupled 3-D constitutive model for isotropic and incompressible dielectric elastomers was proposed by Ghosh and Lopez-Pamies [1]. In the first part of this talk, the governing initial-boundary-value-problem and the constitutive model will be introduced. Following this, a detailed step-by-step procedure of a FEniCSx and Abaqus UEL implementation of the mixed-FE formulation of the governing equations will be spelled out. The scheme makes use of a conforming CrouzeixœRaviart FE discretization of space and a high-order accurate explicit RungeœKutta discretization of time. The combination of these two types of discretizations results into a robust scheme that is capable of handling finite deformations, the incompressibility constraint of the rubber and general time-dependent loading conditions. In the last part, the Abaqus UEL would be deployed to fully describe the electromechanical behavior of VHB 4910 (from 3M) under a complex time-dependent electromechanical load. References 1. Ghosh, K., Lopez-Pamies, O. 2021. On the two-potential constitutive modeling of dielectric elastomers. Meccanica 56, 1505œ1521.

Enhanced Implicit-Iterative Coupling Algorithms for Strong-Coupled Multiphysics

Authored & Presented By
Sethu Subbarayalu (Dassault Systèmes SIMULIA Corp)

Chaired By
Chair Tim Morris

Note: This work was presented in European NAFEMS Multiphysics conference held in Munich in November 2023. I would like the work to reach the audience in Americas. Abstract: In a co-simulation [FEL01], the engineering problem is decomposed into two or more subdomains, with each subdomain solved individually while continually exchanging data on their boundaries. A common use case is fluid-structure interaction, in which computational structural and fluid mechanics solvers are used over respective regions. This partitioned treatment has many merits but also poses stability and accuracy challenges when the physics coupling is strong; for example, when the fluid and structure have similar densities [CAU04], such as in the case of blood flow in an artery or a compliant structure subjected to forces imposed by a heavy and incompressible fluid. In recent years, numerical techniques, called accelerator methods, have been developed [DEG10, UEK16]. These accelerators enhance the coupling by enlarging the convergence radius from a stability point of view; in other words, the accelerators allow us to solve strongly-coupled physics problems that we could not solve before. Furthermore, the accelerators can significantly reduce the computational cost of these simulations. The SIMULIA Co-Simulation Services is an open architecture for coupling two or more solvers for multiphysics and multiscale simulations. Accelerator methods, implemented as Co-Simulation Services, are available for partners and customers. We present convergence studies of several fluid-structure benchmarks [TUR06, FER05], and industrial use cases and guide on using these numerical methods. References: [CAU04] CAUSIN P., GERBEAU J.-F., NOBILE F., Added-mass effect in the design of partitioned algorithms for fluid-structure problems. Computer Methods in Applied Mechanics and Engineering, 2005; 194(42-44):4506-4527. [DEG10] DEGROOTE J., HAELTERMAN R., ANNEREL S., BRUGGEMAN P., VIERENDEELS J., Performance of partitioned procedures in fluid-structure interaction. Computers & Structures, 2010; 88(7-8):446-457. [FEL01] FELIPPA C. A., PARK K.C., FARHAT C., Partitioned analysis of coupled mechanical systems. Computer Methods in Applied Mechanics and Engineering, 2001; 190(24-25):3247-3270. [FER05] FERNANDEZ M. A., MOUBACHIR M., A Newton method using exact Jacobians for solving fluid-structure coupling. Computers & Structures, 2005; 83(2):127-142. [TUR06] TUREK S., HRON J., Proposal for numerical benchmarking of fluid-structure interaction between an elastic object and laminar incompressible flow. In BUNGARTZ H.-J., MEHL M., and SCHAFER M., editors, Fluid Structure Interaction II: Modelling, Simulation, Optimization. Springer, 2006. [UEK16] UEKERMANN B. W., Partitioned Fluid-Structure Interaction on Massively Parallel Systems. PhD thesis, Institut für Informatik, Technische Universität München, 2016.

Integrated Thermal Fluid Simulation Workflows for Electronic Cooling Applications

Authored & Presented By
Devadatta Mukutmoni (Dassault Systèmes SIMULIA Corp)

Chaired By
Chair Tim Morris

A Guide to Digitally Transform Product Innovation with High-Performance Computing

Authored & Presented By
Rod Mach (TotalCAE)

Chaired By
Chair Alana Duma

Predictive Modeling to Enable Innovation in Textile Bonding Adhesives

Authored & Presented By
Liangkai Ma (The Dow Chemical Company)

Chaired By
Chair Steve Arnold

Polymeric adhesives are critical to the textile industry and essential to our daily life. They are gaining in popularity in textile bonding for many applications such as flexible packaging, safe and sustainable mobility, protective garments, sports and recreation equipment. They are advantageous because of optimal performance, ability to achieve higher level of automation, higher throughput, and lower total cost of ownership, when compared to other textile joining alternatives such as traditional sewing. The requirements of adhesives for adhesive textile bonding vary depending on targeted applications. Predictive modeling can be used to gain insight and accelerate new adhesive material development to deliver desired performance for specific applications. In this presentation, a systematic approach is presented to develop and validate material models to ensure prediction accuracy of predictive modeling of real-world applications of textile bonding adhesives. The characterization testing and constitutive modeling of both textile substrate and polymeric adhesive materials are first discussed. Three finite element analysis (FEA) models are then shared in the order of increased complexity: The first model is a single element uniaxial tensile test model developed to validate the adhesive hyperelastic material model and a user subroutine developed for material failure modeling. The second model is a quasi-static model developed for simulating the 180° peeling of bonded fabric samples to validate the accuracy of both the fabric and adhesive material models. The third model is a dynamic model for simulating the pressurization of an inflatable structure constructed by adhesively bonded fabric pieces in real-world applications. The insights gained from the analyses are discussed to demonstrate how predictive modeling can be leveraged to innovate differentiated joining solutions faster based on specific applications needs.

Establishing and Maintaining the Digital Thread of Additively Manufactured Materials and Applications

Authored & Presented By
Brandon Hearley (NASA Glenn Research Center)

Chaired By
Chair Steve Arnold

Robustness and Reliability Studies for Weld Simulation

Authored & Presented By
Gaurish Sharma (ESTECO North America, Inc.)

Chaired By
Chair Steve Arnold

Nonlinear Analysis

Authored & Presented By
Tony Abbey (FETraining)

Convergence of CAE, Data Science and HPC

Authored & Presented By
Fatma Kocer (Altair Engineering, Inc.)

Chaired By
Chair Gregory Westwater

Can I send the abstract after the EDS WG event? I want to see how relevant that talk was before submitting one for Americas event.

Demystifying AI for CAE

Authored & Presented By
Ian Pegler (NVIDIA)

Chaired By
Chair Gregory Westwater

Given the rise in interest in AI there has been a corresponding increase in interest in the applicability of AI approaches for Computer Aided Engineering (CAE). AI offers the prospect of being able to give faster insights into product performance earlier in the design resulting in better products and reduced time to market. Despite the emergence of AI-enhanced techniques, engineering organizations struggle to integrate these capabilities into production settings due to a lack of openness and adaptability in existing tools. The shift towards Engineering Intelligence emphasizes AI as a central paradigm in the design process, with advanced Deep Learning algorithms enabling the transformation from simulation-enabled to AI-enabled design. These algorithms create predictive models that provide post-processed simulation results and optimization suggestions directly from CAD designs. The goal is to simplify processes, emulate simulation engineer expertise, and transfer complex interactions to computers. One of the challenges of applying machine learning techniques is understanding which approach to use and the limitations of that approach. Approaches can vary from completely data driven to fully physics informed. Other challenges include effectively implementing this technology in production, requiring an open and flexible platform that facilitates the synergistic interaction of various engineering intelligence enablers, including new technologies, skilled data science engineers, Large Language Models, cloud HPC services, and other facilitators. In this presentation we will discuss the basics of physics-informed deep learning and data-driven deep learning applied to problems having roots in physics and understand the techniques that help apply deep learning to modeling multi-physics simulations systems. We will discuss these principles in the context of NVIDIA's open-source framework Modulus for building, training, and fine-tuning Physics-ML models using a Python interface. We explore the building blocks provided by the framework, including different architectures like neural operators, GNNs, PINNS, diffusion models, etc. One of NVIDIA's partners Neural Concepts will also discuss real-world use cases and challenges where AI approaches have been applied to industrial scale problems. In summary, this presentation aims to demystify AI approaches in CAE, offering attendees a comprehensive understanding of both physics-informed and data-driven deep learning techniques. Through discussions on Modulus and real-world applications, we hope to empower attendees with the knowledge and tools necessary to leverage AI effectively in the realm of engineering simulation and design optimization.

AI in Design Exploration

Authored & Presented By
Mike Sheh (INTES GmbH)

Chaired By
Chair Gregory Westwater

Abstract Can AI/ML find a design outside of the dataset that is used to train the model? Is an AI-generated digital twin better than a conventional surrogate model such as a response surface, kriging, or RBF model? Will we begin to see AI being implemented in existing commercial CAD/CAM/CAE/PLM software that dramatically increases productivity... and collaboration? Do I need to learn Python to "stay in the game"? As AI/ML gains warp speed recognition that could significantly change how we do things, including engineering design, companies are eager to find their strategy to embrace the new generation technologies. In this presentation, we will discuss our past R&D in the domain of data analytics, surrogate modeling, and optimization. In particular, our attention will be placed in the following two areas: AI-powered "What-if" tools for designers, and an AI-powered exploration algorithm that largely replaces the past generation algorithms. Current real time What-if tools have promised to give designers fast answers to their design changes. The math behind them is typically some type of surrogate model. Historically, this could be a response surface model, Kriging, Radial Basis Function, and many others. These approaches typically suffer when the surrogate model does not correctly predict, within an acceptable level of error, the responses due to design changes. The success rate over the past decades has been marginal, especially when those surrogate models are developed based on a set of simulation data. However, the latest AI/ML approach has greatly improved the ability of these tools to predict responses accurately and consistently. This will be discussed in our presentation. The other aspect is the use of AI in design exploration/optimization, particularly when using simulation in search of better designs. The past generation of exploration tools are mostly math-based. The result is that without the user knowing enough of the underlying mathematical characteristics of the problem, it is hard to pick the suitable algorithm(s) that can derive the best answers in the least amount of time. This is what we used to face in the past: "What algorithm should I pick?" The latest development in using AI in learning-based searches has dramatically changed that. The next generation OASIS.AI Optimizer has shown its promise to bring robustness, versatility, simplicity, and speed to design exploration. We will discuss this advancement and illustrate the new technology's effectiveness using benchmark data.

Deploying an AI Simulation Stack Across Engineering Teams - A Case Study

Presented By
Matt Bemis (SimScale)

Chaired By
Chair Fatma Kocer

Authored By
Matt Bemis (SimScale)
David Heiny (SimScale GmbH)

The advent of artificial intelligence (AI) in engineering simulation begins a transformative era in product development processes. This abstract outlines the substantial impact of AI adoption in engineering simulation, focusing on how AI models, trained past engineering data , can significantly accelerate or even automate future engineering workflows. AI expedites design and innovation by dramatically reducing simulation lead and run times, thereby enabling engineering teams to explore design spaces broader, deeper and faster in pursuit of higher product performance. The influence of AI extends beyond this, however; it has the potential to accelerate every step of the Computer-Aided Engineering (CAE) workflow: -AI's integration begins at the CAD generation phase with generative design, followed by CAD preparation involving simplification and cleanup. -It further streamlines mesh generation by optimizing grid setup based on geometry and physics -Accelerates simulation setup through pre-generated templates, automatic sanity checks, and guidance dependent on the problem -Revolutionizes the numerical solve process with AI-accelerated solutions -Finally, can aid in result interpretation and design optimization, enabling more efficient and effective decision-making. The success of AI models hinges on the availability of consistent, accessible, and large data sets. An integrated solution that combines AI with traditional CAE methods and data management lowers the barriers to AI adoption and enables engineering teams to gradually train, use and improve AI models in their workflows. This approach involves data curation, AI model generation and evaluation, and continual enhancement with newly generated data, all consolidated in a single platform. Such integration facilitates seamless access and collaboration among designers and engineers across all project stages, significantly bolstered by AI capabilities. SimScale is developing a pioneering platform in this domain, offering a solution that amalgamates AI with traditional CAE methods. This platform provides accessible tools for both designers and simulation experts, capturing the entire spectrum of AI-enhanced simulation processes and ensures AI future readiness. This paper presents a case study of deploying AI through a cloud-native engineering simulation platform with global engineering teams in the automotive and manufacturing industries.

Removing the Data science barriers to Deep Learning surrogates

Presented By
Hoss Habib (Key Ward GmbH)

Chaired By
Chair Fatma Kocer

Authored By
Hoss Habib (Key Ward GmbH)
Andrew Fine (Fine Physics)

The convergence of Artificial Intelligence (AI) and engineering disciplines heralds a transformative potential for optimizing design processes, enhancing analytical methodologies, and advancing simulation technologies. Despite these prospects, the integration of AI into engineering practices is impeded by significant challenges, primarily stemming from the handling of vast, unstructured engineering datasets and the inherent interdisciplinary gap between engineering and data science expertise. This paper introduces an automated, engineering-focused framework, Keyward, designed to bridge this divide by facilitating the efficient preparation and utilization of engineering data for AI applications. Leveraging advanced algorithms for data sorting and preprocessing, Keyward enables the seamless conversion of heterogeneous engineering data into AI-ready formats, addressing compatibility issues across various software tools and file formats prevalent in engineering workflows. Furthermore, the framework incorporates intuitive, no-code tools tailored for engineers, alongside comprehensive guidelines and validation checks for data scientists, thereby ensuring the generation of coherent, high-quality datasets for AI modeling. The efficacy of the Keyward framework is demonstrated through a case study within the OpenFOAM community, highlighting its capabilities in significantly reducing the time and expertise required to prepare engineering simulation data for deep learning applications. The results underscore Keyward's potential to not only streamline the AI integration process but also to foster a synergistic collaboration between engineering and data science disciplines, thus accelerating the adoption of AI technologies in engineering practices.

Megacasting 2.0: Transforming the Industry with AI Integration

Authored & Presented By
Milan Raval (Altair Engineering, Inc.)

Chaired By
Chair Liviu Georgescu

Metal casting has its roots in ancient civilization. Starting from the bronze age (around 3300 BCE to 1200 BCE) till the industrialization in the 18th and late 19th centuries, the metal casting process evolved significantly. In the 20th and 21st centuries, metal casting adopted advanced materials, technology, and various casting processes. Computer-aided design (CAD) and simulation (CAE) tools for predictive approaches helped industries to produce more complex castings efficiently and with higher accuracies. Today, metal casting plays a vital role in various industries, including automotive, aerospace, and heavy equipment manufacturing. In the automotive industry, the emerging need for sustainable mobility solutions has inspired designers to create exciting new vehicle concepts and rethink the ways they can minimize weight, lower manufacturing costs, and meet modern performance targets. In particular, megacasting - and its evolution into gigacasting - refers to a new approach for the serial manufacture of large, complex geometry, light alloy metal parts using high-pressure die casting (HPDC). Megacasting is of particular interest for large structures within the body in white (BIW) of electric vehicles (EVs). HPDC is a well-established industrial process for making aluminum- and magnesium-alloy-based automotive parts such as drivetrain casings, smaller precision parts, or other weight-sensitive structures. By increasing cast component size, teams can reduce the amount of required individual parts – which are often stamped and welded and made of different materials – thus saving time and resources. This does not come without risk, because the transition from using wrought products to casting at this scale needs considerable investment in new plants and machinery, along with expertise in optimizing the design and assuring high-quality serial manufacturing of these relatively recent megacastings. Furthermore, such technology advancements came with their own design challenges: manufacturing megacasting parts is not easy and can require a different approach to CAD and CAE product development to overcome additional hurdles like longer computational time, extensive model validation, complex mesh generation, etc. As the whole world is diving deeper into utilizing Artificial Intelligence (AI) algorithms in the most diverse applications, AI can be leveraged to streamline megacasting design manufacturing as well. In one example of an application for battery electric vehicles (BEV) design, we take a deeper dive into a bespoke megacasting workflow to illustrate the symbiosis of AI and simulation in the design of chassis components. Specifically, AI has been used to easily identify the best part designs given specific design and manufacturing criteria, out of an enormous amount of simulation data by applying machine learning clustering, known as AI-driven expert emulation. Furthermore, AI-powered engineering tools can predict faster and make design decisions quicker than traditional casting simulations. This tool can not only save computational time taken by traditional casting simulations but also provide better result predictions on new designs without running the simulation itself. These AI-powered generative design workflows combine AI, generative design, and the latest CAE technologies, providing a proven approach that enables multidisciplinary variant analyses for efficient, production-oriented, and performance-driven megacasting structures.

AI-based optimization of sheet metal stamping for automotive applications

Presented By
Seokpum Kim (Oak Ridge National Laboratory)

Chaired By
Chair Liviu Georgescu

Authored By
Seokpum Kim (Oak Ridge National Laboratory)
Rich Davies
William Halsey
Seokpum Kim
Vincent Paquit (ORNL)
Sathya Dev
Kaiping Li (Stellantis)
Liang Huang
Andrey Ilinich (Ford)
Lu Huang
Krishna Murali
Weidong Wu (GM)
Kidambi Kannan (AutoFORM)

Sheet metal stamping processes are widely used in various industries due to their cost-effectiveness, rapid cycle times, mass production capabilities, and relatively high precision. The quality and efficiency of these processes hinge on several critical factors, including material properties, blank dimension and uniformity, die geometry, and processing parameters such as binder force, lubricants, and spacer utilization. To enhance production efficiency, process simulation tools are commonly employed. A typical optimization procedure for manufacturing requires an iterative process involving parameter setting, execution of computational simulations, and modifying the parameters. The entire process demands substantial computational time, making it impractical for real-time feedback towards rapid corrective actions required for in-line control for running production processes. To overcome this challenge, artificial intelligence (AI) can be leveraged to determine optimal manufacturing parameters within a single manufacturing cycle time. This research proposes an in-line optimization framework incorporating a trained AI model to predict kidney-shaped die forming. Preliminary results indicate that the AI framework can accurately predict draw-in values based on a given parameter set, a process referred to as forward prediction. Furthermore, the AI framework can also predict the optimal parameter set that leads to the desired draw-in values, referred to as backward prediction. This research is performed in collaborations with AutoFORM and USCAR (US Council for Automotive Research). The members of USCAR are Ford, GM, and Stellantis.

Artificial Intelligence to Empower Engineering Organizations: Technology & Applications

Authored & Presented By
Frederick Reis (Neural Concept)

Case Study of Seamless Modeling and Simulation for Electric Drive Engineering

Presented By
Satheesh Kandasamy (Dassault Systèmes SIMULIA Corp)

Authored By
Satheesh Kandasamy (Dassault Systè
mes SIMULIA Corp)
Young-Chang Cho

System Modeling Bridging the Gap with Smart Systems Modeling (SSM)

Authored & Presented By
Alejandro Ayala (Estrateg IA)

Chaired By
Chair Dipak Patil

The main purpose of modeling is to translate physics into a simplified representation, including just the significant variables that determine most of the emergence output. Systems Modeling has been a useful tool to develop complex systems and new technologies. On the other hand, systems modeling has constraints to generate representative models with multiple dynamic and nonlinear variables, therefore there is an incremental gap between the models' capabilities and the growing rate of systems' complexity. This gap is generated by the lack of smart modeling tools to identify the critical factors that generate most of the system's results, under increasing disturbing environments. This work presents the development of Smart Systems Modeling (SSM), powered by Artificial Intelligence (AI), as an evolution of the Smart Model Based Systems Engineering (SMBSE)* approach, to bridge the gap between modeling and the increasingly complex systems. SMBSE presented in 2018, the development of Systems Engineering models, enhanced by automation, optimization, reusability and dynamic simulation functionalities. The AI available tools allow the transformation of SMBSE into more powerful models, enhanced with simultaneous modeling and virtual simulation capabilities, as well as regression techniques to predict results and the required flexibility to adequate the core model for new functions and interfaces. Finally, AI enhances SSM with smart functionalities to determine patterns of relationship and the significant variables across complex big data and non-structured inputs. AI-Machine Learning (ML) will support the development of SSM with learning abilities to improve the models' fidelity until an acceptable correlation level is achieved. This is just the beginning of a promising growing solution to develop powerful design and research models, in conjunction with AI's decision-making processes and learning capabilities. The SSM progress will provide benefits for faster replacement of physical tests by robust virtual model simulation, for new products development, as well as agile research experiments. *"Smart Model Based Systems Engineering", Alejandro Ayala, August 7th 2018, Michigan USA.

Leveraging MODSIM Techniques for enhanced engineering analysis: integration of SOLIDWORKS models with the 3DEXPERIENCE Platform

Authored & Presented By
Omar Zohni (Dassault Systèmes SIMULIA Corp)

Chaired By
Chair Dipak Patil

This paper presents an innovative approach to engineering analysis by harnessing the power of Model Simulation (MODSIM) methodologies in conjunction with industry-leading software tools such as SOLIDWORKS, SIMULIA and the 3DEXPERIENCE Platform. In the contemporary engineering landscape, the demand for efficient and comprehensive simulation techniques has become paramount. By amalgamating MODSIM principles with the capabilities offered by SOLIDWORKS and the 3DEXPERIENCE Platform, engineers can achieve unparalleled insights into complex designs and processes. Through this integration, engineers can seamlessly transfer SOLIDWORKS models into the 3DEXPERIENCE Platform environment, facilitating a streamlined simulation workflow. This transition not only enhances the fidelity of simulations but also enables engineers to leverage advanced simulation functionalities offered by the 3DEXPERIENCE Platform, including multiphysics simulations, optimization algorithms, and collaborative features. The utilization of MODSIM techniques within this integrated framework empowers engineers to conduct thorough analyses across various domains, ranging from structural mechanics and fluid dynamics to electromagnetics and thermal simulations. By capturing the intricate interactions within systems, engineers can make informed design decisions, optimize performance, and mitigate potential risks. Furthermore, the integration of SOLIDWORKS models with the 3DEXPERIENCE Platform fosters collaboration and knowledge sharing among interdisciplinary teams. Engineers can collaborate in real-time, exchange insights, and iterate on designs rapidly, thereby accelerating the product development lifecycle. This paper showcases practical applications of MODSIM techniques integrated with SOLIDWORKS models within the 3DEXPERIENCE Platform utilizing the simulation tools provided by SIMULIA. Case studies will illustrate how this integrated approach has revolutionized engineering analysis, leading to improved product quality, reduced development costs, and shortened time-to-market. In conclusion, the integration of MODSIM methodologies with SOLIDWORKS models and the 3DEXPERIENCE Platform represents a paradigm shift in engineering analysis, enabling engineers to unlock new realms of innovation and efficiency. This paper advocates for the adoption of this integrated approach as a cornerstone in modern engineering practices, driving advancements across industries and shaping the future of product development and design.

Democratizing Simulation Methods to Collaborate Efficiently using the 3DEXPERIENCE Platform

Presented By
Gopi Lanka (Dassault Systèmes UK)

Authored By
Gopi Lanka (Dassault Systè
mes UK)
Dominick LAUZON
Chris OLSON (Dassault Systè
mes)

Both CAD and CAE applications used by designers, analysts and engineers have evolved independently over time, introducing their own 3D modelling capabilities, user interfaces, data formats and security mechanisms. These disconnected technology silos can slow the design process, introduce errors and allow for data loss. The situation gets worse with non-existent or poor data management practices while preventing the ability to capture and deploy best practices across the organization. The 3DEXPERIENCE Platform offers a unified model-based architecture, which eliminates inefficiencies in the design process while propelling virtual engineering to a completely new level. The architecture lays a foundation for a modelling and simulation (MODSIM) approach that seamlessly combines best-in-class design & modelling technologies from CATIA along with superior multi-physics simulation modelling portfolio of SIMULIA. The parametric nature of these models allows for building automated simulation processes that capture the best practices. Simulation processes can later be democratized to CAE non-experts via simulation templates. Process standardization helps in reducing rework while increasing simulation throughput and driving innovation. This presentation will highlight simulation process use cases from different industries and demonstrate the values delivered to individuals and engineering teams. It will show how Designers and CAE engineers can leverage expert simulation methods and perform simulations more efficiently. CAE Managers will observe increased cross-discipline collaboration between teams, increased simulation throughput that helps generate evidence to make better decisions. Finally, it will demonstrate the ability to capture all the generated simulation data in a digital thread, managed seamlessly within the platform, that is easy to search and trace.

Machine Learning for Physics-Accelerated Design of Mixing Tank Processes

Presented By
Mustafa Kaddoura (Ansys, Inc.)

Authored By
Mustafa Kaddoura (Ansys
Inc.)
Matthew Adams
Marc Horner
Sreedevi Krishnan
Hossam Metwally
Ryan O'
Connor (Ansys)

With the escalating prominence of Machine Learning, the convergence of these technologies with physics-simulation methodologies opens unprecedented opportunities to enhance the effectiveness and accuracy of predictive models, particularly evident in the realm of mixing applications within industrial processes and bioreactors. Among the promising machine learning approaches is the data fusion, also known as residual modeling, in which few high-fidelity data points (simulation or experimental) are used to increase the accuracy of low-fidelity simulation data by learning the error trend between the two data sets. Augmenting physics-based models with data-driven models improves solution fidelity without increasing the complexity of the mathematical model form. In this study, we showcase two industrial application case studies for applying the Fusion modeling concept. The first case study aims to enable blend time prediction in mixing tanks efficiently and accurately at distinct design variables. We create a fusion model by augmenting high-fidelity intensive sliding mesh CFD simulation data with the less-accurate Multiple Reference Frame (MRF) CFD simulation data to enhance model fidelity. The design space of the mixing tank in terms of material properties (density and viscosity) as well as operating conditions (agitation speed) was explored using both the high-fidelity approach and the Fusion modeling approach. We then build a system-level fusion-reduced order model (Fusion-ROM) to achieve instantaneous, accurate blend time prediction. Finally, we build a custom user-friendly app on the top of the developed workflow to enable non-experts for running mixing tank simulations. This hybrid approach reduced the root mean square error (RMSE) in MRF model from 32% to 6%, while maintaining quick/instantaneous prediction, and reduced the CPU hours required to construct the design space by approximately 50% versus the purely sliding mesh approach. The second case study aims to build an accurate digital twin for sparged bioreactor applications by augmenting experimental data with CFD simulation data to enhance the prediction of mass transfer coefficient (Kla) and gas hold-up in bioreactors operating at various agitation speeds and flow rates (VVM). Generated fusion models were used on top of fluid flow ROMs as part of a system-level model to design the bioreactor control system and for online/offline digital twin.

Building Digital Twins of Additive Manufacturing Parts for Fatigue Life Prediction

Presented By
Xueyong (Kevin) Qu (The Aerospace Corporation)

Authored By
Xueyong (Kevin) Qu (The Aerospace Corporation)
Vinay Goyal
Jacob Rome
Leland Shimizu (The Aerospace Corporation)

Additive Manufacturing (AM) parts are widely adopted in various industries due to design freedom, manufacturing flexibility, and potential time and cost savings. Porosities produced by AM processes, such as keyholes or lack of fusions, can be crack initiators and can propagate to failure under fatigue environments. While AM has been adopted widely in aerospace applications, predicting failure of AM parts continues to be a challenge. Digital Twins of the AM part can simulate actual defects and can be updated over the service life of the part to support real-time fatigue assessment. This paper employs fracture mechanics to assess fatigue life of a printed component. The number of pores, their sizes and locations in AM parts can be characterized with computed tomography (CT) scans. Pore data, fatigue load spectrum, and material properties are variables considered in deterministic fracture mechanics Digital Twins to predict fatigue life. In parallel, a probabilistic fracture mechanics model is also executed to account for spatial variabilities in the properties. The Digital Twins are validated by simulating fatigue life of an axial fatigue test coupon and then compare to test results. The applications of Digital Twins are demonstrated by in-situ prediction of fatigue life of an AM part during the printing. The influence of pores and build heights on AM part fatigue life are demonstrated by simulating the printing process from bottom to top. First, deterministic pores up to the current built layer are simulated in the model, as if they were detected by in-situ measurements. These deterministic pores are then combined with a probabilistic pore distribution in the unprinted portion characterized by ex-situ CT scans of similar builds to provide a failure life prognosis for the full part. The numerical approach and examples will be presented in the final presentation.

Coupling Manufacturing and Component Design with Systems Engineering

Presented By
Tero Frondelius (Wärtsilä Finland Oy)

Authored By
Tero Frondelius (Wä
rtsilä
Finland Oy)
Isa Banagar
Aapo Hö
lsä

Juho Kö
nnö
(University of Oulu)
Jarkko Laine
Hannu Tienhaara (Wartsila)

In the dynamic landscape of modern engineering, the integration of manufacturing processes, component design, and systems engineering emerges as a critical paradigm for fostering innovation, efficiency, and sustainability. This abstract explores the multifaceted constructive collaboration between manufacturing, component design, and systems engineering, emphasizing the transformative impact on product development and industrial ecosystems. The traditional silos between manufacturing and design are dissipating as collaborative frameworks gain traction. This shift is propelled by advancements in computational tools, simulation techniques, and data-driven approaches. The coupling of manufacturing and component design within the broader context of systems engineering presents an integrated approach to product realization. The integration begins with the co-evolution of design and manufacturing considerations from the project's inception. Early collaboration enables concurrent engineering, allowing for real-time feedback loops between design iterations and manufacturing feasibility assessments. This iterative process optimizes both the product's functionality and manufacturability, streamlining the entire development lifecycle. Systems engineering plays a pivotal role in orchestrating this collaborative dance. It provides the overarching framework to align diverse elements such as requirements, functions, and constraints, ensuring a unified approach. By holistically addressing the interdependencies between design, manufacturing, and broader system requirements, systems engineering enhances product quality, reliability, and performance. The abstract delves into specific case studies illustrating successful implementations of this integrated approach across diverse industries. These cases highlight not only improvements in time-to-market but also cost-effectiveness and sustainability metrics. The integration facilitates the design of components that are inherently manufacturable, reducing waste and optimizing resource utilization. The role of advanced analytics and artificial intelligence in this integration is also examined, highlighting their capacity to analyse vast datasets generated throughout the product lifecycle. These technologies empower engineers to make data-driven decisions, fine-tuning both the design and manufacturing processes iteratively. In conclusion, the abstract underscores the imperative for a change in thinking in engineering methodologies. The coupling of manufacturing and component design with systems engineering heralds a new era of collaborative innovation. Embracing this integrated approach promises not only enhanced product quality and efficiency but also a sustainable and resilient foundation for the future of engineering.

Simulation driven design of a BESS cooling system using CFD and parametric optimization

Authored & Presented By
Benjamin Beckelynck (Optimec Consultants Inc)

Battery Energy Storage System (BESS) utilizes rechargeable batteries such as lithium ion, to store electrical energy in the form of chemical energy. BESS contributes to grid stability by providing services such as frequency regulation, peak shaving, and grid balancing. BESS is often employed to address the intermittent nature of renewable energy sources like solar and wind. By storing excess energy when generation exceeds demand and releasing stored energy during peak demand periods, BESS helps integrate renewables into the grid more effectively. Additionally, portable BESS systems can be used in remote locations or emergency situations where a stable power source is needed. Battery Energy Storage Systems (BESS) require effective heat management to ensure the safe and efficient operation of the battery cells. Heat management is crucial in maintaining the performance, lifespan, and safety of the batteries. Batteries have an optimal operating temperature range for efficient performance. High temperatures can accelerate chemical reactions within the cells, leading to faster degradation and reduced lifespan. Conversely, low temperatures can decrease the efficiency of the battery. Many BESS incorporates active cooling systems, such as liquid cooling or air cooling, to maintain the batteries within the desired temperature range. The study involves an analysis of a BESS comprised of 204 battery modules. The system is arranged in two rows of 6 racks, each containing 17 modules, resulting in a total of 204 modules. Each module generates 126 Watts of power. To cool the batteries, a system of fans is employed to push and direct airflow inside the trays, with each module having its own dedicated fan, totaling 204 fans. Additionally, an HVAC unit with 2 large blower fans and a heat exchanger is utilized to circulate air inside the enclosure and cool it down to a specific temperature of 24°C. A plenum is used to ensure uniform distribution of air to each module, with its design optimized through parametric optimization to maximize airflow uniformity and minimize the pressure drop in the system. To streamline analysis and reduce solve time, porous media are calibrated to represent internal components such as the modules and the heat exchanger, facilitating accurate representation of pressure drops. Thermal analysis is then conducted to estimate the cooling capacity of the HVAC unit and determine the operational temperature of each module at full load. In summary, the work involves conducting a Computational Fluid Dynamics (CFD) analysis on the BESS system to optimize airflow and cooling performance using the powerful tool set of parametric design study and have a simulation driven design.

Advanced Modeling of Filtration Processes: Integrating High-resolution Filter Matrix into CFD-DEM Simulation

Authored & Presented By
Ahmadreza Haghnegahdar (Ansys, Inc.)

Gas flow simulation generated for Internal Arc Fault based on Digital Twins

Authored & Presented By
Mariana S. Bretado de los Ríos (Schneider Electric)

The development of "Digital Twins" models has become popular for describing solutions and optimizing processes and products. With the development of new technologies, digital models can be generated to act as a mirror of a physical system or product, allowing the optimization of designs, processes, and complex industrial systems. The most complete digital twins share three elements: context and data characteristics, real-time operational data, and an information model used to integrate them. The ideal scenario for "Digital twins" is to transfer data in real time between the physical and digital models to generate continuous feedback and optimize the system. However, the complexity of certain physical phenomena makes the synchronization of real and digital data extremely difficult. This is the case in electrical applications to prevent failures due to short circuit or Internal Arc Faults (IAF). This work focuses on the phenomenon of internal arc failure in electrical equipment, for which obtaining experimental data is very difficult, and consequently, comparing them with digital models is challenging. IAF energy is transferred through the air and is generated between two electrified components due to a lack of maintenance or insulation losses. When an IAF occurs in electrical equipment, two important goals must be achieved: 1) the structure and enclosure walls of the equipment must withstand the overpressure, gas blast, and stress waves; 2) the heat flow must be contained and released gradually to avoid thermal damages that could threaten human safety. To prevent IAF damages, experimental tests must be performed according to the standards for electrical equipment. Due to the complexity of IAF tests, simulations become a solution to make the analysis of IAF effects more efficient, saving costs and time. The challenges that IAF digital twin face are mainly: 1) understanding physics, 2) solutions and coupling, 3) technology availability, and 4) validation and verification. This work focuses on characterize the physics behind the gas flow behavior due to the arc fault phenomenon and highlights the importance of understand the model to achieve accurate simulation results. The principal parameters study in this work are energy input, vaporized mass, grills, flaps, or wire meshes. This first step leads to move forward in the development of a IAF digital twins model and consequently, perform a wider analysis to validate and correlate simulation results with experimental tests. A simplified medium-voltage unit is considered as a case study. The results show the gas flow thermal and structural effects with different parameters on the equipment. The validation and verification are made through experimental tests, obtaining an accurate approximation with the simulation. Moreover, parametric assessment is conducted to analyze the effects of vaporized mass, type of grills, and input energy on the gas flow of the IAF. The methodology presented in this work provides accurate results that improve the efficiency of simulations for the development of new products to guarantee the human safety of electrical equipment through the IAF digital twins.

Automating Multi-disciplinary Workflow Collaboration and Simulation Data Management to Accelerate R&D

Authored & Presented By
Bernardo Mendez (Rescale Headquarters)

Achieving Large Scale Detailed Component Level Structural Simulations

Authored & Presented By
Mark Taylor (The Boeing Company)

Chaired By
Chair Bob Tickel

Scale and democratize reduced order modeling techniques within an SPDM framework

Authored & Presented By
Nachiket Kasarekar (ESTECO North America, Inc.)

Chaired By
Chair Bob Tickel

Solving Industry's Data and Process Challenges of Digital Engineering with R&D Clouds

Presented By
Edward Hsu (Rescale)

Chaired By
Chair Bob Tickel

Authored By
Edward Hsu (Rescale)
Alex Kachkovsky
Manager of Virtual Engineering at Daikin Applied Americas

Innovators across industries are embracing new digital engineering methods throughout the R&D process to expedite product development, enhance quality, and mitigate operational risks. With the proliferation of these methods, engineering computing technologies such as CAE and AI-enhanced simulation are generating vast amounts of data that drive product decisions. Concurrently, the development process for new products has become increasingly complex and collaborative, involving more teams and analyses. Companies like Kairos Power and Lear Automotive are competing in highly complex and regulated industries by leveraging purpose-built R&D clouds to optimize their engineering processes with the latest technologies. Today, innovation-focused companies prioritize user workflows and seek purpose-built solutions that meet their specific needs. In the past, engineering computing stacks were built from the ground up, requiring specialized IT expertise and long lead times. As companies look to integrate new use cases such as AI-driven design and hardware-accelerated computing, R&D clouds offer pre-built ecosystem configurations and low-code tools to customize and integrate the latest technologies. This enables R&D teams to advance their product engineering and digital transformation initiatives without requiring expertise in building and managing sophisticated compute and storage infrastructure. A key benefit of purpose-built R&D clouds is their ability to capture and extract critical insights throughout the product design, optimization, and validation process. Engineering teams that can effectively manage large datasets with rich domain and institutional context across all their digital activities (simulation, modeling, AI/ML, etc.) can improve the speed and quality of their product decisions. As data volumes and variety explode with greater usage of digital tools, teams that can easily capture their industry and company's historical information will be able to share, search, and understand context to extract new insights that lead to new breakthroughs in R&D. Additionally, R&D clouds enable organizations to harness cutting-edge technologies to remain competitive. While new domain-specific software and hardware architecture options have multiplied, increased choice and specialization of tools and resulting datasets bring new challenges. R&D Cloud or industry cloud platforms aim to eliminate complexity by simplifying and tailoring the tool stack to empower digital engineers, computational scientists, and researchers. As R&D teams connect multiple tools and build multi-stage processes, effective management of metadata ensures processes can be mapped for reproducibility and optimized for efficiency. Companies leverage Rescale as a composable and customizable R&D cloud to quickly adopt modern engineering use cases including model-based collaboration, knowledge management, product analytics and visualization, and process automation and optimization. Choosing an R&D cloud that is flexible to fit seamlessly with existing workflows ensures engineering and scientific leaders can successfully adopt these use cases in a user-centric experience. These capabilities deliver the agility and control needed to accelerate product decisions and reduce risk and cost across digital engineering activities. Kairos Power, a nuclear company disrupting the energy sector, utilizes Recale to accelerate product delivery timelines with tight collaboration and context-based data management across all simulation activities. Lear Corporation, a leading automotive supplier, orchestrates large-scale engineering computing on Rescale, ensuring their multi-disciplinary teams have shared access to critical context needed to make timely product decisions with confidence. Companies like Kairos and Lear capture institutional knowledge to optimize engineering productivity and process efficiency throughout compute-driven R&D stages. As industry leaders and practitioners pursue their product innovation and digital transformation objectives, R&D cloud platforms and domain-specific engineering solutions will be increasingly differentiated from DIY point solutions. R&D clouds empower companies to balance speed and control with policy-based automation and low-code composability, while focusing on user experiences that drive innovation.

Leveraging Virtual Engineering for Faster Product Development and Cost Efficiency

Authored & Presented By
Prasad Mandava (Visual Collaboration Technologies, Inc.)

Chaired By
Chair VamsiKrishna Pateel

In the field of engineering, where we're always striving to innovate and cut costs, the use of virtual engineering is gaining prominence. Our paper, titled "Leveraging Virtual Engineering for Faster Product Development and Cost Efficiency," explores how virtual engineering can drive innovation, speed up product development, improve quality through design robustness, save cost and help companies launch products faster to the market. Traditionally, building and testing physical prototypes consume a lot of time and cost in product development, hence delaying product launches. But now, we're seeing a shift towards using virtual tools instead. This shift is essential because market demands are pushing for faster, cheaper, and smarter simulation solutions. This paper investigates the evolution of digital technology in product development processes and highlights a critical gap - the underutilization of digital tools in post-processing, reporting, reviewing & communicating phases, such as results interpretation, results reporting, and rapid results reviews. The emergence of a new class of simulation tools addresses this limitation, offering solutions such as a common data structure for integrated multi physics results, intelligent data mining for identifying potential failure modes, advanced 3D visualization, highly compressed simulation results, automated templates for results processing, reporting and the creation of 3D Digital Slides for rapid results reviews/communicating with stake holders. This presentation will focus on the key aspects of democratizing the virtual engineering process in specifics the process of analysis results review. Will highlight automation solutions that are used to quickly extract and visualize analysis results from a broad range of multi-physics solvers, so design decisions can be made faster and with more confidence. Some of the benefits include: 1. Smart automation & efficient multi-format graphic displays (Digital Slides) 2. Ease of sharing graphical results data with the engineering teams using HTML capabilities 3. Considerable reduction of output file sizes so output can be stored for future access. 4. Automatic digital report generation By embracing these new classes of virtual engineering tools, we can speed up product development, launch products faster to the market and save money, all while staying at the forefront of engineering innovation.

Quick and reliable simulation process automation to reduce product development time and cost

Presented By
Karlheinz Peters (Novus Nexus)

Chaired By
Chair VamsiKrishna Pateel

Authored By
Karlheinz Peters (Novus Nexus)
Davis Evans
Bruce Webster (Novus Nexus
Inc)
Mario Felice (virsolTech Engineering Consulting
LLC)

Wide-range use of simulations through all phases of product development is becoming increasingly important to meet the growing challenges of global competition. Simulation not only helps develop competitive products faster but also helps achieve cost efficiencies by reducing the number of necessary prototypes and related physical testing. However, to realize all the achievable benefits of simulation, it needs to be performed systematically, consistently, and efficiently. In an optimal scenario, simulation is directly available to a broader group of users, not just simulation specialists, with designers and analysts collaborating closely. Automated simulation processes for analysts, designers, and product developers are essential for realizing this systematic and efficient use of simulation. Historically, such broader use of automated processes often fails due to the task's difficulty. Automation via common approaches like scripting (Python, etc.) quickly results in a prohibitive effort for implementation and upkeep, especially when complex or vastly differing geometries are involved. Not only is the initial setup complex here, but extensions or modifications down the line can become exceedingly complicated, too often resulting in a complete rebuild of the processes. Therefore, new approaches are needed to overcome the current challenges and secure a fuller range of benefits from simulation, including methods that enable fast, easy, and robust process automation. This presentation explores how a combination of simulation-specific no-code tools (based on abstract modeling), together with universal "low-code" tools, can drastically simplify automation, even and especially when complex products are involved. The impact of systematically automated simulation processes on the overall development cycle is also analyzed. The ability to automate simulation processes without penalty, which means creating an automated process with a similar effort as setting up a traditional one-off simulation, opens more opportunities to benefit from the known advantages of automation, including the ability for non-CAE specialists to initiate reliable simulations, the relief for analysts from unproductive routine work, the preservation of simulation knowledge, and the continuous use of best practices for future simulations.

Parametric modeling of SOEC System in a Grid Energy Storage Application and Democratization for Wider Adoption

Authored & Presented By
Apurva Gokhale (ESTECO North America, Inc.)

Chaired By
Chair VamsiKrishna Pateel