Michael Bogomolny, Ph.D.

The best part of working with Ansys is the vast suite to tools to solve different problems. I did a full 2 way FSI of stress on a boat due to waves close to 20 years ago and it was tough! This is the easy button... and the right approach.

21

1 Comment

Boats and waves. This is a very easy way to simulate the effect to different waves on the durability of floating structures. https://lnkd.in/gE6x_2VM

31

Ever tried to design parts from a 3D scan? Either work with super heavy stl conversion or spend a lot of time create a clean model. I never knew how powerful STL to SubD conversion was until I tried this out in Ansys Discovery! What a powerful tool for working with faceted data! #ansys #stl #3dprinting https://lnkd.in/gHQQuv8V

23

This tool seems very interesting: https://mechinmotion.com/ I am curious if anyone has tried it. I am also curious about calling the APIs from MecSimCalc.

54

Can we use 𝗽𝗵𝘆𝘀𝗶𝗰𝘀-𝗶𝗻𝗳𝗼𝗿𝗺𝗲𝗱 𝗻𝗲𝘂𝗿𝗮𝗹 𝗻𝗲𝘁𝘄𝗼𝗿𝗸𝘀 to predict weld visibility in molded parts when limited experimental data are available? 🔍 Traditional methods for predicting weld line visibility in injection-molded parts usually require 𝗻𝘂𝗺𝗲𝗿𝗼𝘂𝘀 𝗲𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝗮𝗹 𝘁𝗿𝗶𝗮𝗹𝘀, which can be costly and time-consuming. But what if there was a smarter approach? 🧠 Enter physics-informed neural networks (PINNs), which integrate physical laws directly into the learning process. By using physical principles, such as 𝗽𝗮𝗿𝘁𝗶𝗮𝗹 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁𝗶𝗮𝗹 𝗲𝗾𝘂𝗮𝘁𝗶𝗼𝗻𝘀, PINNs help to overcome the lack of extensive experimental data. This allows for the generation of highly 𝗮𝗰𝗰𝘂𝗿𝗮𝘁𝗲 𝘀𝘆𝗻𝘁𝗵𝗲𝘁𝗶𝗰 𝗱𝗮𝘁𝗮𝘀𝗲𝘁𝘀 for training machine learning models, drastically reducing the need for experimental points by up to 78%! 🔄 This novel approach doesn't just predict the temperature gradient within the molded part but associates these values with the visibility of weld lines using a simulation result known as the 𝗙𝗿𝗼𝘇𝗲𝗻 𝗟𝗮𝘆𝗲𝗿 𝗥𝗮𝘁𝗶𝗼. The result? A robust model that 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝘀 𝘄𝗲𝗹𝗹 𝗲𝘃𝗲𝗻 𝘄𝗶𝘁𝗵 𝗹𝗶𝗺𝗶𝘁𝗲𝗱 𝗱𝗮𝘁𝗮, ensuring high-quality production without the traditional costs and efforts associated with extensive data collection. 🚀 Looking forward, the potential applications of PINNs in predicting 𝗼𝘁𝗵𝗲𝗿 𝘁𝘆𝗽𝗲𝘀 𝗼𝗳 𝘀𝘂𝗿𝗳𝗮𝗰𝗲 𝗱𝗲𝗳𝗲𝗰𝘁𝘀, including sink marks and burn marks, are vast. This could provide a powerful tool for defect detection in molded parts across industries. 📖 Interested in diving deeper into this innovative method? Check out our latest research published in the prestigious 𝗝𝗼𝘂𝗿𝗻𝗮𝗹 𝗼𝗳 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝘁 𝗠𝗮𝗻𝘂𝗳𝗮𝗰𝘁𝘂𝗿𝗶𝗻𝗴 to explore how you can apply these insights to your manufacturing processes! https://lnkd.in/dZ9eYGNX Moldex3D ITALIA Moldex3D (CoreTech System Co., Ltd.) Dipartimento di Ingegneria Industriale - DII UNIPD Datapizza Giacomo Ciarlini #Manufacturing #MachineLearning #InjectionMolding #PlasticsEngineering

39

Spray Cooling Using Different DPM Models in ANSYS Fluent In this comprehensive study, three distinct cases were investigated to analyze the effectiveness of spray cooling approaches utilizing ANSYS Fluent software. The simulation setup began with creating detailed geometry using SpaceClaim, followed by generating a high-quality quadrilateral mesh in ANSYS Meshing. Three simulation cases were compared: baseline airflow only (Case 1), water injection with inert particles (Case 2), and droplet interactions and phase change. The DPM configurations employed a two-way coupling approach to capture the complete interaction between the discrete phase and the continuous flow field. https://lnkd.in/dZBxCMRW

1

[Best practices to use bar elements in #FEA] One of the most used (and underestimated) finite elements are bars (or trusses in #Abaqus). From the outside, beams and bars look the same: line elements. But there is a major difference: 🟨 Bars can only transfer axial loads (tensile or compressive). 🟥 Beams can transfer not only axial loads, but also bending moments (and shear forces). Unlike beams, bars have no stiffness in their normal direction!!! ⚠️ Bars do not bend, cannot transfer moments nor rotations. This feature is the source of a few common mistakes in #finiteelement simulations including bar elements: 1️⃣ Do not apply loads normal to a bar (unless there are more bars at that point) 2️⃣ Do not use more than one element along a span of the structure (watch out the element size). 3️⃣ Pay attention to potential "mechanisms". Simulation will be aborted if the model is not statically determined. 4️⃣ Bars cannot be "encastred". They are free to rotate. 5️⃣ The ends of a bar element behave as hinges. Keep it in mind, when connecting beams or other element types. As you can see, most mistakes are related to 'free' degrees of freedom (v,θ), that can 'move' without the application of any force. 😯 Do you have any other recommendations working with bar elements?

98

14 Comments

PyWorkbench is a powerful client library that enhances the Ansys Workbench platform for engineering simulations, allowing users to harness the power of diverse PyAnsys modules. PyAnsys specialist Vikas Namdeo has contributed this guide to walk you through key concepts and practical strategies for using PyWorkbench. Vikas teaches you to make your engineering simulations more efficient and effective, programmatically, using Python’s flexibility to manage and customize your simulation processes: https://ansys.me/3D90sIF #PyAnsys #ANSYS #engineering #simulation #python #workbench

58

Shout out to Evan Pilz & Gabrielle Thelen for this computational model that generates a mask conforming to an individual's face in nTop! It takes as an input a mesh (from a scan) of a face and a few design parameters for the flange and automatically generates the final mask with flange linking the non-custom parts to a face. The mask can also be repositioned & sized based on input points or a target coverage area, or this can be setup to work with a number of different generic mask designs. I'm still impressed with the robustness in using nTop's underlying implicit model + the automation capabilities of the nTop notebook to enable these precise surfaces to just work across a range of inputs -- I guess you could also tweak this file to take in a dogs face if you happen to be designing masks for dogs :) Here's a link to the file: https://lnkd.in/gdrQjMYq

194

18 Comments

Very nice - that’s the right way of designing and evaluating Turbomachinery components: detailed modeling of primary and secondary flow path, combined with automated 3D flow simulation, RANS and URANS. User-friendly, fast, automated, accurate, affordable. (No calculations and „optimizations“ based on loss models or inviscid flow simulation, no bullshit.) #cfturbo #bladerunner #simerics #cfd

59

You are a Mechanical Engineer who took some Machine Learning courses. How can you combine those two now, or more specifically, how can you apply ML to CAE problems? Here is what I would do if I started again: 1. Read the book "Approaching ML problems in CFD and CAE applications" by Justin Hodges, PhD. It will allow you to make the connection between ML and CAE, shifting your mindset about many aspects you have learned in generic ML bootcamps and giving you tons of ideas. You can order the book here: https://geni.us/JaXp 2. Learn about CAE-specific model architectures. There is hundreds of them, but in the beginning you only need to understand the main concepts of how people approach 3D data processing. I recommend reading about Graph Neural Networks (here is a great Medium resource https://lnkd.in/eQm2qniT by Rayan Kanfar) and Occupancy Networks (https://lnkd.in/ePPuWWRs). 3. Get familiar with a 3D data processing library, I recommend PyVista. Don't spend too much time learning it from tutorials, start building things straight away and then find information in the documentation as you progress. After that, build, build and build. The more you build and the more you fail the better. You could try re-creating a project you found online or, if you feel ready, build something of your own. Have you seen the SimJEB dataset? I hope it will give you some ideas: https://simjeb.github.io/ And if you think "I don't even know how 3D data is saved in a computer", here is a super simple explanation of how 3D meshes work.

319

8 Comments

Firearms suppressor simulation. Traditionally difficult to predict the performance until you've machined, build and take it to the firing range. Here is a short blog on how we can quickly simulate this in Ansys CFD! #ansys #firearms #CFD https://lnkd.in/gaJz4wf2

28

If you are designing wire actuated wrists or continuum robots you will be surely disappointed when you build your first few prototypes. what can be expected from these systems when taking into account friction and compliance is often the dirt we sweep under the rug. Our new ASME JMR Journal of Mechanisms and Robotics paper by My students Neel Shihora and Jared Lawson and in collaboration with Long Wang, Paul Moubarak, and Max Reese have tackled these problems. Questions regarding what is the appropriate way to deal with predicting the tool tip position and orientation uncertainty due to frictional losses and compliance are dealt with via a careful modification of the robot Jacobian using a new notion of wire tension zones and wire tension transition maps. if you are a designer working on surgical wrists, this works should also be beneficial to you. ##VISEVANDERBILT

101

New #ApplicationNote: Full 3D Characterization of Lattice Structures in #AdditiveManufacturing In this application note, we focus on AM lattice structures and the detailed 3D insights that CT data can offer. Understanding internal lattice geometry is critical for assessing performance, yet traditional methods often miss key details. Here, we demonstrate how Dragonfly’s advanced analysis tools bring out more information, delivering full 3D characterization for optimized part performance. Check out the full app note here: https://lnkd.in/dHEcFQzB. Don’t miss the “how-to” video link at the end, where we show you the exact steps to perform these analyses. If you’d like to talk more about additive manufacturing and 3D characterization at Formnext - Where ideas take shape, reach out to schedule a time. I’ll be at booth 11.0-B32 with a Dragonfly workstation, ready for live demos and in-depth discussions. The animation below is one of the datasets used in this application note—special thanks to Muofhe Tshibalanganda for sharing the scan used in this example! #3DCharacterization #AdditiveManufacturing #LatticeStructures #CTScanning

113

6 Comments

The latest release of optiSLang software enables users to capture their workflow and publish it to Ansys Minerva simulation process and data management (SPDM) software, and build automatically generated applications. This means that the simulation expert on your team can build out an optiSLang simulation workflow, capture repeatable processes, and publish them so another team member can use the workflow and run simulations. #SPDM #Minerva #optislang

9

Last week we announced our collaboration with Schrödinger to accelerate materials development with multiscale simulation. But what does this mean for teams that want to use Ansys and Schrödinger tools for ICME and multiscale simulation? Here are some resources to get you started: - Blog post with more details about what this brings to you: https://ansys.me/4blORSA - Whitepaper with technical details and use case: https://ansys.me/4auJfV1 #materials #icme #multiscale #simulation

108

🌐 **Harness the Power of a Unified Digital Thread: Elevate Your Engineering with Ansys, MBSE, and More!** 🌐 In the modern engineering landscape, the need for a seamless digital thread is paramount. Integrating tools like Ansys STK, ModelCenter, and Minerva into a cohesive workflow, along with leveraging Model-Based Systems Engineering (MBSE), can significantly enhance your project efficiency and accuracy. 🔧 **Why is a Digital Thread Essential?** 1. **Streamlined Workflow**: By linking STK’s space and satellite modeling, ModelCenter’s design optimization, and Minerva’s data management capabilities, you create a unified workflow that minimizes redundancies and accelerates development. 2. **Enhanced Collaboration**: Integrating MBSE ensures that all stakeholders are working with the same up-to-date data, leading to more informed decisions and fewer miscommunications. 3. **Increased Precision**: A well-integrated digital thread enhances data consistency and accuracy, reducing errors and ensuring that every design decision is based on the most reliable information available. 4. **Agile Innovation**: Real-time data integration and analysis through these tools enable agile responses to changes and innovative solutions to complex challenges. 🔍 **Want to See It in Action?** 🔍 We’re excited to invite you to our upcoming webinar where we’ll explore the transformative potential of a unified digital thread incorporating STK, ModelCenter, Minerva, and MBSE. Discover how this integration can revolutionize your engineering processes and drive your projects to new heights. Don’t miss this opportunity to learn how to optimize your workflows and enhance your engineering capabilities. Register now and be part of the future of integrated simulation and modeling! #DigitalThread #Ansys #STK #ModelCenter #Minerva #MBSE #EngineeringInnovation #Webinar #SystemsEngineering

6

One of the side-effects of parametric modeling with implicits is that your designs update in real-time. Useful for not just interactivity but also simulation. Also, the new rendering in nTop is a major lifestyle upgrade.

64

8 Comments